summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/Makefile17
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu.h22
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c879
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h202
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c18
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h12
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c39
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c24
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_device.c56
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c39
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c32
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c35
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c13
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c55
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c7
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c46
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h36
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c17
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c33
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c8
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_object.c11
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_object.h6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c251
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h14
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c653
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h60
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c66
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c7
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c155
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h42
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c83
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h7
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c69
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c81
-rw-r--r--drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c122
-rw-r--r--drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h30
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atom.c41
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atom.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atombios_crtc.c28
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atombios_dp.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atombios_encoders.c16
-rw-r--r--drivers/gpu/drm/amd/amdgpu/atombios_i2c.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h27
-rw-r--r--drivers/gpu/drm/amd/amdgpu/clearstate_si.h24
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v10_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v11_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v6_0.c22
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v8_0.c22
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c16
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c92
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c20
-rw-r--r--drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c142
-rw-r--r--drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h31
-rw-r--r--drivers/gpu/drm/amd/amdgpu/ih_v7_0.c767
-rw-r--r--drivers/gpu/drm/amd/amdgpu/ih_v7_0.h28
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c16
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h15
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c320
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c570
-rw-r--r--drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c121
-rw-r--r--drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h31
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c87
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/navi10_ih.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v11_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v13_0.c104
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v14_0.c672
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v14_0.h32
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c15
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c72
-rw-r--r--drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/soc15.c22
-rw-r--r--drivers/gpu/drm/amd/amdgpu/soc21.c6
-rw-r--r--drivers/gpu/drm/amd/amdgpu/ta_ras_if.h36
-rw-r--r--drivers/gpu/drm/amd/amdgpu/umc_v12_0.c262
-rw-r--r--drivers/gpu/drm/amd/amdgpu/umc_v12_0.h3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/umc_v6_0.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c1339
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h37
120 files changed, 7650 insertions, 917 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 260e32ef7bae..fa26a4e3a99d 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -80,7 +80,7 @@ amdgpu-y += amdgpu_device.o amdgpu_doorbell_mgr.o amdgpu_kms.o \
amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
amdgpu_fw_attestation.o amdgpu_securedisplay.o \
amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
- amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o
+ amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o
amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o
@@ -98,7 +98,7 @@ amdgpu-y += \
vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o arct_reg_init.o mxgpu_nv.o \
nbio_v7_2.o hdp_v4_0.o hdp_v5_0.o aldebaran_reg_init.o aldebaran.o soc21.o \
sienna_cichlid.o smu_v13_0_10.o nbio_v4_3.o hdp_v6_0.o nbio_v7_7.o hdp_v5_2.o lsdma_v6_0.o \
- nbio_v7_9.o aqua_vanjaram.o nbio_v7_11.o
+ nbio_v7_9.o aqua_vanjaram.o nbio_v7_11.o lsdma_v7_0.o hdp_v7_0.o
# add DF block
amdgpu-y += \
@@ -132,7 +132,8 @@ amdgpu-y += \
vega20_ih.o \
navi10_ih.o \
ih_v6_0.o \
- ih_v6_1.o
+ ih_v6_1.o \
+ ih_v7_0.o
# add PSP block
amdgpu-y += \
@@ -143,7 +144,8 @@ amdgpu-y += \
psp_v11_0_8.o \
psp_v12_0.o \
psp_v13_0.o \
- psp_v13_0_4.o
+ psp_v13_0_4.o \
+ psp_v14_0.o
# add DCE block
amdgpu-y += \
@@ -208,6 +210,7 @@ amdgpu-y += \
vcn_v4_0.o \
vcn_v4_0_3.o \
vcn_v4_0_5.o \
+ vcn_v5_0_0.o \
amdgpu_jpeg.o \
jpeg_v1_0.o \
jpeg_v2_0.o \
@@ -215,7 +218,8 @@ amdgpu-y += \
jpeg_v3_0.o \
jpeg_v4_0.o \
jpeg_v4_0_3.o \
- jpeg_v4_0_5.o
+ jpeg_v4_0_5.o \
+ jpeg_v5_0_0.o
# add VPE block
amdgpu-y += \
@@ -233,7 +237,8 @@ amdgpu-y += \
athub_v1_0.o \
athub_v2_0.o \
athub_v2_1.o \
- athub_v3_0.o
+ athub_v3_0.o \
+ athub_v4_1_0.o
# add SMUIO block
amdgpu-y += \
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 3d8a48f46b01..9246bca0a008 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -107,6 +107,7 @@
#include "amdgpu_smuio.h"
#include "amdgpu_fdinfo.h"
#include "amdgpu_mca.h"
+#include "amdgpu_aca.h"
#include "amdgpu_ras.h"
#include "amdgpu_xcp.h"
#include "amdgpu_seq64.h"
@@ -114,14 +115,12 @@
#define MAX_GPU_INSTANCE 64
-struct amdgpu_gpu_instance
-{
+struct amdgpu_gpu_instance {
struct amdgpu_device *adev;
int mgpu_fan_enabled;
};
-struct amdgpu_mgpu_info
-{
+struct amdgpu_mgpu_info {
struct amdgpu_gpu_instance gpu_ins[MAX_GPU_INSTANCE];
struct mutex mutex;
uint32_t num_gpu;
@@ -140,8 +139,7 @@ enum amdgpu_ss {
AMDGPU_SS_DRV_UNLOAD
};
-struct amdgpu_watchdog_timer
-{
+struct amdgpu_watchdog_timer {
bool timeout_fatal_disable;
uint32_t period; /* maxCycles = (1 << period), the number of cycles before a timeout */
};
@@ -198,8 +196,9 @@ extern int amdgpu_smu_pptable_id;
extern uint amdgpu_dc_feature_mask;
extern uint amdgpu_dc_debug_mask;
extern uint amdgpu_dc_visual_confirm;
-extern uint amdgpu_dm_abm_level;
+extern int amdgpu_dm_abm_level;
extern int amdgpu_backlight;
+extern int amdgpu_damage_clips;
extern struct amdgpu_mgpu_info mgpu_info;
extern int amdgpu_ras_enable;
extern uint amdgpu_ras_mask;
@@ -1045,6 +1044,9 @@ struct amdgpu_device {
/* MCA */
struct amdgpu_mca mca;
+ /* ACA */
+ struct amdgpu_aca aca;
+
struct amdgpu_ip_block ip_blocks[AMDGPU_MAX_IP_NUM];
uint32_t harvest_ip_mask;
int num_ip_blocks;
@@ -1078,6 +1080,8 @@ struct amdgpu_device {
bool in_s3;
bool in_s4;
bool in_s0ix;
+ /* indicate amdgpu suspension status */
+ bool suspend_complete;
enum pp_mp1_state mp1_state;
struct amdgpu_doorbell_index doorbell_index;
@@ -1092,6 +1096,7 @@ struct amdgpu_device {
long sdma_timeout;
long video_timeout;
long compute_timeout;
+ long psp_timeout;
uint64_t unique_id;
uint64_t df_perfmon_config_assign_mask[AMDGPU_MAX_DF_PERFMONS];
@@ -1329,6 +1334,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
#define WREG32_FIELD_OFFSET(reg, offset, field, val) \
WREG32(mm##reg + offset, (RREG32(mm##reg + offset) & ~REG_FIELD_MASK(reg, field)) | (val) << REG_FIELD_SHIFT(reg, field))
+#define AMDGPU_GET_REG_FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> (l))
/*
* BIOS helpers.
*/
@@ -1547,9 +1553,11 @@ static inline int amdgpu_acpi_smart_shift_update(struct drm_device *dev,
#if defined(CONFIG_ACPI) && defined(CONFIG_SUSPEND)
bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev);
bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev);
+void amdgpu_choose_low_power_state(struct amdgpu_device *adev);
#else
static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; }
static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; }
+static inline void amdgpu_choose_low_power_state(struct amdgpu_device *adev) { }
#endif
#if defined(CONFIG_DRM_AMD_DC)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
new file mode 100644
index 000000000000..493982f94649
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.c
@@ -0,0 +1,879 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/list.h>
+#include "amdgpu.h"
+#include "amdgpu_aca.h"
+#include "amdgpu_ras.h"
+
+#define ACA_BANK_HWID(type, hwid, mcatype) [ACA_HWIP_TYPE_##type] = {hwid, mcatype}
+
+typedef int bank_handler_t(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type, void *data);
+
+struct aca_banks {
+ int nr_banks;
+ struct list_head list;
+};
+
+struct aca_hwip {
+ int hwid;
+ int mcatype;
+};
+
+static struct aca_hwip aca_hwid_mcatypes[ACA_HWIP_TYPE_COUNT] = {
+ ACA_BANK_HWID(SMU, 0x01, 0x01),
+ ACA_BANK_HWID(PCS_XGMI, 0x50, 0x00),
+ ACA_BANK_HWID(UMC, 0x96, 0x00),
+};
+
+static void aca_banks_init(struct aca_banks *banks)
+{
+ if (!banks)
+ return;
+
+ memset(banks, 0, sizeof(*banks));
+ INIT_LIST_HEAD(&banks->list);
+}
+
+static int aca_banks_add_bank(struct aca_banks *banks, struct aca_bank *bank)
+{
+ struct aca_bank_node *node;
+
+ if (!bank)
+ return -EINVAL;
+
+ node = kvzalloc(sizeof(*node), GFP_KERNEL);
+ if (!node)
+ return -ENOMEM;
+
+ memcpy(&node->bank, bank, sizeof(*bank));
+
+ INIT_LIST_HEAD(&node->node);
+ list_add_tail(&node->node, &banks->list);
+
+ banks->nr_banks++;
+
+ return 0;
+}
+
+static void aca_banks_release(struct aca_banks *banks)
+{
+ struct aca_bank_node *node, *tmp;
+
+ list_for_each_entry_safe(node, tmp, &banks->list, node) {
+ list_del(&node->node);
+ kvfree(node);
+ }
+}
+
+static int aca_smu_get_valid_aca_count(struct amdgpu_device *adev, enum aca_error_type type, u32 *count)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+ if (!count)
+ return -EINVAL;
+
+ if (!smu_funcs || !smu_funcs->get_valid_aca_count)
+ return -EOPNOTSUPP;
+
+ return smu_funcs->get_valid_aca_count(adev, type, count);
+}
+
+static struct aca_regs_dump {
+ const char *name;
+ int reg_idx;
+} aca_regs[] = {
+ {"CONTROL", ACA_REG_IDX_CTL},
+ {"STATUS", ACA_REG_IDX_STATUS},
+ {"ADDR", ACA_REG_IDX_ADDR},
+ {"MISC", ACA_REG_IDX_MISC0},
+ {"CONFIG", ACA_REG_IDX_CONFG},
+ {"IPID", ACA_REG_IDX_IPID},
+ {"SYND", ACA_REG_IDX_SYND},
+ {"DESTAT", ACA_REG_IDX_DESTAT},
+ {"DEADDR", ACA_REG_IDX_DEADDR},
+ {"CONTROL_MASK", ACA_REG_IDX_CTL_MASK},
+};
+
+static void aca_smu_bank_dump(struct amdgpu_device *adev, int idx, int total, struct aca_bank *bank)
+{
+ int i;
+
+ dev_info(adev->dev, HW_ERR "Accelerator Check Architecture events logged\n");
+ /* plus 1 for output format, e.g: ACA[08/08]: xxxx */
+ for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
+ dev_info(adev->dev, HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n",
+ idx + 1, total, aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]);
+}
+
+static int aca_smu_get_valid_aca_banks(struct amdgpu_device *adev, enum aca_error_type type,
+ int start, int count,
+ struct aca_banks *banks)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+ struct aca_bank bank;
+ int i, max_count, ret;
+
+ if (!count)
+ return 0;
+
+ if (!smu_funcs || !smu_funcs->get_valid_aca_bank)
+ return -EOPNOTSUPP;
+
+ switch (type) {
+ case ACA_ERROR_TYPE_UE:
+ max_count = smu_funcs->max_ue_bank_count;
+ break;
+ case ACA_ERROR_TYPE_CE:
+ max_count = smu_funcs->max_ce_bank_count;
+ break;
+ case ACA_ERROR_TYPE_DEFERRED:
+ default:
+ return -EINVAL;
+ }
+
+ if (start + count >= max_count)
+ return -EINVAL;
+
+ count = min_t(int, count, max_count);
+ for (i = 0; i < count; i++) {
+ memset(&bank, 0, sizeof(bank));
+ ret = smu_funcs->get_valid_aca_bank(adev, type, start + i, &bank);
+ if (ret)
+ return ret;
+
+ aca_smu_bank_dump(adev, i, count, &bank);
+
+ ret = aca_banks_add_bank(banks, &bank);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
+static bool aca_bank_hwip_is_matched(struct aca_bank *bank, enum aca_hwip_type type)
+{
+
+ struct aca_hwip *hwip;
+ int hwid, mcatype;
+ u64 ipid;
+
+ if (!bank || type == ACA_HWIP_TYPE_UNKNOW)
+ return false;
+
+ hwip = &aca_hwid_mcatypes[type];
+ if (!hwip->hwid)
+ return false;
+
+ ipid = bank->regs[ACA_REG_IDX_IPID];
+ hwid = ACA_REG__IPID__HARDWAREID(ipid);
+ mcatype = ACA_REG__IPID__MCATYPE(ipid);
+
+ return hwip->hwid == hwid && hwip->mcatype == mcatype;
+}
+
+static bool aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type)
+{
+ const struct aca_bank_ops *bank_ops = handle->bank_ops;
+
+ if (!aca_bank_hwip_is_matched(bank, handle->hwip))
+ return false;
+
+ if (!bank_ops->aca_bank_is_valid)
+ return true;
+
+ return bank_ops->aca_bank_is_valid(handle, bank, type, handle->data);
+}
+
+static struct aca_bank_error *new_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+ struct aca_bank_error *bank_error;
+
+ bank_error = kvzalloc(sizeof(*bank_error), GFP_KERNEL);
+ if (!bank_error)
+ return NULL;
+
+ INIT_LIST_HEAD(&bank_error->node);
+ memcpy(&bank_error->info, info, sizeof(*info));
+
+ mutex_lock(&aerr->lock);
+ list_add_tail(&bank_error->node, &aerr->list);
+ mutex_unlock(&aerr->lock);
+
+ return bank_error;
+}
+
+static struct aca_bank_error *find_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+ struct aca_bank_error *bank_error = NULL;
+ struct aca_bank_info *tmp_info;
+ bool found = false;
+
+ mutex_lock(&aerr->lock);
+ list_for_each_entry(bank_error, &aerr->list, node) {
+ tmp_info = &bank_error->info;
+ if (tmp_info->socket_id == info->socket_id &&
+ tmp_info->die_id == info->die_id) {
+ found = true;
+ goto out_unlock;
+ }
+ }
+
+out_unlock:
+ mutex_unlock(&aerr->lock);
+
+ return found ? bank_error : NULL;
+}
+
+static void aca_bank_error_remove(struct aca_error *aerr, struct aca_bank_error *bank_error)
+{
+ if (!aerr || !bank_error)
+ return;
+
+ list_del(&bank_error->node);
+ aerr->nr_errors--;
+
+ kvfree(bank_error);
+}
+
+static struct aca_bank_error *get_bank_error(struct aca_error *aerr, struct aca_bank_info *info)
+{
+ struct aca_bank_error *bank_error;
+
+ if (!aerr || !info)
+ return NULL;
+
+ bank_error = find_bank_error(aerr, info);
+ if (bank_error)
+ return bank_error;
+
+ return new_bank_error(aerr, info);
+}
+
+static int aca_log_errors(struct aca_handle *handle, enum aca_error_type type,
+ struct aca_bank_report *report)
+{
+ struct aca_error_cache *error_cache = &handle->error_cache;
+ struct aca_bank_error *bank_error;
+ struct aca_error *aerr;
+
+ if (!handle || !report)
+ return -EINVAL;
+
+ if (!report->count[type])
+ return 0;
+
+ aerr = &error_cache->errors[type];
+ bank_error = get_bank_error(aerr, &report->info);
+ if (!bank_error)
+ return -ENOMEM;
+
+ bank_error->count[type] += report->count[type];
+
+ return 0;
+}
+
+static int aca_generate_bank_report(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, struct aca_bank_report *report)
+{
+ const struct aca_bank_ops *bank_ops = handle->bank_ops;
+
+ if (!bank || !report)
+ return -EINVAL;
+
+ if (!bank_ops->aca_bank_generate_report)
+ return -EOPNOTSUPP;
+
+ memset(report, 0, sizeof(*report));
+ return bank_ops->aca_bank_generate_report(handle, bank, type,
+ report, handle->data);
+}
+
+static int handler_aca_log_bank_error(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, void *data)
+{
+ struct aca_bank_report report;
+ int ret;
+
+ ret = aca_generate_bank_report(handle, bank, type, &report);
+ if (ret)
+ return ret;
+
+ if (!report.count[type])
+ return 0;
+
+ ret = aca_log_errors(handle, type, &report);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int aca_dispatch_bank(struct aca_handle_manager *mgr, struct aca_bank *bank,
+ enum aca_error_type type, bank_handler_t handler, void *data)
+{
+ struct aca_handle *handle;
+ int ret;
+
+ if (list_empty(&mgr->list))
+ return 0;
+
+ list_for_each_entry(handle, &mgr->list, node) {
+ if (!aca_bank_is_valid(handle, bank, type))
+ continue;
+
+ ret = handler(handle, bank, type, data);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
+static int aca_dispatch_banks(struct aca_handle_manager *mgr, struct aca_banks *banks,
+ enum aca_error_type type, bank_handler_t handler, void *data)
+{
+ struct aca_bank_node *node;
+ struct aca_bank *bank;
+ int ret;
+
+ if (!mgr || !banks)
+ return -EINVAL;
+
+ /* pre check to avoid unnecessary operations */
+ if (list_empty(&mgr->list) || list_empty(&banks->list))
+ return 0;
+
+ list_for_each_entry(node, &banks->list, node) {
+ bank = &node->bank;
+
+ ret = aca_dispatch_bank(mgr, bank, type, handler, data);
+ if (ret)
+ return ret;
+ }
+
+ return 0;
+}
+
+static int aca_banks_update(struct amdgpu_device *adev, enum aca_error_type type,
+ bank_handler_t handler, void *data)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ struct aca_banks banks;
+ u32 count = 0;
+ int ret;
+
+ if (list_empty(&aca->mgr.list))
+ return 0;
+
+ /* NOTE: pmfw is only support UE and CE */
+ if (type == ACA_ERROR_TYPE_DEFERRED)
+ type = ACA_ERROR_TYPE_CE;
+
+ ret = aca_smu_get_valid_aca_count(adev, type, &count);
+ if (ret)
+ return ret;
+
+ if (!count)
+ return 0;
+
+ aca_banks_init(&banks);
+
+ ret = aca_smu_get_valid_aca_banks(adev, type, 0, count, &banks);
+ if (ret)
+ goto err_release_banks;
+
+ if (list_empty(&banks.list)) {
+ ret = 0;
+ goto err_release_banks;
+ }
+
+ ret = aca_dispatch_banks(&aca->mgr, &banks, type,
+ handler, data);
+ if (ret)
+ goto err_release_banks;
+
+err_release_banks:
+ aca_banks_release(&banks);
+
+ return ret;
+}
+
+static int aca_log_aca_error_data(struct aca_bank_error *bank_error, enum aca_error_type type, struct ras_err_data *err_data)
+{
+ struct aca_bank_info *info;
+ struct amdgpu_smuio_mcm_config_info mcm_info;
+ u64 count;
+
+ if (type >= ACA_ERROR_TYPE_COUNT)
+ return -EINVAL;
+
+ count = bank_error->count[type];
+ if (!count)
+ return 0;
+
+ info = &bank_error->info;
+ mcm_info.die_id = info->die_id;
+ mcm_info.socket_id = info->socket_id;
+
+ switch (type) {
+ case ACA_ERROR_TYPE_UE:
+ amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, count);
+ break;
+ case ACA_ERROR_TYPE_CE:
+ amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, count);
+ break;
+ case ACA_ERROR_TYPE_DEFERRED:
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+static int aca_log_aca_error(struct aca_handle *handle, enum aca_error_type type, struct ras_err_data *err_data)
+{
+ struct aca_error_cache *error_cache = &handle->error_cache;
+ struct aca_error *aerr = &error_cache->errors[type];
+ struct aca_bank_error *bank_error, *tmp;
+
+ mutex_lock(&aerr->lock);
+
+ if (list_empty(&aerr->list))
+ goto out_unlock;
+
+ list_for_each_entry_safe(bank_error, tmp, &aerr->list, node) {
+ aca_log_aca_error_data(bank_error, type, err_data);
+ aca_bank_error_remove(aerr, bank_error);
+ }
+
+out_unlock:
+ mutex_unlock(&aerr->lock);
+
+ return 0;
+}
+
+static int __aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle, enum aca_error_type type,
+ struct ras_err_data *err_data)
+{
+ int ret;
+
+ /* udpate aca bank to aca source error_cache first */
+ ret = aca_banks_update(adev, type, handler_aca_log_bank_error, NULL);
+ if (ret)
+ return ret;
+
+ return aca_log_aca_error(handle, type, err_data);
+}
+
+static bool aca_handle_is_valid(struct aca_handle *handle)
+{
+ if (!handle->mask || !list_empty(&handle->node))
+ return false;
+
+ return true;
+}
+
+int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle,
+ enum aca_error_type type, void *data)
+{
+ struct ras_err_data *err_data = (struct ras_err_data *)data;
+
+ if (!handle || !err_data)
+ return -EINVAL;
+
+ if (aca_handle_is_valid(handle))
+ return -EOPNOTSUPP;
+
+ if (!(BIT(type) & handle->mask))
+ return 0;
+
+ return __aca_get_error_data(adev, handle, type, err_data);
+}
+
+static void aca_error_init(struct aca_error *aerr, enum aca_error_type type)
+{
+ mutex_init(&aerr->lock);
+ INIT_LIST_HEAD(&aerr->list);
+ aerr->type = type;
+ aerr->nr_errors = 0;
+}
+
+static void aca_init_error_cache(struct aca_handle *handle)
+{
+ struct aca_error_cache *error_cache = &handle->error_cache;
+ int type;
+
+ for (type = ACA_ERROR_TYPE_UE; type < ACA_ERROR_TYPE_COUNT; type++)
+ aca_error_init(&error_cache->errors[type], type);
+}
+
+static void aca_error_fini(struct aca_error *aerr)
+{
+ struct aca_bank_error *bank_error, *tmp;
+
+ mutex_lock(&aerr->lock);
+ list_for_each_entry_safe(bank_error, tmp, &aerr->list, node)
+ aca_bank_error_remove(aerr, bank_error);
+
+ mutex_destroy(&aerr->lock);
+}
+
+static void aca_fini_error_cache(struct aca_handle *handle)
+{
+ struct aca_error_cache *error_cache = &handle->error_cache;
+ int type;
+
+ for (type = ACA_ERROR_TYPE_UE; type < ACA_ERROR_TYPE_COUNT; type++)
+ aca_error_fini(&error_cache->errors[type]);
+}
+
+static int add_aca_handle(struct amdgpu_device *adev, struct aca_handle_manager *mgr, struct aca_handle *handle,
+ const char *name, const struct aca_info *ras_info, void *data)
+{
+ memset(handle, 0, sizeof(*handle));
+
+ handle->adev = adev;
+ handle->mgr = mgr;
+ handle->name = name;
+ handle->hwip = ras_info->hwip;
+ handle->mask = ras_info->mask;
+ handle->bank_ops = ras_info->bank_ops;
+ handle->data = data;
+ aca_init_error_cache(handle);
+
+ INIT_LIST_HEAD(&handle->node);
+ list_add_tail(&handle->node, &mgr->list);
+ mgr->nr_handles++;
+
+ return 0;
+}
+
+static ssize_t aca_sysfs_read(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct aca_handle *handle = container_of(attr, struct aca_handle, aca_attr);
+
+ /* NOTE: the aca cache will be auto cleared once read,
+ * So the driver should unify the query entry point, forward request to ras query interface directly */
+ return amdgpu_ras_aca_sysfs_read(dev, attr, handle, buf, handle->data);
+}
+
+static int add_aca_sysfs(struct amdgpu_device *adev, struct aca_handle *handle)
+{
+ struct device_attribute *aca_attr = &handle->aca_attr;
+
+ snprintf(handle->attr_name, sizeof(handle->attr_name) - 1, "aca_%s", handle->name);
+ aca_attr->show = aca_sysfs_read;
+ aca_attr->attr.name = handle->attr_name;
+ aca_attr->attr.mode = S_IRUGO;
+ sysfs_attr_init(&aca_attr->attr);
+
+ return sysfs_add_file_to_group(&adev->dev->kobj,
+ &aca_attr->attr,
+ "ras");
+}
+
+int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle,
+ const char *name, const struct aca_info *ras_info, void *data)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ int ret;
+
+ if (!amdgpu_aca_is_enabled(adev))
+ return 0;
+
+ ret = add_aca_handle(adev, &aca->mgr, handle, name, ras_info, data);
+ if (ret)
+ return ret;
+
+ return add_aca_sysfs(adev, handle);
+}
+
+static void remove_aca_handle(struct aca_handle *handle)
+{
+ struct aca_handle_manager *mgr = handle->mgr;
+
+ aca_fini_error_cache(handle);
+ list_del(&handle->node);
+ mgr->nr_handles--;
+}
+
+static void remove_aca_sysfs(struct aca_handle *handle)
+{
+ struct amdgpu_device *adev = handle->adev;
+ struct device_attribute *aca_attr = &handle->aca_attr;
+
+ if (adev->dev->kobj.sd)
+ sysfs_remove_file_from_group(&adev->dev->kobj,
+ &aca_attr->attr,
+ "ras");
+}
+
+void amdgpu_aca_remove_handle(struct aca_handle *handle)
+{
+ if (!handle || list_empty(&handle->node))
+ return;
+
+ remove_aca_sysfs(handle);
+ remove_aca_handle(handle);
+}
+
+static int aca_manager_init(struct aca_handle_manager *mgr)
+{
+ INIT_LIST_HEAD(&mgr->list);
+ mgr->nr_handles = 0;
+
+ return 0;
+}
+
+static void aca_manager_fini(struct aca_handle_manager *mgr)
+{
+ struct aca_handle *handle, *tmp;
+
+ list_for_each_entry_safe(handle, tmp, &mgr->list, node)
+ amdgpu_aca_remove_handle(handle);
+}
+
+bool amdgpu_aca_is_enabled(struct amdgpu_device *adev)
+{
+ return adev->aca.is_enabled;
+}
+
+int amdgpu_aca_init(struct amdgpu_device *adev)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ int ret;
+
+ ret = aca_manager_init(&aca->mgr);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+void amdgpu_aca_fini(struct amdgpu_device *adev)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+
+ aca_manager_fini(&aca->mgr);
+}
+
+int amdgpu_aca_reset(struct amdgpu_device *adev)
+{
+ amdgpu_aca_fini(adev);
+
+ return amdgpu_aca_init(adev);
+}
+
+void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+
+ WARN_ON(aca->smu_funcs);
+ aca->smu_funcs = smu_funcs;
+}
+
+int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info)
+{
+ u64 ipid;
+ u32 instidhi, instidlo;
+
+ if (!bank || !info)
+ return -EINVAL;
+
+ ipid = bank->regs[ACA_REG_IDX_IPID];
+ info->hwid = ACA_REG__IPID__HARDWAREID(ipid);
+ info->mcatype = ACA_REG__IPID__MCATYPE(ipid);
+ /*
+ * Unfied DieID Format: SAASS. A:AID, S:Socket.
+ * Unfied DieID[4:4] = InstanceId[0:0]
+ * Unfied DieID[0:3] = InstanceIdHi[0:3]
+ */
+ instidhi = ACA_REG__IPID__INSTANCEIDHI(ipid);
+ instidlo = ACA_REG__IPID__INSTANCEIDLO(ipid);
+ info->die_id = ((instidhi >> 2) & 0x03);
+ info->socket_id = ((instidlo & 0x1) << 2) | (instidhi & 0x03);
+
+ return 0;
+}
+
+static int aca_bank_get_error_code(struct amdgpu_device *adev, struct aca_bank *bank)
+{
+ int error_code;
+
+ switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) {
+ case IP_VERSION(13, 0, 6):
+ if (!(adev->flags & AMD_IS_APU) && adev->pm.fw_version >= 0x00555600) {
+ error_code = ACA_REG__SYND__ERRORINFORMATION(bank->regs[ACA_REG_IDX_SYND]);
+ return error_code & 0xff;
+ }
+ break;
+ default:
+ break;
+ }
+
+ /* NOTE: the true error code is encoded in status.errorcode[0:7] */
+ error_code = ACA_REG__STATUS__ERRORCODE(bank->regs[ACA_REG_IDX_STATUS]);
+
+ return error_code & 0xff;
+}
+
+int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size)
+{
+ int i, error_code;
+
+ if (!bank || !err_codes)
+ return -EINVAL;
+
+ error_code = aca_bank_get_error_code(adev, bank);
+ for (i = 0; i < size; i++) {
+ if (err_codes[i] == error_code)
+ return 0;
+ }
+
+ return -EINVAL;
+}
+
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en)
+{
+ struct amdgpu_aca *aca = &adev->aca;
+ const struct aca_smu_funcs *smu_funcs = aca->smu_funcs;
+
+ if (!smu_funcs || !smu_funcs->set_debug_mode)
+ return -EOPNOTSUPP;
+
+ return smu_funcs->set_debug_mode(adev, en);
+}
+
+#if defined(CONFIG_DEBUG_FS)
+static int amdgpu_aca_smu_debug_mode_set(void *data, u64 val)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)data;
+ int ret;
+
+ ret = amdgpu_ras_set_aca_debug_mode(adev, val ? true : false);
+ if (ret)
+ return ret;
+
+ dev_info(adev->dev, "amdgpu set smu aca debug mode %s success\n", val ? "on" : "off");
+
+ return 0;
+}
+
+static void aca_dump_entry(struct seq_file *m, struct aca_bank *bank, enum aca_error_type type, int idx)
+{
+ struct aca_bank_info info;
+ int i, ret;
+
+ ret = aca_bank_info_decode(bank, &info);
+ if (ret)
+ return;
+
+ seq_printf(m, "aca entry[%d].type: %s\n", idx, type == ACA_ERROR_TYPE_UE ? "UE" : "CE");
+ seq_printf(m, "aca entry[%d].info: socketid:%d aid:%d hwid:0x%03x mcatype:0x%04x\n",
+ idx, info.socket_id, info.die_id, info.hwid, info.mcatype);
+
+ for (i = 0; i < ARRAY_SIZE(aca_regs); i++)
+ seq_printf(m, "aca entry[%d].regs[%d]: 0x%016llx\n", idx, aca_regs[i].reg_idx, bank->regs[aca_regs[i].reg_idx]);
+}
+
+struct aca_dump_context {
+ struct seq_file *m;
+ int idx;
+};
+
+static int handler_aca_bank_dump(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, void *data)
+{
+ struct aca_dump_context *ctx = (struct aca_dump_context *)data;
+
+ aca_dump_entry(ctx->m, bank, type, ctx->idx++);
+
+ return handler_aca_log_bank_error(handle, bank, type, NULL);
+}
+
+static int aca_dump_show(struct seq_file *m, enum aca_error_type type)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)m->private;
+ struct aca_dump_context context = {
+ .m = m,
+ .idx = 0,
+ };
+
+ return aca_banks_update(adev, type, handler_aca_bank_dump, (void *)&context);
+}
+
+static int aca_dump_ce_show(struct seq_file *m, void *unused)
+{
+ return aca_dump_show(m, ACA_ERROR_TYPE_CE);
+}
+
+static int aca_dump_ce_open(struct inode *inode, struct file *file)
+{
+ return single_open(file, aca_dump_ce_show, inode->i_private);
+}
+
+static const struct file_operations aca_ce_dump_debug_fops = {
+ .owner = THIS_MODULE,
+ .open = aca_dump_ce_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+};
+
+static int aca_dump_ue_show(struct seq_file *m, void *unused)
+{
+ return aca_dump_show(m, ACA_ERROR_TYPE_UE);
+}
+
+static int aca_dump_ue_open(struct inode *inode, struct file *file)
+{
+ return single_open(file, aca_dump_ue_show, inode->i_private);
+}
+
+static const struct file_operations aca_ue_dump_debug_fops = {
+ .owner = THIS_MODULE,
+ .open = aca_dump_ue_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+};
+
+DEFINE_DEBUGFS_ATTRIBUTE(aca_debug_mode_fops, NULL, amdgpu_aca_smu_debug_mode_set, "%llu\n");
+#endif
+
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root)
+{
+#if defined(CONFIG_DEBUG_FS)
+ if (!root || adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 6))
+ return;
+
+ debugfs_create_file("aca_debug_mode", 0200, root, adev, &aca_debug_mode_fops);
+ debugfs_create_file("aca_ue_dump", 0400, root, adev, &aca_ue_dump_debug_fops);
+ debugfs_create_file("aca_ce_dump", 0400, root, adev, &aca_ce_dump_debug_fops);
+#endif
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
new file mode 100644
index 000000000000..2da50e095883
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_aca.h
@@ -0,0 +1,202 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __AMDGPU_ACA_H__
+#define __AMDGPU_ACA_H__
+
+#include <linux/list.h>
+
+#define ACA_MAX_REGS_COUNT (16)
+
+#define ACA_REG_FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> l)
+#define ACA_REG__STATUS__VAL(x) ACA_REG_FIELD(x, 63, 63)
+#define ACA_REG__STATUS__OVERFLOW(x) ACA_REG_FIELD(x, 62, 62)
+#define ACA_REG__STATUS__UC(x) ACA_REG_FIELD(x, 61, 61)
+#define ACA_REG__STATUS__EN(x) ACA_REG_FIELD(x, 60, 60)
+#define ACA_REG__STATUS__MISCV(x) ACA_REG_FIELD(x, 59, 59)
+#define ACA_REG__STATUS__ADDRV(x) ACA_REG_FIELD(x, 58, 58)
+#define ACA_REG__STATUS__PCC(x) ACA_REG_FIELD(x, 57, 57)
+#define ACA_REG__STATUS__ERRCOREIDVAL(x) ACA_REG_FIELD(x, 56, 56)
+#define ACA_REG__STATUS__TCC(x) ACA_REG_FIELD(x, 55, 55)
+#define ACA_REG__STATUS__SYNDV(x) ACA_REG_FIELD(x, 53, 53)
+#define ACA_REG__STATUS__CECC(x) ACA_REG_FIELD(x, 46, 46)
+#define ACA_REG__STATUS__UECC(x) ACA_REG_FIELD(x, 45, 45)
+#define ACA_REG__STATUS__DEFERRED(x) ACA_REG_FIELD(x, 44, 44)
+#define ACA_REG__STATUS__POISON(x) ACA_REG_FIELD(x, 43, 43)
+#define ACA_REG__STATUS__SCRUB(x) ACA_REG_FIELD(x, 40, 40)
+#define ACA_REG__STATUS__ERRCOREID(x) ACA_REG_FIELD(x, 37, 32)
+#define ACA_REG__STATUS__ADDRLSB(x) ACA_REG_FIELD(x, 29, 24)
+#define ACA_REG__STATUS__ERRORCODEEXT(x) ACA_REG_FIELD(x, 21, 16)
+#define ACA_REG__STATUS__ERRORCODE(x) ACA_REG_FIELD(x, 15, 0)
+
+#define ACA_REG__IPID__MCATYPE(x) ACA_REG_FIELD(x, 63, 48)
+#define ACA_REG__IPID__INSTANCEIDHI(x) ACA_REG_FIELD(x, 47, 44)
+#define ACA_REG__IPID__HARDWAREID(x) ACA_REG_FIELD(x, 43, 32)
+#define ACA_REG__IPID__INSTANCEIDLO(x) ACA_REG_FIELD(x, 31, 0)
+
+#define ACA_REG__MISC0__VALID(x) ACA_REG_FIELD(x, 63, 63)
+#define ACA_REG__MISC0__OVRFLW(x) ACA_REG_FIELD(x, 48, 48)
+#define ACA_REG__MISC0__ERRCNT(x) ACA_REG_FIELD(x, 43, 32)
+
+#define ACA_REG__SYND__ERRORINFORMATION(x) ACA_REG_FIELD(x, 17, 0)
+
+/* NOTE: The following codes refers to the smu header file */
+#define ACA_EXTERROR_CODE_CE 0x3a
+#define ACA_EXTERROR_CODE_FAULT 0x3b
+
+#define ACA_ERROR_UE_MASK BIT_MASK(ACA_ERROR_TYPE_UE)
+#define ACA_ERROR_CE_MASK BIT_MASK(ACA_ERROR_TYPE_CE)
+#define ACA_ERROR_DEFERRED_MASK BIT_MASK(ACA_ERROR_TYPE_DEFERRED)
+
+enum aca_reg_idx {
+ ACA_REG_IDX_CTL = 0,
+ ACA_REG_IDX_STATUS = 1,
+ ACA_REG_IDX_ADDR = 2,
+ ACA_REG_IDX_MISC0 = 3,
+ ACA_REG_IDX_CONFG = 4,
+ ACA_REG_IDX_IPID = 5,
+ ACA_REG_IDX_SYND = 6,
+ ACA_REG_IDX_DESTAT = 8,
+ ACA_REG_IDX_DEADDR = 9,
+ ACA_REG_IDX_CTL_MASK = 10,
+ ACA_REG_IDX_COUNT = 16,
+};
+
+enum aca_hwip_type {
+ ACA_HWIP_TYPE_UNKNOW = -1,
+ ACA_HWIP_TYPE_PSP = 0,
+ ACA_HWIP_TYPE_UMC,
+ ACA_HWIP_TYPE_SMU,
+ ACA_HWIP_TYPE_PCS_XGMI,
+ ACA_HWIP_TYPE_COUNT,
+};
+
+enum aca_error_type {
+ ACA_ERROR_TYPE_INVALID = -1,
+ ACA_ERROR_TYPE_UE = 0,
+ ACA_ERROR_TYPE_CE,
+ ACA_ERROR_TYPE_DEFERRED,
+ ACA_ERROR_TYPE_COUNT
+};
+
+struct aca_bank {
+ u64 regs[ACA_MAX_REGS_COUNT];
+};
+
+struct aca_bank_node {
+ struct aca_bank bank;
+ struct list_head node;
+};
+
+struct aca_bank_info {
+ int die_id;
+ int socket_id;
+ int hwid;
+ int mcatype;
+};
+
+struct aca_bank_report {
+ struct aca_bank_info info;
+ u64 count[ACA_ERROR_TYPE_COUNT];
+};
+
+struct aca_bank_error {
+ struct list_head node;
+ struct aca_bank_info info;
+ u64 count[ACA_ERROR_TYPE_COUNT];
+};
+
+struct aca_error {
+ struct list_head list;
+ struct mutex lock;
+ enum aca_error_type type;
+ int nr_errors;
+};
+
+struct aca_handle_manager {
+ struct list_head list;
+ int nr_handles;
+};
+
+struct aca_error_cache {
+ struct aca_error errors[ACA_ERROR_TYPE_COUNT];
+};
+
+struct aca_handle {
+ struct list_head node;
+ enum aca_hwip_type hwip;
+ struct amdgpu_device *adev;
+ struct aca_handle_manager *mgr;
+ struct aca_error_cache error_cache;
+ const struct aca_bank_ops *bank_ops;
+ struct device_attribute aca_attr;
+ char attr_name[64];
+ const char *name;
+ u32 mask;
+ void *data;
+};
+
+struct aca_bank_ops {
+ int (*aca_bank_generate_report)(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data);
+ bool (*aca_bank_is_valid)(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type,
+ void *data);
+};
+
+struct aca_smu_funcs {
+ int max_ue_bank_count;
+ int max_ce_bank_count;
+ int (*set_debug_mode)(struct amdgpu_device *adev, bool enable);
+ int (*get_valid_aca_count)(struct amdgpu_device *adev, enum aca_error_type type, u32 *count);
+ int (*get_valid_aca_bank)(struct amdgpu_device *adev, enum aca_error_type type, int idx, struct aca_bank *bank);
+};
+
+struct amdgpu_aca {
+ struct aca_handle_manager mgr;
+ const struct aca_smu_funcs *smu_funcs;
+ bool is_enabled;
+};
+
+struct aca_info {
+ enum aca_hwip_type hwip;
+ const struct aca_bank_ops *bank_ops;
+ u32 mask;
+};
+
+int amdgpu_aca_init(struct amdgpu_device *adev);
+void amdgpu_aca_fini(struct amdgpu_device *adev);
+int amdgpu_aca_reset(struct amdgpu_device *adev);
+void amdgpu_aca_set_smu_funcs(struct amdgpu_device *adev, const struct aca_smu_funcs *smu_funcs);
+bool amdgpu_aca_is_enabled(struct amdgpu_device *adev);
+
+int aca_bank_info_decode(struct aca_bank *bank, struct aca_bank_info *info);
+int aca_bank_check_error_codes(struct amdgpu_device *adev, struct aca_bank *bank, int *err_codes, int size);
+
+int amdgpu_aca_add_handle(struct amdgpu_device *adev, struct aca_handle *handle,
+ const char *name, const struct aca_info *aca_info, void *data);
+void amdgpu_aca_remove_handle(struct aca_handle *handle);
+int amdgpu_aca_get_error_data(struct amdgpu_device *adev, struct aca_handle *handle,
+ enum aca_error_type type, void *data);
+int amdgpu_aca_smu_set_debug_mode(struct amdgpu_device *adev, bool en);
+void amdgpu_aca_smu_debugfs_init(struct amdgpu_device *adev, struct dentry *root);
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index 2deebece810e..7099ff9cf8c5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -1519,4 +1519,22 @@ bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev)
#endif /* CONFIG_AMD_PMC */
}
+/**
+ * amdgpu_choose_low_power_state
+ *
+ * @adev: amdgpu_device_pointer
+ *
+ * Choose the target low power state for the GPU
+ */
+void amdgpu_choose_low_power_state(struct amdgpu_device *adev)
+{
+ if (adev->in_runpm)
+ return;
+
+ if (amdgpu_acpi_is_s0ix_active(adev))
+ adev->in_s0ix = true;
+ else if (amdgpu_acpi_is_s3_active(adev))
+ adev->in_s3 = true;
+}
+
#endif /* CONFIG_SUSPEND */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 41db030ddc4e..190039f14c30 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -742,9 +742,10 @@ void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev)
amdgpu_device_flush_hdp(adev, NULL);
}
-void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, bool reset)
+void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block, bool reset)
{
- amdgpu_umc_poison_handler(adev, reset);
+ amdgpu_umc_poison_handler(adev, block, reset);
}
int amdgpu_amdkfd_send_close_event_drain_irq(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 27c61c535e29..e60f63ccf79a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -193,6 +193,9 @@ struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f);
int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo);
int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
unsigned long cur_seq, struct kgd_mem *mem);
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+ uint32_t domain,
+ struct dma_fence *fence);
#else
static inline
bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
@@ -218,6 +221,13 @@ int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni,
{
return 0;
}
+static inline
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+ uint32_t domain,
+ struct dma_fence *fence)
+{
+ return 0;
+}
#endif
/* Shared API */
int amdgpu_amdkfd_alloc_gtt_mem(struct amdgpu_device *adev, size_t size,
@@ -326,7 +336,7 @@ void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev);
int amdgpu_amdkfd_get_tile_config(struct amdgpu_device *adev,
struct tile_config *config);
void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev,
- bool reset);
+ enum amdgpu_ras_block block, bool reset);
bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem);
void amdgpu_amdkfd_block_mmu_notifications(void *p);
int amdgpu_amdkfd_criu_resume(void *p);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 231fd927dcfb..5cd84f72bf26 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -426,9 +426,9 @@ validate_fail:
return ret;
}
-static int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
- uint32_t domain,
- struct dma_fence *fence)
+int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo,
+ uint32_t domain,
+ struct dma_fence *fence)
{
int ret = amdgpu_bo_reserve(bo, false);
@@ -464,13 +464,15 @@ static int amdgpu_amdkfd_validate_vm_bo(void *_unused, struct amdgpu_bo *bo)
* again. Page directories are only updated after updating page
* tables.
*/
-static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm)
+static int vm_validate_pt_pd_bos(struct amdgpu_vm *vm,
+ struct ww_acquire_ctx *ticket)
{
struct amdgpu_bo *pd = vm->root.bo;
struct amdgpu_device *adev = amdgpu_ttm_adev(pd->tbo.bdev);
int ret;
- ret = amdgpu_vm_validate_pt_bos(adev, vm, amdgpu_amdkfd_validate_vm_bo, NULL);
+ ret = amdgpu_vm_validate(adev, vm, ticket,
+ amdgpu_amdkfd_validate_vm_bo, NULL);
if (ret) {
pr_err("failed to validate PT BOs\n");
return ret;
@@ -1310,14 +1312,15 @@ update_gpuvm_pte_failed:
return ret;
}
-static int process_validate_vms(struct amdkfd_process_info *process_info)
+static int process_validate_vms(struct amdkfd_process_info *process_info,
+ struct ww_acquire_ctx *ticket)
{
struct amdgpu_vm *peer_vm;
int ret;
list_for_each_entry(peer_vm, &process_info->vm_list_head,
vm_list_node) {
- ret = vm_validate_pt_pd_bos(peer_vm);
+ ret = vm_validate_pt_pd_bos(peer_vm, ticket);
if (ret)
return ret;
}
@@ -1402,7 +1405,7 @@ static int init_kfd_vm(struct amdgpu_vm *vm, void **process_info,
ret = amdgpu_bo_reserve(vm->root.bo, true);
if (ret)
goto reserve_pd_fail;
- ret = vm_validate_pt_pd_bos(vm);
+ ret = vm_validate_pt_pd_bos(vm, NULL);
if (ret) {
pr_err("validate_pt_pd_bos() failed\n");
goto validate_pd_fail;
@@ -2043,7 +2046,7 @@ int amdgpu_amdkfd_gpuvm_map_memory_to_gpu(
bo->tbo.resource->mem_type == TTM_PL_SYSTEM)
is_invalid_userptr = true;
- ret = vm_validate_pt_pd_bos(avm);
+ ret = vm_validate_pt_pd_bos(avm, NULL);
if (unlikely(ret))
goto out_unreserve;
@@ -2136,7 +2139,7 @@ int amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(
goto unreserve_out;
}
- ret = vm_validate_pt_pd_bos(avm);
+ ret = vm_validate_pt_pd_bos(avm, NULL);
if (unlikely(ret))
goto unreserve_out;
@@ -2634,7 +2637,7 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
}
}
- ret = process_validate_vms(process_info);
+ ret = process_validate_vms(process_info, NULL);
if (ret)
goto unreserve_out;
@@ -2894,11 +2897,6 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu *
amdgpu_sync_create(&sync_obj);
- /* Validate PDs and PTs */
- ret = process_validate_vms(process_info);
- if (ret)
- goto validate_map_fail;
-
/* Validate BOs and map them to GPUVM (update VM page tables). */
list_for_each_entry(mem, &process_info->kfd_bo_list,
validate_list) {
@@ -2949,6 +2947,13 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu *
if (failed_size)
pr_debug("0x%lx/0x%lx in system\n", failed_size, total_size);
+ /* Validate PDs, PTs and evicted DMABuf imports last. Otherwise BO
+ * validations above would invalidate DMABuf imports again.
+ */
+ ret = process_validate_vms(process_info, &exec.ticket);
+ if (ret)
+ goto validate_map_fail;
+
/* Update mappings not managed by KFD */
list_for_each_entry(peer_vm, &process_info->vm_list_head,
vm_list_node) {
@@ -3020,7 +3025,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu *
&process_info->eviction_fence->base,
DMA_RESV_USAGE_BOOKKEEP);
}
- /* Attach eviction fence to PD / PT BOs */
+ /* Attach eviction fence to PD / PT BOs and DMABuf imports */
list_for_each_entry(peer_vm, &process_info->vm_list_head,
vm_list_node) {
struct amdgpu_bo *bo = peer_vm->root.bo;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
index dce9e7d5e4ec..52b12c1718eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
@@ -1018,7 +1018,8 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
if (clock_type == COMPUTE_ENGINE_PLL_PARAM) {
args.v3.ulClockParams = cpu_to_le32((clock_type << 24) | clock);
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
dividers->post_div = args.v3.ucPostDiv;
dividers->enable_post_div = (args.v3.ucCntlFlag &
@@ -1038,7 +1039,8 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
if (strobe_mode)
args.v5.ucInputFlag = ATOM_PLL_INPUT_FLAG_PLL_STROBE_MODE_EN;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
dividers->post_div = args.v5.ucPostDiv;
dividers->enable_post_div = (args.v5.ucCntlFlag &
@@ -1056,7 +1058,8 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
/* fusion */
args.v4.ulClock = cpu_to_le32(clock); /* 10 khz */
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
dividers->post_divider = dividers->post_div = args.v4.ucPostDiv;
dividers->real_clock = le32_to_cpu(args.v4.ulClock);
@@ -1067,7 +1070,8 @@ int amdgpu_atombios_get_clock_dividers(struct amdgpu_device *adev,
args.v6_in.ulClock.ulComputeClockFlag = clock_type;
args.v6_in.ulClock.ulClockFreq = cpu_to_le32(clock); /* 10 khz */
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
dividers->whole_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDiv);
dividers->frac_fb_div = le16_to_cpu(args.v6_out.ulFbDiv.usFbDivFrac);
@@ -1109,7 +1113,8 @@ int amdgpu_atombios_get_memory_pll_dividers(struct amdgpu_device *adev,
if (strobe_mode)
args.ucInputFlag |= MPLL_INPUT_FLAG_STROBE_MODE_EN;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
mpll_param->clkfrac = le16_to_cpu(args.ulFbDiv.usFbDivFrac);
mpll_param->clkf = le16_to_cpu(args.ulFbDiv.usFbDiv);
@@ -1151,7 +1156,8 @@ void amdgpu_atombios_set_engine_dram_timings(struct amdgpu_device *adev,
if (mem_clock)
args.sReserved.ulClock = cpu_to_le32(mem_clock & SET_CLOCK_FREQ_MASK);
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
}
void amdgpu_atombios_get_default_voltages(struct amdgpu_device *adev,
@@ -1205,7 +1211,8 @@ int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
args.v2.ucVoltageMode = 0;
args.v2.usVoltageLevel = 0;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
*voltage = le16_to_cpu(args.v2.usVoltageLevel);
break;
@@ -1214,7 +1221,8 @@ int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,
args.v3.ucVoltageMode = ATOM_GET_VOLTAGE_LEVEL;
args.v3.usVoltageLevel = cpu_to_le16(voltage_id);
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args,
+ sizeof(args));
*voltage = le16_to_cpu(args.v3.usVoltageLevel);
break;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
index fb2681dd6b33..6857c586ded7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
@@ -941,5 +941,6 @@ int amdgpu_atomfirmware_asic_init(struct amdgpu_device *adev, bool fb_reset)
return -EINVAL;
}
- return amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, (uint32_t *)&asic_init_ps_v2_1);
+ return amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, (uint32_t *)&asic_init_ps_v2_1,
+ sizeof(asic_init_ps_v2_1));
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
index c7eb2caec65a..649b5530d8ae 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
@@ -36,7 +36,7 @@ int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
int amdgpu_atomfirmware_get_gfx_info(struct amdgpu_device *adev);
bool amdgpu_atomfirmware_mem_ecc_supported(struct amdgpu_device *adev);
bool amdgpu_atomfirmware_sram_ecc_supported(struct amdgpu_device *adev);
-bool amdgpu_atomfirmware_ras_rom_addr(struct amdgpu_device *adev, uint8_t* i2c_address);
+bool amdgpu_atomfirmware_ras_rom_addr(struct amdgpu_device *adev, uint8_t *i2c_address);
bool amdgpu_atomfirmware_mem_training_supported(struct amdgpu_device *adev);
bool amdgpu_atomfirmware_dynamic_boot_config_supported(struct amdgpu_device *adev);
int amdgpu_atomfirmware_get_fw_reserved_fb_size(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 6adeddfb3d56..0a4b09709cfb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -952,10 +952,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
p->bytes_moved = 0;
p->bytes_moved_vis = 0;
- r = amdgpu_vm_validate_pt_bos(p->adev, &fpriv->vm,
- amdgpu_cs_bo_validate, p);
+ r = amdgpu_vm_validate(p->adev, &fpriv->vm, NULL,
+ amdgpu_cs_bo_validate, p);
if (r) {
- DRM_ERROR("amdgpu_vm_validate_pt_bos() failed.\n");
+ DRM_ERROR("amdgpu_vm_validate() failed.\n");
goto out_free_user_pages;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
index 796fa6f1420b..cfdf558b48b6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
@@ -28,9 +28,8 @@
uint64_t amdgpu_csa_vaddr(struct amdgpu_device *adev)
{
- uint64_t addr = adev->vm_manager.max_pfn << AMDGPU_GPU_PAGE_SHIFT;
+ uint64_t addr = AMDGPU_VA_RESERVED_CSA_START(adev);
- addr -= AMDGPU_VA_RESERVED_SIZE;
addr = amdgpu_gmc_sign_extend(addr);
return addr;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index fdde7488d0ed..b0ea4ddc8e72 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -96,6 +96,9 @@ MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin");
#define AMDGPU_RESUME_MS 2000
#define AMDGPU_MAX_RETRY_LIMIT 2
#define AMDGPU_RETRY_SRIOV_RESET(r) ((r) == -EBUSY || (r) == -ETIMEDOUT || (r) == -EINVAL)
+#define AMDGPU_PCIE_INDEX_FALLBACK (0x38 >> 2)
+#define AMDGPU_PCIE_INDEX_HI_FALLBACK (0x44 >> 2)
+#define AMDGPU_PCIE_DATA_FALLBACK (0x3C >> 2)
static const struct drm_driver amdgpu_kms_driver;
@@ -781,12 +784,22 @@ u32 amdgpu_device_indirect_rreg_ext(struct amdgpu_device *adev,
void __iomem *pcie_index_hi_offset;
void __iomem *pcie_data_offset;
- pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
- pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
- if ((reg_addr >> 32) && (adev->nbio.funcs->get_pcie_index_hi_offset))
- pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
- else
+ if (unlikely(!adev->nbio.funcs)) {
+ pcie_index = AMDGPU_PCIE_INDEX_FALLBACK;
+ pcie_data = AMDGPU_PCIE_DATA_FALLBACK;
+ } else {
+ pcie_index = adev->nbio.funcs->get_pcie_index_offset(adev);
+ pcie_data = adev->nbio.funcs->get_pcie_data_offset(adev);
+ }
+
+ if (reg_addr >> 32) {
+ if (unlikely(!adev->nbio.funcs))
+ pcie_index_hi = AMDGPU_PCIE_INDEX_HI_FALLBACK;
+ else
+ pcie_index_hi = adev->nbio.funcs->get_pcie_index_hi_offset(adev);
+ } else {
pcie_index_hi = 0;
+ }
spin_lock_irqsave(&adev->pcie_idx_lock, flags);
pcie_index_offset = (void __iomem *)adev->rmmio + pcie_index * 4;
@@ -1218,8 +1231,6 @@ static int amdgpu_device_asic_init(struct amdgpu_device *adev)
amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) {
amdgpu_psp_wait_for_bootloader(adev);
ret = amdgpu_atomfirmware_asic_init(adev, true);
- /* TODO: check the return val and stop device initialization if boot fails */
- amdgpu_psp_query_boot_status(adev);
return ret;
} else {
return amdgpu_atom_asic_init(adev->mode_info.atom_context);
@@ -1442,6 +1453,10 @@ int amdgpu_device_resize_fb_bar(struct amdgpu_device *adev)
if (amdgpu_sriov_vf(adev))
return 0;
+ /* PCI_EXT_CAP_ID_VNDR extended capability is located at 0x100 */
+ if (!pci_find_ext_capability(adev->pdev, PCI_EXT_CAP_ID_VNDR))
+ DRM_WARN("System can't access extended configuration space,please check!!\n");
+
/* skip if the bios has already enabled large BAR */
if (adev->gmc.real_vram_size &&
(pci_resource_len(adev->pdev, 0) >= adev->gmc.real_vram_size))
@@ -4514,13 +4529,15 @@ int amdgpu_device_prepare(struct drm_device *dev)
struct amdgpu_device *adev = drm_to_adev(dev);
int i, r;
+ amdgpu_choose_low_power_state(adev);
+
if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
return 0;
/* Evict the majority of BOs before starting suspend sequence */
r = amdgpu_device_evict_resources(adev);
if (r)
- return r;
+ goto unprepare;
for (i = 0; i < adev->num_ip_blocks; i++) {
if (!adev->ip_blocks[i].status.valid)
@@ -4529,10 +4546,15 @@ int amdgpu_device_prepare(struct drm_device *dev)
continue;
r = adev->ip_blocks[i].version->funcs->prepare_suspend((void *)adev);
if (r)
- return r;
+ goto unprepare;
}
return 0;
+
+unprepare:
+ adev->in_s0ix = adev->in_s3 = false;
+
+ return r;
}
/**
@@ -4569,7 +4591,6 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon)
drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, true);
cancel_delayed_work_sync(&adev->delayed_init_work);
- flush_delayed_work(&adev->gfx.gfx_off_delay_work);
amdgpu_ras_suspend(adev);
@@ -5680,6 +5701,7 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */
/* Aldebaran and gfx_11_0_3 support ras in SRIOV, so need resume ras during reset */
if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
IP_VERSION(9, 4, 2) ||
+ amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3))
amdgpu_ras_resume(adev);
} else {
@@ -6101,6 +6123,20 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
struct amdgpu_reset_context reset_context;
u32 memsize;
struct list_head device_list;
+ struct amdgpu_hive_info *hive;
+ int hive_ras_recovery = 0;
+ struct amdgpu_ras *ras;
+
+ /* PCI error slot reset should be skipped During RAS recovery */
+ hive = amdgpu_get_xgmi_hive(adev);
+ if (hive) {
+ hive_ras_recovery = atomic_read(&hive->ras_recovery);
+ amdgpu_put_xgmi_hive(hive);
+ }
+ ras = amdgpu_ras_get_context(adev);
+ if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) &&
+ ras && (atomic_read(&ras->in_recovery) || hive_ras_recovery))
+ return PCI_ERS_RESULT_RECOVERED;
DRM_INFO("PCI error: slot reset callback!!\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
index c7d60dd0fb97..78588334577a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@@ -27,6 +27,7 @@
#include "amdgpu_discovery.h"
#include "soc15_hw_ip.h"
#include "discovery.h"
+#include "amdgpu_ras.h"
#include "soc15.h"
#include "gfx_v9_0.h"
@@ -63,17 +64,20 @@
#include "hdp_v5_0.h"
#include "hdp_v5_2.h"
#include "hdp_v6_0.h"
+#include "hdp_v7_0.h"
#include "nv.h"
#include "soc21.h"
#include "navi10_ih.h"
#include "ih_v6_0.h"
#include "ih_v6_1.h"
+#include "ih_v7_0.h"
#include "gfx_v10_0.h"
#include "gfx_v11_0.h"
#include "sdma_v5_0.h"
#include "sdma_v5_2.h"
#include "sdma_v6_0.h"
#include "lsdma_v6_0.h"
+#include "lsdma_v7_0.h"
#include "vcn_v2_0.h"
#include "jpeg_v2_0.h"
#include "vcn_v3_0.h"
@@ -92,12 +96,15 @@
#include "smuio_v13_0.h"
#include "smuio_v13_0_3.h"
#include "smuio_v13_0_6.h"
+#include "vcn_v5_0_0.h"
+#include "jpeg_v5_0_0.h"
#include "amdgpu_vpe.h"
#define FIRMWARE_IP_DISCOVERY "amdgpu/ip_discovery.bin"
MODULE_FIRMWARE(FIRMWARE_IP_DISCOVERY);
+#define mmIP_DISCOVERY_VERSION 0x16A00
#define mmRCC_CONFIG_MEMSIZE 0xde3
#define mmMP0_SMN_C2PMSG_33 0x16061
#define mmMM_INDEX 0x0
@@ -518,7 +525,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
out:
kfree(adev->mman.discovery_bin);
adev->mman.discovery_bin = NULL;
-
+ if ((amdgpu_discovery != 2) &&
+ (RREG32(mmIP_DISCOVERY_VERSION) == 4))
+ amdgpu_ras_query_boot_status(adev, 4);
return r;
}
@@ -1278,11 +1287,10 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
* 0b10 : encode is disabled
* 0b01 : decode is disabled
*/
- adev->vcn.vcn_config[adev->vcn.num_vcn_inst] =
- ip->revision & 0xc0;
- ip->revision &= ~0xc0;
if (adev->vcn.num_vcn_inst <
AMDGPU_MAX_VCN_INSTANCES) {
+ adev->vcn.vcn_config[adev->vcn.num_vcn_inst] =
+ ip->revision & 0xc0;
adev->vcn.num_vcn_inst++;
adev->vcn.inst_mask |=
(1U << ip->instance_number);
@@ -1293,6 +1301,7 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
adev->vcn.num_vcn_inst + 1,
AMDGPU_MAX_VCN_INSTANCES);
}
+ ip->revision &= ~0xc0;
}
if (le16_to_cpu(ip->hw_id) == SDMA0_HWID ||
le16_to_cpu(ip->hw_id) == SDMA1_HWID ||
@@ -1763,6 +1772,9 @@ static int amdgpu_discovery_set_ih_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(6, 1, 0):
amdgpu_device_ip_block_add(adev, &ih_v6_1_ip_block);
break;
+ case IP_VERSION(7, 0, 0):
+ amdgpu_device_ip_block_add(adev, &ih_v7_0_ip_block);
+ break;
default:
dev_err(adev->dev,
"Failed to add ih ip block(OSSSYS_HWIP:0x%x)\n",
@@ -1812,11 +1824,16 @@ static int amdgpu_discovery_set_psp_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(13, 0, 10):
case IP_VERSION(13, 0, 11):
case IP_VERSION(14, 0, 0):
+ case IP_VERSION(14, 0, 1):
amdgpu_device_ip_block_add(adev, &psp_v13_0_ip_block);
break;
case IP_VERSION(13, 0, 4):
amdgpu_device_ip_block_add(adev, &psp_v13_0_4_ip_block);
break;
+ case IP_VERSION(14, 0, 2):
+ case IP_VERSION(14, 0, 3):
+ amdgpu_device_ip_block_add(adev, &psp_v14_0_ip_block);
+ break;
default:
dev_err(adev->dev,
"Failed to add psp ip block(MP0_HWIP:0x%x)\n",
@@ -2033,6 +2050,7 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(6, 0, 2):
case IP_VERSION(6, 0, 3):
case IP_VERSION(6, 1, 0):
+ case IP_VERSION(6, 1, 1):
amdgpu_device_ip_block_add(adev, &sdma_v6_0_ip_block);
break;
default:
@@ -2122,6 +2140,10 @@ static int amdgpu_discovery_set_mm_ip_blocks(struct amdgpu_device *adev)
amdgpu_device_ip_block_add(adev, &vcn_v4_0_5_ip_block);
amdgpu_device_ip_block_add(adev, &jpeg_v4_0_5_ip_block);
break;
+ case IP_VERSION(5, 0, 0):
+ amdgpu_device_ip_block_add(adev, &vcn_v5_0_0_ip_block);
+ amdgpu_device_ip_block_add(adev, &jpeg_v5_0_0_ip_block);
+ break;
default:
dev_err(adev->dev,
"Failed to add vcn/jpeg ip block(UVD_HWIP:0x%x)\n",
@@ -2493,6 +2515,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
adev->nbio.hdp_flush_reg = &nbio_v7_9_hdp_flush_reg;
break;
case IP_VERSION(7, 11, 0):
+ case IP_VERSION(7, 11, 1):
adev->nbio.funcs = &nbio_v7_11_funcs;
adev->nbio.hdp_flush_reg = &nbio_v7_11_hdp_flush_reg;
break;
@@ -2560,6 +2583,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(6, 1, 0):
adev->hdp.funcs = &hdp_v6_0_funcs;
break;
+ case IP_VERSION(7, 0, 0):
+ adev->hdp.funcs = &hdp_v7_0_funcs;
+ break;
default:
break;
}
@@ -2624,6 +2650,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(13, 0, 6):
case IP_VERSION(13, 0, 8):
case IP_VERSION(14, 0, 0):
+ case IP_VERSION(14, 0, 1):
adev->smuio.funcs = &smuio_v13_0_6_funcs;
break;
default:
@@ -2637,6 +2664,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev)
case IP_VERSION(6, 0, 3):
adev->lsdma.funcs = &lsdma_v6_0_funcs;
break;
+ case IP_VERSION(7, 0, 0):
+ case IP_VERSION(7, 0, 1):
+ adev->lsdma.funcs = &lsdma_v7_0_funcs;
+ break;
default:
break;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
index decbbe3d4f06..055ba2ea4c12 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
@@ -377,6 +377,10 @@ amdgpu_dma_buf_move_notify(struct dma_buf_attachment *attach)
struct amdgpu_vm_bo_base *bo_base;
int r;
+ /* FIXME: This should be after the "if", but needs a fix to make sure
+ * DMABuf imports are initialized in the right VM list.
+ */
+ amdgpu_vm_bo_invalidate(adev, bo, false);
if (!bo->tbo.resource || bo->tbo.resource->mem_type == TTM_PL_SYSTEM)
return;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 971acf01bea6..af7fae7907d7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -211,6 +211,7 @@ int amdgpu_seamless = -1; /* auto */
uint amdgpu_debug_mask;
int amdgpu_agp = -1; /* auto */
int amdgpu_wbrf = -1;
+int amdgpu_damage_clips = -1; /* auto */
static void amdgpu_drv_delayed_reset_work_handler(struct work_struct *work);
@@ -366,7 +367,7 @@ module_param_named(aspm, amdgpu_aspm, int, 0444);
* Setting the value to 0 disables this functionality.
* Setting the value to -2 is auto enabled with power down when displays are attached.
*/
-MODULE_PARM_DESC(runpm, "PX runtime pm (2 = force enable with BAMACO, 1 = force enable with BACO, 0 = disable, -1 = auto, -2 = autowith displays)");
+MODULE_PARM_DESC(runpm, "PX runtime pm (2 = force enable with BAMACO, 1 = force enable with BACO, 0 = disable, -1 = auto, -2 = auto with displays)");
module_param_named(runpm, amdgpu_runtime_pm, int, 0444);
/**
@@ -593,7 +594,7 @@ module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644);
#ifdef CONFIG_DRM_AMDGPU_SI
#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_si_support = 0;
+int amdgpu_si_support;
MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))");
#else
int amdgpu_si_support = 1;
@@ -612,7 +613,7 @@ module_param_named(si_support, amdgpu_si_support, int, 0444);
#ifdef CONFIG_DRM_AMDGPU_CIK
#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE)
-int amdgpu_cik_support = 0;
+int amdgpu_cik_support;
MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))");
#else
int amdgpu_cik_support = 1;
@@ -848,18 +849,31 @@ module_param_named(visualconfirm, amdgpu_dc_visual_confirm, uint, 0444);
* the ABM algorithm, with 1 being the least reduction and 4 being the most
* reduction.
*
- * Defaults to 0, or disabled. Userspace can still override this level later
- * after boot.
+ * Defaults to -1, or disabled. Userspace can only override this level after
+ * boot if it's set to auto.
*/
-uint amdgpu_dm_abm_level;
-MODULE_PARM_DESC(abmlevel, "ABM level (0 = off (default), 1-4 = backlight reduction level) ");
-module_param_named(abmlevel, amdgpu_dm_abm_level, uint, 0444);
+int amdgpu_dm_abm_level = -1;
+MODULE_PARM_DESC(abmlevel,
+ "ABM level (0 = off, 1-4 = backlight reduction level, -1 auto (default))");
+module_param_named(abmlevel, amdgpu_dm_abm_level, int, 0444);
int amdgpu_backlight = -1;
MODULE_PARM_DESC(backlight, "Backlight control (0 = pwm, 1 = aux, -1 auto (default))");
module_param_named(backlight, amdgpu_backlight, bint, 0444);
/**
+ * DOC: damageclips (int)
+ * Enable or disable damage clips support. If damage clips support is disabled,
+ * we will force full frame updates, irrespective of what user space sends to
+ * us.
+ *
+ * Defaults to -1 (where it is enabled unless a PSR-SU display is detected).
+ */
+MODULE_PARM_DESC(damageclips,
+ "Damage clips support (0 = disable, 1 = enable, -1 auto (default))");
+module_param_named(damageclips, amdgpu_damage_clips, int, 0444);
+
+/**
* DOC: tmz (int)
* Trusted Memory Zone (TMZ) is a method to protect data being written
* to or read from memory.
@@ -2476,6 +2490,7 @@ static int amdgpu_pmops_suspend(struct device *dev)
struct drm_device *drm_dev = dev_get_drvdata(dev);
struct amdgpu_device *adev = drm_to_adev(drm_dev);
+ adev->suspend_complete = false;
if (amdgpu_acpi_is_s0ix_active(adev))
adev->in_s0ix = true;
else if (amdgpu_acpi_is_s3_active(adev))
@@ -2490,6 +2505,7 @@ static int amdgpu_pmops_suspend_noirq(struct device *dev)
struct drm_device *drm_dev = dev_get_drvdata(dev);
struct amdgpu_device *adev = drm_to_adev(drm_dev);
+ adev->suspend_complete = true;
if (amdgpu_acpi_should_gpu_reset(adev))
return amdgpu_asic_reset(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
index 5706b282a0c7..c7df7fa3459f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fdinfo.c
@@ -97,6 +97,10 @@ void amdgpu_show_fdinfo(struct drm_printer *p, struct drm_file *file)
stats.requested_visible_vram/1024UL);
drm_printf(p, "amd-requested-gtt:\t%llu KiB\n",
stats.requested_gtt/1024UL);
+ drm_printf(p, "drm-shared-vram:\t%llu KiB\n", stats.vram_shared/1024UL);
+ drm_printf(p, "drm-shared-gtt:\t%llu KiB\n", stats.gtt_shared/1024UL);
+ drm_printf(p, "drm-shared-cpu:\t%llu KiB\n", stats.cpu_shared/1024UL);
+
for (hw_ip = 0; hw_ip < AMDGPU_HW_IP_NUM; ++hw_ip) {
if (!usage[hw_ip])
continue;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 49a5f1c73b3e..22aeee8adb71 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -187,7 +187,34 @@ static int amdgpu_gem_object_open(struct drm_gem_object *obj,
else
++bo_va->ref_count;
amdgpu_bo_unreserve(abo);
- return 0;
+
+ /* Validate and add eviction fence to DMABuf imports with dynamic
+ * attachment in compute VMs. Re-validation will be done by
+ * amdgpu_vm_validate. Fences are on the reservation shared with the
+ * export, which is currently required to be validated and fenced
+ * already by amdgpu_amdkfd_gpuvm_restore_process_bos.
+ *
+ * Nested locking below for the case that a GEM object is opened in
+ * kfd_mem_export_dmabuf. Since the lock below is only taken for imports,
+ * but not for export, this is a different lock class that cannot lead to
+ * circular lock dependencies.
+ */
+ if (!vm->is_compute_context || !vm->process_info)
+ return 0;
+ if (!obj->import_attach ||
+ !dma_buf_is_dynamic(obj->import_attach->dmabuf))
+ return 0;
+ mutex_lock_nested(&vm->process_info->lock, 1);
+ if (!WARN_ON(!vm->process_info->eviction_fence)) {
+ r = amdgpu_amdkfd_bo_validate_and_fence(abo, AMDGPU_GEM_DOMAIN_GTT,
+ &vm->process_info->eviction_fence->base);
+ if (r)
+ dev_warn(adev->dev, "%d: validate_and_fence failed: %d\n",
+ vm->task_info.pid, r);
+ }
+ mutex_unlock(&vm->process_info->lock);
+
+ return r;
}
static void amdgpu_gem_object_close(struct drm_gem_object *obj,
@@ -682,10 +709,10 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
uint64_t vm_size;
int r = 0;
- if (args->va_address < AMDGPU_VA_RESERVED_SIZE) {
+ if (args->va_address < AMDGPU_VA_RESERVED_BOTTOM) {
dev_dbg(dev->dev,
"va_address 0x%llx is in reserved area 0x%llx\n",
- args->va_address, AMDGPU_VA_RESERVED_SIZE);
+ args->va_address, AMDGPU_VA_RESERVED_BOTTOM);
return -EINVAL;
}
@@ -701,7 +728,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
args->va_address &= AMDGPU_GMC_HOLE_MASK;
vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
- vm_size -= AMDGPU_VA_RESERVED_SIZE;
+ vm_size -= AMDGPU_VA_RESERVED_TOP;
if (args->va_address + args->map_size > vm_size) {
dev_dbg(dev->dev,
"va_address 0x%llx is in top reserved area 0x%llx\n",
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
index 82b4b2019fca..f04803a44b44 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
@@ -643,8 +643,8 @@ int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev, int xcc_id)
kiq->pmf->kiq_set_resources(kiq_ring, queue_mask);
for (i = 0; i < adev->gfx.num_compute_rings; i++) {
j = i + xcc_id * adev->gfx.num_compute_rings;
- kiq->pmf->kiq_map_queues(kiq_ring,
- &adev->gfx.compute_ring[j]);
+ kiq->pmf->kiq_map_queues(kiq_ring,
+ &adev->gfx.compute_ring[j]);
}
r = amdgpu_ring_test_helper(kiq_ring);
@@ -724,8 +724,15 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
if (adev->gfx.gfx_off_req_count == 0 &&
!adev->gfx.gfx_off_state) {
- schedule_delayed_work(&adev->gfx.gfx_off_delay_work,
+ /* If going to s2idle, no need to wait */
+ if (adev->in_s0ix) {
+ if (!amdgpu_dpm_set_powergating_by_smu(adev,
+ AMD_IP_BLOCK_TYPE_GFX, true))
+ adev->gfx.gfx_off_state = true;
+ } else {
+ schedule_delayed_work(&adev->gfx.gfx_off_delay_work,
delay);
+ }
}
} else {
if (adev->gfx.gfx_off_req_count == 0) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
index 55784a9f26c4..d4a848c51a83 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
@@ -52,7 +52,7 @@ int amdgpu_gmc_pdb0_alloc(struct amdgpu_device *adev)
struct amdgpu_bo_param bp;
u64 vram_size = adev->gmc.xgmi.node_segment_size * adev->gmc.xgmi.num_physical_nodes;
uint32_t pde0_page_shift = adev->gmc.vmid0_page_table_block_size + 21;
- uint32_t npdes = (vram_size + (1ULL << pde0_page_shift) -1) >> pde0_page_shift;
+ uint32_t npdes = (vram_size + (1ULL << pde0_page_shift) - 1) >> pde0_page_shift;
memset(&bp, 0, sizeof(bp));
bp.size = PAGE_ALIGN((npdes + 1) * 8);
@@ -746,6 +746,59 @@ error_unlock_reset:
return r;
}
+void amdgpu_gmc_fw_reg_write_reg_wait(struct amdgpu_device *adev,
+ uint32_t reg0, uint32_t reg1,
+ uint32_t ref, uint32_t mask,
+ uint32_t xcc_inst)
+{
+ struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_inst];
+ struct amdgpu_ring *ring = &kiq->ring;
+ signed long r, cnt = 0;
+ unsigned long flags;
+ uint32_t seq;
+
+ if (adev->mes.ring.sched.ready) {
+ amdgpu_mes_reg_write_reg_wait(adev, reg0, reg1,
+ ref, mask);
+ return;
+ }
+
+ spin_lock_irqsave(&kiq->ring_lock, flags);
+ amdgpu_ring_alloc(ring, 32);
+ amdgpu_ring_emit_reg_write_reg_wait(ring, reg0, reg1,
+ ref, mask);
+ r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
+ if (r)
+ goto failed_undo;
+
+ amdgpu_ring_commit(ring);
+ spin_unlock_irqrestore(&kiq->ring_lock, flags);
+
+ r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+
+ /* don't wait anymore for IRQ context */
+ if (r < 1 && in_interrupt())
+ goto failed_kiq;
+
+ might_sleep();
+ while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
+
+ msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
+ r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
+ }
+
+ if (cnt > MAX_KIQ_REG_TRY)
+ goto failed_kiq;
+
+ return;
+
+failed_undo:
+ amdgpu_ring_undo(ring);
+ spin_unlock_irqrestore(&kiq->ring_lock, flags);
+failed_kiq:
+ dev_err(adev->dev, "failed to write reg %x wait reg %x\n", reg0, reg1);
+}
+
/**
* amdgpu_gmc_tmz_set -- check and set if a device supports TMZ
* @adev: amdgpu_device pointer
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
index e699d1ca8deb..17f40ea1104b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -417,6 +417,10 @@ void amdgpu_gmc_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid,
uint32_t flush_type, bool all_hub,
uint32_t inst);
+void amdgpu_gmc_fw_reg_write_reg_wait(struct amdgpu_device *adev,
+ uint32_t reg0, uint32_t reg1,
+ uint32_t ref, uint32_t mask,
+ uint32_t xcc_inst);
extern void amdgpu_gmc_tmz_set(struct amdgpu_device *adev);
extern void amdgpu_gmc_noretry_set(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
index ddd0891da116..3d7fcdeaf8cf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
@@ -62,9 +62,8 @@ int amdgpu_pasid_alloc(unsigned int bits)
int pasid = -EINVAL;
for (bits = min(bits, 31U); bits > 0; bits--) {
- pasid = ida_simple_get(&amdgpu_pasid_ida,
- 1U << (bits - 1), 1U << bits,
- GFP_KERNEL);
+ pasid = ida_alloc_range(&amdgpu_pasid_ida, 1U << (bits - 1),
+ (1U << bits) - 1, GFP_KERNEL);
if (pasid != -ENOSPC)
break;
}
@@ -82,7 +81,7 @@ int amdgpu_pasid_alloc(unsigned int bits)
void amdgpu_pasid_free(u32 pasid)
{
trace_amdgpu_pasid_freed(pasid);
- ida_simple_remove(&amdgpu_pasid_ida, pasid);
+ ida_free(&amdgpu_pasid_ida, pasid);
}
static void amdgpu_pasid_free_cb(struct dma_fence *fence,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
index 2ff2897fd1db..6df99cb00d9a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c
@@ -36,10 +36,35 @@ static void amdgpu_jpeg_idle_work_handler(struct work_struct *work);
int amdgpu_jpeg_sw_init(struct amdgpu_device *adev)
{
+ int i, r;
+
INIT_DELAYED_WORK(&adev->jpeg.idle_work, amdgpu_jpeg_idle_work_handler);
mutex_init(&adev->jpeg.jpeg_pg_lock);
atomic_set(&adev->jpeg.total_submission_cnt, 0);
+ if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) &&
+ (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG))
+ adev->jpeg.indirect_sram = true;
+
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) {
+ if (adev->jpeg.harvest_config & (1 << i))
+ continue;
+
+ if (adev->jpeg.indirect_sram) {
+ r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
+ AMDGPU_GEM_DOMAIN_VRAM |
+ AMDGPU_GEM_DOMAIN_GTT,
+ &adev->jpeg.inst[i].dpg_sram_bo,
+ &adev->jpeg.inst[i].dpg_sram_gpu_addr,
+ &adev->jpeg.inst[i].dpg_sram_cpu_addr);
+ if (r) {
+ dev_err(adev->dev,
+ "JPEG %d (%d) failed to allocate DPG bo\n", i, r);
+ return r;
+ }
+ }
+ }
+
return 0;
}
@@ -51,6 +76,11 @@ int amdgpu_jpeg_sw_fini(struct amdgpu_device *adev)
if (adev->jpeg.harvest_config & (1 << i))
continue;
+ amdgpu_bo_free_kernel(
+ &adev->jpeg.inst[i].dpg_sram_bo,
+ &adev->jpeg.inst[i].dpg_sram_gpu_addr,
+ (void **)&adev->jpeg.inst[i].dpg_sram_cpu_addr);
+
for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j)
amdgpu_ring_fini(&adev->jpeg.inst[i].ring_dec[j]);
}
@@ -210,12 +240,15 @@ int amdgpu_jpeg_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout)
} else {
r = 0;
}
+
if (!amdgpu_sriov_vf(adev)) {
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32(adev->jpeg.inst[ring->me].external.jpeg_pitch[ring->pipe]);
if (tmp == 0xDEADBEEF)
break;
udelay(1);
+ if (amdgpu_emu_mode == 1)
+ udelay(10);
}
if (i >= adev->usec_timeout)
@@ -296,3 +329,16 @@ int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev)
return 0;
}
+
+int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+ enum AMDGPU_UCODE_ID ucode_id)
+{
+ struct amdgpu_firmware_info ucode = {
+ .ucode_id = AMDGPU_UCODE_ID_JPEG_RAM,
+ .mc_addr = adev->jpeg.inst[inst_idx].dpg_sram_gpu_addr,
+ .ucode_size = ((uintptr_t)adev->jpeg.inst[inst_idx].dpg_sram_curr_addr -
+ (uintptr_t)adev->jpeg.inst[inst_idx].dpg_sram_cpu_addr),
+ };
+
+ return psp_execute_ip_fw_load(&adev->psp, &ucode);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
index ffe47e9f5bf2..aea31d61d991 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h
@@ -32,6 +32,34 @@
#define AMDGPU_JPEG_HARVEST_JPEG0 (1 << 0)
#define AMDGPU_JPEG_HARVEST_JPEG1 (1 << 1)
+#define WREG32_SOC15_JPEG_DPG_MODE(inst_idx, offset, value, indirect) \
+ do { \
+ if (!indirect) { \
+ WREG32_SOC15(JPEG, GET_INST(JPEG, inst_idx), \
+ mmUVD_DPG_LMA_DATA, value); \
+ WREG32_SOC15( \
+ JPEG, GET_INST(JPEG, inst_idx), \
+ mmUVD_DPG_LMA_CTL, \
+ (UVD_DPG_LMA_CTL__READ_WRITE_MASK | \
+ offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT | \
+ indirect << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); \
+ } else { \
+ *adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ = \
+ offset; \
+ *adev->jpeg.inst[inst_idx].dpg_sram_curr_addr++ = \
+ value; \
+ } \
+ } while (0)
+
+#define RREG32_SOC15_JPEG_DPG_MODE(inst_idx, offset, mask_en) \
+ ({ \
+ WREG32_SOC15(JPEG, inst_idx, mmUVD_DPG_LMA_CTL, \
+ (0x0 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT | \
+ mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT | \
+ offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT)); \
+ RREG32_SOC15(JPEG, inst_idx, mmUVD_DPG_LMA_DATA); \
+ })
+
struct amdgpu_jpeg_reg{
unsigned jpeg_pitch[AMDGPU_MAX_JPEG_RINGS];
};
@@ -41,6 +69,11 @@ struct amdgpu_jpeg_inst {
struct amdgpu_irq_src irq;
struct amdgpu_irq_src ras_poison_irq;
struct amdgpu_jpeg_reg external;
+ struct amdgpu_bo *dpg_sram_bo;
+ struct dpg_pause_state pause_state;
+ void *dpg_sram_cpu_addr;
+ uint64_t dpg_sram_gpu_addr;
+ uint32_t *dpg_sram_curr_addr;
uint8_t aid_id;
};
@@ -63,6 +96,7 @@ struct amdgpu_jpeg {
uint16_t inst_mask;
uint8_t num_inst_per_aid;
+ bool indirect_sram;
};
int amdgpu_jpeg_sw_init(struct amdgpu_device *adev);
@@ -82,5 +116,7 @@ int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev,
int amdgpu_jpeg_ras_late_init(struct amdgpu_device *adev,
struct ras_common_if *ras_block);
int amdgpu_jpeg_ras_sw_init(struct amdgpu_device *adev);
+int amdgpu_jpeg_psp_update_sram(struct amdgpu_device *adev, int inst_idx,
+ enum AMDGPU_UCODE_ID ucode_id);
#endif /*__AMDGPU_JPEG_H__*/
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index bf4f48fe438d..a2df3025a754 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -894,14 +894,14 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
dev_info->ids_flags |= AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD;
vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
- vm_size -= AMDGPU_VA_RESERVED_SIZE;
+ vm_size -= AMDGPU_VA_RESERVED_TOP;
/* Older VCE FW versions are buggy and can handle only 40bits */
if (adev->vce.fw_version &&
adev->vce.fw_version < AMDGPU_VCE_FW_53_45)
vm_size = min(vm_size, 1ULL << 40);
- dev_info->virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
+ dev_info->virtual_address_offset = AMDGPU_VA_RESERVED_BOTTOM;
dev_info->virtual_address_max =
min(vm_size, AMDGPU_GMC_HOLE_START);
@@ -1114,6 +1114,15 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
}
ui32 >>= 8;
break;
+ case AMDGPU_INFO_SENSOR_GPU_INPUT_POWER:
+ /* get input GPU power */
+ if (amdgpu_dpm_read_sensor(adev,
+ AMDGPU_PP_SENSOR_GPU_INPUT_POWER,
+ (void *)&ui32, &ui32_size)) {
+ return -EINVAL;
+ }
+ ui32 >>= 8;
+ break;
case AMDGPU_INFO_SENSOR_VDDNB:
/* get VDDNB in millivolts */
if (amdgpu_dpm_read_sensor(adev,
@@ -1370,6 +1379,10 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
goto error_vm;
}
+ r = amdgpu_seq64_map(adev, &fpriv->vm, &fpriv->seq64_va);
+ if (r)
+ goto error_vm;
+
mutex_init(&fpriv->bo_list_lock);
idr_init_base(&fpriv->bo_list_handles, 1);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
index 59fafb8392e0..24ad4b97177b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.c
@@ -27,6 +27,16 @@
#include "umc/umc_6_7_0_offset.h"
#include "umc/umc_6_7_0_sh_mask.h"
+static bool amdgpu_mca_is_deferred_error(struct amdgpu_device *adev,
+ uint64_t mc_status)
+{
+ if (adev->umc.ras->check_ecc_err_status)
+ return adev->umc.ras->check_ecc_err_status(adev,
+ AMDGPU_MCA_ERROR_TYPE_DE, &mc_status);
+
+ return false;
+}
+
void amdgpu_mca_query_correctable_error_count(struct amdgpu_device *adev,
uint64_t mc_status_addr,
unsigned long *error_count)
@@ -202,16 +212,16 @@ int amdgpu_mca_smu_set_debug_mode(struct amdgpu_device *adev, bool enable)
static void amdgpu_mca_smu_mca_bank_dump(struct amdgpu_device *adev, int idx, struct mca_bank_entry *entry)
{
- dev_info(adev->dev, "[Hardware error] Accelerator Check Architecture events logged\n");
- dev_info(adev->dev, "[Hardware error] aca entry[%02d].STATUS=0x%016llx\n",
+ dev_info(adev->dev, HW_ERR "Accelerator Check Architecture events logged\n");
+ dev_info(adev->dev, HW_ERR "aca entry[%02d].STATUS=0x%016llx\n",
idx, entry->regs[MCA_REG_IDX_STATUS]);
- dev_info(adev->dev, "[Hardware error] aca entry[%02d].ADDR=0x%016llx\n",
+ dev_info(adev->dev, HW_ERR "aca entry[%02d].ADDR=0x%016llx\n",
idx, entry->regs[MCA_REG_IDX_ADDR]);
- dev_info(adev->dev, "[Hardware error] aca entry[%02d].MISC0=0x%016llx\n",
+ dev_info(adev->dev, HW_ERR "aca entry[%02d].MISC0=0x%016llx\n",
idx, entry->regs[MCA_REG_IDX_MISC0]);
- dev_info(adev->dev, "[Hardware error] aca entry[%02d].IPID=0x%016llx\n",
+ dev_info(adev->dev, HW_ERR "aca entry[%02d].IPID=0x%016llx\n",
idx, entry->regs[MCA_REG_IDX_IPID]);
- dev_info(adev->dev, "[Hardware error] aca entry[%02d].SYND=0x%016llx\n",
+ dev_info(adev->dev, HW_ERR "aca entry[%02d].SYND=0x%016llx\n",
idx, entry->regs[MCA_REG_IDX_SYND]);
}
@@ -256,9 +266,14 @@ int amdgpu_mca_smu_log_ras_error(struct amdgpu_device *adev, enum amdgpu_ras_blo
if (type == AMDGPU_MCA_ERROR_TYPE_UE)
amdgpu_ras_error_statistic_ue_count(err_data,
&mcm_info, &err_addr, (uint64_t)count);
- else
- amdgpu_ras_error_statistic_ce_count(err_data,
- &mcm_info, &err_addr, (uint64_t)count);
+ else {
+ if (amdgpu_mca_is_deferred_error(adev, entry->regs[MCA_REG_IDX_STATUS]))
+ amdgpu_ras_error_statistic_de_count(err_data,
+ &mcm_info, &err_addr, (uint64_t)count);
+ else
+ amdgpu_ras_error_statistic_ce_count(err_data,
+ &mcm_info, &err_addr, (uint64_t)count);
+ }
}
out_mca_release:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
index b399f1b62887..b964110ed1e0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mca.h
@@ -65,6 +65,7 @@ enum amdgpu_mca_ip {
enum amdgpu_mca_error_type {
AMDGPU_MCA_ERROR_TYPE_UE = 0,
AMDGPU_MCA_ERROR_TYPE_CE,
+ AMDGPU_MCA_ERROR_TYPE_DE,
};
struct amdgpu_mca_ras_block {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
index da48b6da0107..a98e03e0a51f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c
@@ -1398,7 +1398,7 @@ int amdgpu_mes_self_test(struct amdgpu_device *adev)
goto error_fini;
}
- ctx_data.meta_data_gpu_addr = AMDGPU_VA_RESERVED_SIZE;
+ ctx_data.meta_data_gpu_addr = AMDGPU_VA_RESERVED_BOTTOM;
r = amdgpu_mes_ctx_map_meta_data(adev, vm, &ctx_data);
if (r) {
DRM_ERROR("failed to map ctx meta data\n");
@@ -1565,9 +1565,9 @@ void amdgpu_debugfs_mes_event_log_init(struct amdgpu_device *adev)
#if defined(CONFIG_DEBUG_FS)
struct drm_minor *minor = adev_to_drm(adev)->primary;
struct dentry *root = minor->debugfs_root;
-
- debugfs_create_file("amdgpu_mes_event_log", 0444, root,
- adev, &amdgpu_debugfs_mes_event_log_fops);
+ if (adev->enable_mes)
+ debugfs_create_file("amdgpu_mes_event_log", 0444, root,
+ adev, &amdgpu_debugfs_mes_event_log_fops);
#endif
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index b671b0665492..010b0cb7693c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -1273,25 +1273,36 @@ void amdgpu_bo_get_memory(struct amdgpu_bo *bo,
struct amdgpu_mem_stats *stats)
{
uint64_t size = amdgpu_bo_size(bo);
+ struct drm_gem_object *obj;
unsigned int domain;
+ bool shared;
/* Abort if the BO doesn't currently have a backing store */
if (!bo->tbo.resource)
return;
+ obj = &bo->tbo.base;
+ shared = drm_gem_object_is_shared_for_memory_stats(obj);
+
domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type);
switch (domain) {
case AMDGPU_GEM_DOMAIN_VRAM:
stats->vram += size;
if (amdgpu_bo_in_cpu_visible_vram(bo))
stats->visible_vram += size;
+ if (shared)
+ stats->vram_shared += size;
break;
case AMDGPU_GEM_DOMAIN_GTT:
stats->gtt += size;
+ if (shared)
+ stats->gtt_shared += size;
break;
case AMDGPU_GEM_DOMAIN_CPU:
default:
stats->cpu += size;
+ if (shared)
+ stats->cpu_shared += size;
break;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index a3ea8a82db23..be679c42b0b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -138,12 +138,18 @@ struct amdgpu_bo_vm {
struct amdgpu_mem_stats {
/* current VRAM usage, includes visible VRAM */
uint64_t vram;
+ /* current shared VRAM usage, includes visible VRAM */
+ uint64_t vram_shared;
/* current visible VRAM usage */
uint64_t visible_vram;
/* current GTT usage */
uint64_t gtt;
+ /* current shared GTT usage */
+ uint64_t gtt_shared;
/* current system memory usage */
uint64_t cpu;
+ /* current shared system memory usage */
+ uint64_t cpu_shared;
/* sum of evicted buffers, includes visible VRAM */
uint64_t evicted_vram;
/* sum of evicted buffers due to CPU access */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index 0328616473f8..3c2b1413058b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -38,6 +38,7 @@
#include "psp_v12_0.h"
#include "psp_v13_0.h"
#include "psp_v13_0_4.h"
+#include "psp_v14_0.h"
#include "amdgpu_ras.h"
#include "amdgpu_securedisplay.h"
@@ -162,20 +163,26 @@ static int psp_early_init(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct psp_context *psp = &adev->psp;
+ psp->autoload_supported = true;
+ psp->boot_time_tmr = true;
+
switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
case IP_VERSION(9, 0, 0):
psp_v3_1_set_psp_funcs(psp);
psp->autoload_supported = false;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(10, 0, 0):
case IP_VERSION(10, 0, 1):
psp_v10_0_set_psp_funcs(psp);
psp->autoload_supported = false;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(11, 0, 2):
case IP_VERSION(11, 0, 4):
psp_v11_0_set_psp_funcs(psp);
psp->autoload_supported = false;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(11, 0, 0):
case IP_VERSION(11, 0, 7):
@@ -188,15 +195,20 @@ static int psp_early_init(void *handle)
case IP_VERSION(11, 0, 12):
case IP_VERSION(11, 0, 13):
psp_v11_0_set_psp_funcs(psp);
- psp->autoload_supported = true;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(11, 0, 3):
case IP_VERSION(12, 0, 1):
psp_v12_0_set_psp_funcs(psp);
+ psp->autoload_supported = false;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(13, 0, 2):
+ psp->boot_time_tmr = false;
+ fallthrough;
case IP_VERSION(13, 0, 6):
psp_v13_0_set_psp_funcs(psp);
+ psp->autoload_supported = false;
break;
case IP_VERSION(13, 0, 1):
case IP_VERSION(13, 0, 3):
@@ -204,25 +216,31 @@ static int psp_early_init(void *handle)
case IP_VERSION(13, 0, 8):
case IP_VERSION(13, 0, 11):
case IP_VERSION(14, 0, 0):
+ case IP_VERSION(14, 0, 1):
psp_v13_0_set_psp_funcs(psp);
- psp->autoload_supported = true;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(11, 0, 8):
if (adev->apu_flags & AMD_APU_IS_CYAN_SKILLFISH2) {
psp_v11_0_8_set_psp_funcs(psp);
- psp->autoload_supported = false;
}
+ psp->autoload_supported = false;
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(13, 0, 0):
case IP_VERSION(13, 0, 7):
case IP_VERSION(13, 0, 10):
psp_v13_0_set_psp_funcs(psp);
- psp->autoload_supported = true;
adev->psp.sup_ifwi_up = !amdgpu_sriov_vf(adev);
+ psp->boot_time_tmr = false;
break;
case IP_VERSION(13, 0, 4):
psp_v13_0_4_set_psp_funcs(psp);
- psp->autoload_supported = true;
+ psp->boot_time_tmr = false;
+ break;
+ case IP_VERSION(14, 0, 2):
+ case IP_VERSION(14, 0, 3):
+ psp_v14_0_set_psp_funcs(psp);
break;
default:
return -EINVAL;
@@ -230,6 +248,8 @@ static int psp_early_init(void *handle)
psp->adev = adev;
+ adev->psp_timeout = 20000;
+
psp_check_pmfw_centralized_cstate_management(psp);
if (amdgpu_sriov_vf(adev))
@@ -291,21 +311,22 @@ static int psp_memory_training_init(struct psp_context *psp)
struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
if (ctx->init != PSP_MEM_TRAIN_RESERVE_SUCCESS) {
- DRM_DEBUG("memory training is not supported!\n");
+ dev_dbg(psp->adev->dev, "memory training is not supported!\n");
return 0;
}
ctx->sys_cache = kzalloc(ctx->train_data_size, GFP_KERNEL);
if (ctx->sys_cache == NULL) {
- DRM_ERROR("alloc mem_train_ctx.sys_cache failed!\n");
+ dev_err(psp->adev->dev, "alloc mem_train_ctx.sys_cache failed!\n");
ret = -ENOMEM;
goto Err_out;
}
- DRM_DEBUG("train_data_size:%llx,p2c_train_data_offset:%llx,c2p_train_data_offset:%llx.\n",
- ctx->train_data_size,
- ctx->p2c_train_data_offset,
- ctx->c2p_train_data_offset);
+ dev_dbg(psp->adev->dev,
+ "train_data_size:%llx,p2c_train_data_offset:%llx,c2p_train_data_offset:%llx.\n",
+ ctx->train_data_size,
+ ctx->p2c_train_data_offset,
+ ctx->c2p_train_data_offset);
ctx->init = PSP_MEM_TRAIN_INIT_SUCCESS;
return 0;
@@ -407,7 +428,7 @@ static int psp_sw_init(void *handle)
psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
if (!psp->cmd) {
- DRM_ERROR("Failed to allocate memory to command buffer!\n");
+ dev_err(adev->dev, "Failed to allocate memory to command buffer!\n");
ret = -ENOMEM;
}
@@ -454,13 +475,13 @@ static int psp_sw_init(void *handle)
if (mem_training_ctx->enable_mem_training) {
ret = psp_memory_training_init(psp);
if (ret) {
- DRM_ERROR("Failed to initialize memory training!\n");
+ dev_err(adev->dev, "Failed to initialize memory training!\n");
return ret;
}
ret = psp_mem_training(psp, PSP_MEM_TRAIN_COLD_BOOT);
if (ret) {
- DRM_ERROR("Failed to process memory training!\n");
+ dev_err(adev->dev, "Failed to process memory training!\n");
return ret;
}
}
@@ -626,7 +647,7 @@ psp_cmd_submit_buf(struct psp_context *psp,
{
int ret;
int index;
- int timeout = 20000;
+ int timeout = psp->adev->psp_timeout;
bool ras_intr = false;
bool skip_unsupport = false;
@@ -675,9 +696,11 @@ psp_cmd_submit_buf(struct psp_context *psp,
*/
if (!skip_unsupport && (psp->cmd_buf_mem->resp.status || !timeout) && !ras_intr) {
if (ucode)
- DRM_WARN("failed to load ucode %s(0x%X) ",
- amdgpu_ucode_name(ucode->ucode_id), ucode->ucode_id);
- DRM_WARN("psp gfx command %s(0x%X) failed and response status is (0x%X)\n",
+ dev_warn(psp->adev->dev,
+ "failed to load ucode %s(0x%X) ",
+ amdgpu_ucode_name(ucode->ucode_id), ucode->ucode_id);
+ dev_warn(psp->adev->dev,
+ "psp gfx command %s(0x%X) failed and response status is (0x%X)\n",
psp_gfx_cmd_name(psp->cmd_buf_mem->cmd_id), psp->cmd_buf_mem->cmd_id,
psp->cmd_buf_mem->resp.status);
/* If any firmware (including CAP) load fails under SRIOV, it should
@@ -771,16 +794,6 @@ static int psp_load_toc(struct psp_context *psp,
return ret;
}
-static bool psp_boottime_tmr(struct psp_context *psp)
-{
- switch (amdgpu_ip_version(psp->adev, MP0_HWIP, 0)) {
- case IP_VERSION(13, 0, 6):
- return true;
- default:
- return false;
- }
-}
-
/* Set up Trusted Memory Region */
static int psp_tmr_init(struct psp_context *psp)
{
@@ -807,12 +820,12 @@ static int psp_tmr_init(struct psp_context *psp)
psp->fw_pri_buf) {
ret = psp_load_toc(psp, &tmr_size);
if (ret) {
- DRM_ERROR("Failed to load toc\n");
+ dev_err(psp->adev->dev, "Failed to load toc\n");
return ret;
}
}
- if (!psp->tmr_bo) {
+ if (!psp->tmr_bo && !psp->boot_time_tmr) {
pptr = amdgpu_sriov_vf(psp->adev) ? &tmr_buf : NULL;
ret = amdgpu_bo_create_kernel(psp->adev, tmr_size,
PSP_TMR_ALIGNMENT,
@@ -855,7 +868,7 @@ static int psp_tmr_load(struct psp_context *psp)
psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, psp->tmr_bo);
if (psp->tmr_bo)
- DRM_INFO("reserve 0x%lx from 0x%llx for PSP TMR\n",
+ dev_info(psp->adev->dev, "reserve 0x%lx from 0x%llx for PSP TMR\n",
amdgpu_bo_size(psp->tmr_bo), psp->tmr_mc_addr);
ret = psp_cmd_submit_buf(psp, NULL, cmd,
@@ -1113,7 +1126,7 @@ int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg,
psp_prep_reg_prog_cmd_buf(cmd, reg, value);
ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr);
if (ret)
- DRM_ERROR("PSP failed to program reg id %d", reg);
+ dev_err(psp->adev->dev, "PSP failed to program reg id %d\n", reg);
release_psp_cmd_buf(psp);
@@ -1526,22 +1539,22 @@ static void psp_ras_ta_check_status(struct psp_context *psp)
switch (ras_cmd->ras_status) {
case TA_RAS_STATUS__ERROR_UNSUPPORTED_IP:
dev_warn(psp->adev->dev,
- "RAS WARNING: cmd failed due to unsupported ip\n");
+ "RAS WARNING: cmd failed due to unsupported ip\n");
break;
case TA_RAS_STATUS__ERROR_UNSUPPORTED_ERROR_INJ:
dev_warn(psp->adev->dev,
- "RAS WARNING: cmd failed due to unsupported error injection\n");
+ "RAS WARNING: cmd failed due to unsupported error injection\n");
break;
case TA_RAS_STATUS__SUCCESS:
break;
case TA_RAS_STATUS__TEE_ERROR_ACCESS_DENIED:
if (ras_cmd->cmd_id == TA_RAS_COMMAND__TRIGGER_ERROR)
dev_warn(psp->adev->dev,
- "RAS WARNING: Inject error to critical region is not allowed\n");
+ "RAS WARNING: Inject error to critical region is not allowed\n");
break;
default:
dev_warn(psp->adev->dev,
- "RAS WARNING: ras status = 0x%X\n", ras_cmd->ras_status);
+ "RAS WARNING: ras status = 0x%X\n", ras_cmd->ras_status);
break;
}
}
@@ -1565,7 +1578,7 @@ int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id)
return ret;
if (ras_cmd->if_version > RAS_TA_HOST_IF_VER) {
- DRM_WARN("RAS: Unsupported Interface");
+ dev_warn(psp->adev->dev, "RAS: Unsupported Interface\n");
return -EINVAL;
}
@@ -1715,7 +1728,7 @@ int psp_ras_initialize(struct psp_context *psp)
psp->ras_context.context.initialized = true;
else {
if (ras_cmd->ras_status)
- dev_warn(psp->adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status);
+ dev_warn(adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status);
/* fail to load RAS TA */
psp->ras_context.context.initialized = false;
@@ -1779,6 +1792,31 @@ int psp_ras_trigger_error(struct psp_context *psp,
return 0;
}
+
+int psp_ras_query_address(struct psp_context *psp,
+ struct ta_ras_query_address_input *addr_in,
+ struct ta_ras_query_address_output *addr_out)
+{
+ struct ta_ras_shared_memory *ras_cmd;
+ int ret;
+
+ if (!psp->ras_context.context.initialized)
+ return -EINVAL;
+
+ ras_cmd = (struct ta_ras_shared_memory *)psp->ras_context.context.mem_context.shared_buf;
+ memset(ras_cmd, 0, sizeof(struct ta_ras_shared_memory));
+
+ ras_cmd->cmd_id = TA_RAS_COMMAND__QUERY_ADDRESS;
+ ras_cmd->ras_in_message.address = *addr_in;
+
+ ret = psp_ras_invoke(psp, ras_cmd->cmd_id);
+ if (ret || ras_cmd->ras_status || psp->cmd_buf_mem->resp.status)
+ return -EINVAL;
+
+ *addr_out = ras_cmd->ras_out_message.address;
+
+ return 0;
+}
// ras end
// HDCP start
@@ -2125,19 +2163,14 @@ int amdgpu_psp_wait_for_bootloader(struct amdgpu_device *adev)
return ret;
}
-int amdgpu_psp_query_boot_status(struct amdgpu_device *adev)
+bool amdgpu_psp_get_ras_capability(struct psp_context *psp)
{
- struct psp_context *psp = &adev->psp;
- int ret = 0;
-
- if (amdgpu_sriov_vf(adev) || (adev->flags & AMD_IS_APU))
- return 0;
-
if (psp->funcs &&
- psp->funcs->query_boot_status)
- ret = psp->funcs->query_boot_status(psp);
-
- return ret;
+ psp->funcs->get_ras_capability) {
+ return psp->funcs->get_ras_capability(psp);
+ } else {
+ return false;
+ }
}
static int psp_hw_start(struct psp_context *psp)
@@ -2150,7 +2183,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_kdb != NULL)) {
ret = psp_bootloader_load_kdb(psp);
if (ret) {
- DRM_ERROR("PSP load kdb failed!\n");
+ dev_err(adev->dev, "PSP load kdb failed!\n");
return ret;
}
}
@@ -2159,7 +2192,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_spl != NULL)) {
ret = psp_bootloader_load_spl(psp);
if (ret) {
- DRM_ERROR("PSP load spl failed!\n");
+ dev_err(adev->dev, "PSP load spl failed!\n");
return ret;
}
}
@@ -2168,7 +2201,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_sysdrv != NULL)) {
ret = psp_bootloader_load_sysdrv(psp);
if (ret) {
- DRM_ERROR("PSP load sys drv failed!\n");
+ dev_err(adev->dev, "PSP load sys drv failed!\n");
return ret;
}
}
@@ -2177,7 +2210,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_soc_drv != NULL)) {
ret = psp_bootloader_load_soc_drv(psp);
if (ret) {
- DRM_ERROR("PSP load soc drv failed!\n");
+ dev_err(adev->dev, "PSP load soc drv failed!\n");
return ret;
}
}
@@ -2186,7 +2219,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_intf_drv != NULL)) {
ret = psp_bootloader_load_intf_drv(psp);
if (ret) {
- DRM_ERROR("PSP load intf drv failed!\n");
+ dev_err(adev->dev, "PSP load intf drv failed!\n");
return ret;
}
}
@@ -2195,7 +2228,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_dbg_drv != NULL)) {
ret = psp_bootloader_load_dbg_drv(psp);
if (ret) {
- DRM_ERROR("PSP load dbg drv failed!\n");
+ dev_err(adev->dev, "PSP load dbg drv failed!\n");
return ret;
}
}
@@ -2204,7 +2237,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_ras_drv != NULL)) {
ret = psp_bootloader_load_ras_drv(psp);
if (ret) {
- DRM_ERROR("PSP load ras_drv failed!\n");
+ dev_err(adev->dev, "PSP load ras_drv failed!\n");
return ret;
}
}
@@ -2213,7 +2246,7 @@ static int psp_hw_start(struct psp_context *psp)
(psp->funcs->bootloader_load_sos != NULL)) {
ret = psp_bootloader_load_sos(psp);
if (ret) {
- DRM_ERROR("PSP load sos failed!\n");
+ dev_err(adev->dev, "PSP load sos failed!\n");
return ret;
}
}
@@ -2221,17 +2254,17 @@ static int psp_hw_start(struct psp_context *psp)
ret = psp_ring_create(psp, PSP_RING_TYPE__KM);
if (ret) {
- DRM_ERROR("PSP create ring failed!\n");
+ dev_err(adev->dev, "PSP create ring failed!\n");
return ret;
}
if (amdgpu_sriov_vf(adev) && amdgpu_in_reset(adev))
goto skip_pin_bo;
- if (!psp_boottime_tmr(psp)) {
+ if (!psp->boot_time_tmr || psp->autoload_supported) {
ret = psp_tmr_init(psp);
if (ret) {
- DRM_ERROR("PSP tmr init failed!\n");
+ dev_err(adev->dev, "PSP tmr init failed!\n");
return ret;
}
}
@@ -2248,10 +2281,12 @@ skip_pin_bo:
return ret;
}
- ret = psp_tmr_load(psp);
- if (ret) {
- DRM_ERROR("PSP load tmr failed!\n");
- return ret;
+ if (!psp->boot_time_tmr || !psp->autoload_supported) {
+ ret = psp_tmr_load(psp);
+ if (ret) {
+ dev_err(adev->dev, "PSP load tmr failed!\n");
+ return ret;
+ }
}
return 0;
@@ -2462,6 +2497,9 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
case AMDGPU_UCODE_ID_P2S_TABLE:
*type = GFX_FW_TYPE_P2S_TABLE;
break;
+ case AMDGPU_UCODE_ID_JPEG_RAM:
+ *type = GFX_FW_TYPE_JPEG_RAM;
+ break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
@@ -2518,7 +2556,8 @@ static void psp_print_fw_hdr(struct psp_context *psp,
}
}
-static int psp_prep_load_ip_fw_cmd_buf(struct amdgpu_firmware_info *ucode,
+static int psp_prep_load_ip_fw_cmd_buf(struct psp_context *psp,
+ struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd)
{
int ret;
@@ -2531,7 +2570,7 @@ static int psp_prep_load_ip_fw_cmd_buf(struct amdgpu_firmware_info *ucode,
ret = psp_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
if (ret)
- DRM_ERROR("Unknown firmware type\n");
+ dev_err(psp->adev->dev, "Unknown firmware type\n");
return ret;
}
@@ -2542,7 +2581,7 @@ int psp_execute_ip_fw_load(struct psp_context *psp,
int ret = 0;
struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp);
- ret = psp_prep_load_ip_fw_cmd_buf(ucode, cmd);
+ ret = psp_prep_load_ip_fw_cmd_buf(psp, ucode, cmd);
if (!ret) {
ret = psp_cmd_submit_buf(psp, ucode, cmd,
psp->fence_buf_mc_addr);
@@ -2601,13 +2640,13 @@ static int psp_load_smu_fw(struct psp_context *psp)
amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(11, 0, 2)))) {
ret = amdgpu_dpm_set_mp1_state(adev, PP_MP1_STATE_UNLOAD);
if (ret)
- DRM_WARN("Failed to set MP1 state prepare for reload\n");
+ dev_err(adev->dev, "Failed to set MP1 state prepare for reload\n");
}
ret = psp_execute_ip_fw_load(psp, ucode);
if (ret)
- DRM_ERROR("PSP load smu failed!\n");
+ dev_err(adev->dev, "PSP load smu failed!\n");
return ret;
}
@@ -2712,7 +2751,7 @@ static int psp_load_non_psp_fw(struct psp_context *psp)
adev->virt.autoload_ucode_id : AMDGPU_UCODE_ID_RLC_G)) {
ret = psp_rlc_autoload_start(psp);
if (ret) {
- DRM_ERROR("Failed to start rlc autoload\n");
+ dev_err(adev->dev, "Failed to start rlc autoload\n");
return ret;
}
}
@@ -2734,7 +2773,7 @@ static int psp_load_fw(struct amdgpu_device *adev)
ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
if (ret) {
- DRM_ERROR("PSP ring init failed!\n");
+ dev_err(adev->dev, "PSP ring init failed!\n");
goto failed;
}
}
@@ -2749,13 +2788,13 @@ static int psp_load_fw(struct amdgpu_device *adev)
ret = psp_asd_initialize(psp);
if (ret) {
- DRM_ERROR("PSP load asd failed!\n");
+ dev_err(adev->dev, "PSP load asd failed!\n");
goto failed1;
}
ret = psp_rl_load(adev);
if (ret) {
- DRM_ERROR("PSP load RL failed!\n");
+ dev_err(adev->dev, "PSP load RL failed!\n");
goto failed1;
}
@@ -2775,7 +2814,7 @@ static int psp_load_fw(struct amdgpu_device *adev)
ret = psp_ras_initialize(psp);
if (ret)
dev_err(psp->adev->dev,
- "RAS: Failed to initialize RAS\n");
+ "RAS: Failed to initialize RAS\n");
ret = psp_hdcp_initialize(psp);
if (ret)
@@ -2828,7 +2867,7 @@ static int psp_hw_init(void *handle)
ret = psp_load_fw(adev);
if (ret) {
- DRM_ERROR("PSP firmware loading failed\n");
+ dev_err(adev->dev, "PSP firmware loading failed\n");
goto failed;
}
@@ -2875,7 +2914,7 @@ static int psp_suspend(void *handle)
psp->xgmi_context.context.initialized) {
ret = psp_xgmi_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate xgmi ta\n");
+ dev_err(adev->dev, "Failed to terminate xgmi ta\n");
goto out;
}
}
@@ -2883,46 +2922,46 @@ static int psp_suspend(void *handle)
if (psp->ta_fw) {
ret = psp_ras_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate ras ta\n");
+ dev_err(adev->dev, "Failed to terminate ras ta\n");
goto out;
}
ret = psp_hdcp_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate hdcp ta\n");
+ dev_err(adev->dev, "Failed to terminate hdcp ta\n");
goto out;
}
ret = psp_dtm_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate dtm ta\n");
+ dev_err(adev->dev, "Failed to terminate dtm ta\n");
goto out;
}
ret = psp_rap_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate rap ta\n");
+ dev_err(adev->dev, "Failed to terminate rap ta\n");
goto out;
}
ret = psp_securedisplay_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate securedisplay ta\n");
+ dev_err(adev->dev, "Failed to terminate securedisplay ta\n");
goto out;
}
}
ret = psp_asd_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate asd\n");
+ dev_err(adev->dev, "Failed to terminate asd\n");
goto out;
}
ret = psp_tmr_terminate(psp);
if (ret) {
- DRM_ERROR("Failed to terminate tmr\n");
+ dev_err(adev->dev, "Failed to terminate tmr\n");
goto out;
}
ret = psp_ring_stop(psp, PSP_RING_TYPE__KM);
if (ret)
- DRM_ERROR("PSP ring stop failed\n");
+ dev_err(adev->dev, "PSP ring stop failed\n");
out:
return ret;
@@ -2934,12 +2973,12 @@ static int psp_resume(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct psp_context *psp = &adev->psp;
- DRM_INFO("PSP is resuming...\n");
+ dev_info(adev->dev, "PSP is resuming...\n");
if (psp->mem_train_ctx.enable_mem_training) {
ret = psp_mem_training(psp, PSP_MEM_TRAIN_RESUME);
if (ret) {
- DRM_ERROR("Failed to process memory training!\n");
+ dev_err(adev->dev, "Failed to process memory training!\n");
return ret;
}
}
@@ -2956,7 +2995,7 @@ static int psp_resume(void *handle)
ret = psp_asd_initialize(psp);
if (ret) {
- DRM_ERROR("PSP load asd failed!\n");
+ dev_err(adev->dev, "PSP load asd failed!\n");
goto failed;
}
@@ -2980,7 +3019,7 @@ static int psp_resume(void *handle)
ret = psp_ras_initialize(psp);
if (ret)
dev_err(psp->adev->dev,
- "RAS: Failed to initialize RAS\n");
+ "RAS: Failed to initialize RAS\n");
ret = psp_hdcp_initialize(psp);
if (ret)
@@ -3008,7 +3047,7 @@ static int psp_resume(void *handle)
return 0;
failed:
- DRM_ERROR("PSP resume failed\n");
+ dev_err(adev->dev, "PSP resume failed\n");
mutex_unlock(&adev->firmware.mutex);
return ret;
}
@@ -3069,9 +3108,11 @@ int psp_ring_cmd_submit(struct psp_context *psp,
write_frame = ring_buffer_start + (psp_write_ptr_reg / rb_frame_size_dw);
/* Check invalid write_frame ptr address */
if ((write_frame < ring_buffer_start) || (ring_buffer_end < write_frame)) {
- DRM_ERROR("ring_buffer_start = %p; ring_buffer_end = %p; write_frame = %p\n",
- ring_buffer_start, ring_buffer_end, write_frame);
- DRM_ERROR("write_frame is pointing to address out of bounds\n");
+ dev_err(adev->dev,
+ "ring_buffer_start = %p; ring_buffer_end = %p; write_frame = %p\n",
+ ring_buffer_start, ring_buffer_end, write_frame);
+ dev_err(adev->dev,
+ "write_frame is pointing to address out of bounds\n");
return -EINVAL;
}
@@ -3597,7 +3638,7 @@ static ssize_t psp_usbc_pd_fw_sysfs_read(struct device *dev,
int ret;
if (!adev->ip_blocks[AMD_IP_BLOCK_TYPE_PSP].status.late_initialized) {
- DRM_INFO("PSP block is not ready yet.");
+ dev_info(adev->dev, "PSP block is not ready yet\n.");
return -EBUSY;
}
@@ -3606,7 +3647,7 @@ static ssize_t psp_usbc_pd_fw_sysfs_read(struct device *dev,
mutex_unlock(&adev->psp.mutex);
if (ret) {
- DRM_ERROR("Failed to read USBC PD FW, err = %d", ret);
+ dev_err(adev->dev, "Failed to read USBC PD FW, err = %d\n", ret);
return ret;
}
@@ -3628,7 +3669,7 @@ static ssize_t psp_usbc_pd_fw_sysfs_write(struct device *dev,
void *fw_pri_cpu_addr;
if (!adev->ip_blocks[AMD_IP_BLOCK_TYPE_PSP].status.late_initialized) {
- DRM_INFO("PSP block is not ready yet.");
+ dev_err(adev->dev, "PSP block is not ready yet.");
return -EBUSY;
}
@@ -3661,7 +3702,7 @@ rel_buf:
release_firmware(usbc_pd_fw);
fail:
if (ret) {
- DRM_ERROR("Failed to load USBC PD FW, err = %d", ret);
+ dev_err(adev->dev, "Failed to load USBC PD FW, err = %d", ret);
count = ret;
}
@@ -3708,7 +3749,7 @@ static ssize_t amdgpu_psp_vbflash_write(struct file *filp, struct kobject *kobj,
/* Safeguard against memory drain */
if (adev->psp.vbflash_image_size > AMD_VBIOS_FILE_MAX_SIZE_B) {
- dev_err(adev->dev, "File size cannot exceed %u", AMD_VBIOS_FILE_MAX_SIZE_B);
+ dev_err(adev->dev, "File size cannot exceed %u\n", AMD_VBIOS_FILE_MAX_SIZE_B);
kvfree(adev->psp.vbflash_tmp_buf);
adev->psp.vbflash_tmp_buf = NULL;
adev->psp.vbflash_image_size = 0;
@@ -3727,7 +3768,7 @@ static ssize_t amdgpu_psp_vbflash_write(struct file *filp, struct kobject *kobj,
adev->psp.vbflash_image_size += count;
mutex_unlock(&adev->psp.mutex);
- dev_dbg(adev->dev, "IFWI staged for update");
+ dev_dbg(adev->dev, "IFWI staged for update\n");
return count;
}
@@ -3747,7 +3788,7 @@ static ssize_t amdgpu_psp_vbflash_read(struct file *filp, struct kobject *kobj,
if (adev->psp.vbflash_image_size == 0)
return -EINVAL;
- dev_dbg(adev->dev, "PSP IFWI flash process initiated");
+ dev_dbg(adev->dev, "PSP IFWI flash process initiated\n");
ret = amdgpu_bo_create_kernel(adev, adev->psp.vbflash_image_size,
AMDGPU_GPU_PAGE_SIZE,
@@ -3772,11 +3813,11 @@ rel_buf:
adev->psp.vbflash_image_size = 0;
if (ret) {
- dev_err(adev->dev, "Failed to load IFWI, err = %d", ret);
+ dev_err(adev->dev, "Failed to load IFWI, err = %d\n", ret);
return ret;
}
- dev_dbg(adev->dev, "PSP IFWI flash process done");
+ dev_dbg(adev->dev, "PSP IFWI flash process done\n");
return 0;
}
@@ -3930,3 +3971,11 @@ const struct amdgpu_ip_block_version psp_v13_0_4_ip_block = {
.rev = 4,
.funcs = &psp_ip_funcs,
};
+
+const struct amdgpu_ip_block_version psp_v14_0_ip_block = {
+ .type = AMD_IP_BLOCK_TYPE_PSP,
+ .major = 14,
+ .minor = 0,
+ .rev = 0,
+ .funcs = &psp_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
index c4d9cbde55b9..ee16f134ae92 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
@@ -134,7 +134,7 @@ struct psp_funcs {
int (*update_spirom)(struct psp_context *psp, uint64_t fw_pri_mc_addr);
int (*vbflash_stat)(struct psp_context *psp);
int (*fatal_error_recovery_quirk)(struct psp_context *psp);
- int (*query_boot_status)(struct psp_context *psp);
+ bool (*get_ras_capability)(struct psp_context *psp);
};
struct ta_funcs {
@@ -203,7 +203,7 @@ struct psp_ras_context {
#define GDDR6_MEM_TRAINING_DATA_SIZE_IN_BYTES 0x1000
#define GDDR6_MEM_TRAINING_OFFSET 0x8000
/*Define the VRAM size that will be encroached by BIST training.*/
-#define GDDR6_MEM_TRAINING_ENCROACHED_SIZE 0x2000000
+#define BIST_MEM_TRAINING_ENCROACHED_SIZE 0x2000000
enum psp_memory_training_init_flag {
PSP_MEM_TRAIN_NOT_SUPPORT = 0x0,
@@ -364,6 +364,8 @@ struct psp_context {
atomic_t fence_value;
/* flag to mark whether gfx fw autoload is supported or not */
bool autoload_supported;
+ /* flag to mark whether psp use runtime TMR or boottime TMR */
+ bool boot_time_tmr;
/* flag to mark whether df cstate management centralized to PMFW */
bool pmfw_centralized_cstate_management;
@@ -463,6 +465,7 @@ extern const struct amdgpu_ip_block_version psp_v11_0_8_ip_block;
extern const struct amdgpu_ip_block_version psp_v12_0_ip_block;
extern const struct amdgpu_ip_block_version psp_v13_0_ip_block;
extern const struct amdgpu_ip_block_version psp_v13_0_4_ip_block;
+extern const struct amdgpu_ip_block_version psp_v14_0_ip_block;
extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
uint32_t field_val, uint32_t mask, bool check_changed);
@@ -502,6 +505,9 @@ int psp_ras_enable_features(struct psp_context *psp,
int psp_ras_trigger_error(struct psp_context *psp,
struct ta_ras_trigger_error_input *info, uint32_t instance_mask);
int psp_ras_terminate(struct psp_context *psp);
+int psp_ras_query_address(struct psp_context *psp,
+ struct ta_ras_query_address_input *addr_in,
+ struct ta_ras_query_address_output *addr_out);
int psp_hdcp_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
int psp_dtm_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
@@ -538,7 +544,5 @@ int psp_spatial_partition(struct psp_context *psp, int mode);
int is_psp_fw_valid(struct psp_bin_desc bin);
int amdgpu_psp_wait_for_bootloader(struct amdgpu_device *adev);
-
-int amdgpu_psp_query_boot_status(struct amdgpu_device *adev);
-
+bool amdgpu_psp_get_ras_capability(struct psp_context *psp);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
index 468a67b302d4..ca5c86e5f7cd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c
@@ -362,7 +362,7 @@ static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf, size
}
}
- if (copy_to_user((char *)buf, context->mem_context.shared_buf, shared_buf_len))
+ if (copy_to_user((char *)&buf[copy_pos], context->mem_context.shared_buf, shared_buf_len))
ret = -EFAULT;
err_free_shared_buf:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 31823a30dea2..46f3d1013e8c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -39,6 +39,7 @@
#include "nbio_v7_9.h"
#include "atom.h"
#include "amdgpu_reset.h"
+#include "amdgpu_psp.h"
#ifdef CONFIG_X86_MCE_AMD
#include <asm/mce.h>
@@ -73,6 +74,8 @@ const char *ras_block_string[] = {
"mca",
"vcn",
"jpeg",
+ "ih",
+ "mpio",
};
const char *ras_mca_block_string[] = {
@@ -94,7 +97,8 @@ const char *get_ras_block_str(struct ras_common_if *ras_block)
if (!ras_block)
return "NULL";
- if (ras_block->block >= AMDGPU_RAS_BLOCK_COUNT)
+ if (ras_block->block >= AMDGPU_RAS_BLOCK_COUNT ||
+ ras_block->block >= ARRAY_SIZE(ras_block_string))
return "OUT OF RANGE";
if (ras_block->block == AMDGPU_RAS_BLOCK__MCA)
@@ -116,6 +120,8 @@ const char *get_ras_block_str(struct ras_common_if *ras_block)
/* typical ECC bad page rate is 1 bad page per 100MB VRAM */
#define RAS_BAD_PAGE_COVER (100 * 1024 * 1024ULL)
+#define MAX_UMC_POISON_POLLING_TIME_ASYNC 100 //ms
+
enum amdgpu_ras_retire_page_reservation {
AMDGPU_RAS_RETIRE_PAGE_RESERVED,
AMDGPU_RAS_RETIRE_PAGE_PENDING,
@@ -628,8 +634,12 @@ static ssize_t amdgpu_ras_sysfs_read(struct device *dev,
dev_warn(obj->adev->dev, "Failed to reset error counter and error status");
}
- return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count,
- "ce", info.ce_count);
+ if (info.head.block == AMDGPU_RAS_BLOCK__UMC)
+ return sysfs_emit(buf, "%s: %lu\n%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+ "ce", info.ce_count, "de", info.de_count);
+ else
+ return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+ "ce", info.ce_count);
}
/* obj begin */
@@ -1036,7 +1046,8 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
struct ras_manager *ras_mgr,
struct ras_err_data *err_data,
const char *blk_name,
- bool is_ue)
+ bool is_ue,
+ bool is_de)
{
struct amdgpu_smuio_mcm_config_info *mcm_info;
struct ras_err_node *err_node;
@@ -1065,25 +1076,50 @@ static void amdgpu_ras_error_print_error_data(struct amdgpu_device *adev,
}
} else {
- for_each_ras_error(err_node, err_data) {
- err_info = &err_node->err_info;
- mcm_info = &err_info->mcm_info;
- if (err_info->ce_count) {
+ if (is_de) {
+ for_each_ras_error(err_node, err_data) {
+ err_info = &err_node->err_info;
+ mcm_info = &err_info->mcm_info;
+ if (err_info->de_count) {
+ dev_info(adev->dev, "socket: %d, die: %d, "
+ "%lld new deferred hardware errors detected in %s block\n",
+ mcm_info->socket_id,
+ mcm_info->die_id,
+ err_info->de_count,
+ blk_name);
+ }
+ }
+
+ for_each_ras_error(err_node, &ras_mgr->err_data) {
+ err_info = &err_node->err_info;
+ mcm_info = &err_info->mcm_info;
dev_info(adev->dev, "socket: %d, die: %d, "
- "%lld new correctable hardware errors detected in %s block\n",
- mcm_info->socket_id,
- mcm_info->die_id,
- err_info->ce_count,
- blk_name);
+ "%lld deferred hardware errors detected in total in %s block\n",
+ mcm_info->socket_id, mcm_info->die_id,
+ err_info->de_count, blk_name);
+ }
+ } else {
+ for_each_ras_error(err_node, err_data) {
+ err_info = &err_node->err_info;
+ mcm_info = &err_info->mcm_info;
+ if (err_info->ce_count) {
+ dev_info(adev->dev, "socket: %d, die: %d, "
+ "%lld new correctable hardware errors detected in %s block\n",
+ mcm_info->socket_id,
+ mcm_info->die_id,
+ err_info->ce_count,
+ blk_name);
+ }
}
- }
- for_each_ras_error(err_node, &ras_mgr->err_data) {
- err_info = &err_node->err_info;
- mcm_info = &err_info->mcm_info;
- dev_info(adev->dev, "socket: %d, die: %d, "
- "%lld correctable hardware errors detected in total in %s block\n",
- mcm_info->socket_id, mcm_info->die_id, err_info->ce_count, blk_name);
+ for_each_ras_error(err_node, &ras_mgr->err_data) {
+ err_info = &err_node->err_info;
+ mcm_info = &err_info->mcm_info;
+ dev_info(adev->dev, "socket: %d, die: %d, "
+ "%lld correctable hardware errors detected in total in %s block\n",
+ mcm_info->socket_id, mcm_info->die_id,
+ err_info->ce_count, blk_name);
+ }
}
}
}
@@ -1102,7 +1138,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
if (err_data->ce_count) {
if (err_data_has_source_info(err_data)) {
- amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, false);
+ amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data,
+ blk_name, false, false);
} else if (!adev->aid_mask &&
adev->smuio.funcs &&
adev->smuio.funcs->get_socket_id &&
@@ -1124,7 +1161,8 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
if (err_data->ue_count) {
if (err_data_has_source_info(err_data)) {
- amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data, blk_name, true);
+ amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data,
+ blk_name, true, false);
} else if (!adev->aid_mask &&
adev->smuio.funcs &&
adev->smuio.funcs->get_socket_id &&
@@ -1144,6 +1182,28 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
}
}
+ if (err_data->de_count) {
+ if (err_data_has_source_info(err_data)) {
+ amdgpu_ras_error_print_error_data(adev, ras_mgr, err_data,
+ blk_name, false, true);
+ } else if (!adev->aid_mask &&
+ adev->smuio.funcs &&
+ adev->smuio.funcs->get_socket_id &&
+ adev->smuio.funcs->get_die_id) {
+ dev_info(adev->dev, "socket: %d, die: %d "
+ "%ld deferred hardware errors "
+ "detected in %s block\n",
+ adev->smuio.funcs->get_socket_id(adev),
+ adev->smuio.funcs->get_die_id(adev),
+ ras_mgr->err_data.de_count,
+ blk_name);
+ } else {
+ dev_info(adev->dev, "%ld deferred hardware errors "
+ "detected in %s block\n",
+ ras_mgr->err_data.de_count,
+ blk_name);
+ }
+ }
}
static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, struct ras_err_data *err_data)
@@ -1154,7 +1214,8 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s
if (err_data_has_source_info(err_data)) {
for_each_ras_error(err_node, err_data) {
err_info = &err_node->err_info;
-
+ amdgpu_ras_error_statistic_de_count(&obj->err_data,
+ &err_info->mcm_info, NULL, err_info->de_count);
amdgpu_ras_error_statistic_ce_count(&obj->err_data,
&err_info->mcm_info, NULL, err_info->ce_count);
amdgpu_ras_error_statistic_ue_count(&obj->err_data,
@@ -1164,9 +1225,72 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s
/* for legacy asic path which doesn't has error source info */
obj->err_data.ue_count += err_data->ue_count;
obj->err_data.ce_count += err_data->ce_count;
+ obj->err_data.de_count += err_data->de_count;
}
}
+static struct ras_manager *get_ras_manager(struct amdgpu_device *adev, enum amdgpu_ras_block blk)
+{
+ struct ras_common_if head;
+
+ memset(&head, 0, sizeof(head));
+ head.block = blk;
+
+ return amdgpu_ras_find_obj(adev, &head);
+}
+
+int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+ const struct aca_info *aca_info, void *data)
+{
+ struct ras_manager *obj;
+
+ obj = get_ras_manager(adev, blk);
+ if (!obj)
+ return -EINVAL;
+
+ return amdgpu_aca_add_handle(adev, &obj->aca_handle, ras_block_str(blk), aca_info, data);
+}
+
+int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk)
+{
+ struct ras_manager *obj;
+
+ obj = get_ras_manager(adev, blk);
+ if (!obj)
+ return -EINVAL;
+
+ amdgpu_aca_remove_handle(&obj->aca_handle);
+
+ return 0;
+}
+
+static int amdgpu_aca_log_ras_error_data(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+ enum aca_error_type type, struct ras_err_data *err_data)
+{
+ struct ras_manager *obj;
+
+ obj = get_ras_manager(adev, blk);
+ if (!obj)
+ return -EINVAL;
+
+ return amdgpu_aca_get_error_data(adev, &obj->aca_handle, type, err_data);
+}
+
+ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr,
+ struct aca_handle *handle, char *buf, void *data)
+{
+ struct ras_manager *obj = container_of(handle, struct ras_manager, aca_handle);
+ struct ras_query_if info = {
+ .head = obj->head,
+ };
+
+ if (amdgpu_ras_query_error_status(obj->adev, &info))
+ return -EINVAL;
+
+ return sysfs_emit(buf, "%s: %lu\n%s: %lu\n", "ue", info.ue_count,
+ "ce", info.ce_count);
+}
+
static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev,
struct ras_query_if *info,
struct ras_err_data *err_data,
@@ -1174,6 +1298,7 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev,
{
enum amdgpu_ras_block blk = info ? info->head.block : AMDGPU_RAS_BLOCK_COUNT;
struct amdgpu_ras_block_object *block_obj = NULL;
+ int ret;
if (blk == AMDGPU_RAS_BLOCK_COUNT)
return -EINVAL;
@@ -1203,9 +1328,19 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev,
}
}
} else {
- /* FIXME: add code to check return value later */
- amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data);
- amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data);
+ if (amdgpu_aca_is_enabled(adev)) {
+ ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_UE, err_data);
+ if (ret)
+ return ret;
+
+ ret = amdgpu_aca_log_ras_error_data(adev, blk, ACA_ERROR_TYPE_CE, err_data);
+ if (ret)
+ return ret;
+ } else {
+ /* FIXME: add code to check return value later */
+ amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_UE, err_data);
+ amdgpu_mca_smu_log_ras_error(adev, blk, AMDGPU_MCA_ERROR_TYPE_CE, err_data);
+ }
}
return 0;
@@ -1239,6 +1374,7 @@ int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_i
info->ue_count = obj->err_data.ue_count;
info->ce_count = obj->err_data.ce_count;
+ info->de_count = obj->err_data.de_count;
amdgpu_ras_error_generate_report(adev, info, &err_data);
@@ -1254,6 +1390,7 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, block, 0);
struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+ const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
struct amdgpu_hive_info *hive;
int hive_ras_recovery = 0;
@@ -1264,7 +1401,7 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
}
if (!amdgpu_ras_is_supported(adev, block) ||
- !amdgpu_ras_get_mca_debug_mode(adev))
+ !amdgpu_ras_get_aca_debug_mode(adev))
return -EOPNOTSUPP;
hive = amdgpu_get_xgmi_hive(adev);
@@ -1276,7 +1413,8 @@ int amdgpu_ras_reset_error_count(struct amdgpu_device *adev,
/* skip ras error reset in gpu reset */
if ((amdgpu_in_reset(adev) || atomic_read(&ras->in_recovery) ||
hive_ras_recovery) &&
- mca_funcs && mca_funcs->mca_set_debug_mode)
+ ((smu_funcs && smu_funcs->set_debug_mode) ||
+ (mca_funcs && mca_funcs->mca_set_debug_mode)))
return -EOPNOTSUPP;
if (block_obj->hw_ops->reset_ras_error_count)
@@ -1772,7 +1910,10 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
}
}
- amdgpu_mca_smu_debugfs_init(adev, dir);
+ if (amdgpu_aca_is_enabled(adev))
+ amdgpu_aca_smu_debugfs_init(adev, dir);
+ else
+ amdgpu_mca_smu_debugfs_init(adev, dir);
}
/* debugfs end */
@@ -1900,7 +2041,7 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager *
}
}
- amdgpu_umc_poison_handler(adev, false);
+ amdgpu_umc_poison_handler(adev, obj->head.block, false);
if (block_obj->hw_ops && block_obj->hw_ops->handle_poison_consumption)
poison_stat = block_obj->hw_ops->handle_poison_consumption(adev);
@@ -1951,6 +2092,7 @@ static void amdgpu_ras_interrupt_umc_handler(struct ras_manager *obj,
*/
obj->err_data.ue_count += err_data.ue_count;
obj->err_data.ce_count += err_data.ce_count;
+ obj->err_data.de_count += err_data.de_count;
}
amdgpu_ras_error_data_fini(&err_data);
@@ -2520,6 +2662,32 @@ static void amdgpu_ras_validate_threshold(struct amdgpu_device *adev,
}
}
+static int amdgpu_ras_page_retirement_thread(void *param)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)param;
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+ while (!kthread_should_stop()) {
+
+ wait_event_interruptible(con->page_retirement_wq,
+ kthread_should_stop() ||
+ atomic_read(&con->page_retirement_req_cnt));
+
+ if (kthread_should_stop())
+ break;
+
+ dev_info(adev->dev, "Start processing page retirement. request:%d\n",
+ atomic_read(&con->page_retirement_req_cnt));
+
+ atomic_dec(&con->page_retirement_req_cnt);
+
+ amdgpu_umc_bad_page_polling_timeout(adev,
+ false, MAX_UMC_POISON_POLLING_TIME_ASYNC);
+ }
+
+ return 0;
+}
+
int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
@@ -2583,6 +2751,16 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
}
}
+ mutex_init(&con->page_retirement_lock);
+ init_waitqueue_head(&con->page_retirement_wq);
+ atomic_set(&con->page_retirement_req_cnt, 0);
+ con->page_retirement_thread =
+ kthread_run(amdgpu_ras_page_retirement_thread, adev, "umc_page_retirement");
+ if (IS_ERR(con->page_retirement_thread)) {
+ con->page_retirement_thread = NULL;
+ dev_warn(adev->dev, "Failed to create umc_page_retirement thread!!!\n");
+ }
+
#ifdef CONFIG_X86_MCE_AMD
if ((adev->asic_type == CHIP_ALDEBARAN) &&
(adev->gmc.xgmi.connected_to_cpu))
@@ -2618,6 +2796,11 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
if (!data)
return 0;
+ if (con->page_retirement_thread)
+ kthread_stop(con->page_retirement_thread);
+
+ atomic_set(&con->page_retirement_req_cnt, 0);
+
cancel_work_sync(&con->recovery_work);
mutex_lock(&con->recovery_lock);
@@ -2679,6 +2862,87 @@ static void amdgpu_ras_get_quirks(struct amdgpu_device *adev)
adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__GFX);
}
+/* Query ras capablity via atomfirmware interface */
+static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev)
+{
+ /* mem_ecc cap */
+ if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
+ dev_info(adev->dev, "MEM ECC is active.\n");
+ adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__UMC |
+ 1 << AMDGPU_RAS_BLOCK__DF);
+ } else {
+ dev_info(adev->dev, "MEM ECC is not presented.\n");
+ }
+
+ /* sram_ecc cap */
+ if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
+ dev_info(adev->dev, "SRAM ECC is active.\n");
+ if (!amdgpu_sriov_vf(adev))
+ adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
+ 1 << AMDGPU_RAS_BLOCK__DF);
+ else
+ adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF |
+ 1 << AMDGPU_RAS_BLOCK__SDMA |
+ 1 << AMDGPU_RAS_BLOCK__GFX);
+
+ /*
+ * VCN/JPEG RAS can be supported on both bare metal and
+ * SRIOV environment
+ */
+ if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) ||
+ amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) ||
+ amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3))
+ adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
+ 1 << AMDGPU_RAS_BLOCK__JPEG);
+ else
+ adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
+ 1 << AMDGPU_RAS_BLOCK__JPEG);
+
+ /*
+ * XGMI RAS is not supported if xgmi num physical nodes
+ * is zero
+ */
+ if (!adev->gmc.xgmi.num_physical_nodes)
+ adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL);
+ } else {
+ dev_info(adev->dev, "SRAM ECC is not presented.\n");
+ }
+}
+
+/* Query poison mode from umc/df IP callbacks */
+static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev)
+{
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ bool df_poison, umc_poison;
+
+ /* poison setting is useless on SRIOV guest */
+ if (amdgpu_sriov_vf(adev) || !con)
+ return;
+
+ /* Init poison supported flag, the default value is false */
+ if (adev->gmc.xgmi.connected_to_cpu ||
+ adev->gmc.is_app_apu) {
+ /* enabled by default when GPU is connected to CPU */
+ con->poison_supported = true;
+ } else if (adev->df.funcs &&
+ adev->df.funcs->query_ras_poison_mode &&
+ adev->umc.ras &&
+ adev->umc.ras->query_ras_poison_mode) {
+ df_poison =
+ adev->df.funcs->query_ras_poison_mode(adev);
+ umc_poison =
+ adev->umc.ras->query_ras_poison_mode(adev);
+
+ /* Only poison is set in both DF and UMC, we can support it */
+ if (df_poison && umc_poison)
+ con->poison_supported = true;
+ else if (df_poison != umc_poison)
+ dev_warn(adev->dev,
+ "Poison setting is inconsistent in DF/UMC(%d:%d)!\n",
+ df_poison, umc_poison);
+ }
+}
+
/*
* check hardware's ras ability which will be saved in hw_supported.
* if hardware does not support ras, we can skip some ras initializtion and
@@ -2695,49 +2959,13 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
if (!amdgpu_ras_asic_supported(adev))
return;
- if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
- if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
- dev_info(adev->dev, "MEM ECC is active.\n");
- adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__UMC |
- 1 << AMDGPU_RAS_BLOCK__DF);
- } else {
- dev_info(adev->dev, "MEM ECC is not presented.\n");
- }
-
- if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
- dev_info(adev->dev, "SRAM ECC is active.\n");
- if (!amdgpu_sriov_vf(adev))
- adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
- 1 << AMDGPU_RAS_BLOCK__DF);
- else
- adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__PCIE_BIF |
- 1 << AMDGPU_RAS_BLOCK__SDMA |
- 1 << AMDGPU_RAS_BLOCK__GFX);
-
- /* VCN/JPEG RAS can be supported on both bare metal and
- * SRIOV environment
- */
- if (amdgpu_ip_version(adev, VCN_HWIP, 0) ==
- IP_VERSION(2, 6, 0) ||
- amdgpu_ip_version(adev, VCN_HWIP, 0) ==
- IP_VERSION(4, 0, 0) ||
- amdgpu_ip_version(adev, VCN_HWIP, 0) ==
- IP_VERSION(4, 0, 3))
- adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN |
- 1 << AMDGPU_RAS_BLOCK__JPEG);
- else
- adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__VCN |
- 1 << AMDGPU_RAS_BLOCK__JPEG);
+ /* query ras capability from psp */
+ if (amdgpu_psp_get_ras_capability(&adev->psp))
+ goto init_ras_enabled_flag;
- /*
- * XGMI RAS is not supported if xgmi num physical nodes
- * is zero
- */
- if (!adev->gmc.xgmi.num_physical_nodes)
- adev->ras_hw_enabled &= ~(1 << AMDGPU_RAS_BLOCK__XGMI_WAFL);
- } else {
- dev_info(adev->dev, "SRAM ECC is not presented.\n");
- }
+ /* query ras capablity from bios */
+ if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) {
+ amdgpu_ras_query_ras_capablity_from_vbios(adev);
} else {
/* driver only manages a few IP blocks RAS feature
* when GPU is connected cpu through XGMI */
@@ -2746,13 +2974,21 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
1 << AMDGPU_RAS_BLOCK__MMHUB);
}
+ /* apply asic specific settings (vega20 only for now) */
amdgpu_ras_get_quirks(adev);
+ /* query poison mode from umc/df ip callback */
+ amdgpu_ras_query_poison_mode(adev);
+
+init_ras_enabled_flag:
/* hw_supported needs to be aligned with RAS block mask. */
adev->ras_hw_enabled &= AMDGPU_RAS_BLOCK_MASK;
adev->ras_enabled = amdgpu_ras_enable == 0 ? 0 :
adev->ras_hw_enabled & amdgpu_ras_mask;
+
+ /* aca is disabled by default */
+ adev->aca.is_enabled = false;
}
static void amdgpu_ras_counte_dw(struct work_struct *work)
@@ -2780,39 +3016,6 @@ Out:
pm_runtime_put_autosuspend(dev->dev);
}
-static void amdgpu_ras_query_poison_mode(struct amdgpu_device *adev)
-{
- struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
- bool df_poison, umc_poison;
-
- /* poison setting is useless on SRIOV guest */
- if (amdgpu_sriov_vf(adev) || !con)
- return;
-
- /* Init poison supported flag, the default value is false */
- if (adev->gmc.xgmi.connected_to_cpu ||
- adev->gmc.is_app_apu) {
- /* enabled by default when GPU is connected to CPU */
- con->poison_supported = true;
- } else if (adev->df.funcs &&
- adev->df.funcs->query_ras_poison_mode &&
- adev->umc.ras &&
- adev->umc.ras->query_ras_poison_mode) {
- df_poison =
- adev->df.funcs->query_ras_poison_mode(adev);
- umc_poison =
- adev->umc.ras->query_ras_poison_mode(adev);
-
- /* Only poison is set in both DF and UMC, we can support it */
- if (df_poison && umc_poison)
- con->poison_supported = true;
- else if (df_poison != umc_poison)
- dev_warn(adev->dev,
- "Poison setting is inconsistent in DF/UMC(%d:%d)!\n",
- df_poison, umc_poison);
- }
-}
-
static int amdgpu_get_ras_schema(struct amdgpu_device *adev)
{
return amdgpu_ras_is_poison_mode_supported(adev) ? AMDGPU_RAS_ERROR__POISON : 0 |
@@ -2917,12 +3120,11 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
goto release_con;
}
- amdgpu_ras_query_poison_mode(adev);
-
/* Packed socket_id to ras feature mask bits[31:29] */
if (adev->smuio.funcs &&
adev->smuio.funcs->get_socket_id)
- con->features |= ((adev->smuio.funcs->get_socket_id(adev)) << 29);
+ con->features |= ((adev->smuio.funcs->get_socket_id(adev)) <<
+ AMDGPU_RAS_FEATURES_SOCKETID_SHIFT);
/* Get RAS schema for particular SOC */
con->schema = amdgpu_get_ras_schema(adev);
@@ -3128,7 +3330,7 @@ void amdgpu_ras_suspend(struct amdgpu_device *adev)
amdgpu_ras_disable_all_features(adev, 0);
/* Make sure all ras objects are disabled. */
- if (con->features)
+ if (AMDGPU_RAS_GET_FEATURES(con->features))
amdgpu_ras_disable_all_features(adev, 1);
}
@@ -3142,15 +3344,29 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev)
if (amdgpu_sriov_vf(adev))
return 0;
- amdgpu_ras_set_mca_debug_mode(adev, false);
+ if (amdgpu_aca_is_enabled(adev)) {
+ if (amdgpu_in_reset(adev))
+ r = amdgpu_aca_reset(adev);
+ else
+ r = amdgpu_aca_init(adev);
+ if (r)
+ return r;
+
+ amdgpu_ras_set_aca_debug_mode(adev, false);
+ } else {
+ amdgpu_ras_set_mca_debug_mode(adev, false);
+ }
list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
- if (!node->ras_obj) {
+ obj = node->ras_obj;
+ if (!obj) {
dev_warn(adev->dev, "Warning: abnormal ras list node.\n");
continue;
}
- obj = node->ras_obj;
+ if (!amdgpu_ras_is_supported(adev, obj->ras_comm.block))
+ continue;
+
if (obj->ras_late_init) {
r = obj->ras_late_init(adev, &obj->ras_comm);
if (r) {
@@ -3175,7 +3391,7 @@ int amdgpu_ras_pre_fini(struct amdgpu_device *adev)
/* Need disable ras on all IPs here before ip [hw/sw]fini */
- if (con->features)
+ if (AMDGPU_RAS_GET_FEATURES(con->features))
amdgpu_ras_disable_all_features(adev, 0);
amdgpu_ras_recovery_fini(adev);
return 0;
@@ -3208,10 +3424,13 @@ int amdgpu_ras_fini(struct amdgpu_device *adev)
amdgpu_ras_fs_fini(adev);
amdgpu_ras_interrupt_remove_all(adev);
- WARN(con->features, "Feature mask is not cleared");
+ if (amdgpu_aca_is_enabled(adev))
+ amdgpu_aca_fini(adev);
- if (con->features)
- amdgpu_ras_disable_all_features(adev, 1);
+ WARN(AMDGPU_RAS_GET_FEATURES(con->features), "Feature mask is not cleared");
+
+ if (AMDGPU_RAS_GET_FEATURES(con->features))
+ amdgpu_ras_disable_all_features(adev, 0);
cancel_delayed_work_sync(&con->ras_counte_delay_work);
@@ -3425,22 +3644,41 @@ int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable)
if (con) {
ret = amdgpu_mca_smu_set_debug_mode(adev, enable);
if (!ret)
- con->is_mca_debug_mode = enable;
+ con->is_aca_debug_mode = enable;
+ }
+
+ return ret;
+}
+
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable)
+{
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ int ret = 0;
+
+ if (con) {
+ if (amdgpu_aca_is_enabled(adev))
+ ret = amdgpu_aca_smu_set_debug_mode(adev, enable);
+ else
+ ret = amdgpu_mca_smu_set_debug_mode(adev, enable);
+ if (!ret)
+ con->is_aca_debug_mode = enable;
}
return ret;
}
-bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev)
+bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
if (!con)
return false;
- if (mca_funcs && mca_funcs->mca_set_debug_mode)
- return con->is_mca_debug_mode;
+ if ((amdgpu_aca_is_enabled(adev) && smu_funcs && smu_funcs->set_debug_mode) ||
+ (!amdgpu_aca_is_enabled(adev) && mca_funcs && mca_funcs->mca_set_debug_mode))
+ return con->is_aca_debug_mode;
else
return true;
}
@@ -3450,15 +3688,16 @@ bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
const struct amdgpu_mca_smu_funcs *mca_funcs = adev->mca.mca_funcs;
+ const struct aca_smu_funcs *smu_funcs = adev->aca.smu_funcs;
if (!con) {
*error_query_mode = AMDGPU_RAS_INVALID_ERROR_QUERY;
return false;
}
- if (mca_funcs && mca_funcs->mca_set_debug_mode)
+ if ((smu_funcs && smu_funcs->set_debug_mode) || (mca_funcs && mca_funcs->mca_set_debug_mode))
*error_query_mode =
- (con->is_mca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY;
+ (con->is_aca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY;
else
*error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY;
@@ -3699,8 +3938,7 @@ static int ras_err_info_cmp(void *priv, const struct list_head *a, const struct
}
static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_data,
- struct amdgpu_smuio_mcm_config_info *mcm_info,
- struct ras_err_addr *err_addr)
+ struct amdgpu_smuio_mcm_config_info *mcm_info)
{
struct ras_err_node *err_node;
@@ -3712,10 +3950,9 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d
if (!err_node)
return NULL;
- memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info));
+ INIT_LIST_HEAD(&err_node->err_info.err_addr_list);
- if (err_addr)
- memcpy(&err_node->err_info.err_addr, err_addr, sizeof(*err_addr));
+ memcpy(&err_node->err_info.mcm_info, mcm_info, sizeof(*mcm_info));
err_data->err_list_count++;
list_add_tail(&err_node->node, &err_data->err_node_list);
@@ -3724,6 +3961,29 @@ static struct ras_err_info *amdgpu_ras_error_get_info(struct ras_err_data *err_d
return &err_node->err_info;
}
+void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info, struct ras_err_addr *err_addr)
+{
+ struct ras_err_addr *mca_err_addr;
+
+ mca_err_addr = kzalloc(sizeof(*mca_err_addr), GFP_KERNEL);
+ if (!mca_err_addr)
+ return;
+
+ INIT_LIST_HEAD(&mca_err_addr->node);
+
+ mca_err_addr->err_status = err_addr->err_status;
+ mca_err_addr->err_ipid = err_addr->err_ipid;
+ mca_err_addr->err_addr = err_addr->err_addr;
+
+ list_add_tail(&mca_err_addr->node, &err_info->err_addr_list);
+}
+
+void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info, struct ras_err_addr *mca_err_addr)
+{
+ list_del(&mca_err_addr->node);
+ kfree(mca_err_addr);
+}
+
int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
struct amdgpu_smuio_mcm_config_info *mcm_info,
struct ras_err_addr *err_addr, u64 count)
@@ -3736,10 +3996,13 @@ int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
if (!count)
return 0;
- err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr);
+ err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
if (!err_info)
return -EINVAL;
+ if (err_addr && err_addr->err_status)
+ amdgpu_ras_add_mca_err_addr(err_info, err_addr);
+
err_info->ue_count += count;
err_data->ue_count += count;
@@ -3758,7 +4021,7 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
if (!count)
return 0;
- err_info = amdgpu_ras_error_get_info(err_data, mcm_info, err_addr);
+ err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
if (!err_info)
return -EINVAL;
@@ -3767,3 +4030,135 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
return 0;
}
+
+int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data,
+ struct amdgpu_smuio_mcm_config_info *mcm_info,
+ struct ras_err_addr *err_addr, u64 count)
+{
+ struct ras_err_info *err_info;
+
+ if (!err_data || !mcm_info)
+ return -EINVAL;
+
+ if (!count)
+ return 0;
+
+ err_info = amdgpu_ras_error_get_info(err_data, mcm_info);
+ if (!err_info)
+ return -EINVAL;
+
+ if (err_addr && err_addr->err_status)
+ amdgpu_ras_add_mca_err_addr(err_info, err_addr);
+
+ err_info->de_count += count;
+ err_data->de_count += count;
+
+ return 0;
+}
+
+#define mmMP0_SMN_C2PMSG_92 0x1609C
+#define mmMP0_SMN_C2PMSG_126 0x160BE
+static void amdgpu_ras_boot_time_error_reporting(struct amdgpu_device *adev,
+ u32 instance, u32 boot_error)
+{
+ u32 socket_id, aid_id, hbm_id;
+ u32 reg_data;
+ u64 reg_addr;
+
+ socket_id = AMDGPU_RAS_GPU_ERR_SOCKET_ID(boot_error);
+ aid_id = AMDGPU_RAS_GPU_ERR_AID_ID(boot_error);
+ hbm_id = AMDGPU_RAS_GPU_ERR_HBM_ID(boot_error);
+
+ /* The pattern for smn addressing in other SOC could be different from
+ * the one for aqua_vanjaram. We should revisit the code if the pattern
+ * is changed. In such case, replace the aqua_vanjaram implementation
+ * with more common helper */
+ reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) +
+ aqua_vanjaram_encode_ext_smn_addressing(instance);
+
+ reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+ dev_err(adev->dev, "socket: %d, aid: %d, firmware boot failed, fw status is 0x%x\n",
+ socket_id, aid_id, reg_data);
+
+ if (AMDGPU_RAS_GPU_ERR_MEM_TRAINING(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, memory training failed\n",
+ socket_id, aid_id, hbm_id);
+
+ if (AMDGPU_RAS_GPU_ERR_FW_LOAD(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, firmware load failed at boot time\n",
+ socket_id, aid_id);
+
+ if (AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, wafl link training failed\n",
+ socket_id, aid_id);
+
+ if (AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, xgmi link training failed\n",
+ socket_id, aid_id);
+
+ if (AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, usr cp link training failed\n",
+ socket_id, aid_id);
+
+ if (AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, usr dp link training failed\n",
+ socket_id, aid_id);
+
+ if (AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm memory test failed\n",
+ socket_id, aid_id, hbm_id);
+
+ if (AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(boot_error))
+ dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm bist test failed\n",
+ socket_id, aid_id, hbm_id);
+}
+
+static int amdgpu_ras_wait_for_boot_complete(struct amdgpu_device *adev,
+ u32 instance, u32 *boot_error)
+{
+ u32 reg_addr;
+ u32 reg_data;
+ int retry_loop;
+
+ reg_addr = (mmMP0_SMN_C2PMSG_92 << 2) +
+ aqua_vanjaram_encode_ext_smn_addressing(instance);
+
+ for (retry_loop = 0; retry_loop < AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT; retry_loop++) {
+ reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+ if ((reg_data & AMDGPU_RAS_BOOT_STATUS_MASK) == AMDGPU_RAS_BOOT_STEADY_STATUS) {
+ *boot_error = AMDGPU_RAS_BOOT_SUCEESS;
+ return 0;
+ }
+ msleep(1);
+ }
+
+ /* The pattern for smn addressing in other SOC could be different from
+ * the one for aqua_vanjaram. We should revisit the code if the pattern
+ * is changed. In such case, replace the aqua_vanjaram implementation
+ * with more common helper */
+ reg_addr = (mmMP0_SMN_C2PMSG_126 << 2) +
+ aqua_vanjaram_encode_ext_smn_addressing(instance);
+
+ for (retry_loop = 0; retry_loop < AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT; retry_loop++) {
+ reg_data = amdgpu_device_indirect_rreg_ext(adev, reg_addr);
+ if (AMDGPU_RAS_GPU_ERR_BOOT_STATUS(reg_data)) {
+ *boot_error = reg_data;
+ return 0;
+ }
+ msleep(1);
+ }
+
+ *boot_error = reg_data;
+ return -ETIME;
+}
+
+void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances)
+{
+ u32 boot_error = 0;
+ u32 i;
+
+ for (i = 0; i < num_instances; i++) {
+ if (amdgpu_ras_wait_for_boot_complete(adev, i, &boot_error))
+ amdgpu_ras_boot_time_error_reporting(adev, i, boot_error);
+ }
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
index 76fb85628716..d10e5bb0e52f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
@@ -29,9 +29,28 @@
#include "ta_ras_if.h"
#include "amdgpu_ras_eeprom.h"
#include "amdgpu_smuio.h"
+#include "amdgpu_aca.h"
struct amdgpu_iv_entry;
+#define AMDGPU_RAS_GPU_ERR_MEM_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 0, 0)
+#define AMDGPU_RAS_GPU_ERR_FW_LOAD(x) AMDGPU_GET_REG_FIELD(x, 1, 1)
+#define AMDGPU_RAS_GPU_ERR_WAFL_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 2, 2)
+#define AMDGPU_RAS_GPU_ERR_XGMI_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 3, 3)
+#define AMDGPU_RAS_GPU_ERR_USR_CP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 4, 4)
+#define AMDGPU_RAS_GPU_ERR_USR_DP_LINK_TRAINING(x) AMDGPU_GET_REG_FIELD(x, 5, 5)
+#define AMDGPU_RAS_GPU_ERR_HBM_MEM_TEST(x) AMDGPU_GET_REG_FIELD(x, 6, 6)
+#define AMDGPU_RAS_GPU_ERR_HBM_BIST_TEST(x) AMDGPU_GET_REG_FIELD(x, 7, 7)
+#define AMDGPU_RAS_GPU_ERR_SOCKET_ID(x) AMDGPU_GET_REG_FIELD(x, 10, 8)
+#define AMDGPU_RAS_GPU_ERR_AID_ID(x) AMDGPU_GET_REG_FIELD(x, 12, 11)
+#define AMDGPU_RAS_GPU_ERR_HBM_ID(x) AMDGPU_GET_REG_FIELD(x, 13, 13)
+#define AMDGPU_RAS_GPU_ERR_BOOT_STATUS(x) AMDGPU_GET_REG_FIELD(x, 31, 31)
+
+#define AMDGPU_RAS_BOOT_STATUS_POLLING_LIMIT 1000
+#define AMDGPU_RAS_BOOT_STEADY_STATUS 0xBA
+#define AMDGPU_RAS_BOOT_STATUS_MASK 0xFF
+#define AMDGPU_RAS_BOOT_SUCEESS 0x80000000
+
#define AMDGPU_RAS_FLAG_INIT_BY_VBIOS (0x1 << 0)
/* position of instance value in sub_block_index of
* ta_ras_trigger_error_input, the sub block uses lower 12 bits
@@ -39,6 +58,12 @@ struct amdgpu_iv_entry;
#define AMDGPU_RAS_INST_MASK 0xfffff000
#define AMDGPU_RAS_INST_SHIFT 0xc
+#define AMDGPU_RAS_FEATURES_SOCKETID_SHIFT 29
+#define AMDGPU_RAS_FEATURES_SOCKETID_MASK 0xe0000000
+
+/* The high three bits indicates socketid */
+#define AMDGPU_RAS_GET_FEATURES(val) ((val) & ~AMDGPU_RAS_FEATURES_SOCKETID_MASK)
+
enum amdgpu_ras_block {
AMDGPU_RAS_BLOCK__UMC = 0,
AMDGPU_RAS_BLOCK__SDMA,
@@ -57,6 +82,8 @@ enum amdgpu_ras_block {
AMDGPU_RAS_BLOCK__MCA,
AMDGPU_RAS_BLOCK__VCN,
AMDGPU_RAS_BLOCK__JPEG,
+ AMDGPU_RAS_BLOCK__IH,
+ AMDGPU_RAS_BLOCK__MPIO,
AMDGPU_RAS_BLOCK__LAST
};
@@ -441,10 +468,15 @@ struct amdgpu_ras {
/* Indicates smu whether need update bad channel info */
bool update_channel_flag;
/* Record status of smu mca debug mode */
- bool is_mca_debug_mode;
+ bool is_aca_debug_mode;
/* Record special requirements of gpu reset caller */
uint32_t gpu_reset_flags;
+
+ struct task_struct *page_retirement_thread;
+ wait_queue_head_t page_retirement_wq;
+ struct mutex page_retirement_lock;
+ atomic_t page_retirement_req_cnt;
};
struct ras_fs_data {
@@ -453,6 +485,7 @@ struct ras_fs_data {
};
struct ras_err_addr {
+ struct list_head node;
uint64_t err_status;
uint64_t err_ipid;
uint64_t err_addr;
@@ -462,7 +495,8 @@ struct ras_err_info {
struct amdgpu_smuio_mcm_config_info mcm_info;
u64 ce_count;
u64 ue_count;
- struct ras_err_addr err_addr;
+ u64 de_count;
+ struct list_head err_addr_list;
};
struct ras_err_node {
@@ -473,6 +507,7 @@ struct ras_err_node {
struct ras_err_data {
unsigned long ue_count;
unsigned long ce_count;
+ unsigned long de_count;
unsigned long err_addr_cnt;
struct eeprom_table_record *err_addr;
u32 err_list_count;
@@ -529,6 +564,8 @@ struct ras_manager {
struct ras_ih_data ih_data;
struct ras_err_data err_data;
+
+ struct aca_handle aca_handle;
};
struct ras_badpage {
@@ -548,6 +585,7 @@ struct ras_query_if {
struct ras_common_if head;
unsigned long ue_count;
unsigned long ce_count;
+ unsigned long de_count;
};
struct ras_inject_if {
@@ -781,7 +819,8 @@ struct amdgpu_ras* amdgpu_ras_get_context(struct amdgpu_device *adev);
int amdgpu_ras_set_context(struct amdgpu_device *adev, struct amdgpu_ras *ras_con);
int amdgpu_ras_set_mca_debug_mode(struct amdgpu_device *adev, bool enable);
-bool amdgpu_ras_get_mca_debug_mode(struct amdgpu_device *adev);
+int amdgpu_ras_set_aca_debug_mode(struct amdgpu_device *adev, bool enable);
+bool amdgpu_ras_get_aca_debug_mode(struct amdgpu_device *adev);
bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
unsigned int *mode);
@@ -818,5 +857,20 @@ int amdgpu_ras_error_statistic_ce_count(struct ras_err_data *err_data,
int amdgpu_ras_error_statistic_ue_count(struct ras_err_data *err_data,
struct amdgpu_smuio_mcm_config_info *mcm_info,
struct ras_err_addr *err_addr, u64 count);
+int amdgpu_ras_error_statistic_de_count(struct ras_err_data *err_data,
+ struct amdgpu_smuio_mcm_config_info *mcm_info,
+ struct ras_err_addr *err_addr, u64 count);
+void amdgpu_ras_query_boot_status(struct amdgpu_device *adev, u32 num_instances);
+int amdgpu_ras_bind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk,
+ const struct aca_info *aca_info, void *data);
+int amdgpu_ras_unbind_aca(struct amdgpu_device *adev, enum amdgpu_ras_block blk);
+
+ssize_t amdgpu_ras_aca_sysfs_read(struct device *dev, struct device_attribute *attr,
+ struct aca_handle *handle, char *buf, void *data);
+
+void amdgpu_ras_add_mca_err_addr(struct ras_err_info *err_info,
+ struct ras_err_addr *err_addr);
+void amdgpu_ras_del_mca_err_addr(struct ras_err_info *err_info,
+ struct ras_err_addr *mca_err_addr);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
index 2fde93b00cab..b12808c0c331 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
@@ -735,6 +735,9 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control)
control->tbl_rai.rma_status = GPU_RETIRED__ECC_REACH_THRESHOLD;
control->tbl_rai.health_percent = 0;
}
+
+ /* ignore the -ENOTSUPP return value */
+ amdgpu_dpm_send_rma_reason(adev);
}
if (control->tbl_hdr.version == RAS_TABLE_VER_V2_1)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
index 2c3675d91614..db5791e1a7ce 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.c
@@ -241,7 +241,7 @@ void amdgpu_gfx_rlc_setup_cp_table(struct amdgpu_device *adev)
table_size = le32_to_cpu(hdr->jt_size);
}
- for (i = 0; i < table_size; i ++) {
+ for (i = 0; i < table_size; i++) {
dst_ptr[bo_offset + i] =
cpu_to_le32(le32_to_cpu(fw_data[table_offset + i]));
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
index b591d33af264..5a17e0ff2ab8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
@@ -169,7 +169,7 @@ struct amdgpu_rlc_funcs {
void (*stop)(struct amdgpu_device *adev);
void (*reset)(struct amdgpu_device *adev);
void (*start)(struct amdgpu_device *adev);
- void (*update_spm_vmid)(struct amdgpu_device *adev, unsigned vmid);
+ void (*update_spm_vmid)(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid);
bool (*is_rlcg_access_range)(struct amdgpu_device *adev, uint32_t reg);
};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
index 7a6a67275404..e22cb2b5cd92 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c
@@ -36,13 +36,24 @@
*/
/**
+ * amdgpu_seq64_get_va_base - Get the seq64 va base address
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns:
+ * va base address on success
+ */
+static inline u64 amdgpu_seq64_get_va_base(struct amdgpu_device *adev)
+{
+ return AMDGPU_VA_RESERVED_SEQ64_START(adev);
+}
+
+/**
* amdgpu_seq64_map - Map the seq64 memory to VM
*
* @adev: amdgpu_device pointer
* @vm: vm pointer
* @bo_va: bo_va pointer
- * @seq64_addr: seq64 vaddr start address
- * @size: seq64 pool size
*
* Map the seq64 memory to the given VM.
*
@@ -50,11 +61,11 @@
* 0 on success or a negative error code on failure
*/
int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
- struct amdgpu_bo_va **bo_va, u64 seq64_addr,
- uint32_t size)
+ struct amdgpu_bo_va **bo_va)
{
struct amdgpu_bo *bo;
struct drm_exec exec;
+ u64 seq64_addr;
int r;
bo = adev->seq64.sbo;
@@ -77,9 +88,9 @@ int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
goto error;
}
- r = amdgpu_vm_bo_map(adev, *bo_va, seq64_addr, 0, size,
- AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE |
- AMDGPU_PTE_EXECUTABLE);
+ seq64_addr = amdgpu_seq64_get_va_base(adev);
+ r = amdgpu_vm_bo_map(adev, *bo_va, seq64_addr, 0, AMDGPU_VA_RESERVED_SEQ64_SIZE,
+ AMDGPU_PTE_READABLE);
if (r) {
DRM_ERROR("failed to do bo_map on userq sem, err=%d\n", r);
amdgpu_vm_bo_del(adev, *bo_va);
@@ -144,31 +155,25 @@ error:
* amdgpu_seq64_alloc - Allocate a 64 bit memory
*
* @adev: amdgpu_device pointer
- * @gpu_addr: allocated gpu VA start address
- * @cpu_addr: allocated cpu VA start address
+ * @va: VA to access the seq in process address space
+ * @cpu_addr: CPU address to access the seq
*
* Alloc a 64 bit memory from seq64 pool.
*
* Returns:
* 0 on success or a negative error code on failure
*/
-int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *gpu_addr,
- u64 **cpu_addr)
+int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *va, u64 **cpu_addr)
{
unsigned long bit_pos;
- u32 offset;
bit_pos = find_first_zero_bit(adev->seq64.used, adev->seq64.num_sem);
+ if (bit_pos >= adev->seq64.num_sem)
+ return -ENOSPC;
- if (bit_pos < adev->seq64.num_sem) {
- __set_bit(bit_pos, adev->seq64.used);
- offset = bit_pos << 6; /* convert to qw offset */
- } else {
- return -EINVAL;
- }
-
- *gpu_addr = offset + AMDGPU_SEQ64_VADDR_START;
- *cpu_addr = offset + adev->seq64.cpu_base_addr;
+ __set_bit(bit_pos, adev->seq64.used);
+ *va = bit_pos * sizeof(u64) + amdgpu_seq64_get_va_base(adev);
+ *cpu_addr = bit_pos + adev->seq64.cpu_base_addr;
return 0;
}
@@ -177,20 +182,17 @@ int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *gpu_addr,
* amdgpu_seq64_free - Free the given 64 bit memory
*
* @adev: amdgpu_device pointer
- * @gpu_addr: gpu start address to be freed
+ * @va: gpu start address to be freed
*
* Free the given 64 bit memory from seq64 pool.
- *
*/
-void amdgpu_seq64_free(struct amdgpu_device *adev, u64 gpu_addr)
+void amdgpu_seq64_free(struct amdgpu_device *adev, u64 va)
{
- u32 offset;
-
- offset = gpu_addr - AMDGPU_SEQ64_VADDR_START;
+ unsigned long bit_pos;
- offset >>= 6;
- if (offset < adev->seq64.num_sem)
- __clear_bit(offset, adev->seq64.used);
+ bit_pos = (va - amdgpu_seq64_get_va_base(adev)) / sizeof(u64);
+ if (bit_pos < adev->seq64.num_sem)
+ __clear_bit(bit_pos, adev->seq64.used);
}
/**
@@ -229,7 +231,7 @@ int amdgpu_seq64_init(struct amdgpu_device *adev)
* AMDGPU_MAX_SEQ64_SLOTS * sizeof(u64) * 8 = AMDGPU_MAX_SEQ64_SLOTS
* 64bit slots
*/
- r = amdgpu_bo_create_kernel(adev, AMDGPU_SEQ64_SIZE,
+ r = amdgpu_bo_create_kernel(adev, AMDGPU_VA_RESERVED_SEQ64_SIZE,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
&adev->seq64.sbo, NULL,
(void **)&adev->seq64.cpu_base_addr);
@@ -238,7 +240,7 @@ int amdgpu_seq64_init(struct amdgpu_device *adev)
return r;
}
- memset(adev->seq64.cpu_base_addr, 0, AMDGPU_SEQ64_SIZE);
+ memset(adev->seq64.cpu_base_addr, 0, AMDGPU_VA_RESERVED_SEQ64_SIZE);
adev->seq64.num_sem = AMDGPU_MAX_SEQ64_SLOTS;
memset(&adev->seq64.used, 0, sizeof(adev->seq64.used));
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
index 2196e72be508..4203b2ab318d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.h
@@ -25,10 +25,9 @@
#ifndef __AMDGPU_SEQ64_H__
#define __AMDGPU_SEQ64_H__
-#define AMDGPU_SEQ64_SIZE (2ULL << 20)
-#define AMDGPU_MAX_SEQ64_SLOTS (AMDGPU_SEQ64_SIZE / (sizeof(u64) * 8))
-#define AMDGPU_SEQ64_VADDR_OFFSET 0x50000
-#define AMDGPU_SEQ64_VADDR_START (AMDGPU_VA_RESERVED_SIZE + AMDGPU_SEQ64_VADDR_OFFSET)
+#include "amdgpu_vm.h"
+
+#define AMDGPU_MAX_SEQ64_SLOTS (AMDGPU_VA_RESERVED_SEQ64_SIZE / sizeof(u64))
struct amdgpu_seq64 {
struct amdgpu_bo *sbo;
@@ -42,7 +41,7 @@ int amdgpu_seq64_init(struct amdgpu_device *adev);
int amdgpu_seq64_alloc(struct amdgpu_device *adev, u64 *gpu_addr, u64 **cpu_addr);
void amdgpu_seq64_free(struct amdgpu_device *adev, u64 gpu_addr);
int amdgpu_seq64_map(struct amdgpu_device *adev, struct amdgpu_vm *vm,
- struct amdgpu_bo_va **bo_va, u64 seq64_addr, uint32_t size);
+ struct amdgpu_bo_va **bo_va);
void amdgpu_seq64_unmap(struct amdgpu_device *adev, struct amdgpu_fpriv *fpriv);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
index 3e12763e477a..0867fd9e15ba 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -556,6 +556,8 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
default:
if (!load_type)
return AMDGPU_FW_LOAD_DIRECT;
+ else if (load_type == 3)
+ return AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO;
else
return AMDGPU_FW_LOAD_PSP;
}
@@ -678,6 +680,8 @@ const char *amdgpu_ucode_name(enum AMDGPU_UCODE_ID ucode_id)
return "UMSCH_MM_DATA";
case AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER:
return "UMSCH_MM_CMD_BUFFER";
+ case AMDGPU_UCODE_ID_JPEG_RAM:
+ return "JPEG";
default:
return "UNKNOWN UCODE";
}
@@ -1060,7 +1064,8 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
int amdgpu_ucode_create_bo(struct amdgpu_device *adev)
{
- if (adev->firmware.load_type != AMDGPU_FW_LOAD_DIRECT) {
+ if ((adev->firmware.load_type != AMDGPU_FW_LOAD_DIRECT) &&
+ (adev->firmware.load_type != AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO)) {
amdgpu_bo_create_kernel(adev, adev->firmware.fw_size, PAGE_SIZE,
(amdgpu_sriov_vf(adev) || adev->debug_use_vram_fw_buf) ?
AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
index 4244a13f9f22..619445760037 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
@@ -511,6 +511,7 @@ enum AMDGPU_UCODE_ID {
AMDGPU_UCODE_ID_UMSCH_MM_DATA,
AMDGPU_UCODE_ID_UMSCH_MM_CMD_BUFFER,
AMDGPU_UCODE_ID_P2S_TABLE,
+ AMDGPU_UCODE_ID_JPEG_RAM,
AMDGPU_UCODE_ID_MAXIMUM,
};
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
index d65e21914d8c..20436f81856a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@@ -23,6 +23,7 @@
#include "amdgpu.h"
#include "umc_v6_7.h"
+#define MAX_UMC_POISON_POLLING_TIME_SYNC 20 //ms
static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev,
struct ras_err_data *err_data, uint64_t err_addr,
@@ -85,18 +86,21 @@ out_fini_err_data:
return ret;
}
-static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
- void *ras_error_status,
- struct amdgpu_iv_entry *entry,
- bool reset)
+static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
+ void *ras_error_status)
{
struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ unsigned int error_query_mode;
int ret = 0;
+ unsigned long err_count;
- kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+ amdgpu_ras_get_error_query_mode(adev, &error_query_mode);
+
+ mutex_lock(&con->page_retirement_lock);
ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc));
- if (ret == -EOPNOTSUPP) {
+ if (ret == -EOPNOTSUPP &&
+ error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status);
@@ -120,7 +124,8 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
*/
adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status);
}
- } else if (!ret) {
+ } else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY ||
+ (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) {
if (adev->umc.ras &&
adev->umc.ras->ecc_info_query_ras_error_count)
adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status);
@@ -147,16 +152,13 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
}
/* only uncorrectable error needs gpu reset */
- if (err_data->ue_count) {
- dev_info(adev->dev, "%ld uncorrectable hardware errors "
- "detected in UMC block\n",
- err_data->ue_count);
-
+ if (err_data->ue_count || err_data->de_count) {
+ err_count = err_data->ue_count + err_data->de_count;
if ((amdgpu_bad_page_threshold != 0) &&
err_data->err_addr_cnt) {
amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
err_data->err_addr_cnt);
- amdgpu_ras_save_bad_pages(adev, &(err_data->ue_count));
+ amdgpu_ras_save_bad_pages(adev, &err_count);
amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);
@@ -165,20 +167,87 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
con->update_channel_flag = false;
}
}
-
- if (reset) {
- /* use mode-2 reset for poison consumption */
- if (!entry)
- con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
- amdgpu_ras_reset_gpu(adev);
- }
}
kfree(err_data->err_addr);
+
+ mutex_unlock(&con->page_retirement_lock);
+}
+
+static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
+ void *ras_error_status,
+ struct amdgpu_iv_entry *entry,
+ bool reset)
+{
+ struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+ kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+ amdgpu_umc_handle_bad_pages(adev, ras_error_status);
+
+ if (err_data->ue_count && reset) {
+ /* use mode-2 reset for poison consumption */
+ if (!entry)
+ con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+ amdgpu_ras_reset_gpu(adev);
+ }
+
return AMDGPU_RAS_SUCCESS;
}
-int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset)
+int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev,
+ bool reset, uint32_t timeout_ms)
+{
+ struct ras_err_data err_data;
+ struct ras_common_if head = {
+ .block = AMDGPU_RAS_BLOCK__UMC,
+ };
+ struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+ uint32_t timeout = timeout_ms;
+
+ memset(&err_data, 0, sizeof(err_data));
+ amdgpu_ras_error_data_init(&err_data);
+
+ do {
+
+ amdgpu_umc_handle_bad_pages(adev, &err_data);
+
+ if (timeout && !err_data.de_count) {
+ msleep(1);
+ timeout--;
+ }
+
+ } while (timeout && !err_data.de_count);
+
+ if (!timeout)
+ dev_warn(adev->dev, "Can't find bad pages\n");
+
+ if (err_data.de_count)
+ dev_info(adev->dev, "%ld new deferred hardware errors detected\n", err_data.de_count);
+
+ if (obj) {
+ obj->err_data.ue_count += err_data.ue_count;
+ obj->err_data.ce_count += err_data.ce_count;
+ obj->err_data.de_count += err_data.de_count;
+ }
+
+ amdgpu_ras_error_data_fini(&err_data);
+
+ kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
+
+ if (reset) {
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+
+ /* use mode-2 reset for poison consumption */
+ con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
+ amdgpu_ras_reset_gpu(adev);
+ }
+
+ return 0;
+}
+
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block, bool reset)
{
int ret = AMDGPU_RAS_SUCCESS;
@@ -195,27 +264,41 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset)
}
if (!amdgpu_sriov_vf(adev)) {
- struct ras_err_data err_data;
- struct ras_common_if head = {
- .block = AMDGPU_RAS_BLOCK__UMC,
- };
- struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+ if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
+ struct ras_err_data err_data;
+ struct ras_common_if head = {
+ .block = AMDGPU_RAS_BLOCK__UMC,
+ };
+ struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+
+ ret = amdgpu_ras_error_data_init(&err_data);
+ if (ret)
+ return ret;
- ret = amdgpu_ras_error_data_init(&err_data);
- if (ret)
- return ret;
+ ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
- ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
+ if (ret == AMDGPU_RAS_SUCCESS && obj) {
+ obj->err_data.ue_count += err_data.ue_count;
+ obj->err_data.ce_count += err_data.ce_count;
+ obj->err_data.de_count += err_data.de_count;
+ }
- if (ret == AMDGPU_RAS_SUCCESS && obj) {
- obj->err_data.ue_count += err_data.ue_count;
- obj->err_data.ce_count += err_data.ce_count;
- }
+ amdgpu_ras_error_data_fini(&err_data);
+ } else {
+ if (reset) {
+ amdgpu_umc_bad_page_polling_timeout(adev,
+ reset, MAX_UMC_POISON_POLLING_TIME_SYNC);
+ } else {
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
- amdgpu_ras_error_data_fini(&err_data);
+ atomic_inc(&con->page_retirement_req_cnt);
+
+ wake_up(&con->page_retirement_wq);
+ }
+ }
} else {
if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
- adev->virt.ops->ras_poison_handler(adev);
+ adev->virt.ops->ras_poison_handler(adev, block);
else
dev_warn(adev->dev,
"No ras_poison_handler interface in SRIOV!\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
index 417a6726c71b..26d2ae498daf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h
@@ -21,7 +21,7 @@
#ifndef __AMDGPU_UMC_H__
#define __AMDGPU_UMC_H__
#include "amdgpu_ras.h"
-
+#include "amdgpu_mca.h"
/*
* (addr / 256) * 4096, the higher 26 bits in ErrorAddr
* is the index of 4KB block
@@ -64,6 +64,8 @@ struct amdgpu_umc_ras {
void *ras_error_status);
void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev,
void *ras_error_status);
+ bool (*check_ecc_err_status)(struct amdgpu_device *adev,
+ enum amdgpu_mca_error_type type, void *ras_error_status);
/* support different eeprom table version for different asic */
void (*set_eeprom_table_version)(struct amdgpu_ras_eeprom_table_header *hdr);
};
@@ -100,7 +102,8 @@ struct amdgpu_umc {
int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev);
int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block);
-int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset);
+int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block, bool reset);
int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry);
@@ -118,4 +121,7 @@ int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
umc_func func, void *data);
+
+int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev,
+ bool reset, uint32_t timeout_ms);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h
index 107f9bb0e24f..5b27fc41ffbf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umr.h
@@ -69,12 +69,12 @@ struct amdgpu_debugfs_gprwave_data {
};
enum AMDGPU_DEBUGFS_REGS2_CMDS {
- AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE=0,
+ AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE = 0,
AMDGPU_DEBUGFS_REGS2_CMD_SET_STATE_V2,
};
enum AMDGPU_DEBUGFS_GPRWAVE_CMDS {
- AMDGPU_DEBUGFS_GPRWAVE_CMD_SET_STATE=0,
+ AMDGPU_DEBUGFS_GPRWAVE_CMD_SET_STATE = 0,
};
//reg2 interface
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
index bfbf59326ee1..ab820cf52668 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umsch_mm.c
@@ -358,7 +358,7 @@ static int setup_umsch_mm_test(struct amdgpu_device *adev,
memset(test->ring_data_cpu_addr, 0, sizeof(struct umsch_mm_test_ring_data));
- test->ring_data_gpu_addr = AMDGPU_VA_RESERVED_SIZE;
+ test->ring_data_gpu_addr = AMDGPU_VA_RESERVED_BOTTOM;
r = map_ring_data(adev, test->vm, test->ring_data_obj, &test->bo_va,
test->ring_data_gpu_addr, sizeof(struct umsch_mm_test_ring_data));
if (r)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
index f4963330c772..eb2a88991206 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
@@ -59,6 +59,7 @@
#define FIRMWARE_VCN4_0_3 "amdgpu/vcn_4_0_3.bin"
#define FIRMWARE_VCN4_0_4 "amdgpu/vcn_4_0_4.bin"
#define FIRMWARE_VCN4_0_5 "amdgpu/vcn_4_0_5.bin"
+#define FIRMWARE_VCN5_0_0 "amdgpu/vcn_5_0_0.bin"
MODULE_FIRMWARE(FIRMWARE_RAVEN);
MODULE_FIRMWARE(FIRMWARE_PICASSO);
@@ -82,6 +83,7 @@ MODULE_FIRMWARE(FIRMWARE_VCN4_0_2);
MODULE_FIRMWARE(FIRMWARE_VCN4_0_3);
MODULE_FIRMWARE(FIRMWARE_VCN4_0_4);
MODULE_FIRMWARE(FIRMWARE_VCN4_0_5);
+MODULE_FIRMWARE(FIRMWARE_VCN5_0_0);
static void amdgpu_vcn_idle_work_handler(struct work_struct *work);
@@ -1189,7 +1191,7 @@ int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev,
amdgpu_ras_interrupt_dispatch(adev, &ih_data);
} else {
if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
- adev->virt.ops->ras_poison_handler(adev);
+ adev->virt.ops->ras_poison_handler(adev, ras_if->block);
else
dev_warn(adev->dev,
"No ras_poison_handler interface in SRIOV for VCN!\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
index 514c98ea144f..1985f71b4373 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
@@ -160,6 +160,48 @@
} \
} while (0)
+#define SOC24_DPG_MODE_OFFSET(ip, inst_idx, reg) \
+ ({ \
+ uint32_t internal_reg_offset, addr; \
+ bool video_range, aon_range; \
+ \
+ addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg); \
+ addr <<= 2; \
+ video_range = ((((0xFFFFF & addr) >= (VCN_VID_SOC_ADDRESS)) && \
+ ((0xFFFFF & addr) < ((VCN_VID_SOC_ADDRESS + 0x2600))))); \
+ aon_range = ((((0xFFFFF & addr) >= (VCN_AON_SOC_ADDRESS)) && \
+ ((0xFFFFF & addr) < ((VCN_AON_SOC_ADDRESS + 0x600))))); \
+ if (video_range) \
+ internal_reg_offset = ((0xFFFFF & addr) - (VCN_VID_SOC_ADDRESS) + \
+ (VCN_VID_IP_ADDRESS)); \
+ else if (aon_range) \
+ internal_reg_offset = ((0xFFFFF & addr) - (VCN_AON_SOC_ADDRESS) + \
+ (VCN_AON_IP_ADDRESS)); \
+ else \
+ internal_reg_offset = (0xFFFFF & addr); \
+ \
+ internal_reg_offset >>= 2; \
+ })
+
+#define WREG32_SOC24_DPG_MODE(inst_idx, offset, value, mask_en, indirect) \
+ do { \
+ if (!indirect) { \
+ WREG32_SOC15(VCN, GET_INST(VCN, inst_idx), \
+ regUVD_DPG_LMA_DATA, value); \
+ WREG32_SOC15( \
+ VCN, GET_INST(VCN, inst_idx), \
+ regUVD_DPG_LMA_CTL, \
+ (0x1 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT | \
+ mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT | \
+ offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT)); \
+ } else { \
+ *adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ = \
+ offset; \
+ *adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ = \
+ value; \
+ } \
+ } while (0)
+
#define AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE (1 << 2)
#define AMDGPU_FW_SHARED_FLAG_0_DRM_KEY_INJECT (1 << 4)
#define AMDGPU_VCN_FW_SHARED_FLAG_0_RB (1 << 6)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
index 0dcff2889e25..6ff7d3fb2008 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
@@ -71,59 +71,6 @@ void amdgpu_virt_init_setting(struct amdgpu_device *adev)
amdgpu_num_kcq = 2;
}
-void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev,
- uint32_t reg0, uint32_t reg1,
- uint32_t ref, uint32_t mask,
- uint32_t xcc_inst)
-{
- struct amdgpu_kiq *kiq = &adev->gfx.kiq[xcc_inst];
- struct amdgpu_ring *ring = &kiq->ring;
- signed long r, cnt = 0;
- unsigned long flags;
- uint32_t seq;
-
- if (adev->mes.ring.sched.ready) {
- amdgpu_mes_reg_write_reg_wait(adev, reg0, reg1,
- ref, mask);
- return;
- }
-
- spin_lock_irqsave(&kiq->ring_lock, flags);
- amdgpu_ring_alloc(ring, 32);
- amdgpu_ring_emit_reg_write_reg_wait(ring, reg0, reg1,
- ref, mask);
- r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT);
- if (r)
- goto failed_undo;
-
- amdgpu_ring_commit(ring);
- spin_unlock_irqrestore(&kiq->ring_lock, flags);
-
- r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
-
- /* don't wait anymore for IRQ context */
- if (r < 1 && in_interrupt())
- goto failed_kiq;
-
- might_sleep();
- while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) {
-
- msleep(MAX_KIQ_REG_BAILOUT_INTERVAL);
- r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT);
- }
-
- if (cnt > MAX_KIQ_REG_TRY)
- goto failed_kiq;
-
- return;
-
-failed_undo:
- amdgpu_ring_undo(ring);
- spin_unlock_irqrestore(&kiq->ring_lock, flags);
-failed_kiq:
- dev_err(adev->dev, "failed to write reg %x wait reg %x\n", reg0, reg1);
-}
-
/**
* amdgpu_virt_request_full_gpu() - request full gpu access
* @adev: amdgpu device.
@@ -303,11 +250,11 @@ static int amdgpu_virt_init_ras_err_handler_data(struct amdgpu_device *adev)
if (!*data)
goto data_failure;
- bps = kmalloc_array(align_space, sizeof((*data)->bps), GFP_KERNEL);
+ bps = kmalloc_array(align_space, sizeof(*(*data)->bps), GFP_KERNEL);
if (!bps)
goto bps_failure;
- bps_bo = kmalloc_array(align_space, sizeof((*data)->bps_bo), GFP_KERNEL);
+ bps_bo = kmalloc_array(align_space, sizeof(*(*data)->bps_bo), GFP_KERNEL);
if (!bps_bo)
goto bps_bo_failure;
@@ -340,8 +287,10 @@ static void amdgpu_virt_ras_release_bp(struct amdgpu_device *adev)
for (i = data->last_reserved - 1; i >= 0; i--) {
bo = data->bps_bo[i];
- amdgpu_bo_free_kernel(&bo, NULL, NULL);
- data->bps_bo[i] = bo;
+ if (bo) {
+ amdgpu_bo_free_kernel(&bo, NULL, NULL);
+ data->bps_bo[i] = bo;
+ }
data->last_reserved = i;
}
}
@@ -381,6 +330,8 @@ static void amdgpu_virt_ras_reserve_bps(struct amdgpu_device *adev)
{
struct amdgpu_virt *virt = &adev->virt;
struct amdgpu_virt_ras_err_handler_data *data = virt->virt_eh_data;
+ struct amdgpu_vram_mgr *mgr = &adev->mman.vram_mgr;
+ struct ttm_resource_manager *man = &mgr->manager;
struct amdgpu_bo *bo = NULL;
uint64_t bp;
int i;
@@ -396,12 +347,18 @@ static void amdgpu_virt_ras_reserve_bps(struct amdgpu_device *adev)
* 2) a ras bad page has been reserved (duplicate error injection
* for one page);
*/
- if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT,
- AMDGPU_GPU_PAGE_SIZE,
- &bo, NULL))
- DRM_DEBUG("RAS WARN: reserve vram for retired page %llx fail\n", bp);
-
- data->bps_bo[i] = bo;
+ if (ttm_resource_manager_used(man)) {
+ amdgpu_vram_mgr_reserve_range(&adev->mman.vram_mgr,
+ bp << AMDGPU_GPU_PAGE_SHIFT,
+ AMDGPU_GPU_PAGE_SIZE);
+ data->bps_bo[i] = NULL;
+ } else {
+ if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT,
+ AMDGPU_GPU_PAGE_SIZE,
+ &bo, NULL))
+ DRM_DEBUG("RAS WARN: reserve vram for retired page %llx fail\n", bp);
+ data->bps_bo[i] = bo;
+ }
data->last_reserved = i + 1;
bo = NULL;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
index d4207e44141f..fa7be5f277b9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
@@ -88,7 +88,8 @@ struct amdgpu_virt_ops {
int (*wait_reset)(struct amdgpu_device *adev);
void (*trans_msg)(struct amdgpu_device *adev, enum idh_request req,
u32 data1, u32 data2, u32 data3);
- void (*ras_poison_handler)(struct amdgpu_device *adev);
+ void (*ras_poison_handler)(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block);
};
/*
@@ -332,10 +333,6 @@ static inline bool is_virtual_machine(void)
((adev)->virt.gim_feature & AMDGIM_FEATURE_VCN_RB_DECOUPLE)
bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev);
void amdgpu_virt_init_setting(struct amdgpu_device *adev);
-void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev,
- uint32_t reg0, uint32_t rreg1,
- uint32_t ref, uint32_t mask,
- uint32_t xcc_inst);
int amdgpu_virt_request_full_gpu(struct amdgpu_device *adev, bool init);
int amdgpu_virt_release_full_gpu(struct amdgpu_device *adev, bool init);
int amdgpu_virt_reset_gpu(struct amdgpu_device *adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
index 453a4b786cfc..8baa2e0935cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c
@@ -660,8 +660,7 @@ static const struct amd_ip_funcs amdgpu_vkms_ip_funcs = {
.set_powergating_state = amdgpu_vkms_set_powergating_state,
};
-const struct amdgpu_ip_block_version amdgpu_vkms_ip_block =
-{
+const struct amdgpu_ip_block_version amdgpu_vkms_ip_block = {
.type = AMD_IP_BLOCK_TYPE_DCE,
.major = 1,
.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index b8fcb6c55698..ed4a8c5d26d7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -234,6 +234,22 @@ static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
}
/**
+ * amdgpu_vm_bo_evicted_user - vm_bo is evicted
+ *
+ * @vm_bo: vm_bo which is evicted
+ *
+ * State for BOs used by user mode queues which are not at the location they
+ * should be.
+ */
+static void amdgpu_vm_bo_evicted_user(struct amdgpu_vm_bo_base *vm_bo)
+{
+ vm_bo->moved = true;
+ spin_lock(&vm_bo->vm->status_lock);
+ list_move(&vm_bo->vm_status, &vm_bo->vm->evicted_user);
+ spin_unlock(&vm_bo->vm->status_lock);
+}
+
+/**
* amdgpu_vm_bo_relocated - vm_bo is reloacted
*
* @vm_bo: vm_bo which is relocated
@@ -427,21 +443,25 @@ uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm)
}
/**
- * amdgpu_vm_validate_pt_bos - validate the page table BOs
+ * amdgpu_vm_validate - validate evicted BOs tracked in the VM
*
* @adev: amdgpu device pointer
* @vm: vm providing the BOs
+ * @ticket: optional reservation ticket used to reserve the VM
* @validate: callback to do the validation
* @param: parameter for the validation callback
*
- * Validate the page table BOs on command submission if neccessary.
+ * Validate the page table BOs and per-VM BOs on command submission if
+ * necessary. If a ticket is given, also try to validate evicted user queue
+ * BOs. They must already be reserved with the given ticket.
*
* Returns:
* Validation result.
*/
-int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
- int (*validate)(void *p, struct amdgpu_bo *bo),
- void *param)
+int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+ struct ww_acquire_ctx *ticket,
+ int (*validate)(void *p, struct amdgpu_bo *bo),
+ void *param)
{
struct amdgpu_vm_bo_base *bo_base;
struct amdgpu_bo *shadow;
@@ -484,6 +504,28 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
}
spin_lock(&vm->status_lock);
}
+ while (ticket && !list_empty(&vm->evicted_user)) {
+ bo_base = list_first_entry(&vm->evicted_user,
+ struct amdgpu_vm_bo_base,
+ vm_status);
+ spin_unlock(&vm->status_lock);
+
+ bo = bo_base->bo;
+
+ if (dma_resv_locking_ctx(bo->tbo.base.resv) != ticket) {
+ pr_warn_ratelimited("Evicted user BO is not reserved in pid %d\n",
+ vm->task_info.pid);
+ return -EINVAL;
+ }
+
+ r = validate(param, bo);
+ if (r)
+ return r;
+
+ amdgpu_vm_bo_invalidated(bo_base);
+
+ spin_lock(&vm->status_lock);
+ }
spin_unlock(&vm->status_lock);
amdgpu_vm_eviction_lock(vm);
@@ -651,7 +693,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job,
amdgpu_gmc_emit_pasid_mapping(ring, job->vmid, job->pasid);
if (spm_update_needed && adev->gfx.rlc.funcs->update_spm_vmid)
- adev->gfx.rlc.funcs->update_spm_vmid(adev, job->vmid);
+ adev->gfx.rlc.funcs->update_spm_vmid(adev, ring, job->vmid);
if (!ring->is_mes_queue && ring->funcs->emit_gds_switch &&
gds_switch_needed) {
@@ -1426,11 +1468,21 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
}
r = amdgpu_vm_bo_update(adev, bo_va, clear);
- if (r)
- return r;
if (unlock)
dma_resv_unlock(resv);
+ if (r)
+ return r;
+
+ /* Remember evicted DMABuf imports in compute VMs for later
+ * validation
+ */
+ if (vm->is_compute_context &&
+ bo_va->base.bo->tbo.base.import_attach &&
+ (!bo_va->base.bo->tbo.resource ||
+ bo_va->base.bo->tbo.resource->mem_type == TTM_PL_SYSTEM))
+ amdgpu_vm_bo_evicted_user(&bo_va->base);
+
spin_lock(&vm->status_lock);
}
spin_unlock(&vm->status_lock);
@@ -2196,6 +2248,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
for (i = 0; i < AMDGPU_MAX_VMHUBS; i++)
vm->reserved_vmid[i] = NULL;
INIT_LIST_HEAD(&vm->evicted);
+ INIT_LIST_HEAD(&vm->evicted_user);
INIT_LIST_HEAD(&vm->relocated);
INIT_LIST_HEAD(&vm->moved);
INIT_LIST_HEAD(&vm->idle);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 4740dd65b99d..42f6ddec50c1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -135,8 +135,21 @@ struct amdgpu_mem_stats;
#define AMDGPU_IS_MMHUB0(x) ((x) >= AMDGPU_MMHUB0_START && (x) < AMDGPU_MMHUB1_START)
#define AMDGPU_IS_MMHUB1(x) ((x) >= AMDGPU_MMHUB1_START && (x) < AMDGPU_MAX_VMHUBS)
-/* Reserve 2MB at top/bottom of address space for kernel use */
-#define AMDGPU_VA_RESERVED_SIZE (2ULL << 20)
+/* Reserve space at top/bottom of address space for kernel use */
+#define AMDGPU_VA_RESERVED_CSA_SIZE (2ULL << 20)
+#define AMDGPU_VA_RESERVED_CSA_START(adev) (((adev)->vm_manager.max_pfn \
+ << AMDGPU_GPU_PAGE_SHIFT) \
+ - AMDGPU_VA_RESERVED_CSA_SIZE)
+#define AMDGPU_VA_RESERVED_SEQ64_SIZE (2ULL << 20)
+#define AMDGPU_VA_RESERVED_SEQ64_START(adev) (AMDGPU_VA_RESERVED_CSA_START(adev) \
+ - AMDGPU_VA_RESERVED_SEQ64_SIZE)
+#define AMDGPU_VA_RESERVED_TRAP_SIZE (2ULL << 12)
+#define AMDGPU_VA_RESERVED_TRAP_START(adev) (AMDGPU_VA_RESERVED_SEQ64_START(adev) \
+ - AMDGPU_VA_RESERVED_TRAP_SIZE)
+#define AMDGPU_VA_RESERVED_BOTTOM (1ULL << 16)
+#define AMDGPU_VA_RESERVED_TOP (AMDGPU_VA_RESERVED_TRAP_SIZE + \
+ AMDGPU_VA_RESERVED_SEQ64_SIZE + \
+ AMDGPU_VA_RESERVED_CSA_SIZE)
/* See vm_update_mode */
#define AMDGPU_VM_USE_CPU_FOR_GFX (1 << 0)
@@ -288,9 +301,12 @@ struct amdgpu_vm {
/* Lock to protect vm_bo add/del/move on all lists of vm */
spinlock_t status_lock;
- /* BOs who needs a validation */
+ /* Per-VM and PT BOs who needs a validation */
struct list_head evicted;
+ /* BOs for user mode queues that need a validation */
+ struct list_head evicted_user;
+
/* PT BOs which relocated and their parent need an update */
struct list_head relocated;
@@ -434,9 +450,10 @@ int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec,
unsigned int num_fences);
bool amdgpu_vm_ready(struct amdgpu_vm *vm);
uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm);
-int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
- int (*callback)(void *p, struct amdgpu_bo *bo),
- void *param);
+int amdgpu_vm_validate(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+ struct ww_acquire_ctx *ticket,
+ int (*callback)(void *p, struct amdgpu_bo *bo),
+ void *param);
int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync);
int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
struct amdgpu_vm *vm, bool immediate);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index a6c88f2fe6e5..20d51f6c9bb8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -1035,15 +1035,74 @@ int amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
return 0;
}
+static int xgmi_v6_4_0_aca_bank_generate_report(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data)
+{
+ struct amdgpu_device *adev = handle->adev;
+ const char *error_str;
+ u64 status;
+ int ret, ext_error_code;
+
+ ret = aca_bank_info_decode(bank, &report->info);
+ if (ret)
+ return ret;
+
+ status = bank->regs[ACA_REG_IDX_STATUS];
+ ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status);
+
+ error_str = ext_error_code < ARRAY_SIZE(xgmi_v6_4_0_ras_error_code_ext) ?
+ xgmi_v6_4_0_ras_error_code_ext[ext_error_code] : NULL;
+ if (error_str)
+ dev_info(adev->dev, "%s detected\n", error_str);
+
+ if ((type == ACA_ERROR_TYPE_UE && ext_error_code == 0) ||
+ (type == ACA_ERROR_TYPE_CE && ext_error_code == 6))
+ report->count[type] = ACA_REG__MISC0__ERRCNT(bank->regs[ACA_REG_IDX_MISC0]);
+
+ return 0;
+}
+
+static const struct aca_bank_ops xgmi_v6_4_0_aca_bank_ops = {
+ .aca_bank_generate_report = xgmi_v6_4_0_aca_bank_generate_report,
+};
+
+static const struct aca_info xgmi_v6_4_0_aca_info = {
+ .hwip = ACA_HWIP_TYPE_PCS_XGMI,
+ .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+ .bank_ops = &xgmi_v6_4_0_aca_bank_ops,
+};
+
static int amdgpu_xgmi_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
{
+ int r;
+
if (!adev->gmc.xgmi.supported ||
adev->gmc.xgmi.num_physical_nodes == 0)
return 0;
amdgpu_ras_reset_error_count(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL);
- return amdgpu_ras_block_late_init(adev, ras_block);
+ r = amdgpu_ras_block_late_init(adev, ras_block);
+ if (r)
+ return r;
+
+ switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) {
+ case IP_VERSION(6, 4, 0):
+ r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__XGMI_WAFL,
+ &xgmi_v6_4_0_aca_info, NULL);
+ if (r)
+ goto late_fini;
+ break;
+ default:
+ break;
+ }
+
+ return 0;
+
+late_fini:
+ amdgpu_ras_block_late_fini(adev, ras_block);
+
+ return r;
}
uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev,
@@ -1099,7 +1158,7 @@ static void amdgpu_xgmi_legacy_reset_ras_error_count(struct amdgpu_device *adev)
static void __xgmi_v6_4_0_reset_error_count(struct amdgpu_device *adev, int xgmi_inst, u64 mca_base)
{
- WREG64_MCA(xgmi_inst, mca_base, MCA_REG_IDX_STATUS, 0ULL);
+ WREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS, 0ULL);
}
static void xgmi_v6_4_0_reset_error_count(struct amdgpu_device *adev, int xgmi_inst)
@@ -1277,12 +1336,12 @@ static void amdgpu_xgmi_legacy_query_ras_error_count(struct amdgpu_device *adev,
err_data->ce_count += ce_cnt;
}
-static enum amdgpu_mca_error_type xgmi_v6_4_0_pcs_mca_get_error_type(struct amdgpu_device *adev, u64 status)
+static enum aca_error_type xgmi_v6_4_0_pcs_mca_get_error_type(struct amdgpu_device *adev, u64 status)
{
const char *error_str;
int ext_error_code;
- ext_error_code = MCA_REG__STATUS__ERRORCODEEXT(status);
+ ext_error_code = ACA_REG__STATUS__ERRORCODEEXT(status);
error_str = ext_error_code < ARRAY_SIZE(xgmi_v6_4_0_ras_error_code_ext) ?
xgmi_v6_4_0_ras_error_code_ext[ext_error_code] : NULL;
@@ -1291,9 +1350,9 @@ static enum amdgpu_mca_error_type xgmi_v6_4_0_pcs_mca_get_error_type(struct amdg
switch (ext_error_code) {
case 0:
- return AMDGPU_MCA_ERROR_TYPE_UE;
+ return ACA_ERROR_TYPE_UE;
case 6:
- return AMDGPU_MCA_ERROR_TYPE_CE;
+ return ACA_ERROR_TYPE_CE;
default:
return -EINVAL;
}
@@ -1307,22 +1366,22 @@ static void __xgmi_v6_4_0_query_error_count(struct amdgpu_device *adev, struct a
int xgmi_inst = mcm_info->die_id;
u64 status = 0;
- status = RREG64_MCA(xgmi_inst, mca_base, MCA_REG_IDX_STATUS);
- if (!MCA_REG__STATUS__VAL(status))
+ status = RREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS);
+ if (!ACA_REG__STATUS__VAL(status))
return;
switch (xgmi_v6_4_0_pcs_mca_get_error_type(adev, status)) {
- case AMDGPU_MCA_ERROR_TYPE_UE:
+ case ACA_ERROR_TYPE_UE:
amdgpu_ras_error_statistic_ue_count(err_data, mcm_info, NULL, 1ULL);
break;
- case AMDGPU_MCA_ERROR_TYPE_CE:
+ case ACA_ERROR_TYPE_CE:
amdgpu_ras_error_statistic_ce_count(err_data, mcm_info, NULL, 1ULL);
break;
default:
break;
}
- WREG64_MCA(xgmi_inst, mca_base, MCA_REG_IDX_STATUS, 0ULL);
+ WREG64_MCA(xgmi_inst, mca_base, ACA_REG_IDX_STATUS, 0ULL);
}
static void xgmi_v6_4_0_query_error_count(struct amdgpu_device *adev, int xgmi_inst, struct ras_err_data *err_data)
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c
new file mode 100644
index 000000000000..8a0773b80864
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.c
@@ -0,0 +1,122 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "athub_v4_1_0.h"
+#include "athub/athub_4_1_0_offset.h"
+#include "athub/athub_4_1_0_sh_mask.h"
+#include "soc15_common.h"
+
+static uint32_t athub_v4_1_0_get_cg_cntl(struct amdgpu_device *adev)
+{
+ uint32_t data;
+
+ switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+ case IP_VERSION(4, 1, 0):
+ data = RREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL);
+ break;
+ default:
+ data = 0;
+ break;
+ }
+ return data;
+}
+
+static void athub_v4_1_0_set_cg_cntl(struct amdgpu_device *adev, uint32_t data)
+{
+ switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+ case IP_VERSION(4, 1, 0):
+ WREG32_SOC15(ATHUB, 0, regATHUB_MISC_CNTL, data);
+ break;
+ default:
+ break;
+ }
+}
+
+static void
+athub_v4_1_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t def, data;
+
+ def = data = athub_v4_1_0_get_cg_cntl(adev);
+
+ if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_MGCG))
+ data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+ else
+ data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
+
+ if (def != data)
+ athub_v4_1_0_set_cg_cntl(adev, data);
+}
+
+static void
+athub_v4_1_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t def, data;
+
+ def = data = athub_v4_1_0_get_cg_cntl(adev);
+
+ if (enable && (adev->cg_flags & AMD_CG_SUPPORT_ATHUB_LS))
+ data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+ else
+ data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
+
+ if (def != data)
+ athub_v4_1_0_set_cg_cntl(adev, data);
+}
+
+int athub_v4_1_0_set_clockgating(struct amdgpu_device *adev,
+ enum amd_clockgating_state state)
+{
+ if (amdgpu_sriov_vf(adev))
+ return 0;
+
+ switch (amdgpu_ip_version(adev, ATHUB_HWIP, 0)) {
+ case IP_VERSION(4, 1, 0):
+ athub_v4_1_0_update_medium_grain_clock_gating(adev,
+ state == AMD_CG_STATE_GATE);
+ athub_v4_1_0_update_medium_grain_light_sleep(adev,
+ state == AMD_CG_STATE_GATE);
+ break;
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+void athub_v4_1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags)
+{
+ int data;
+
+ /* AMD_CG_SUPPORT_ATHUB_MGCG */
+ data = athub_v4_1_0_get_cg_cntl(adev);
+ if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
+ *flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
+
+ /* AMD_CG_SUPPORT_ATHUB_LS */
+ if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
+ *flags |= AMD_CG_SUPPORT_ATHUB_LS;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h
new file mode 100644
index 000000000000..4d18d0998fa8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/athub_v4_1_0.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __ATHUB_V4_1_0_H__
+#define __ATHUB_V4_1_0_H__
+
+int athub_v4_1_0_set_clockgating(struct amdgpu_device *adev,
+ enum amd_clockgating_state state);
+void athub_v4_1_0_get_clockgating(struct amdgpu_device *adev, u64 *flags);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.c b/drivers/gpu/drm/amd/amdgpu/atom.c
index a33e890c70d9..b888613f653f 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.c
+++ b/drivers/gpu/drm/amd/amdgpu/atom.c
@@ -62,6 +62,7 @@
typedef struct {
struct atom_context *ctx;
uint32_t *ps, *ws;
+ int ps_size, ws_size;
int ps_shift;
uint16_t start;
unsigned last_jump;
@@ -70,8 +71,8 @@ typedef struct {
} atom_exec_context;
int amdgpu_atom_debug;
-static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params);
-int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params);
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params, int params_size);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size);
static uint32_t atom_arg_mask[8] =
{ 0xFFFFFFFF, 0xFFFF, 0xFFFF00, 0xFFFF0000, 0xFF, 0xFF00, 0xFF0000,
@@ -223,7 +224,10 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
(*ptr)++;
/* get_unaligned_le32 avoids unaligned accesses from atombios
* tables, noticed on a DEC Alpha. */
- val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+ if (idx < ctx->ps_size)
+ val = get_unaligned_le32((u32 *)&ctx->ps[idx]);
+ else
+ pr_info("PS index out of range: %i > %i\n", idx, ctx->ps_size);
if (print)
DEBUG("PS[0x%02X,0x%04X]", idx, val);
break;
@@ -261,7 +265,10 @@ static uint32_t atom_get_src_int(atom_exec_context *ctx, uint8_t attr,
val = gctx->reg_block;
break;
default:
- val = ctx->ws[idx];
+ if (idx < ctx->ws_size)
+ val = ctx->ws[idx];
+ else
+ pr_info("WS index out of range: %i > %i\n", idx, ctx->ws_size);
}
break;
case ATOM_ARG_ID:
@@ -495,6 +502,10 @@ static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
idx = U8(*ptr);
(*ptr)++;
DEBUG("PS[0x%02X]", idx);
+ if (idx >= ctx->ps_size) {
+ pr_info("PS index out of range: %i > %i\n", idx, ctx->ps_size);
+ return;
+ }
ctx->ps[idx] = cpu_to_le32(val);
break;
case ATOM_ARG_WS:
@@ -527,6 +538,10 @@ static void atom_put_dst(atom_exec_context *ctx, int arg, uint8_t attr,
gctx->reg_block = val;
break;
default:
+ if (idx >= ctx->ws_size) {
+ pr_info("WS index out of range: %i > %i\n", idx, ctx->ws_size);
+ return;
+ }
ctx->ws[idx] = val;
}
break;
@@ -624,7 +639,7 @@ static void atom_op_calltable(atom_exec_context *ctx, int *ptr, int arg)
else
SDEBUG(" table: %d\n", idx);
if (U16(ctx->ctx->cmd_table + 4 + 2 * idx))
- r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift);
+ r = amdgpu_atom_execute_table_locked(ctx->ctx, idx, ctx->ps + ctx->ps_shift, ctx->ps_size - ctx->ps_shift);
if (r) {
ctx->abort = true;
}
@@ -1203,7 +1218,7 @@ static struct {
atom_op_div32, ATOM_ARG_WS},
};
-static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params)
+static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index, uint32_t *params, int params_size)
{
int base = CU16(ctx->cmd_table + 4 + 2 * index);
int len, ws, ps, ptr;
@@ -1225,12 +1240,16 @@ static int amdgpu_atom_execute_table_locked(struct atom_context *ctx, int index,
ectx.ps_shift = ps / 4;
ectx.start = base;
ectx.ps = params;
+ ectx.ps_size = params_size;
ectx.abort = false;
ectx.last_jump = 0;
- if (ws)
+ if (ws) {
ectx.ws = kcalloc(4, ws, GFP_KERNEL);
- else
+ ectx.ws_size = ws;
+ } else {
ectx.ws = NULL;
+ ectx.ws_size = 0;
+ }
debug_depth++;
while (1) {
@@ -1264,7 +1283,7 @@ free:
return ret;
}
-int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params)
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size)
{
int r;
@@ -1280,7 +1299,7 @@ int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *par
/* reset divmul */
ctx->divmul[0] = 0;
ctx->divmul[1] = 0;
- r = amdgpu_atom_execute_table_locked(ctx, index, params);
+ r = amdgpu_atom_execute_table_locked(ctx, index, params, params_size);
mutex_unlock(&ctx->mutex);
return r;
}
@@ -1552,7 +1571,7 @@ int amdgpu_atom_asic_init(struct atom_context *ctx)
if (!CU16(ctx->cmd_table + 4 + 2 * ATOM_CMD_INIT))
return 1;
- ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps);
+ ret = amdgpu_atom_execute_table(ctx, ATOM_CMD_INIT, ps, 16);
if (ret)
return ret;
diff --git a/drivers/gpu/drm/amd/amdgpu/atom.h b/drivers/gpu/drm/amd/amdgpu/atom.h
index c11cf18a0f18..b807f6639a4c 100644
--- a/drivers/gpu/drm/amd/amdgpu/atom.h
+++ b/drivers/gpu/drm/amd/amdgpu/atom.h
@@ -156,7 +156,7 @@ struct atom_context {
extern int amdgpu_atom_debug;
struct atom_context *amdgpu_atom_parse(struct card_info *card, void *bios);
-int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params);
+int amdgpu_atom_execute_table(struct atom_context *ctx, int index, uint32_t *params, int params_size);
int amdgpu_atom_asic_init(struct atom_context *ctx);
void amdgpu_atom_destroy(struct atom_context *ctx);
bool amdgpu_atom_parse_data_header(struct atom_context *ctx, int index, uint16_t *size,
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
index 10098fdd33fc..3dfc28840a7d 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
@@ -77,7 +77,7 @@ void amdgpu_atombios_crtc_overscan_setup(struct drm_crtc *crtc,
args.usOverscanTop = cpu_to_le16(amdgpu_crtc->v_border);
break;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
@@ -106,7 +106,7 @@ void amdgpu_atombios_crtc_scaler_setup(struct drm_crtc *crtc)
args.ucEnable = ATOM_SCALER_DISABLE;
break;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
@@ -123,7 +123,7 @@ void amdgpu_atombios_crtc_lock(struct drm_crtc *crtc, int lock)
args.ucCRTC = amdgpu_crtc->crtc_id;
args.ucEnable = lock;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
@@ -139,7 +139,7 @@ void amdgpu_atombios_crtc_enable(struct drm_crtc *crtc, int state)
args.ucCRTC = amdgpu_crtc->crtc_id;
args.ucEnable = state;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
@@ -155,7 +155,7 @@ void amdgpu_atombios_crtc_blank(struct drm_crtc *crtc, int state)
args.ucCRTC = amdgpu_crtc->crtc_id;
args.ucBlanking = state;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
@@ -171,7 +171,7 @@ void amdgpu_atombios_crtc_powergate(struct drm_crtc *crtc, int state)
args.ucDispPipeId = amdgpu_crtc->crtc_id;
args.ucEnable = state;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
@@ -183,7 +183,7 @@ void amdgpu_atombios_crtc_powergate_init(struct amdgpu_device *adev)
args.ucEnable = ATOM_INIT;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
@@ -228,7 +228,7 @@ void amdgpu_atombios_crtc_set_dtd_timing(struct drm_crtc *crtc,
args.susModeMiscInfo.usAccess = cpu_to_le16(misc);
args.ucCRTC = amdgpu_crtc->crtc_id;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
union atom_enable_ss {
@@ -293,7 +293,7 @@ static void amdgpu_atombios_crtc_program_ss(struct amdgpu_device *adev,
args.v3.usSpreadSpectrumStep = cpu_to_le16(ss->step);
args.v3.ucEnable = enable;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
union adjust_pixel_clock {
@@ -395,7 +395,7 @@ static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
ADJUST_DISPLAY_CONFIG_SS_ENABLE;
amdgpu_atom_execute_table(adev->mode_info.atom_context,
- index, (uint32_t *)&args);
+ index, (uint32_t *)&args, sizeof(args));
adjusted_clock = le16_to_cpu(args.v1.usPixelClock) * 10;
break;
case 3:
@@ -428,7 +428,7 @@ static u32 amdgpu_atombios_crtc_adjust_pll(struct drm_crtc *crtc,
args.v3.sInput.ucExtTransmitterID = 0;
amdgpu_atom_execute_table(adev->mode_info.atom_context,
- index, (uint32_t *)&args);
+ index, (uint32_t *)&args, sizeof(args));
adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10;
if (args.v3.sOutput.ucRefDiv) {
amdgpu_crtc->pll_flags |= AMDGPU_PLL_USE_FRAC_FB_DIV;
@@ -514,7 +514,7 @@ void amdgpu_atombios_crtc_set_disp_eng_pll(struct amdgpu_device *adev,
DRM_ERROR("Unknown table version %d %d\n", frev, crev);
return;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
union set_dce_clock {
@@ -544,7 +544,7 @@ u32 amdgpu_atombios_crtc_set_dce_clock(struct amdgpu_device *adev,
args.v2_1.asParam.ulDCEClkFreq = cpu_to_le32(freq); /* 10kHz units */
args.v2_1.asParam.ucDCEClkType = clk_type;
args.v2_1.asParam.ucDCEClkSrc = clk_src;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
ret_freq = le32_to_cpu(args.v2_1.asParam.ulDCEClkFreq) * 10;
break;
default:
@@ -740,7 +740,7 @@ void amdgpu_atombios_crtc_program_pll(struct drm_crtc *crtc,
return;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
int amdgpu_atombios_crtc_prepare_pll(struct drm_crtc *crtc,
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
index 87c41e0e9b7c..622634c08c7b 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_dp.c
@@ -83,7 +83,7 @@ static int amdgpu_atombios_dp_process_aux_ch(struct amdgpu_i2c_chan *chan,
args.v2.ucDelay = delay / 10;
args.v2.ucHPD_ID = chan->rec.hpd;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
*ack = args.v2.ucReplyStatus;
@@ -301,7 +301,7 @@ static u8 amdgpu_atombios_dp_encoder_service(struct amdgpu_device *adev,
args.ucLaneNum = lane_num;
args.ucStatus = 0;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
return args.ucStatus;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
index 7672abe6c140..25feab188dfe 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c
@@ -335,7 +335,7 @@ amdgpu_atombios_encoder_setup_dac(struct drm_encoder *encoder, int action)
args.ucDacStandard = ATOM_DAC1_PS2;
args.usPixelClock = cpu_to_le16(amdgpu_encoder->pixel_clock / 10);
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
@@ -432,7 +432,7 @@ amdgpu_atombios_encoder_setup_dvo(struct drm_encoder *encoder, int action)
break;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
int amdgpu_atombios_encoder_get_encoder_mode(struct drm_encoder *encoder)
@@ -732,7 +732,7 @@ amdgpu_atombios_encoder_setup_dig_encoder(struct drm_encoder *encoder,
break;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
@@ -1136,7 +1136,7 @@ amdgpu_atombios_encoder_setup_dig_transmitter(struct drm_encoder *encoder, int a
break;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
bool
@@ -1164,7 +1164,7 @@ amdgpu_atombios_encoder_set_edp_panel_power(struct drm_connector *connector,
args.v1.ucAction = action;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
/* wait for the panel to power up */
if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) {
@@ -1288,7 +1288,7 @@ amdgpu_atombios_encoder_setup_external_encoder(struct drm_encoder *encoder,
DRM_ERROR("Unknown table version: %d, %d\n", frev, crev);
return;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
static void
@@ -1633,7 +1633,7 @@ amdgpu_atombios_encoder_set_crtc_source(struct drm_encoder *encoder)
return;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
/* This only needs to be called once at startup */
@@ -1706,7 +1706,7 @@ amdgpu_atombios_encoder_dac_load_detect(struct drm_encoder *encoder,
args.sDacload.ucMisc = DAC_LOAD_MISC_YPrPb;
}
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
return true;
} else
diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
index af0335535f82..a6501114322f 100644
--- a/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
+++ b/drivers/gpu/drm/amd/amdgpu/atombios_i2c.c
@@ -86,7 +86,7 @@ static int amdgpu_atombios_i2c_process_i2c_ch(struct amdgpu_i2c_chan *chan,
args.ucSlaveAddr = slave_addr << 1;
args.ucLineNumber = chan->rec.i2c_id;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
/* error */
if (args.ucStatus != HW_ASSISTED_I2C_STATUS_SUCCESS) {
@@ -172,5 +172,5 @@ void amdgpu_atombios_i2c_channel_trans(struct amdgpu_device *adev, u8 slave_addr
args.ucSlaveAddr = slave_addr;
args.ucLineNumber = line_number;
- amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args);
+ amdgpu_atom_execute_table(adev->mode_info.atom_context, index, (uint32_t *)&args, sizeof(args));
}
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
index 567a904804bc..9c85ca6358c1 100644
--- a/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
@@ -21,8 +21,7 @@
*
*/
-static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_1[] = {
0x00000000, // DB_RENDER_CONTROL
0x00000000, // DB_COUNT_CONTROL
0x00000000, // DB_DEPTH_VIEW
@@ -236,8 +235,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_1[] =
0x00000000, // PA_SC_VPORT_ZMIN_15
0x3f800000, // PA_SC_VPORT_ZMAX_15
};
-static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_2[] = {
0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
0x00000000, // PA_SC_TILE_STEERING_OVERRIDE
0x00000000, // CP_PERFMON_CNTX_CNTL
@@ -521,15 +519,13 @@ static const unsigned int gfx9_SECT_CONTEXT_def_2[] =
0x00000000, // CB_MRT6_EPITCH
0x00000000, // CB_MRT7_EPITCH
};
-static const unsigned int gfx9_SECT_CONTEXT_def_3[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_3[] = {
0x00000000, // PA_CL_POINT_X_RAD
0x00000000, // PA_CL_POINT_Y_RAD
0x00000000, // PA_CL_POINT_SIZE
0x00000000, // PA_CL_POINT_CULL_RAD
};
-static const unsigned int gfx9_SECT_CONTEXT_def_4[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_4[] = {
0x00000000, // DB_DEPTH_CONTROL
0x00000000, // DB_EQAA
0x00000000, // CB_COLOR_CONTROL
@@ -688,17 +684,14 @@ static const unsigned int gfx9_SECT_CONTEXT_def_4[] =
0x00000000, // VGT_GS_OUT_PRIM_TYPE
0x00000000, // IA_ENHANCE
};
-static const unsigned int gfx9_SECT_CONTEXT_def_5[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_5[] = {
0x00000000, // WD_ENHANCE
0x00000000, // VGT_PRIMITIVEID_EN
};
-static const unsigned int gfx9_SECT_CONTEXT_def_6[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_6[] = {
0x00000000, // VGT_PRIMITIVEID_RESET
};
-static const unsigned int gfx9_SECT_CONTEXT_def_7[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_7[] = {
0x00000000, // VGT_GS_MAX_PRIMS_PER_SUBGROUP
0x00000000, // VGT_DRAW_PAYLOAD_CNTL
0, // HOLE
@@ -766,8 +759,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_7[] =
0x00000000, // VGT_STRMOUT_CONFIG
0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
};
-static const unsigned int gfx9_SECT_CONTEXT_def_8[] =
-{
+static const unsigned int gfx9_SECT_CONTEXT_def_8[] = {
0x00000000, // PA_SC_CENTROID_PRIORITY_0
0x00000000, // PA_SC_CENTROID_PRIORITY_1
0x00001000, // PA_SC_LINE_CNTL
@@ -924,8 +916,7 @@ static const unsigned int gfx9_SECT_CONTEXT_def_8[] =
0x00000000, // CB_COLOR7_DCC_BASE
0x00000000, // CB_COLOR7_DCC_BASE_EXT
};
-static const struct cs_extent_def gfx9_SECT_CONTEXT_defs[] =
-{
+static const struct cs_extent_def gfx9_SECT_CONTEXT_defs[] = {
{gfx9_SECT_CONTEXT_def_1, 0x0000a000, 212 },
{gfx9_SECT_CONTEXT_def_2, 0x0000a0d6, 282 },
{gfx9_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
diff --git a/drivers/gpu/drm/amd/amdgpu/clearstate_si.h b/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
index 66e39cdb5cb0..5fd96ddd7f0f 100644
--- a/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
+++ b/drivers/gpu/drm/amd/amdgpu/clearstate_si.h
@@ -21,8 +21,7 @@
*
*/
-static const u32 si_SECT_CONTEXT_def_1[] =
-{
+static const u32 si_SECT_CONTEXT_def_1[] = {
0x00000000, // DB_RENDER_CONTROL
0x00000000, // DB_COUNT_CONTROL
0x00000000, // DB_DEPTH_VIEW
@@ -236,8 +235,7 @@ static const u32 si_SECT_CONTEXT_def_1[] =
0x00000000, // PA_SC_VPORT_ZMIN_15
0x3f800000, // PA_SC_VPORT_ZMAX_15
};
-static const u32 si_SECT_CONTEXT_def_2[] =
-{
+static const u32 si_SECT_CONTEXT_def_2[] = {
0x00000000, // CP_PERFMON_CNTX_CNTL
0x00000000, // CP_RINGID
0x00000000, // CP_VMID
@@ -511,8 +509,7 @@ static const u32 si_SECT_CONTEXT_def_2[] =
0x00000000, // CB_BLEND6_CONTROL
0x00000000, // CB_BLEND7_CONTROL
};
-static const u32 si_SECT_CONTEXT_def_3[] =
-{
+static const u32 si_SECT_CONTEXT_def_3[] = {
0x00000000, // PA_CL_POINT_X_RAD
0x00000000, // PA_CL_POINT_Y_RAD
0x00000000, // PA_CL_POINT_SIZE
@@ -520,8 +517,7 @@ static const u32 si_SECT_CONTEXT_def_3[] =
0x00000000, // VGT_DMA_BASE_HI
0x00000000, // VGT_DMA_BASE
};
-static const u32 si_SECT_CONTEXT_def_4[] =
-{
+static const u32 si_SECT_CONTEXT_def_4[] = {
0x00000000, // DB_DEPTH_CONTROL
0x00000000, // DB_EQAA
0x00000000, // CB_COLOR_CONTROL
@@ -680,16 +676,13 @@ static const u32 si_SECT_CONTEXT_def_4[] =
0x00000000, // VGT_GS_OUT_PRIM_TYPE
0x00000000, // IA_ENHANCE
};
-static const u32 si_SECT_CONTEXT_def_5[] =
-{
+static const u32 si_SECT_CONTEXT_def_5[] = {
0x00000000, // VGT_PRIMITIVEID_EN
};
-static const u32 si_SECT_CONTEXT_def_6[] =
-{
+static const u32 si_SECT_CONTEXT_def_6[] = {
0x00000000, // VGT_PRIMITIVEID_RESET
};
-static const u32 si_SECT_CONTEXT_def_7[] =
-{
+static const u32 si_SECT_CONTEXT_def_7[] = {
0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
0, // HOLE
0, // HOLE
@@ -924,8 +917,7 @@ static const u32 si_SECT_CONTEXT_def_7[] =
0x00000000, // CB_COLOR7_CLEAR_WORD0
0x00000000, // CB_COLOR7_CLEAR_WORD1
};
-static const struct cs_extent_def si_SECT_CONTEXT_defs[] =
-{
+static const struct cs_extent_def si_SECT_CONTEXT_defs[] = {
{si_SECT_CONTEXT_def_1, 0x0000a000, 212 },
{si_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
{si_SECT_CONTEXT_def_3, 0x0000a1f5, 6 },
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 587ee632a3b8..221af054d874 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -52,6 +52,7 @@
static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev);
static void dce_v10_0_set_irq_funcs(struct amdgpu_device *adev);
+static void dce_v10_0_hpd_int_ack(struct amdgpu_device *adev, int hpd);
static const u32 crtc_offsets[] = {
CRTC0_REGISTER_OFFSET,
@@ -364,6 +365,7 @@ static void dce_v10_0_hpd_init(struct amdgpu_device *adev)
AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+ dce_v10_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
dce_v10_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
amdgpu_irq_get(adev, &adev->hpd_irq,
amdgpu_connector->hpd.hpd);
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index f22ec27365bd..69e8b0db6cf7 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -52,6 +52,7 @@
static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev);
static void dce_v11_0_set_irq_funcs(struct amdgpu_device *adev);
+static void dce_v11_0_hpd_int_ack(struct amdgpu_device *adev, int hpd);
static const u32 crtc_offsets[] =
{
@@ -388,6 +389,7 @@ static void dce_v11_0_hpd_init(struct amdgpu_device *adev)
AMDGPU_HPD_DISCONNECT_INT_DELAY_IN_MS);
WREG32(mmDC_HPD_TOGGLE_FILT_CNTL + hpd_offsets[amdgpu_connector->hpd.hpd], tmp);
+ dce_v11_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
dce_v11_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
index 4dbe9b3259b5..60d40201fdd1 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
@@ -273,6 +273,21 @@ static void dce_v6_0_hpd_set_polarity(struct amdgpu_device *adev,
WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
}
+static void dce_v6_0_hpd_int_ack(struct amdgpu_device *adev,
+ int hpd)
+{
+ u32 tmp;
+
+ if (hpd >= adev->mode_info.num_hpd) {
+ DRM_DEBUG("invalid hdp %d\n", hpd);
+ return;
+ }
+
+ tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+ tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+ WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
/**
* dce_v6_0_hpd_init - hpd setup callback.
*
@@ -312,6 +327,7 @@ static void dce_v6_0_hpd_init(struct amdgpu_device *adev)
continue;
}
+ dce_v6_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
dce_v6_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
}
@@ -3089,7 +3105,7 @@ static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
{
- uint32_t disp_int, mask, tmp;
+ uint32_t disp_int, mask;
unsigned hpd;
if (entry->src_data[0] >= adev->mode_info.num_hpd) {
@@ -3102,9 +3118,7 @@ static int dce_v6_0_hpd_irq(struct amdgpu_device *adev,
mask = interrupt_status_offsets[hpd].hpd;
if (disp_int & mask) {
- tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
- tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
- WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+ dce_v6_0_hpd_int_ack(adev, hpd);
schedule_delayed_work(&adev->hotplug_work, 0);
DRM_DEBUG("IH: HPD%d\n", hpd + 1);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index 05bcce23385e..5a5fcc45e452 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -265,6 +265,21 @@ static void dce_v8_0_hpd_set_polarity(struct amdgpu_device *adev,
WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
}
+static void dce_v8_0_hpd_int_ack(struct amdgpu_device *adev,
+ int hpd)
+{
+ u32 tmp;
+
+ if (hpd >= adev->mode_info.num_hpd) {
+ DRM_DEBUG("invalid hdp %d\n", hpd);
+ return;
+ }
+
+ tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
+ tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
+ WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+}
+
/**
* dce_v8_0_hpd_init - hpd setup callback.
*
@@ -304,6 +319,7 @@ static void dce_v8_0_hpd_init(struct amdgpu_device *adev)
continue;
}
+ dce_v8_0_hpd_int_ack(adev, amdgpu_connector->hpd.hpd);
dce_v8_0_hpd_set_polarity(adev, amdgpu_connector->hpd.hpd);
amdgpu_irq_get(adev, &adev->hpd_irq, amdgpu_connector->hpd.hpd);
}
@@ -3177,7 +3193,7 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
{
- uint32_t disp_int, mask, tmp;
+ uint32_t disp_int, mask;
unsigned hpd;
if (entry->src_data[0] >= adev->mode_info.num_hpd) {
@@ -3190,9 +3206,7 @@ static int dce_v8_0_hpd_irq(struct amdgpu_device *adev,
mask = interrupt_status_offsets[hpd].hpd;
if (disp_int & mask) {
- tmp = RREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd]);
- tmp |= DC_HPD1_INT_CONTROL__DC_HPD1_INT_ACK_MASK;
- WREG32(mmDC_HPD1_INT_CONTROL + hpd_offsets[hpd], tmp);
+ dce_v8_0_hpd_int_ack(adev, hpd);
schedule_delayed_work(&adev->hotplug_work, 0);
DRM_DEBUG("IH: HPD%d\n", hpd + 1);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index dcdecb18b230..b02d63328f1c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -7947,7 +7947,7 @@ static void gfx_v10_0_update_spm_vmid_internal(struct amdgpu_device *adev,
WREG32_SOC15_NO_KIQ(GC, 0, mmRLC_SPM_MC_CNTL, data);
}
-static void gfx_v10_0_update_spm_vmid(struct amdgpu_device *adev, unsigned int vmid)
+static void gfx_v10_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int vmid)
{
amdgpu_gfx_off_ctrl(adev, false);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index 4f3bfdc75b37..2fb1342d5bd9 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -727,7 +727,7 @@ static int gfx_v11_0_rlc_init(struct amdgpu_device *adev)
/* init spm vmid with 0xf */
if (adev->gfx.rlc.funcs->update_spm_vmid)
- adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf);
+ adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
return 0;
}
@@ -5027,7 +5027,7 @@ static int gfx_v11_0_update_gfx_clock_gating(struct amdgpu_device *adev,
return 0;
}
-static void gfx_v11_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
+static void gfx_v11_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
{
u32 data;
@@ -5041,6 +5041,14 @@ static void gfx_v11_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
WREG32_SOC15_NO_KIQ(GC, 0, regRLC_SPM_MC_CNTL, data);
amdgpu_gfx_off_ctrl(adev, true);
+
+ if (ring
+ && amdgpu_sriov_is_pp_one_vf(adev)
+ && ((ring->funcs->type == AMDGPU_RING_TYPE_GFX)
+ || (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE))) {
+ uint32_t reg = SOC15_REG_OFFSET(GC, 0, regRLC_SPM_MC_CNTL);
+ amdgpu_ring_emit_wreg(ring, reg, data);
+ }
}
static const struct amdgpu_rlc_funcs gfx_v11_0_rlc_funcs = {
@@ -6104,7 +6112,8 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = {
.get_rptr = gfx_v11_0_ring_get_rptr_gfx,
.get_wptr = gfx_v11_0_ring_get_wptr_gfx,
.set_wptr = gfx_v11_0_ring_set_wptr_gfx,
- .emit_frame_size = /* totally 242 maximum if 16 IBs */
+ .emit_frame_size = /* totally 247 maximum if 16 IBs */
+ 5 + /* update_spm_vmid */
5 + /* COND_EXEC */
9 + /* SET_Q_PREEMPTION_MODE */
7 + /* PIPELINE_SYNC */
@@ -6154,6 +6163,7 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_compute = {
.get_wptr = gfx_v11_0_ring_get_wptr_compute,
.set_wptr = gfx_v11_0_ring_set_wptr_compute,
.emit_frame_size =
+ 5 + /* update_spm_vmid */
20 + /* gfx_v11_0_ring_emit_gds_switch */
7 + /* gfx_v11_0_ring_emit_hdp_flush */
5 + /* hdp invalidate */
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
index 26d6286d86c9..9e7ce1e6bc06 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0_3.c
@@ -69,7 +69,7 @@ static int gfx_v11_0_3_rlc_gc_fed_irq(struct amdgpu_device *adev,
amdgpu_ras_interrupt_dispatch(adev, &ih_data);
} else {
if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
- adev->virt.ops->ras_poison_handler(adev);
+ adev->virt.ops->ras_poison_handler(adev, ras_if->block);
else
dev_warn(adev->dev,
"No ras_poison_handler interface in SRIOV for %s!\n", ras_if->name);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index c2faf6b4c2fc..86a4865b1ae5 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -3274,7 +3274,7 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
/* init spm vmid with 0xf */
if (adev->gfx.rlc.funcs->update_spm_vmid)
- adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf);
+ adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
return 0;
}
@@ -3500,7 +3500,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
return 0;
}
-static void gfx_v7_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
+static void gfx_v7_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
{
u32 data;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 1943beb135c4..ea174b76ee70 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -1288,7 +1288,7 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
/* init spm vmid with 0xf */
if (adev->gfx.rlc.funcs->update_spm_vmid)
- adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf);
+ adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
return 0;
}
@@ -5579,7 +5579,7 @@ static void gfx_v8_0_unset_safe_mode(struct amdgpu_device *adev, int xcc_id)
}
}
-static void gfx_v8_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
+static void gfx_v8_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned vmid)
{
u32 data;
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 69c500910746..169d45268ef6 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -3034,6 +3034,14 @@ static int gfx_v9_0_cp_gfx_start(struct amdgpu_device *adev)
gfx_v9_0_cp_gfx_enable(adev, true);
+ /* Now only limit the quirk on the APU gfx9 series and already
+ * confirmed that the APU gfx10/gfx11 needn't such update.
+ */
+ if (adev->flags & AMD_IS_APU &&
+ adev->in_s3 && !adev->suspend_complete) {
+ DRM_INFO(" Will skip the CSB packet resubmit\n");
+ return 0;
+ }
r = amdgpu_ring_alloc(ring, gfx_v9_0_get_csb_size(adev) + 4 + 3);
if (r) {
DRM_ERROR("amdgpu: cp failed to lock ring (%d).\n", r);
@@ -4894,7 +4902,7 @@ static void gfx_v9_0_update_spm_vmid_internal(struct amdgpu_device *adev,
WREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL, data);
}
-static void gfx_v9_0_update_spm_vmid(struct amdgpu_device *adev, unsigned int vmid)
+static void gfx_v9_0_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring, unsigned int vmid)
{
amdgpu_gfx_off_ctrl(adev, false);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
index bc8416afb62c..f53b379d8971 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
@@ -970,8 +970,9 @@ static void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_DSM_INDEX, 255);
}
-static const struct soc15_reg_entry gfx_v9_4_ea_err_status_regs =
- { SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32 };
+static const struct soc15_reg_entry gfx_v9_4_ea_err_status_regs = {
+ SOC15_REG_ENTRY(GC, 0, mmGCEA_ERR_STATUS), 0, 1, 32
+};
static void gfx_v9_4_query_ras_error_status(struct amdgpu_device *adev)
{
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
index 131cddbdda0d..aace4594a603 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -38,6 +38,7 @@
#include "gfx_v9_4_3.h"
#include "amdgpu_xcp.h"
+#include "amdgpu_aca.h"
MODULE_FIRMWARE("amdgpu/gc_9_4_3_mec.bin");
MODULE_FIRMWARE("amdgpu/gc_9_4_3_rlc.bin");
@@ -48,6 +49,10 @@ MODULE_FIRMWARE("amdgpu/gc_9_4_3_rlc.bin");
#define GOLDEN_GB_ADDR_CONFIG 0x2a114042
#define CP_HQD_PERSISTENT_STATE_DEFAULT 0xbe05301
+#define mmSMNAID_XCD0_MCA_SMU 0x36430400 /* SMN AID XCD0 */
+#define mmSMNAID_XCD1_MCA_SMU 0x38430400 /* SMN AID XCD1 */
+#define mmSMNXCD_XCD0_MCA_SMU 0x40430400 /* SMN XCD XCD0 */
+
struct amdgpu_gfx_ras gfx_v9_4_3_ras;
static void gfx_v9_4_3_set_ring_funcs(struct amdgpu_device *adev);
@@ -675,6 +680,66 @@ static const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs = {
.ih_node_to_logical_xcc = &gfx_v9_4_3_ih_to_xcc_inst,
};
+static int gfx_v9_4_3_aca_bank_generate_report(struct aca_handle *handle,
+ struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data)
+{
+ u64 status, misc0;
+ u32 instlo;
+ int ret;
+
+ status = bank->regs[ACA_REG_IDX_STATUS];
+ if ((type == ACA_ERROR_TYPE_UE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) ||
+ (type == ACA_ERROR_TYPE_CE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) {
+
+ ret = aca_bank_info_decode(bank, &report->info);
+ if (ret)
+ return ret;
+
+ /* NOTE: overwrite info.die_id with xcd id for gfx */
+ instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+ instlo &= GENMASK(31, 1);
+ report->info.die_id = instlo == mmSMNAID_XCD0_MCA_SMU ? 0 : 1;
+
+ misc0 = bank->regs[ACA_REG_IDX_MISC0];
+ report->count[type] = ACA_REG__MISC0__ERRCNT(misc0);
+ }
+
+ return 0;
+}
+
+static bool gfx_v9_4_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, void *data)
+{
+ u32 instlo;
+
+ instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+ instlo &= GENMASK(31, 1);
+ switch (instlo) {
+ case mmSMNAID_XCD0_MCA_SMU:
+ case mmSMNAID_XCD1_MCA_SMU:
+ case mmSMNXCD_XCD0_MCA_SMU:
+ return true;
+ default:
+ break;
+ }
+
+ return false;
+}
+
+static const struct aca_bank_ops gfx_v9_4_3_aca_bank_ops = {
+ .aca_bank_generate_report = gfx_v9_4_3_aca_bank_generate_report,
+ .aca_bank_is_valid = gfx_v9_4_3_aca_bank_is_valid,
+};
+
+static const struct aca_info gfx_v9_4_3_aca_info = {
+ .hwip = ACA_HWIP_TYPE_SMU,
+ .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+ .bank_ops = &gfx_v9_4_3_aca_bank_ops,
+};
+
static int gfx_v9_4_3_gpu_early_init(struct amdgpu_device *adev)
{
u32 gb_addr_config;
@@ -1109,7 +1174,7 @@ static int gfx_v9_4_3_rlc_init(struct amdgpu_device *adev)
{
/* init spm vmid with 0xf */
if (adev->gfx.rlc.funcs->update_spm_vmid)
- adev->gfx.rlc.funcs->update_spm_vmid(adev, 0xf);
+ adev->gfx.rlc.funcs->update_spm_vmid(adev, NULL, 0xf);
return 0;
}
@@ -1320,7 +1385,7 @@ static int gfx_v9_4_3_rlc_resume(struct amdgpu_device *adev)
return 0;
}
-static void gfx_v9_4_3_update_spm_vmid(struct amdgpu_device *adev,
+static void gfx_v9_4_3_update_spm_vmid(struct amdgpu_device *adev, struct amdgpu_ring *ring,
unsigned vmid)
{
u32 reg, data;
@@ -4242,9 +4307,32 @@ struct amdgpu_ras_block_hw_ops gfx_v9_4_3_ras_ops = {
.reset_ras_error_count = &gfx_v9_4_3_reset_ras_error_count,
};
+static int gfx_v9_4_3_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+ int r;
+
+ r = amdgpu_ras_block_late_init(adev, ras_block);
+ if (r)
+ return r;
+
+ r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__GFX,
+ &gfx_v9_4_3_aca_info,
+ NULL);
+ if (r)
+ goto late_fini;
+
+ return 0;
+
+late_fini:
+ amdgpu_ras_block_late_fini(adev, ras_block);
+
+ return r;
+}
+
struct amdgpu_gfx_ras gfx_v9_4_3_ras = {
.ras_block = {
.hw_ops = &gfx_v9_4_3_ras_ops,
+ .ras_late_init = &gfx_v9_4_3_ras_late_init,
},
.enable_watchdog_timer = &gfx_v9_4_3_enable_watchdog_timer,
};
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
index 6c5185608854..db89d13bd80d 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@@ -262,16 +262,17 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
/* flush hdp cache */
adev->hdp.funcs->flush_hdp(adev, NULL);
- /* For SRIOV run time, driver shouldn't access the register through MMIO
- * Directly use kiq to do the vm invalidation instead
+ /* This is necessary for SRIOV as well as for GFXOFF to function
+ * properly under bare metal
*/
if (adev->gfx.kiq[0].ring.sched.ready && !adev->enable_mes &&
(amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
- amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, inv_req,
- 1 << vmid, GET_INST(GC, 0));
+ amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+ 1 << vmid, GET_INST(GC, 0));
return;
}
+ /* This path is needed before KIQ/MES/GFXOFF are set up */
hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? GC_HWIP : MMHUB_HWIP;
spin_lock(&adev->gmc.invalidate_lock);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
index c9c653cfc765..998daa702b44 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c
@@ -223,16 +223,17 @@ static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
/* flush hdp cache */
adev->hdp.funcs->flush_hdp(adev, NULL);
- /* For SRIOV run time, driver shouldn't access the register through MMIO
- * Directly use kiq to do the vm invalidation instead
+ /* This is necessary for SRIOV as well as for GFXOFF to function
+ * properly under bare metal
*/
if ((adev->gfx.kiq[0].ring.sched.ready || adev->mes.ring.sched.ready) &&
(amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
- amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, inv_req,
- 1 << vmid, GET_INST(GC, 0));
+ amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+ 1 << vmid, GET_INST(GC, 0));
return;
}
+ /* This path is needed before KIQ/MES/GFXOFF are set up */
hub_ip = (vmhub == AMDGPU_GFXHUB(0)) ? GC_HWIP : MMHUB_HWIP;
spin_lock(&adev->gmc.invalidate_lock);
@@ -570,6 +571,7 @@ static void gmc_v11_0_set_mmhub_funcs(struct amdgpu_device *adev)
adev->mmhub.funcs = &mmhub_v3_0_2_funcs;
break;
case IP_VERSION(3, 3, 0):
+ case IP_VERSION(3, 3, 1):
adev->mmhub.funcs = &mmhub_v3_3_funcs;
break;
default:
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
index 59d9215e5556..23b478639921 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
@@ -435,9 +435,10 @@ static void gmc_v6_0_set_prt(struct amdgpu_device *adev, bool enable)
WREG32(mmVM_PRT_CNTL, tmp);
if (enable) {
- uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+ uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+ AMDGPU_GPU_PAGE_SHIFT;
uint32_t high = adev->vm_manager.max_pfn -
- (AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+ (AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
index 45a2f8e031a2..3da7b6a2b00d 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
@@ -563,9 +563,10 @@ static void gmc_v7_0_set_prt(struct amdgpu_device *adev, bool enable)
WREG32(mmVM_PRT_CNTL, tmp);
if (enable) {
- uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+ uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+ AMDGPU_GPU_PAGE_SHIFT;
uint32_t high = adev->vm_manager.max_pfn -
- (AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+ (AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
index 4422b27a3cc2..969a9e867170 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
@@ -777,9 +777,10 @@ static void gmc_v8_0_set_prt(struct amdgpu_device *adev, bool enable)
WREG32(mmVM_PRT_CNTL, tmp);
if (enable) {
- uint32_t low = AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT;
+ uint32_t low = AMDGPU_VA_RESERVED_BOTTOM >>
+ AMDGPU_GPU_PAGE_SHIFT;
uint32_t high = adev->vm_manager.max_pfn -
- (AMDGPU_VA_RESERVED_SIZE >> AMDGPU_GPU_PAGE_SHIFT);
+ (AMDGPU_VA_RESERVED_TOP >> AMDGPU_GPU_PAGE_SHIFT);
WREG32(mmVM_PRT_APERTURE0_LOW_ADDR, low);
WREG32(mmVM_PRT_APERTURE1_LOW_ADDR, low);
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 40a00ea0009f..d442ae85162d 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -829,23 +829,25 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
- /* This is necessary for a HW workaround under SRIOV as well
- * as GFXOFF under bare metal
- */
if (vmhub >= AMDGPU_MMHUB0(0))
inst = GET_INST(GC, 0);
else
inst = vmhub;
+
+ /* This is necessary for SRIOV as well as for GFXOFF to function
+ * properly under bare metal
+ */
if (adev->gfx.kiq[inst].ring.sched.ready &&
(amdgpu_sriov_runtime(adev) || !amdgpu_sriov_vf(adev))) {
uint32_t req = hub->vm_inv_eng0_req + hub->eng_distance * eng;
uint32_t ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng;
- amdgpu_virt_kiq_reg_write_reg_wait(adev, req, ack, inv_req,
- 1 << vmid, inst);
+ amdgpu_gmc_fw_reg_write_reg_wait(adev, req, ack, inv_req,
+ 1 << vmid, inst);
return;
}
+ /* This path is needed before KIQ/MES/GFXOFF are set up */
spin_lock(&adev->gmc.invalidate_lock);
/*
@@ -1947,14 +1949,6 @@ static int gmc_v9_0_init_mem_ranges(struct amdgpu_device *adev)
static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev)
{
- static const u32 regBIF_BIOS_SCRATCH_4 = 0x50;
- u32 vram_info;
-
- /* Only for dGPU, vendor informaton is reliable */
- if (!amdgpu_sriov_vf(adev) && !(adev->flags & AMD_IS_APU)) {
- vram_info = RREG32(regBIF_BIOS_SCRATCH_4);
- adev->gmc.vram_vendor = vram_info & 0xF;
- }
adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM;
adev->gmc.vram_width = 128 * 64;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
new file mode 100644
index 000000000000..8d7d0813e331
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
@@ -0,0 +1,142 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include "amdgpu.h"
+#include "amdgpu_atombios.h"
+#include "hdp_v7_0.h"
+
+#include "hdp/hdp_7_0_0_offset.h"
+#include "hdp/hdp_7_0_0_sh_mask.h"
+#include <uapi/linux/kfd_ioctl.h>
+
+static void hdp_v7_0_flush_hdp(struct amdgpu_device *adev,
+ struct amdgpu_ring *ring)
+{
+ if (!ring || !ring->funcs->emit_wreg)
+ WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+ else
+ amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+}
+
+static void hdp_v7_0_update_clock_gating(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t hdp_clk_cntl, hdp_clk_cntl1;
+ uint32_t hdp_mem_pwr_cntl;
+
+ if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
+ AMD_CG_SUPPORT_HDP_DS |
+ AMD_CG_SUPPORT_HDP_SD)))
+ return;
+
+ hdp_clk_cntl = hdp_clk_cntl1 = RREG32_SOC15(HDP, 0,regHDP_CLK_CNTL);
+ hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+
+ /* Before doing clock/power mode switch,
+ * forced on IPH & RC clock */
+ hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+ RC_MEM_CLK_SOFT_OVERRIDE, 1);
+ WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+
+ /* disable clock and power gating before any changing */
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_CTRL_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_LS_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_DS_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_SD_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_CTRL_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_LS_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_DS_EN, 0);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_SD_EN, 0);
+ WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+
+ /* Already disabled above. The actions below are for "enabled" only */
+ if (enable) {
+ /* only one clock gating mode (LS/DS/SD) can be enabled */
+ if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_SD_EN, 1);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_SD_EN, 1);
+ } else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_LS_EN, 1);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_LS_EN, 1);
+ } else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_DS_EN, 1);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
+ HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_DS_EN, 1);
+ }
+
+ /* confirmed that IPH_MEM_POWER_CTRL_EN and RC_MEM_POWER_CTRL_EN have to
+ * be set for SRAM LS/DS/SD */
+ if (adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS | AMD_CG_SUPPORT_HDP_DS |
+ AMD_CG_SUPPORT_HDP_SD)) {
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ ATOMIC_MEM_POWER_CTRL_EN, 1);
+ hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
+ RC_MEM_POWER_CTRL_EN, 1);
+ WREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
+ }
+ }
+
+ /* disable IPH & RC clock override after clock/power mode changing */
+ hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
+ RC_MEM_CLK_SOFT_OVERRIDE, 0);
+ WREG32_SOC15(HDP, 0, regHDP_CLK_CNTL, hdp_clk_cntl);
+}
+
+static void hdp_v7_0_get_clockgating_state(struct amdgpu_device *adev,
+ u64 *flags)
+{
+ uint32_t tmp;
+
+ /* AMD_CG_SUPPORT_HDP_LS/DS/SD */
+ tmp = RREG32_SOC15(HDP, 0, regHDP_MEM_POWER_CTRL);
+ if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_LS_EN_MASK)
+ *flags |= AMD_CG_SUPPORT_HDP_LS;
+ else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_DS_EN_MASK)
+ *flags |= AMD_CG_SUPPORT_HDP_DS;
+ else if (tmp & HDP_MEM_POWER_CTRL__ATOMIC_MEM_POWER_SD_EN_MASK)
+ *flags |= AMD_CG_SUPPORT_HDP_SD;
+}
+
+const struct amdgpu_hdp_funcs hdp_v7_0_funcs = {
+ .flush_hdp = hdp_v7_0_flush_hdp,
+ .update_clock_gating = hdp_v7_0_update_clock_gating,
+ .get_clock_gating_state = hdp_v7_0_get_clockgating_state,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h
new file mode 100644
index 000000000000..25b69201402d
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __HDP_V7_0_H__
+#define __HDP_V7_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_hdp_funcs hdp_v7_0_funcs;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
new file mode 100644
index 000000000000..16fe428c0722
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.c
@@ -0,0 +1,767 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/pci.h>
+
+#include "amdgpu.h"
+#include "amdgpu_ih.h"
+
+#include "oss/osssys_7_0_0_offset.h"
+#include "oss/osssys_7_0_0_sh_mask.h"
+
+#include "soc15_common.h"
+#include "ih_v7_0.h"
+
+#define MAX_REARM_RETRY 10
+
+static void ih_v7_0_set_interrupt_funcs(struct amdgpu_device *adev);
+
+/**
+ * ih_v7_0_init_register_offset - Initialize register offset for ih rings
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Initialize register offset ih rings (IH_V7_0).
+ */
+static void ih_v7_0_init_register_offset(struct amdgpu_device *adev)
+{
+ struct amdgpu_ih_regs *ih_regs;
+
+ /* ih ring 2 is removed
+ * ih ring and ih ring 1 are available */
+ if (adev->irq.ih.ring_size) {
+ ih_regs = &adev->irq.ih.ih_regs;
+ ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE);
+ ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI);
+ ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL);
+ ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR);
+ ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR);
+ ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR);
+ ih_regs->ih_rb_wptr_addr_lo = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_LO);
+ ih_regs->ih_rb_wptr_addr_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_ADDR_HI);
+ ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL;
+ }
+
+ if (adev->irq.ih1.ring_size) {
+ ih_regs = &adev->irq.ih1.ih_regs;
+ ih_regs->ih_rb_base = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_RING1);
+ ih_regs->ih_rb_base_hi = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_BASE_HI_RING1);
+ ih_regs->ih_rb_cntl = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_CNTL_RING1);
+ ih_regs->ih_rb_wptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_WPTR_RING1);
+ ih_regs->ih_rb_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_RB_RPTR_RING1);
+ ih_regs->ih_doorbell_rptr = SOC15_REG_OFFSET(OSSSYS, 0, regIH_DOORBELL_RPTR_RING1);
+ ih_regs->psp_reg_id = PSP_REG_IH_RB_CNTL_RING1;
+ }
+}
+
+/**
+ * force_update_wptr_for_self_int - Force update the wptr for self interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @threshold: threshold to trigger the wptr reporting
+ * @timeout: timeout to trigger the wptr reporting
+ * @enabled: Enable/disable timeout flush mechanism
+ *
+ * threshold input range: 0 ~ 15, default 0,
+ * real_threshold = 2^threshold
+ * timeout input range: 0 ~ 20, default 8,
+ * real_timeout = (2^timeout) * 1024 / (socclk_freq)
+ *
+ * Force update wptr for self interrupt ( >= SIENNA_CICHLID).
+ */
+static void
+force_update_wptr_for_self_int(struct amdgpu_device *adev,
+ u32 threshold, u32 timeout, bool enabled)
+{
+ u32 ih_cntl, ih_rb_cntl;
+
+ ih_cntl = RREG32_SOC15(OSSSYS, 0, regIH_CNTL2);
+ ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1);
+
+ ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+ SELF_IV_FORCE_WPTR_UPDATE_TIMEOUT, timeout);
+ ih_cntl = REG_SET_FIELD(ih_cntl, IH_CNTL2,
+ SELF_IV_FORCE_WPTR_UPDATE_ENABLE, enabled);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
+ RB_USED_INT_THRESHOLD, threshold);
+
+ if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+ if (psp_reg_program(&adev->psp, PSP_REG_IH_RB_CNTL_RING1, ih_rb_cntl))
+ return;
+ } else {
+ WREG32_SOC15(OSSSYS, 0, regIH_RB_CNTL_RING1, ih_rb_cntl);
+ }
+
+ WREG32_SOC15(OSSSYS, 0, regIH_CNTL2, ih_cntl);
+}
+
+/**
+ * ih_v7_0_toggle_ring_interrupts - toggle the interrupt ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointet
+ * @enable: true - enable the interrupts, false - disable the interrupts
+ *
+ * Toggle the interrupt ring buffer (IH_V7_0)
+ */
+static int ih_v7_0_toggle_ring_interrupts(struct amdgpu_device *adev,
+ struct amdgpu_ih_ring *ih,
+ bool enable)
+{
+ struct amdgpu_ih_regs *ih_regs;
+ uint32_t tmp;
+
+ ih_regs = &ih->ih_regs;
+
+ tmp = RREG32(ih_regs->ih_rb_cntl);
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_ENABLE, (enable ? 1 : 0));
+ /* enable_intr field is only valid in ring0 */
+ if (ih == &adev->irq.ih)
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, ENABLE_INTR, (enable ? 1 : 0));
+
+ if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+ if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp))
+ return -ETIMEDOUT;
+ } else {
+ WREG32(ih_regs->ih_rb_cntl, tmp);
+ }
+
+ if (enable) {
+ ih->enabled = true;
+ } else {
+ /* set rptr, wptr to 0 */
+ WREG32(ih_regs->ih_rb_rptr, 0);
+ WREG32(ih_regs->ih_rb_wptr, 0);
+ ih->enabled = false;
+ ih->rptr = 0;
+ }
+
+ return 0;
+}
+
+/**
+ * ih_v7_0_toggle_interrupts - Toggle all the available interrupt ring buffers
+ *
+ * @adev: amdgpu_device pointer
+ * @enable: enable or disable interrupt ring buffers
+ *
+ * Toggle all the available interrupt ring buffers (IH_V7_0).
+ */
+static int ih_v7_0_toggle_interrupts(struct amdgpu_device *adev, bool enable)
+{
+ struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+ int i;
+ int r;
+
+ for (i = 0; i < ARRAY_SIZE(ih); i++) {
+ if (ih[i]->ring_size) {
+ r = ih_v7_0_toggle_ring_interrupts(adev, ih[i], enable);
+ if (r)
+ return r;
+ }
+ }
+
+ return 0;
+}
+
+static uint32_t ih_v7_0_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
+{
+ int rb_bufsz = order_base_2(ih->ring_size / 4);
+
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+ MC_SPACE, ih->use_bus_addr ? 2 : 4);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+ WPTR_OVERFLOW_CLEAR, 1);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+ WPTR_OVERFLOW_ENABLE, 1);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
+ /* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
+ * value is written to memory
+ */
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
+ WPTR_WRITEBACK_ENABLE, 1);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
+ ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
+
+ return ih_rb_cntl;
+}
+
+static uint32_t ih_v7_0_doorbell_rptr(struct amdgpu_ih_ring *ih)
+{
+ u32 ih_doorbell_rtpr = 0;
+
+ if (ih->use_doorbell) {
+ ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+ IH_DOORBELL_RPTR, OFFSET,
+ ih->doorbell_index);
+ ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+ IH_DOORBELL_RPTR,
+ ENABLE, 1);
+ } else {
+ ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
+ IH_DOORBELL_RPTR,
+ ENABLE, 0);
+ }
+ return ih_doorbell_rtpr;
+}
+
+/**
+ * ih_v7_0_enable_ring - enable an ih ring buffer
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: amdgpu_ih_ring pointer
+ *
+ * Enable an ih ring buffer (IH_V7_0)
+ */
+static int ih_v7_0_enable_ring(struct amdgpu_device *adev,
+ struct amdgpu_ih_ring *ih)
+{
+ struct amdgpu_ih_regs *ih_regs;
+ uint32_t tmp;
+
+ ih_regs = &ih->ih_regs;
+
+ /* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
+ WREG32(ih_regs->ih_rb_base, ih->gpu_addr >> 8);
+ WREG32(ih_regs->ih_rb_base_hi, (ih->gpu_addr >> 40) & 0xff);
+
+ tmp = RREG32(ih_regs->ih_rb_cntl);
+ tmp = ih_v7_0_rb_cntl(ih, tmp);
+ if (ih == &adev->irq.ih)
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RPTR_REARM, !!adev->irq.msi_enabled);
+ if (ih == &adev->irq.ih1) {
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 0);
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, RB_FULL_DRAIN_ENABLE, 1);
+ }
+
+ if (amdgpu_sriov_vf(adev) && amdgpu_sriov_reg_indirect_ih(adev)) {
+ if (psp_reg_program(&adev->psp, ih_regs->psp_reg_id, tmp)) {
+ DRM_ERROR("PSP program IH_RB_CNTL failed!\n");
+ return -ETIMEDOUT;
+ }
+ } else {
+ WREG32(ih_regs->ih_rb_cntl, tmp);
+ }
+
+ if (ih == &adev->irq.ih) {
+ /* set the ih ring 0 writeback address whether it's enabled or not */
+ WREG32(ih_regs->ih_rb_wptr_addr_lo, lower_32_bits(ih->wptr_addr));
+ WREG32(ih_regs->ih_rb_wptr_addr_hi, upper_32_bits(ih->wptr_addr) & 0xFFFF);
+ }
+
+ /* set rptr, wptr to 0 */
+ WREG32(ih_regs->ih_rb_wptr, 0);
+ WREG32(ih_regs->ih_rb_rptr, 0);
+
+ WREG32(ih_regs->ih_doorbell_rptr, ih_v7_0_doorbell_rptr(ih));
+
+ return 0;
+}
+
+/**
+ * ih_v7_0_irq_init - init and enable the interrupt ring
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Allocate a ring buffer for the interrupt controller,
+ * enable the RLC, disable interrupts, enable the IH
+ * ring buffer and enable it.
+ * Called at device load and reume.
+ * Returns 0 for success, errors for failure.
+ */
+static int ih_v7_0_irq_init(struct amdgpu_device *adev)
+{
+ struct amdgpu_ih_ring *ih[] = {&adev->irq.ih, &adev->irq.ih1};
+ u32 ih_chicken;
+ u32 tmp;
+ int ret;
+ int i;
+
+ /* disable irqs */
+ ret = ih_v7_0_toggle_interrupts(adev, false);
+ if (ret)
+ return ret;
+
+ adev->nbio.funcs->ih_control(adev);
+
+ if (unlikely((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) ||
+ (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO))) {
+ if (ih[0]->use_bus_addr) {
+ ih_chicken = RREG32_SOC15(OSSSYS, 0, regIH_CHICKEN);
+ ih_chicken = REG_SET_FIELD(ih_chicken,
+ IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
+ WREG32_SOC15(OSSSYS, 0, regIH_CHICKEN, ih_chicken);
+ }
+ }
+
+ for (i = 0; i < ARRAY_SIZE(ih); i++) {
+ if (ih[i]->ring_size) {
+ ret = ih_v7_0_enable_ring(adev, ih[i]);
+ if (ret)
+ return ret;
+ }
+ }
+
+ /* update doorbell range for ih ring 0 */
+ adev->nbio.funcs->ih_doorbell_range(adev, ih[0]->use_doorbell,
+ ih[0]->doorbell_index);
+
+ tmp = RREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL);
+ tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
+ CLIENT18_IS_STORM_CLIENT, 1);
+ WREG32_SOC15(OSSSYS, 0, regIH_STORM_CLIENT_LIST_CNTL, tmp);
+
+ tmp = RREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL);
+ tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
+ WREG32_SOC15(OSSSYS, 0, regIH_INT_FLOOD_CNTL, tmp);
+
+ /* GC/MMHUB UTCL2 page fault interrupts are configured as
+ * MSI storm capable interrupts by deafult. The delay is
+ * used to avoid ISR being called too frequently
+ * when page fault happens on several continuous page
+ * and thus avoid MSI storm */
+ tmp = RREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL);
+ tmp = REG_SET_FIELD(tmp, IH_MSI_STORM_CTRL,
+ DELAY, 3);
+ WREG32_SOC15(OSSSYS, 0, regIH_MSI_STORM_CTRL, tmp);
+
+ pci_set_master(adev->pdev);
+
+ /* enable interrupts */
+ ret = ih_v7_0_toggle_interrupts(adev, true);
+ if (ret)
+ return ret;
+ /* enable wptr force update for self int */
+ force_update_wptr_for_self_int(adev, 0, 8, true);
+
+ if (adev->irq.ih_soft.ring_size)
+ adev->irq.ih_soft.enabled = true;
+
+ return 0;
+}
+
+/**
+ * ih_v7_0_irq_disable - disable interrupts
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Disable interrupts on the hw.
+ */
+static void ih_v7_0_irq_disable(struct amdgpu_device *adev)
+{
+ force_update_wptr_for_self_int(adev, 0, 8, false);
+ ih_v7_0_toggle_interrupts(adev, false);
+
+ /* Wait and acknowledge irq */
+ mdelay(1);
+}
+
+/**
+ * ih_v7_0_get_wptr() - get the IH ring buffer wptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to fetch wptr
+ *
+ * Get the IH ring buffer wptr from either the register
+ * or the writeback memory buffer. Also check for
+ * ring buffer overflow and deal with it.
+ * Returns the value of the wptr.
+ */
+static u32 ih_v7_0_get_wptr(struct amdgpu_device *adev,
+ struct amdgpu_ih_ring *ih)
+{
+ u32 wptr, tmp;
+ struct amdgpu_ih_regs *ih_regs;
+
+ wptr = le32_to_cpu(*ih->wptr_cpu);
+ ih_regs = &ih->ih_regs;
+
+ if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+ goto out;
+
+ wptr = RREG32_NO_KIQ(ih_regs->ih_rb_wptr);
+ if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
+ goto out;
+ wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
+
+ /* When a ring buffer overflow happen start parsing interrupt
+ * from the last not overwritten vector (wptr + 32). Hopefully
+ * this should allow us to catch up.
+ */
+ tmp = (wptr + 32) & ih->ptr_mask;
+ dev_warn(adev->dev, "IH ring buffer overflow "
+ "(0x%08X, 0x%08X, 0x%08X)\n",
+ wptr, ih->rptr, tmp);
+ ih->rptr = tmp;
+
+ tmp = RREG32_NO_KIQ(ih_regs->ih_rb_cntl);
+ tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
+ WREG32_NO_KIQ(ih_regs->ih_rb_cntl, tmp);
+out:
+ return (wptr & ih->ptr_mask);
+}
+
+/**
+ * ih_v7_0_irq_rearm - rearm IRQ if lost
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring to match
+ *
+ */
+static void ih_v7_0_irq_rearm(struct amdgpu_device *adev,
+ struct amdgpu_ih_ring *ih)
+{
+ uint32_t v = 0;
+ uint32_t i = 0;
+ struct amdgpu_ih_regs *ih_regs;
+
+ ih_regs = &ih->ih_regs;
+
+ /* Rearm IRQ / re-write doorbell if doorbell write is lost */
+ for (i = 0; i < MAX_REARM_RETRY; i++) {
+ v = RREG32_NO_KIQ(ih_regs->ih_rb_rptr);
+ if ((v < ih->ring_size) && (v != ih->rptr))
+ WDOORBELL32(ih->doorbell_index, ih->rptr);
+ else
+ break;
+ }
+}
+
+/**
+ * ih_v7_0_set_rptr - set the IH ring buffer rptr
+ *
+ * @adev: amdgpu_device pointer
+ * @ih: IH ring buffer to set rptr
+ */
+static void ih_v7_0_set_rptr(struct amdgpu_device *adev,
+ struct amdgpu_ih_ring *ih)
+{
+ struct amdgpu_ih_regs *ih_regs;
+
+ if (ih->use_doorbell) {
+ /* XXX check if swapping is necessary on BE */
+ *ih->rptr_cpu = ih->rptr;
+ WDOORBELL32(ih->doorbell_index, ih->rptr);
+
+ if (amdgpu_sriov_vf(adev))
+ ih_v7_0_irq_rearm(adev, ih);
+ } else {
+ ih_regs = &ih->ih_regs;
+ WREG32(ih_regs->ih_rb_rptr, ih->rptr);
+ }
+}
+
+/**
+ * ih_v7_0_self_irq - dispatch work for ring 1
+ *
+ * @adev: amdgpu_device pointer
+ * @source: irq source
+ * @entry: IV with WPTR update
+ *
+ * Update the WPTR from the IV and schedule work to handle the entries.
+ */
+static int ih_v7_0_self_irq(struct amdgpu_device *adev,
+ struct amdgpu_irq_src *source,
+ struct amdgpu_iv_entry *entry)
+{
+ uint32_t wptr = cpu_to_le32(entry->src_data[0]);
+
+ switch (entry->ring_id) {
+ case 1:
+ *adev->irq.ih1.wptr_cpu = wptr;
+ schedule_work(&adev->irq.ih1_work);
+ break;
+ default: break;
+ }
+ return 0;
+}
+
+static const struct amdgpu_irq_src_funcs ih_v7_0_self_irq_funcs = {
+ .process = ih_v7_0_self_irq,
+};
+
+static void ih_v7_0_set_self_irq_funcs(struct amdgpu_device *adev)
+{
+ adev->irq.self_irq.num_types = 0;
+ adev->irq.self_irq.funcs = &ih_v7_0_self_irq_funcs;
+}
+
+static int ih_v7_0_early_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ ih_v7_0_set_interrupt_funcs(adev);
+ ih_v7_0_set_self_irq_funcs(adev);
+ return 0;
+}
+
+static int ih_v7_0_sw_init(void *handle)
+{
+ int r;
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ bool use_bus_addr;
+
+ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_IH, 0,
+ &adev->irq.self_irq);
+
+ if (r)
+ return r;
+
+ /* use gpu virtual address for ih ring
+ * until ih_checken is programmed to allow
+ * use bus address for ih ring by psp bl */
+ use_bus_addr =
+ (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
+ r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
+ if (r)
+ return r;
+
+ adev->irq.ih.use_doorbell = true;
+ adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
+
+ adev->irq.ih1.ring_size = 0;
+ adev->irq.ih2.ring_size = 0;
+
+ /* initialize ih control register offset */
+ ih_v7_0_init_register_offset(adev);
+
+ r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, PAGE_SIZE, true);
+ if (r)
+ return r;
+
+ r = amdgpu_irq_init(adev);
+
+ return r;
+}
+
+static int ih_v7_0_sw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ amdgpu_irq_fini_sw(adev);
+
+ return 0;
+}
+
+static int ih_v7_0_hw_init(void *handle)
+{
+ int r;
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ r = ih_v7_0_irq_init(adev);
+ if (r)
+ return r;
+
+ return 0;
+}
+
+static int ih_v7_0_hw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ ih_v7_0_irq_disable(adev);
+
+ return 0;
+}
+
+static int ih_v7_0_suspend(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ return ih_v7_0_hw_fini(adev);
+}
+
+static int ih_v7_0_resume(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ return ih_v7_0_hw_init(adev);
+}
+
+static bool ih_v7_0_is_idle(void *handle)
+{
+ /* todo */
+ return true;
+}
+
+static int ih_v7_0_wait_for_idle(void *handle)
+{
+ /* todo */
+ return -ETIMEDOUT;
+}
+
+static int ih_v7_0_soft_reset(void *handle)
+{
+ /* todo */
+ return 0;
+}
+
+static void ih_v7_0_update_clockgating_state(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t data, def, field_val;
+
+ if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
+ def = data = RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL);
+ field_val = enable ? 0 : 1;
+ data = REG_SET_FIELD(data, IH_CLK_CTRL,
+ DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
+ data = REG_SET_FIELD(data, IH_CLK_CTRL,
+ OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
+ data = REG_SET_FIELD(data, IH_CLK_CTRL,
+ LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
+ data = REG_SET_FIELD(data, IH_CLK_CTRL,
+ DYN_CLK_SOFT_OVERRIDE, field_val);
+ data = REG_SET_FIELD(data, IH_CLK_CTRL,
+ REG_CLK_SOFT_OVERRIDE, field_val);
+ if (def != data)
+ WREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL, data);
+ }
+
+ return;
+}
+
+static int ih_v7_0_set_clockgating_state(void *handle,
+ enum amd_clockgating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ ih_v7_0_update_clockgating_state(adev,
+ state == AMD_CG_STATE_GATE);
+ return 0;
+}
+
+static void ih_v7_0_update_ih_mem_power_gating(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t ih_mem_pwr_cntl;
+
+ /* Disable ih sram power cntl before switch powergating mode */
+ ih_mem_pwr_cntl = RREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_CTRL_EN, 0);
+ WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+
+ /* It is recommended to set mem powergating mode to DS mode */
+ if (enable) {
+ /* mem power mode */
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_LS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_DS_EN, 1);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_SD_EN, 0);
+ /* cam mem power mode */
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 1);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+ /* re-enable power cntl */
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+ } else {
+ /* mem power mode */
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_LS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_DS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_SD_EN, 0);
+ /* cam mem power mode */
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_LS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_DS_EN, 0);
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_RETRY_INT_CAM_MEM_POWER_SD_EN, 0);
+ /* re-enable power cntl*/
+ ih_mem_pwr_cntl = REG_SET_FIELD(ih_mem_pwr_cntl, IH_MEM_POWER_CTRL,
+ IH_BUFFER_MEM_POWER_CTRL_EN, 1);
+ }
+
+ WREG32_SOC15(OSSSYS, 0, regIH_MEM_POWER_CTRL, ih_mem_pwr_cntl);
+}
+
+static int ih_v7_0_set_powergating_state(void *handle,
+ enum amd_powergating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ bool enable = (state == AMD_PG_STATE_GATE);
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_IH_SRAM_PG)
+ ih_v7_0_update_ih_mem_power_gating(adev, enable);
+
+ return 0;
+}
+
+static void ih_v7_0_get_clockgating_state(void *handle, u64 *flags)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ if (!RREG32_SOC15(OSSSYS, 0, regIH_CLK_CTRL))
+ *flags |= AMD_CG_SUPPORT_IH_CG;
+
+ return;
+}
+
+static const struct amd_ip_funcs ih_v7_0_ip_funcs = {
+ .name = "ih_v7_0",
+ .early_init = ih_v7_0_early_init,
+ .late_init = NULL,
+ .sw_init = ih_v7_0_sw_init,
+ .sw_fini = ih_v7_0_sw_fini,
+ .hw_init = ih_v7_0_hw_init,
+ .hw_fini = ih_v7_0_hw_fini,
+ .suspend = ih_v7_0_suspend,
+ .resume = ih_v7_0_resume,
+ .is_idle = ih_v7_0_is_idle,
+ .wait_for_idle = ih_v7_0_wait_for_idle,
+ .soft_reset = ih_v7_0_soft_reset,
+ .set_clockgating_state = ih_v7_0_set_clockgating_state,
+ .set_powergating_state = ih_v7_0_set_powergating_state,
+ .get_clockgating_state = ih_v7_0_get_clockgating_state,
+};
+
+static const struct amdgpu_ih_funcs ih_v7_0_funcs = {
+ .get_wptr = ih_v7_0_get_wptr,
+ .decode_iv = amdgpu_ih_decode_iv_helper,
+ .decode_iv_ts = amdgpu_ih_decode_iv_ts_helper,
+ .set_rptr = ih_v7_0_set_rptr
+};
+
+static void ih_v7_0_set_interrupt_funcs(struct amdgpu_device *adev)
+{
+ adev->irq.ih_funcs = &ih_v7_0_funcs;
+}
+
+const struct amdgpu_ip_block_version ih_v7_0_ip_block =
+{
+ .type = AMD_IP_BLOCK_TYPE_IH,
+ .major = 7,
+ .minor = 0,
+ .rev = 0,
+ .funcs = &ih_v7_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h
new file mode 100644
index 000000000000..af9dcbc451fd
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/ih_v7_0.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __IH_V7_0_IH_H__
+#define __IH_V7_0_IH_H__
+
+extern const struct amdgpu_ip_block_version ih_v7_0_ip_block;
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
index e67a337457ed..99cd49ee8ef6 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
@@ -551,7 +551,7 @@ static int jpeg_v2_5_set_powergating_state(void *handle,
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
int ret;
- if(state == adev->jpeg.cur_state)
+ if (state == adev->jpeg.cur_state)
return 0;
if (state == AMD_PG_STATE_GATE)
@@ -559,7 +559,7 @@ static int jpeg_v2_5_set_powergating_state(void *handle,
else
ret = jpeg_v2_5_start(adev);
- if(!ret)
+ if (!ret)
adev->jpeg.cur_state = state;
return ret;
@@ -754,8 +754,7 @@ static void jpeg_v2_5_set_irq_funcs(struct amdgpu_device *adev)
}
}
-const struct amdgpu_ip_block_version jpeg_v2_5_ip_block =
-{
+const struct amdgpu_ip_block_version jpeg_v2_5_ip_block = {
.type = AMD_IP_BLOCK_TYPE_JPEG,
.major = 2,
.minor = 5,
@@ -763,8 +762,7 @@ const struct amdgpu_ip_block_version jpeg_v2_5_ip_block =
.funcs = &jpeg_v2_5_ip_funcs,
};
-const struct amdgpu_ip_block_version jpeg_v2_6_ip_block =
-{
+const struct amdgpu_ip_block_version jpeg_v2_6_ip_block = {
.type = AMD_IP_BLOCK_TYPE_JPEG,
.major = 2,
.minor = 6,
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
index bc38b90f8cf8..88ea58d5c4ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c
@@ -674,14 +674,6 @@ static int jpeg_v4_0_set_powergating_state(void *handle,
return ret;
}
-static int jpeg_v4_0_set_interrupt_state(struct amdgpu_device *adev,
- struct amdgpu_irq_src *source,
- unsigned type,
- enum amdgpu_interrupt_state state)
-{
- return 0;
-}
-
static int jpeg_v4_0_set_ras_interrupt_state(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
unsigned int type,
@@ -765,7 +757,6 @@ static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev)
}
static const struct amdgpu_irq_src_funcs jpeg_v4_0_irq_funcs = {
- .set = jpeg_v4_0_set_interrupt_state,
.process = jpeg_v4_0_process_interrupt,
};
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
index 82b6b62c170b..32caeb37cef9 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c
@@ -652,7 +652,7 @@ static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring)
*
* Write a start command to the ring.
*/
-static void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring)
+void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring)
{
if (!amdgpu_sriov_vf(ring->adev)) {
amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
@@ -672,7 +672,7 @@ static void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring)
*
* Write a end command to the ring.
*/
-static void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring)
+void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring)
{
if (!amdgpu_sriov_vf(ring->adev)) {
amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
@@ -695,7 +695,7 @@ static void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring)
*
* Write a fence and a trap command to the ring.
*/
-static void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
unsigned int flags)
{
WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
@@ -764,7 +764,7 @@ static void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
*
* Write ring commands to execute the indirect buffer.
*/
-static void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
+void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
uint32_t flags)
@@ -815,7 +815,7 @@ static void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
amdgpu_ring_write(ring, 0x2);
}
-static void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
uint32_t val, uint32_t mask)
{
uint32_t reg_offset = (reg << 2);
@@ -842,7 +842,7 @@ static void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_
amdgpu_ring_write(ring, mask);
}
-static void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
unsigned int vmid, uint64_t pd_addr)
{
struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub];
@@ -857,7 +857,7 @@ static void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
jpeg_v4_0_3_dec_ring_emit_reg_wait(ring, data0, data1, mask);
}
-static void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
+void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
{
uint32_t reg_offset = (reg << 2);
@@ -875,7 +875,7 @@ static void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t re
amdgpu_ring_write(ring, val);
}
-static void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
+void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count)
{
int i;
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
index 22483dc66351..747a3e5f6856 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h
@@ -48,4 +48,19 @@
extern const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block;
+void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring,
+ struct amdgpu_job *job,
+ struct amdgpu_ib *ib,
+ uint32_t flags);
+void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
+ unsigned int flags);
+void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
+ unsigned int vmid, uint64_t pd_addr);
+void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count);
+void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring);
+void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring);
+void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
+void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
+ uint32_t val, uint32_t mask);
+
#endif /* __JPEG_V4_0_3_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
index 6ede85b28cc8..8d1754e35605 100644
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_5.c
@@ -34,7 +34,17 @@
#include "vcn/vcn_4_0_5_sh_mask.h"
#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
-#define regUVD_JPEG_PITCH_INTERNAL_OFFSET 0x401f
+#define mmUVD_DPG_LMA_CTL regUVD_DPG_LMA_CTL
+#define mmUVD_DPG_LMA_CTL_BASE_IDX regUVD_DPG_LMA_CTL_BASE_IDX
+#define mmUVD_DPG_LMA_DATA regUVD_DPG_LMA_DATA
+#define mmUVD_DPG_LMA_DATA_BASE_IDX regUVD_DPG_LMA_DATA_BASE_IDX
+
+#define regUVD_JPEG_PITCH_INTERNAL_OFFSET 0x401f
+#define regJPEG_DEC_GFX10_ADDR_CONFIG_INTERNAL_OFFSET 0x4026
+#define regJPEG_SYS_INT_EN_INTERNAL_OFFSET 0x4141
+#define regJPEG_CGC_CTRL_INTERNAL_OFFSET 0x4161
+#define regJPEG_CGC_GATE_INTERNAL_OFFSET 0x4160
+#define regUVD_NO_OP_INTERNAL_OFFSET 0x0029
static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev);
static void jpeg_v4_0_5_set_irq_funcs(struct amdgpu_device *adev);
@@ -155,11 +165,18 @@ static int jpeg_v4_0_5_hw_init(void *handle)
struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
int r;
+ // TODO: Enable ring test with DPG support
+ if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+ DRM_DEV_INFO(adev->dev, "JPEG decode initialized successfully under DPG Mode");
+ return 0;
+ }
+
r = amdgpu_ring_test_helper(ring);
if (r)
return r;
- DRM_DEV_INFO(adev->dev, "JPEG decode initialized successfully.\n");
+ if (!r)
+ DRM_INFO("JPEG decode initialized successfully under SPG Mode\n");
return 0;
}
@@ -181,7 +198,6 @@ static int jpeg_v4_0_5_hw_fini(void *handle)
RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS))
jpeg_v4_0_5_set_powergating_state(adev, AMD_PG_STATE_GATE);
}
- amdgpu_irq_put(adev, &adev->jpeg.inst->irq, 0);
return 0;
}
@@ -228,11 +244,11 @@ static int jpeg_v4_0_5_resume(void *handle)
return r;
}
-static void jpeg_v4_0_5_disable_clock_gating(struct amdgpu_device *adev)
+static void jpeg_v4_0_5_disable_clock_gating(struct amdgpu_device *adev, int inst)
{
uint32_t data = 0;
- data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+ data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL);
if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
data &= (~JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK);
@@ -242,21 +258,21 @@ static void jpeg_v4_0_5_disable_clock_gating(struct amdgpu_device *adev)
data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
- WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+ WREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL, data);
- data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+ data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE);
data &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
| JPEG_CGC_GATE__JPEG2_DEC_MASK
| JPEG_CGC_GATE__JMCIF_MASK
| JPEG_CGC_GATE__JRBBM_MASK);
- WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+ WREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE, data);
}
-static void jpeg_v4_0_5_enable_clock_gating(struct amdgpu_device *adev)
+static void jpeg_v4_0_5_enable_clock_gating(struct amdgpu_device *adev, int inst)
{
uint32_t data = 0;
- data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+ data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL);
if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG) {
data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
data |= JPEG_CGC_CTRL__JPEG_DEC_MODE_MASK;
@@ -266,47 +282,66 @@ static void jpeg_v4_0_5_enable_clock_gating(struct amdgpu_device *adev)
data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
- WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+ WREG32_SOC15(JPEG, inst, regJPEG_CGC_CTRL, data);
- data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+ data = RREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE);
data |= (JPEG_CGC_GATE__JPEG_DEC_MASK
|JPEG_CGC_GATE__JPEG2_DEC_MASK
|JPEG_CGC_GATE__JMCIF_MASK
|JPEG_CGC_GATE__JRBBM_MASK);
- WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+ WREG32_SOC15(JPEG, inst, regJPEG_CGC_GATE, data);
+}
+
+static void jpeg_engine_4_0_5_dpg_clock_gating_mode(struct amdgpu_device *adev,
+ int inst_idx, uint8_t indirect)
+{
+ uint32_t data = 0;
+
+ if (adev->cg_flags & AMD_CG_SUPPORT_JPEG_MGCG)
+ data |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+ else
+ data |= 0 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
+
+ data |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
+ data |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
+ WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_CGC_CTRL_INTERNAL_OFFSET, data, indirect);
+
+ data = 0;
+ WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_CGC_GATE_INTERNAL_OFFSET,
+ data, indirect);
}
-static int jpeg_v4_0_5_disable_static_power_gating(struct amdgpu_device *adev)
+static int jpeg_v4_0_5_disable_static_power_gating(struct amdgpu_device *adev, int inst)
{
if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
- WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_IPX_DLDO_CONFIG),
+ WREG32(SOC15_REG_OFFSET(JPEG, inst, regUVD_IPX_DLDO_CONFIG),
1 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
- SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS,
+ SOC15_WAIT_ON_RREG(JPEG, inst, regUVD_IPX_DLDO_STATUS,
0, UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
}
/* disable anti hang mechanism */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+ WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS), 0,
~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
/* keep the JPEG in static PG mode */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+ WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS), 0,
~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
return 0;
}
-static int jpeg_v4_0_5_enable_static_power_gating(struct amdgpu_device *adev)
+static int jpeg_v4_0_5_enable_static_power_gating(struct amdgpu_device *adev, int inst)
{
/* enable anti hang mechanism */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+ WREG32_P(SOC15_REG_OFFSET(JPEG, inst, regUVD_JPEG_POWER_STATUS),
UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
- WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_IPX_DLDO_CONFIG),
+ WREG32(SOC15_REG_OFFSET(JPEG, inst, regUVD_IPX_DLDO_CONFIG),
2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
- SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS,
+ SOC15_WAIT_ON_RREG(JPEG, inst, regUVD_IPX_DLDO_STATUS,
1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
}
@@ -315,61 +350,149 @@ static int jpeg_v4_0_5_enable_static_power_gating(struct amdgpu_device *adev)
}
/**
- * jpeg_v4_0_5_start - start JPEG block
+ * jpeg_v4_0_5_start_dpg_mode - Jpeg start with dpg mode
*
* @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
*
- * Setup and start the JPEG block
+ * Start JPEG block with dpg mode
*/
-static int jpeg_v4_0_5_start(struct amdgpu_device *adev)
+static void jpeg_v4_0_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect)
{
- struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
- int r;
+ struct amdgpu_ring *ring = adev->jpeg.inst[inst_idx].ring_dec;
+ uint32_t reg_data = 0;
- if (adev->pm.dpm_enabled)
- amdgpu_dpm_enable_jpeg(adev, true);
+ /* enable anti hang mechanism */
+ reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+ reg_data &= ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK;
+ reg_data |= 0x1;
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
- /* doorbell programming is done for every playback */
- adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
- (adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+ if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+ WREG32(SOC15_REG_OFFSET(JPEG, inst_idx, regUVD_IPX_DLDO_CONFIG),
+ 2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+ SOC15_WAIT_ON_RREG(JPEG, inst_idx, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+ }
- WREG32_SOC15(VCN, 0, regVCN_JPEG_DB_CTRL,
- ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
- VCN_JPEG_DB_CTRL__EN_MASK);
+ reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+ reg_data |= UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
- /* disable power gating */
- r = jpeg_v4_0_5_disable_static_power_gating(adev);
- if (r)
- return r;
+ if (indirect)
+ adev->jpeg.inst[inst_idx].dpg_sram_curr_addr =
+ (uint32_t *)adev->jpeg.inst[inst_idx].dpg_sram_cpu_addr;
- /* JPEG disable CGC */
- jpeg_v4_0_5_disable_clock_gating(adev);
+ jpeg_engine_4_0_5_dpg_clock_gating_mode(adev, inst_idx, indirect);
/* MJPEG global tiling registers */
- WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
- adev->gfx.config.gb_addr_config);
-
+ WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_DEC_GFX10_ADDR_CONFIG_INTERNAL_OFFSET,
+ adev->gfx.config.gb_addr_config, indirect);
+ /* enable System Interrupt for JRBC */
+ WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regJPEG_SYS_INT_EN_INTERNAL_OFFSET,
+ JPEG_SYS_INT_EN__DJRBC_MASK, indirect);
- /* enable JMI channel */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL), 0,
- ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+ /* add nop to workaround PSP size check */
+ WREG32_SOC15_JPEG_DPG_MODE(inst_idx, regUVD_NO_OP_INTERNAL_OFFSET, 0, indirect);
- /* enable System Interrupt for JRBC */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regJPEG_SYS_INT_EN),
- JPEG_SYS_INT_EN__DJRBC_MASK,
- ~JPEG_SYS_INT_EN__DJRBC_MASK);
+ if (indirect)
+ amdgpu_jpeg_psp_update_sram(adev, inst_idx, 0);
- WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_VMID, 0);
- WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
- WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+ WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_VMID, 0);
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+ WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
lower_32_bits(ring->gpu_addr));
- WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+ WREG32_SOC15(JPEG, inst_idx, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
upper_32_bits(ring->gpu_addr));
- WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR, 0);
- WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, 0);
- WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, 0x00000002L);
- WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
- ring->wptr = RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_RPTR, 0);
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR, 0);
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_CNTL, 0x00000002L);
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+ ring->wptr = RREG32_SOC15(JPEG, inst_idx, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v4_0_5_stop_dpg_mode - Jpeg stop with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ *
+ * Stop JPEG block with dpg mode
+ */
+static void jpeg_v4_0_5_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx)
+{
+ uint32_t reg_data = 0;
+
+ reg_data = RREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS);
+ reg_data &= ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK;
+ WREG32_SOC15(JPEG, inst_idx, regUVD_JPEG_POWER_STATUS, reg_data);
+
+}
+
+/**
+ * jpeg_v4_0_5_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v4_0_5_start(struct amdgpu_device *adev)
+{
+ struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+ int r, i;
+
+ if (adev->pm.dpm_enabled)
+ amdgpu_dpm_enable_jpeg(adev, true);
+
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ /* doorbell programming is done for every playback */
+ adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+ (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i, i);
+
+ WREG32_SOC15(VCN, i, regVCN_JPEG_DB_CTRL,
+ ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+ VCN_JPEG_DB_CTRL__EN_MASK);
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
+ jpeg_v4_0_5_start_dpg_mode(adev, i, adev->jpeg.indirect_sram);
+ continue;
+ }
+
+ /* disable power gating */
+ r = jpeg_v4_0_5_disable_static_power_gating(adev, i);
+ if (r)
+ return r;
+
+ /* JPEG disable CGC */
+ jpeg_v4_0_5_disable_clock_gating(adev, i);
+
+ /* MJPEG global tiling registers */
+ WREG32_SOC15(JPEG, i, regJPEG_DEC_GFX10_ADDR_CONFIG,
+ adev->gfx.config.gb_addr_config);
+
+ /* enable JMI channel */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, i, regUVD_JMI_CNTL), 0,
+ ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+ /* enable System Interrupt for JRBC */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, i, regJPEG_SYS_INT_EN),
+ JPEG_SYS_INT_EN__DJRBC_MASK,
+ ~JPEG_SYS_INT_EN__DJRBC_MASK);
+
+ WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_VMID, 0);
+ WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+ WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+ lower_32_bits(ring->gpu_addr));
+ WREG32_SOC15(JPEG, i, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+ upper_32_bits(ring->gpu_addr));
+ WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_RPTR, 0);
+ WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_WPTR, 0);
+ WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_CNTL, 0x00000002L);
+ WREG32_SOC15(JPEG, i, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+ ring->wptr = RREG32_SOC15(JPEG, i, regUVD_JRBC_RB_WPTR);
+ }
return 0;
}
@@ -383,19 +506,26 @@ static int jpeg_v4_0_5_start(struct amdgpu_device *adev)
*/
static int jpeg_v4_0_5_stop(struct amdgpu_device *adev)
{
- int r;
+ int r, i;
- /* reset JMI */
- WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL),
- UVD_JMI_CNTL__SOFT_RESET_MASK,
- ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ if (adev->pg_flags & AMD_PG_SUPPORT_JPEG_DPG) {
- jpeg_v4_0_5_enable_clock_gating(adev);
+ jpeg_v4_0_5_stop_dpg_mode(adev, i);
+ continue;
+ }
+ /* reset JMI */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, i, regUVD_JMI_CNTL),
+ UVD_JMI_CNTL__SOFT_RESET_MASK,
+ ~UVD_JMI_CNTL__SOFT_RESET_MASK);
- /* enable power gating */
- r = jpeg_v4_0_5_enable_static_power_gating(adev);
- if (r)
- return r;
+ jpeg_v4_0_5_enable_clock_gating(adev, i);
+
+ /* enable power gating */
+ r = jpeg_v4_0_5_enable_static_power_gating(adev, i);
+ if (r)
+ return r;
+ }
if (adev->pm.dpm_enabled)
amdgpu_dpm_enable_jpeg(adev, false);
@@ -479,13 +609,20 @@ static int jpeg_v4_0_5_set_clockgating_state(void *handle,
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+ int i;
- if (enable) {
- if (!jpeg_v4_0_5_is_idle(handle))
- return -EBUSY;
- jpeg_v4_0_5_enable_clock_gating(adev);
- } else {
- jpeg_v4_0_5_disable_clock_gating(adev);
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ if (adev->jpeg.harvest_config & (1 << i))
+ continue;
+
+ if (enable) {
+ if (!jpeg_v4_0_5_is_idle(handle))
+ return -EBUSY;
+
+ jpeg_v4_0_5_enable_clock_gating(adev, i);
+ } else {
+ jpeg_v4_0_5_disable_clock_gating(adev, i);
+ }
}
return 0;
@@ -516,14 +653,6 @@ static int jpeg_v4_0_5_set_powergating_state(void *handle,
return ret;
}
-static int jpeg_v4_0_5_set_interrupt_state(struct amdgpu_device *adev,
- struct amdgpu_irq_src *source,
- unsigned type,
- enum amdgpu_interrupt_state state)
-{
- return 0;
-}
-
static int jpeg_v4_0_5_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
@@ -598,19 +727,32 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_5_dec_ring_vm_funcs = {
static void jpeg_v4_0_5_set_dec_ring_funcs(struct amdgpu_device *adev)
{
- adev->jpeg.inst->ring_dec->funcs = &jpeg_v4_0_5_dec_ring_vm_funcs;
- DRM_DEV_INFO(adev->dev, "JPEG decode is enabled in VM mode\n");
+ int i;
+
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ if (adev->jpeg.harvest_config & (1 << i))
+ continue;
+
+ adev->jpeg.inst[i].ring_dec->funcs = &jpeg_v4_0_5_dec_ring_vm_funcs;
+ DRM_DEV_INFO(adev->dev, "JPEG%d decode is enabled in VM mode\n", i);
+ }
}
static const struct amdgpu_irq_src_funcs jpeg_v4_0_5_irq_funcs = {
- .set = jpeg_v4_0_5_set_interrupt_state,
.process = jpeg_v4_0_5_process_interrupt,
};
static void jpeg_v4_0_5_set_irq_funcs(struct amdgpu_device *adev)
{
- adev->jpeg.inst->irq.num_types = 1;
- adev->jpeg.inst->irq.funcs = &jpeg_v4_0_5_irq_funcs;
+ int i;
+
+ for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) {
+ if (adev->jpeg.harvest_config & (1 << i))
+ continue;
+
+ adev->jpeg.inst[i].irq.num_types = 1;
+ adev->jpeg.inst[i].irq.funcs = &jpeg_v4_0_5_irq_funcs;
+ }
}
const struct amdgpu_ip_block_version jpeg_v4_0_5_ip_block = {
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
new file mode 100644
index 000000000000..e70200f97555
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.c
@@ -0,0 +1,570 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "amdgpu.h"
+#include "amdgpu_jpeg.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "jpeg_v4_0_3.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+
+static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev);
+static void jpeg_v5_0_0_set_irq_funcs(struct amdgpu_device *adev);
+static int jpeg_v5_0_0_set_powergating_state(void *handle,
+ enum amd_powergating_state state);
+
+/**
+ * jpeg_v5_0_0_early_init - set function pointers
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Set ring and irq function pointers
+ */
+static int jpeg_v5_0_0_early_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ adev->jpeg.num_jpeg_inst = 1;
+ adev->jpeg.num_jpeg_rings = 1;
+
+ jpeg_v5_0_0_set_dec_ring_funcs(adev);
+ jpeg_v5_0_0_set_irq_funcs(adev);
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_sw_init - sw init for JPEG block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Load firmware and sw initialization
+ */
+static int jpeg_v5_0_0_sw_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ struct amdgpu_ring *ring;
+ int r;
+
+ /* JPEG TRAP */
+ r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
+ VCN_4_0__SRCID__JPEG_DECODE, &adev->jpeg.inst->irq);
+ if (r)
+ return r;
+
+ r = amdgpu_jpeg_sw_init(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_jpeg_resume(adev);
+ if (r)
+ return r;
+
+ ring = adev->jpeg.inst->ring_dec;
+ ring->use_doorbell = true;
+ ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
+ ring->vm_hub = AMDGPU_MMHUB0(0);
+
+ sprintf(ring->name, "jpeg_dec");
+ r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0,
+ AMDGPU_RING_PRIO_DEFAULT, NULL);
+ if (r)
+ return r;
+
+ adev->jpeg.internal.jpeg_pitch[0] = regUVD_JPEG_PITCH_INTERNAL_OFFSET;
+ adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_PITCH);
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_sw_fini - sw fini for JPEG block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * JPEG suspend and free up sw allocation
+ */
+static int jpeg_v5_0_0_sw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int r;
+
+ r = amdgpu_jpeg_suspend(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_jpeg_sw_fini(adev);
+
+ return r;
+}
+
+/**
+ * jpeg_v5_0_0_hw_init - start and test JPEG block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ */
+static int jpeg_v5_0_0_hw_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+ int r;
+
+ adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+ (adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0);
+
+ WREG32_SOC15(VCN, 0, regVCN_JPEG_DB_CTRL,
+ ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT |
+ VCN_JPEG_DB_CTRL__EN_MASK);
+
+ r = amdgpu_ring_test_helper(ring);
+ if (r)
+ return r;
+
+ DRM_DEV_INFO(adev->dev, "JPEG decode initialized successfully.\n");
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_hw_fini - stop the hardware block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Stop the JPEG block, mark ring as not ready any more
+ */
+static int jpeg_v5_0_0_hw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ cancel_delayed_work_sync(&adev->vcn.idle_work);
+
+ if (adev->jpeg.cur_state != AMD_PG_STATE_GATE &&
+ RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS))
+ jpeg_v5_0_0_set_powergating_state(adev, AMD_PG_STATE_GATE);
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_suspend - suspend JPEG block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * HW fini and suspend JPEG block
+ */
+static int jpeg_v5_0_0_suspend(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int r;
+
+ r = jpeg_v5_0_0_hw_fini(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_jpeg_suspend(adev);
+
+ return r;
+}
+
+/**
+ * jpeg_v5_0_0_resume - resume JPEG block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Resume firmware and hw init JPEG block
+ */
+static int jpeg_v5_0_0_resume(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int r;
+
+ r = amdgpu_jpeg_resume(adev);
+ if (r)
+ return r;
+
+ r = jpeg_v5_0_0_hw_init(adev);
+
+ return r;
+}
+
+static void jpeg_v5_0_0_disable_clock_gating(struct amdgpu_device *adev)
+{
+ uint32_t data = 0;
+
+ WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+
+ data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+ data &= ~(JPEG_CGC_CTRL__JPEG0_DEC_MODE_MASK
+ | JPEG_CGC_CTRL__JPEG_ENC_MODE_MASK);
+ WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+}
+
+static void jpeg_v5_0_0_enable_clock_gating(struct amdgpu_device *adev)
+{
+ uint32_t data = 0;
+
+ data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL);
+
+ data |= 1 << JPEG_CGC_CTRL__JPEG0_DEC_MODE__SHIFT;
+ WREG32_SOC15(JPEG, 0, regJPEG_CGC_CTRL, data);
+
+ data = RREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE);
+ data |= (JPEG_CGC_GATE__JPEG0_DEC_MASK
+ |JPEG_CGC_GATE__JPEG_ENC_MASK
+ |JPEG_CGC_GATE__JMCIF_MASK
+ |JPEG_CGC_GATE__JRBBM_MASK);
+ WREG32_SOC15(JPEG, 0, regJPEG_CGC_GATE, data);
+}
+
+static int jpeg_v5_0_0_disable_static_power_gating(struct amdgpu_device *adev)
+{
+ uint32_t data = 0;
+
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(JPEG, 0, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+
+ /* disable anti hang mechanism */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+ ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+ /* keep the JPEG in static PG mode */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS), 0,
+ ~UVD_JPEG_POWER_STATUS__JPEG_PG_MODE_MASK);
+
+ return 0;
+}
+
+static int jpeg_v5_0_0_enable_static_power_gating(struct amdgpu_device *adev)
+{
+ /* enable anti hang mechanism */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JPEG_POWER_STATUS),
+ UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK,
+ ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK);
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_JPEG) {
+ WREG32(SOC15_REG_OFFSET(JPEG, 0, regUVD_IPX_DLDO_CONFIG),
+ 2 << UVD_IPX_DLDO_CONFIG__ONO1_PWR_CONFIG__SHIFT);
+ SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO1_PWR_STATUS_MASK);
+ }
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_start - start JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Setup and start the JPEG block
+ */
+static int jpeg_v5_0_0_start(struct amdgpu_device *adev)
+{
+ struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec;
+ int r;
+
+ if (adev->pm.dpm_enabled)
+ amdgpu_dpm_enable_jpeg(adev, true);
+
+ /* disable power gating */
+ r = jpeg_v5_0_0_disable_static_power_gating(adev);
+ if (r)
+ return r;
+
+ /* JPEG disable CGC */
+ jpeg_v5_0_0_disable_clock_gating(adev);
+
+ /* MJPEG global tiling registers */
+ WREG32_SOC15(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG,
+ adev->gfx.config.gb_addr_config);
+
+
+ /* enable JMI channel */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL), 0,
+ ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+ /* enable System Interrupt for JRBC */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regJPEG_SYS_INT_EN),
+ JPEG_SYS_INT_EN__DJRBC0_MASK,
+ ~JPEG_SYS_INT_EN__DJRBC0_MASK);
+
+ WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_VMID, 0);
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
+ WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
+ lower_32_bits(ring->gpu_addr));
+ WREG32_SOC15(JPEG, 0, regUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
+ upper_32_bits(ring->gpu_addr));
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR, 0);
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, 0);
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_CNTL, 0x00000002L);
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_SIZE, ring->ring_size / 4);
+ ring->wptr = RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_stop - stop JPEG block
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * stop the JPEG block
+ */
+static int jpeg_v5_0_0_stop(struct amdgpu_device *adev)
+{
+ int r;
+
+ /* reset JMI */
+ WREG32_P(SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI_CNTL),
+ UVD_JMI_CNTL__SOFT_RESET_MASK,
+ ~UVD_JMI_CNTL__SOFT_RESET_MASK);
+
+ jpeg_v5_0_0_enable_clock_gating(adev);
+
+ /* enable power gating */
+ r = jpeg_v5_0_0_enable_static_power_gating(adev);
+ if (r)
+ return r;
+
+ if (adev->pm.dpm_enabled)
+ amdgpu_dpm_enable_jpeg(adev, false);
+
+ return 0;
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_get_rptr - get read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware read pointer
+ */
+static uint64_t jpeg_v5_0_0_dec_ring_get_rptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_RPTR);
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_get_wptr - get write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware write pointer
+ */
+static uint64_t jpeg_v5_0_0_dec_ring_get_wptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ if (ring->use_doorbell)
+ return *ring->wptr_cpu_addr;
+ else
+ return RREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR);
+}
+
+/**
+ * jpeg_v5_0_0_dec_ring_set_wptr - set write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the write pointer to the hardware
+ */
+static void jpeg_v5_0_0_dec_ring_set_wptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ if (ring->use_doorbell) {
+ *ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+ WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+ } else {
+ WREG32_SOC15(JPEG, 0, regUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
+ }
+}
+
+static bool jpeg_v5_0_0_is_idle(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int ret = 1;
+
+ ret &= (((RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS) &
+ UVD_JRBC_STATUS__RB_JOB_DONE_MASK) ==
+ UVD_JRBC_STATUS__RB_JOB_DONE_MASK));
+
+ return ret;
+}
+
+static int jpeg_v5_0_0_wait_for_idle(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ return SOC15_WAIT_ON_RREG(JPEG, 0, regUVD_JRBC_STATUS,
+ UVD_JRBC_STATUS__RB_JOB_DONE_MASK,
+ UVD_JRBC_STATUS__RB_JOB_DONE_MASK);
+}
+
+static int jpeg_v5_0_0_set_clockgating_state(void *handle,
+ enum amd_clockgating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+
+ if (enable) {
+ if (!jpeg_v5_0_0_is_idle(handle))
+ return -EBUSY;
+ jpeg_v5_0_0_enable_clock_gating(adev);
+ } else {
+ jpeg_v5_0_0_disable_clock_gating(adev);
+ }
+
+ return 0;
+}
+
+static int jpeg_v5_0_0_set_powergating_state(void *handle,
+ enum amd_powergating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int ret;
+
+ if (state == adev->jpeg.cur_state)
+ return 0;
+
+ if (state == AMD_PG_STATE_GATE)
+ ret = jpeg_v5_0_0_stop(adev);
+ else
+ ret = jpeg_v5_0_0_start(adev);
+
+ if (!ret)
+ adev->jpeg.cur_state = state;
+
+ return ret;
+}
+
+static int jpeg_v5_0_0_set_interrupt_state(struct amdgpu_device *adev,
+ struct amdgpu_irq_src *source,
+ unsigned int type,
+ enum amdgpu_interrupt_state state)
+{
+ return 0;
+}
+
+static int jpeg_v5_0_0_process_interrupt(struct amdgpu_device *adev,
+ struct amdgpu_irq_src *source,
+ struct amdgpu_iv_entry *entry)
+{
+ DRM_DEBUG("IH: JPEG TRAP\n");
+
+ switch (entry->src_id) {
+ case VCN_4_0__SRCID__JPEG_DECODE:
+ amdgpu_fence_process(adev->jpeg.inst->ring_dec);
+ break;
+ default:
+ DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n",
+ entry->src_id, entry->src_data[0]);
+ break;
+ }
+
+ return 0;
+}
+
+static const struct amd_ip_funcs jpeg_v5_0_0_ip_funcs = {
+ .name = "jpeg_v5_0_0",
+ .early_init = jpeg_v5_0_0_early_init,
+ .late_init = NULL,
+ .sw_init = jpeg_v5_0_0_sw_init,
+ .sw_fini = jpeg_v5_0_0_sw_fini,
+ .hw_init = jpeg_v5_0_0_hw_init,
+ .hw_fini = jpeg_v5_0_0_hw_fini,
+ .suspend = jpeg_v5_0_0_suspend,
+ .resume = jpeg_v5_0_0_resume,
+ .is_idle = jpeg_v5_0_0_is_idle,
+ .wait_for_idle = jpeg_v5_0_0_wait_for_idle,
+ .check_soft_reset = NULL,
+ .pre_soft_reset = NULL,
+ .soft_reset = NULL,
+ .post_soft_reset = NULL,
+ .set_clockgating_state = jpeg_v5_0_0_set_clockgating_state,
+ .set_powergating_state = jpeg_v5_0_0_set_powergating_state,
+};
+
+static const struct amdgpu_ring_funcs jpeg_v5_0_0_dec_ring_vm_funcs = {
+ .type = AMDGPU_RING_TYPE_VCN_JPEG,
+ .align_mask = 0xf,
+ .get_rptr = jpeg_v5_0_0_dec_ring_get_rptr,
+ .get_wptr = jpeg_v5_0_0_dec_ring_get_wptr,
+ .set_wptr = jpeg_v5_0_0_dec_ring_set_wptr,
+ .emit_frame_size =
+ SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
+ SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
+ 8 + /* jpeg_v5_0_0_dec_ring_emit_vm_flush */
+ 22 + 22 + /* jpeg_v5_0_0_dec_ring_emit_fence x2 vm fence */
+ 8 + 16,
+ .emit_ib_size = 22, /* jpeg_v5_0_0_dec_ring_emit_ib */
+ .emit_ib = jpeg_v4_0_3_dec_ring_emit_ib,
+ .emit_fence = jpeg_v4_0_3_dec_ring_emit_fence,
+ .emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush,
+ .test_ring = amdgpu_jpeg_dec_ring_test_ring,
+ .test_ib = amdgpu_jpeg_dec_ring_test_ib,
+ .insert_nop = jpeg_v4_0_3_dec_ring_nop,
+ .insert_start = jpeg_v4_0_3_dec_ring_insert_start,
+ .insert_end = jpeg_v4_0_3_dec_ring_insert_end,
+ .pad_ib = amdgpu_ring_generic_pad_ib,
+ .begin_use = amdgpu_jpeg_ring_begin_use,
+ .end_use = amdgpu_jpeg_ring_end_use,
+ .emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg,
+ .emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait,
+ .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+static void jpeg_v5_0_0_set_dec_ring_funcs(struct amdgpu_device *adev)
+{
+ adev->jpeg.inst->ring_dec->funcs = &jpeg_v5_0_0_dec_ring_vm_funcs;
+ DRM_DEV_INFO(adev->dev, "JPEG decode is enabled in VM mode\n");
+}
+
+static const struct amdgpu_irq_src_funcs jpeg_v5_0_0_irq_funcs = {
+ .set = jpeg_v5_0_0_set_interrupt_state,
+ .process = jpeg_v5_0_0_process_interrupt,
+};
+
+static void jpeg_v5_0_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+ adev->jpeg.inst->irq.num_types = 1;
+ adev->jpeg.inst->irq.funcs = &jpeg_v5_0_0_irq_funcs;
+}
+
+const struct amdgpu_ip_block_version jpeg_v5_0_0_ip_block = {
+ .type = AMD_IP_BLOCK_TYPE_JPEG,
+ .major = 5,
+ .minor = 0,
+ .rev = 0,
+ .funcs = &jpeg_v5_0_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h
new file mode 100644
index 000000000000..bd348336b215
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_0.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __JPEG_V5_0_0_H__
+#define __JPEG_V5_0_0_H__
+
+extern const struct amdgpu_ip_block_version jpeg_v5_0_0_ip_block;
+
+#endif /* __JPEG_V5_0_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c
new file mode 100644
index 000000000000..396262044ea8
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.c
@@ -0,0 +1,121 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/delay.h>
+#include "amdgpu.h"
+#include "lsdma_v7_0.h"
+#include "amdgpu_lsdma.h"
+
+#include "lsdma/lsdma_7_0_0_offset.h"
+#include "lsdma/lsdma_7_0_0_sh_mask.h"
+
+static int lsdma_v7_0_wait_pio_status(struct amdgpu_device *adev)
+{
+ return amdgpu_lsdma_wait_for(adev, SOC15_REG_OFFSET(LSDMA, 0, regLSDMA_PIO_STATUS),
+ LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK,
+ LSDMA_PIO_STATUS__PIO_IDLE_MASK | LSDMA_PIO_STATUS__PIO_FIFO_EMPTY_MASK);
+}
+
+static int lsdma_v7_0_copy_mem(struct amdgpu_device *adev,
+ uint64_t src_addr,
+ uint64_t dst_addr,
+ uint64_t size)
+{
+ int ret;
+ uint32_t tmp;
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_LO, lower_32_bits(src_addr));
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_SRC_ADDR_HI, upper_32_bits(src_addr));
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+ tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 0);
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+ ret = lsdma_v7_0_wait_pio_status(adev);
+ if (ret)
+ dev_err(adev->dev, "LSDMA PIO failed to copy memory!\n");
+
+ return ret;
+}
+
+static int lsdma_v7_0_fill_mem(struct amdgpu_device *adev,
+ uint64_t dst_addr,
+ uint32_t data,
+ uint64_t size)
+{
+ int ret;
+ uint32_t tmp;
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONSTFILL_DATA, data);
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_LO, lower_32_bits(dst_addr));
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_DST_ADDR_HI, upper_32_bits(dst_addr));
+
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_CONTROL, 0x0);
+
+ tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, BYTE_COUNT, size);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_LOCATION, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_LOCATION, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, SRC_ADDR_INC, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, DST_ADDR_INC, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, OVERLAP_DISABLE, 0);
+ tmp = REG_SET_FIELD(tmp, LSDMA_PIO_COMMAND, CONSTANT_FILL, 1);
+ WREG32_SOC15(LSDMA, 0, regLSDMA_PIO_COMMAND, tmp);
+
+ ret = lsdma_v7_0_wait_pio_status(adev);
+ if (ret)
+ dev_err(adev->dev, "LSDMA PIO failed to fill memory!\n");
+
+ return ret;
+}
+
+static void lsdma_v7_0_update_memory_power_gating(struct amdgpu_device *adev,
+ bool enable)
+{
+ uint32_t tmp;
+
+ tmp = RREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL);
+ tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, 0);
+ WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+
+ tmp = REG_SET_FIELD(tmp, LSDMA_MEM_POWER_CTRL, MEM_POWER_CTRL_EN, enable);
+ WREG32_SOC15(LSDMA, 0, regLSDMA_MEM_POWER_CTRL, tmp);
+}
+
+const struct amdgpu_lsdma_funcs lsdma_v7_0_funcs = {
+ .copy_mem = lsdma_v7_0_copy_mem,
+ .fill_mem = lsdma_v7_0_fill_mem,
+ .update_memory_power_gating = lsdma_v7_0_update_memory_power_gating
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h
new file mode 100644
index 000000000000..52b4485cdd98
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/lsdma_v7_0.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __LSDMA_V7_0_H__
+#define __LSDMA_V7_0_H__
+
+#include "soc15_common.h"
+
+extern const struct amdgpu_lsdma_funcs lsdma_v7_0_funcs;
+
+#endif /* __LSDMA_V7_0_H__ */
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c
index fb53aacdcba2..c0fc44cdd658 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c
@@ -33,6 +33,7 @@
#define regVM_L2_CNTL3_DEFAULT 0x80100007
#define regVM_L2_CNTL4_DEFAULT 0x000000c1
+#define mmSMNAID_AID0_MCA_SMU 0x03b30400
static u64 mmhub_v1_8_get_fb_location(struct amdgpu_device *adev)
{
@@ -705,8 +706,94 @@ static const struct amdgpu_ras_block_hw_ops mmhub_v1_8_ras_hw_ops = {
.reset_ras_error_count = mmhub_v1_8_reset_ras_error_count,
};
+static int mmhub_v1_8_aca_bank_generate_report(struct aca_handle *handle,
+ struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data)
+{
+ u64 status, misc0;
+ int ret;
+
+ status = bank->regs[ACA_REG_IDX_STATUS];
+ if ((type == ACA_ERROR_TYPE_UE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) ||
+ (type == ACA_ERROR_TYPE_CE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) {
+
+ ret = aca_bank_info_decode(bank, &report->info);
+ if (ret)
+ return ret;
+
+ misc0 = bank->regs[ACA_REG_IDX_MISC0];
+ report->count[type] = ACA_REG__MISC0__ERRCNT(misc0);
+ }
+
+ return 0;
+}
+
+/* reference to smu driver if header file */
+static int mmhub_v1_8_err_codes[] = {
+ 0, 1, 2, 3, 4, /* CODE_DAGB0 - 4 */
+ 5, 6, 7, 8, 9, /* CODE_EA0 - 4 */
+ 10, /* CODE_UTCL2_ROUTER */
+ 11, /* CODE_VML2 */
+ 12, /* CODE_VML2_WALKER */
+ 13, /* CODE_MMCANE */
+};
+
+static bool mmhub_v1_8_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, void *data)
+{
+ u32 instlo;
+
+ instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+ instlo &= GENMASK(31, 1);
+
+ if (instlo != mmSMNAID_AID0_MCA_SMU)
+ return false;
+
+ if (aca_bank_check_error_codes(handle->adev, bank,
+ mmhub_v1_8_err_codes,
+ ARRAY_SIZE(mmhub_v1_8_err_codes)))
+ return false;
+
+ return true;
+}
+
+static const struct aca_bank_ops mmhub_v1_8_aca_bank_ops = {
+ .aca_bank_generate_report = mmhub_v1_8_aca_bank_generate_report,
+ .aca_bank_is_valid = mmhub_v1_8_aca_bank_is_valid,
+};
+
+static const struct aca_info mmhub_v1_8_aca_info = {
+ .hwip = ACA_HWIP_TYPE_SMU,
+ .mask = ACA_ERROR_UE_MASK,
+ .bank_ops = &mmhub_v1_8_aca_bank_ops,
+};
+
+static int mmhub_v1_8_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+ int r;
+
+ r = amdgpu_ras_block_late_init(adev, ras_block);
+ if (r)
+ return r;
+
+ r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__MMHUB,
+ &mmhub_v1_8_aca_info, NULL);
+ if (r)
+ goto late_fini;
+
+ return 0;
+
+late_fini:
+ amdgpu_ras_block_late_fini(adev, ras_block);
+
+ return r;
+}
+
struct amdgpu_mmhub_ras mmhub_v1_8_ras = {
.ras_block = {
.hw_ops = &mmhub_v1_8_ras_hw_ops,
+ .ras_late_init = mmhub_v1_8_ras_late_init,
},
};
diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c
index dc4812ecc98d..b3961968c10c 100644
--- a/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v3_3.c
@@ -98,6 +98,7 @@ mmhub_v3_3_print_l2_protection_fault_status(struct amdgpu_device *adev,
switch (amdgpu_ip_version(adev, MMHUB_HWIP, 0)) {
case IP_VERSION(3, 3, 0):
+ case IP_VERSION(3, 3, 1):
mmhub_cid = mmhub_client_ids_v3_3[cid][rw];
break;
default:
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
index 63725b2ebc03..a2bd2c3b1ef9 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
@@ -404,7 +404,8 @@ static int xgpu_ai_request_init_data(struct amdgpu_device *adev)
return xgpu_ai_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA);
}
-static void xgpu_ai_ras_poison_handler(struct amdgpu_device *adev)
+static void xgpu_ai_ras_poison_handler(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block)
{
xgpu_ai_send_access_requests(adev, IDH_RAS_POISON);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
index 6a68ee946f1c..77f5b55decf9 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
@@ -152,14 +152,14 @@ static void xgpu_nv_mailbox_trans_msg (struct amdgpu_device *adev,
xgpu_nv_mailbox_set_valid(adev, false);
}
-static int xgpu_nv_send_access_requests(struct amdgpu_device *adev,
- enum idh_request req)
+static int xgpu_nv_send_access_requests_with_param(struct amdgpu_device *adev,
+ enum idh_request req, u32 data1, u32 data2, u32 data3)
{
int r, retry = 1;
enum idh_event event = -1;
send_request:
- xgpu_nv_mailbox_trans_msg(adev, req, 0, 0, 0);
+ xgpu_nv_mailbox_trans_msg(adev, req, data1, data2, data3);
switch (req) {
case IDH_REQ_GPU_INIT_ACCESS:
@@ -170,6 +170,10 @@ send_request:
case IDH_REQ_GPU_INIT_DATA:
event = IDH_REQ_GPU_INIT_DATA_READY;
break;
+ case IDH_RAS_POISON:
+ if (data1 != 0)
+ event = IDH_RAS_POISON_READY;
+ break;
default:
break;
}
@@ -206,6 +210,13 @@ send_request:
return 0;
}
+static int xgpu_nv_send_access_requests(struct amdgpu_device *adev,
+ enum idh_request req)
+{
+ return xgpu_nv_send_access_requests_with_param(adev,
+ req, 0, 0, 0);
+}
+
static int xgpu_nv_request_reset(struct amdgpu_device *adev)
{
int ret, i = 0;
@@ -424,9 +435,17 @@ void xgpu_nv_mailbox_put_irq(struct amdgpu_device *adev)
amdgpu_irq_put(adev, &adev->virt.rcv_irq, 0);
}
-static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev)
+static void xgpu_nv_ras_poison_handler(struct amdgpu_device *adev,
+ enum amdgpu_ras_block block)
{
- xgpu_nv_send_access_requests(adev, IDH_RAS_POISON);
+ if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
+ xgpu_nv_send_access_requests(adev, IDH_RAS_POISON);
+ } else {
+ amdgpu_virt_fini_data_exchange(adev);
+ xgpu_nv_send_access_requests_with_param(adev,
+ IDH_RAS_POISON, block, 0, 0);
+ amdgpu_virt_init_data_exchange(adev);
+ }
}
const struct amdgpu_virt_ops xgpu_nv_virt_ops = {
diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h
index d0221ce08769..1e8fd90cab43 100644
--- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h
+++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.h
@@ -51,6 +51,7 @@ enum idh_event {
IDH_FAIL,
IDH_QUERY_ALIVE,
IDH_REQ_GPU_INIT_DATA_READY,
+ IDH_RAS_POISON_READY,
IDH_TEXT_MESSAGE = 255,
};
diff --git a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
index de93614726c9..4178f4e5dad7 100644
--- a/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/navi10_ih.c
@@ -728,8 +728,7 @@ static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
adev->irq.ih_funcs = &navi10_ih_funcs;
}
-const struct amdgpu_ip_block_version navi10_ih_ip_block =
-{
+const struct amdgpu_ip_block_version navi10_ih_ip_block = {
.type = AMD_IP_BLOCK_TYPE_IH,
.major = 5,
.minor = 0,
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c
index 1f52b4b1db03..05020141c0ae 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c
@@ -89,7 +89,9 @@ static void nbio_v7_11_vpe_doorbell_range(struct amdgpu_device *adev, int instan
bool use_doorbell, int doorbell_index,
int doorbell_size)
{
- u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VPE_DOORBELL_RANGE);
+ u32 reg = instance == 0 ?
+ SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VPE_DOORBELL_RANGE) :
+ SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VPE1_DOORBELL_RANGE);
u32 doorbell_range = RREG32_PCIE_PORT(reg);
if (use_doorbell) {
@@ -112,7 +114,10 @@ static void nbio_v7_11_vcn_doorbell_range(struct amdgpu_device *adev,
bool use_doorbell,
int doorbell_index, int instance)
{
- u32 reg = SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE);
+ u32 reg = instance == 0 ?
+ SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN0_DOORBELL_RANGE):
+ SOC15_REG_OFFSET(NBIO, 0, regGDC0_BIF_VCN1_DOORBELL_RANGE);
+
u32 doorbell_range = RREG32_PCIE_PORT(reg);
if (use_doorbell) {
diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
index e90f33780803..b4723d68eab0 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
@@ -431,6 +431,12 @@ static void nbio_v7_9_init_registers(struct amdgpu_device *adev)
u32 inst_mask;
int i;
+ if (amdgpu_sriov_vf(adev))
+ adev->rmmio_remap.reg_offset =
+ SOC15_REG_OFFSET(
+ NBIO, 0,
+ regBIF_BX_DEV0_EPF0_VF0_HDP_MEM_COHERENCY_FLUSH_CNTL)
+ << 2;
WREG32_SOC15(NBIO, 0, regXCC_DOORBELL_FENCE,
0xff & ~(adev->gfx.xcc_mask));
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
index 4bb5e10217bb..7566973ed8f5 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
@@ -296,6 +296,7 @@ enum psp_gfx_fw_type {
GFX_FW_TYPE_VPEC_FW1 = 100, /* VPEC FW1 To Save VPE */
GFX_FW_TYPE_VPEC_FW2 = 101, /* VPEC FW2 To Save VPE */
GFX_FW_TYPE_VPE = 102,
+ GFX_FW_TYPE_JPEG_RAM = 128, /**< JPEG Command buffer */
GFX_FW_TYPE_P2S_TABLE = 129,
GFX_FW_TYPE_MAX
};
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
index efa37e3b7931..2395f1856962 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
@@ -506,7 +506,7 @@ static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops)
* before training, and restore it after training to avoid
* VRAM corruption.
*/
- sz = GDDR6_MEM_TRAINING_ENCROACHED_SIZE;
+ sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
DRM_ERROR("visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
index df1844d0800f..0da50ea46eaf 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
@@ -27,6 +27,7 @@
#include "amdgpu_ucode.h"
#include "soc15_common.h"
#include "psp_v13_0.h"
+#include "amdgpu_ras.h"
#include "mp/mp_13_0_2_offset.h"
#include "mp/mp_13_0_2_sh_mask.h"
@@ -52,6 +53,8 @@ MODULE_FIRMWARE("amdgpu/psp_13_0_6_sos.bin");
MODULE_FIRMWARE("amdgpu/psp_13_0_6_ta.bin");
MODULE_FIRMWARE("amdgpu/psp_14_0_0_toc.bin");
MODULE_FIRMWARE("amdgpu/psp_14_0_0_ta.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_1_toc.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_1_ta.bin");
/* For large FW files the time to complete can be very long */
#define USBC_PD_POLLING_LIMIT_S 240
@@ -100,6 +103,7 @@ static int psp_v13_0_init_microcode(struct psp_context *psp)
case IP_VERSION(13, 0, 8):
case IP_VERSION(13, 0, 11):
case IP_VERSION(14, 0, 0):
+ case IP_VERSION(14, 0, 1):
err = psp_init_toc_microcode(psp, ucode_prefix);
if (err)
return err;
@@ -187,11 +191,18 @@ static int psp_v13_0_wait_for_bootloader(struct psp_context *psp)
static int psp_v13_0_wait_for_bootloader_steady_state(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
+ int ret;
if (amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6)) {
- psp_v13_0_wait_for_vmbx_ready(psp);
+ ret = psp_v13_0_wait_for_vmbx_ready(psp);
+ if (ret)
+ amdgpu_ras_query_boot_status(adev, 4);
+
+ ret = psp_v13_0_wait_for_bootloader(psp);
+ if (ret)
+ amdgpu_ras_query_boot_status(adev, 4);
- return psp_v13_0_wait_for_bootloader(psp);
+ return ret;
}
return 0;
@@ -553,7 +564,7 @@ static int psp_v13_0_memory_training(struct psp_context *psp, uint32_t ops)
* before training, and restore it after training to avoid
* VRAM corruption.
*/
- sz = GDDR6_MEM_TRAINING_ENCROACHED_SIZE;
+ sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
dev_err(adev->dev, "visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
@@ -763,81 +774,28 @@ static int psp_v13_0_fatal_error_recovery_quirk(struct psp_context *psp)
return 0;
}
-
-static void psp_v13_0_boot_error_reporting(struct amdgpu_device *adev,
- uint32_t inst,
- uint32_t boot_error)
-{
- uint32_t socket_id;
- uint32_t aid_id;
- uint32_t hbm_id;
- uint32_t reg_data;
-
- socket_id = REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, SOCKET_ID);
- aid_id = REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, AID_ID);
- hbm_id = REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, HBM_ID);
-
- reg_data = RREG32_SOC15(MP0, inst, regMP0_SMN_C2PMSG_109);
- dev_info(adev->dev, "socket: %d, aid: %d, firmware boot failed, fw status is 0x%x\n",
- socket_id, aid_id, reg_data);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_MEM_TRAINING))
- dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, memory training failed\n",
- socket_id, aid_id, hbm_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_FW_LOAD))
- dev_info(adev->dev, "socket: %d, aid: %d, firmware load failed at boot time\n",
- socket_id, aid_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_WAFL_LINK_TRAINING))
- dev_info(adev->dev, "socket: %d, aid: %d, wafl link training failed\n",
- socket_id, aid_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_XGMI_LINK_TRAINING))
- dev_info(adev->dev, "socket: %d, aid: %d, xgmi link training failed\n",
- socket_id, aid_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_USR_CP_LINK_TRAINING))
- dev_info(adev->dev, "socket: %d, aid: %d, usr cp link training failed\n",
- socket_id, aid_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_USR_DP_LINK_TRAINING))
- dev_info(adev->dev, "socket: %d, aid: %d, usr dp link training failed\n",
- socket_id, aid_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_HBM_MEM_TEST))
- dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm memory test failed\n",
- socket_id, aid_id, hbm_id);
-
- if (REG_GET_FIELD(boot_error, MP0_SMN_C2PMSG_126, GPU_ERR_HBM_BIST_TEST))
- dev_info(adev->dev, "socket: %d, aid: %d, hbm: %d, hbm bist test failed\n",
- socket_id, aid_id, hbm_id);
-}
-
-static int psp_v13_0_query_boot_status(struct psp_context *psp)
+static bool psp_v13_0_get_ras_capability(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
- int inst_mask = adev->aid_mask;
- uint32_t reg_data;
- uint32_t i;
- int ret = 0;
+ struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+ u32 reg_data;
- if (amdgpu_ip_version(adev, MP0_HWIP, 0) != IP_VERSION(13, 0, 6))
- return 0;
+ /* query ras cap should be done from host side */
+ if (amdgpu_sriov_vf(adev))
+ return false;
- if (RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_59) < 0x00a10109)
- return 0;
+ if (!con)
+ return false;
- for_each_inst(i, inst_mask) {
- reg_data = RREG32_SOC15(MP0, i, regMP0_SMN_C2PMSG_126);
- if (!REG_GET_FIELD(reg_data, MP0_SMN_C2PMSG_126, BOOT_STATUS)) {
- psp_v13_0_boot_error_reporting(adev, i, reg_data);
- ret = -EINVAL;
- break;
- }
+ if ((amdgpu_ip_version(adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6)) &&
+ (!(adev->flags & AMD_IS_APU))) {
+ reg_data = RREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_127);
+ adev->ras_hw_enabled = (reg_data & GENMASK_ULL(23, 0));
+ con->poison_supported = ((reg_data & GENMASK_ULL(24, 24)) >> 24) ? true : false;
+ return true;
+ } else {
+ return false;
}
-
- return ret;
}
static const struct psp_funcs psp_v13_0_funcs = {
@@ -862,7 +820,7 @@ static const struct psp_funcs psp_v13_0_funcs = {
.update_spirom = psp_v13_0_update_spirom,
.vbflash_stat = psp_v13_0_vbflash_status,
.fatal_error_recovery_quirk = psp_v13_0_fatal_error_recovery_quirk,
- .query_boot_status = psp_v13_0_query_boot_status,
+ .get_ras_capability = psp_v13_0_get_ras_capability,
};
void psp_v13_0_set_psp_funcs(struct psp_context *psp)
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c
new file mode 100644
index 000000000000..78a95f8f370b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c
@@ -0,0 +1,672 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#include <drm/drm_drv.h>
+#include <linux/vmalloc.h>
+#include "amdgpu.h"
+#include "amdgpu_psp.h"
+#include "amdgpu_ucode.h"
+#include "soc15_common.h"
+#include "psp_v14_0.h"
+
+#include "mp/mp_14_0_2_offset.h"
+#include "mp/mp_14_0_2_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/psp_14_0_2_sos.bin");
+MODULE_FIRMWARE("amdgpu/psp_14_0_3_sos.bin");
+
+/* For large FW files the time to complete can be very long */
+#define USBC_PD_POLLING_LIMIT_S 240
+
+/* Read USB-PD from LFB */
+#define GFX_CMD_USB_PD_USE_LFB 0x480
+
+/* VBIOS gfl defines */
+#define MBOX_READY_MASK 0x80000000
+#define MBOX_STATUS_MASK 0x0000FFFF
+#define MBOX_COMMAND_MASK 0x00FF0000
+#define MBOX_READY_FLAG 0x80000000
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO 0x2
+#define C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI 0x3
+#define C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE 0x4
+
+/* memory training timeout define */
+#define MEM_TRAIN_SEND_MSG_TIMEOUT_US 3000000
+
+static int psp_v14_0_init_microcode(struct psp_context *psp)
+{
+ struct amdgpu_device *adev = psp->adev;
+ char ucode_prefix[30];
+ int err = 0;
+
+ amdgpu_ucode_ip_version_decode(adev, MP0_HWIP, ucode_prefix, sizeof(ucode_prefix));
+
+ switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) {
+ case IP_VERSION(14, 0, 2):
+ case IP_VERSION(14, 0, 3):
+ err = psp_init_sos_microcode(psp, ucode_prefix);
+ if (err)
+ return err;
+ break;
+ default:
+ BUG();
+ }
+
+ return 0;
+}
+
+static bool psp_v14_0_is_sos_alive(struct psp_context *psp)
+{
+ struct amdgpu_device *adev = psp->adev;
+ uint32_t sol_reg;
+
+ sol_reg = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_81);
+
+ return sol_reg != 0x0;
+}
+
+static int psp_v14_0_wait_for_bootloader(struct psp_context *psp)
+{
+ struct amdgpu_device *adev = psp->adev;
+
+ int ret;
+ int retry_loop;
+
+ for (retry_loop = 0; retry_loop < 10; retry_loop++) {
+ /* Wait for bootloader to signify that is
+ ready having bit 31 of C2PMSG_35 set to 1 */
+ ret = psp_wait_for(psp,
+ SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+ 0x80000000,
+ 0x80000000,
+ false);
+
+ if (ret == 0)
+ return 0;
+ }
+
+ return ret;
+}
+
+static int psp_v14_0_bootloader_load_component(struct psp_context *psp,
+ struct psp_bin_desc *bin_desc,
+ enum psp_bootloader_cmd bl_cmd)
+{
+ int ret;
+ uint32_t psp_gfxdrv_command_reg = 0;
+ struct amdgpu_device *adev = psp->adev;
+
+ /* Check tOS sign of life register to confirm sys driver and sOS
+ * are already been loaded.
+ */
+ if (psp_v14_0_is_sos_alive(psp))
+ return 0;
+
+ ret = psp_v14_0_wait_for_bootloader(psp);
+ if (ret)
+ return ret;
+
+ memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+ /* Copy PSP KDB binary to memory */
+ memcpy(psp->fw_pri_buf, bin_desc->start_addr, bin_desc->size_bytes);
+
+ /* Provide the PSP KDB to bootloader */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36,
+ (uint32_t)(psp->fw_pri_mc_addr >> 20));
+ psp_gfxdrv_command_reg = bl_cmd;
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35,
+ psp_gfxdrv_command_reg);
+
+ ret = psp_v14_0_wait_for_bootloader(psp);
+
+ return ret;
+}
+
+static int psp_v14_0_bootloader_load_kdb(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_KEY_DATABASE);
+}
+
+static int psp_v14_0_bootloader_load_spl(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->kdb, PSP_BL__LOAD_TOS_SPL_TABLE);
+}
+
+static int psp_v14_0_bootloader_load_sysdrv(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->sys, PSP_BL__LOAD_SYSDRV);
+}
+
+static int psp_v14_0_bootloader_load_soc_drv(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->soc_drv, PSP_BL__LOAD_SOCDRV);
+}
+
+static int psp_v14_0_bootloader_load_intf_drv(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->intf_drv, PSP_BL__LOAD_INTFDRV);
+}
+
+static int psp_v14_0_bootloader_load_dbg_drv(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->dbg_drv, PSP_BL__LOAD_DBGDRV);
+}
+
+static int psp_v14_0_bootloader_load_ras_drv(struct psp_context *psp)
+{
+ return psp_v14_0_bootloader_load_component(psp, &psp->ras_drv, PSP_BL__LOAD_RASDRV);
+}
+
+
+static int psp_v14_0_bootloader_load_sos(struct psp_context *psp)
+{
+ int ret;
+ unsigned int psp_gfxdrv_command_reg = 0;
+ struct amdgpu_device *adev = psp->adev;
+
+ /* Check sOS sign of life register to confirm sys driver and sOS
+ * are already been loaded.
+ */
+ if (psp_v14_0_is_sos_alive(psp))
+ return 0;
+
+ ret = psp_v14_0_wait_for_bootloader(psp);
+ if (ret)
+ return ret;
+
+ memset(psp->fw_pri_buf, 0, PSP_1_MEG);
+
+ /* Copy Secure OS binary to PSP memory */
+ memcpy(psp->fw_pri_buf, psp->sos.start_addr, psp->sos.size_bytes);
+
+ /* Provide the PSP secure OS to bootloader */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36,
+ (uint32_t)(psp->fw_pri_mc_addr >> 20));
+ psp_gfxdrv_command_reg = PSP_BL__LOAD_SOSDRV;
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35,
+ psp_gfxdrv_command_reg);
+
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_81),
+ RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_81),
+ 0, true);
+
+ return ret;
+}
+
+static int psp_v14_0_ring_stop(struct psp_context *psp,
+ enum psp_ring_type ring_type)
+{
+ int ret = 0;
+ struct amdgpu_device *adev = psp->adev;
+
+ if (amdgpu_sriov_vf(adev)) {
+ /* Write the ring destroy command*/
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+ GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+ /* Wait for response flag (bit 31) */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101),
+ 0x80000000, 0x80000000, false);
+ } else {
+ /* Write the ring destroy command*/
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_64,
+ GFX_CTRL_CMD_ID_DESTROY_RINGS);
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+ /* Wait for response flag (bit 31) */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+ 0x80000000, 0x80000000, false);
+ }
+
+ return ret;
+}
+
+static int psp_v14_0_ring_create(struct psp_context *psp,
+ enum psp_ring_type ring_type)
+{
+ int ret = 0;
+ unsigned int psp_ring_reg = 0;
+ struct psp_ring *ring = &psp->km_ring;
+ struct amdgpu_device *adev = psp->adev;
+
+ if (amdgpu_sriov_vf(adev)) {
+ ret = psp_v14_0_ring_stop(psp, ring_type);
+ if (ret) {
+ DRM_ERROR("psp_v14_0_ring_stop_sriov failed!\n");
+ return ret;
+ }
+
+ /* Write low address of the ring to C2PMSG_102 */
+ psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102, psp_ring_reg);
+ /* Write high address of the ring to C2PMSG_103 */
+ psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_103, psp_ring_reg);
+
+ /* Write the ring initialization command to C2PMSG_101 */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+ GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+
+ /* Wait for response flag (bit 31) in C2PMSG_101 */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_101),
+ 0x80000000, 0x8000FFFF, false);
+
+ } else {
+ /* Wait for sOS ready for ring creation */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+ 0x80000000, 0x80000000, false);
+ if (ret) {
+ DRM_ERROR("Failed to wait for trust OS ready for ring creation\n");
+ return ret;
+ }
+
+ /* Write low address of the ring to C2PMSG_69 */
+ psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_69, psp_ring_reg);
+ /* Write high address of the ring to C2PMSG_70 */
+ psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_70, psp_ring_reg);
+ /* Write size of ring to C2PMSG_71 */
+ psp_ring_reg = ring->ring_size;
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_71, psp_ring_reg);
+ /* Write the ring initialization command to C2PMSG_64 */
+ psp_ring_reg = ring_type;
+ psp_ring_reg = psp_ring_reg << 16;
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_64, psp_ring_reg);
+
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+
+ /* Wait for response flag (bit 31) in C2PMSG_64 */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_64),
+ 0x80000000, 0x8000FFFF, false);
+ }
+
+ return ret;
+}
+
+static int psp_v14_0_ring_destroy(struct psp_context *psp,
+ enum psp_ring_type ring_type)
+{
+ int ret = 0;
+ struct psp_ring *ring = &psp->km_ring;
+ struct amdgpu_device *adev = psp->adev;
+
+ ret = psp_v14_0_ring_stop(psp, ring_type);
+ if (ret)
+ DRM_ERROR("Fail to stop psp ring\n");
+
+ amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+ &ring->ring_mem_mc_addr,
+ (void **)&ring->ring_mem);
+
+ return ret;
+}
+
+static uint32_t psp_v14_0_ring_get_wptr(struct psp_context *psp)
+{
+ uint32_t data;
+ struct amdgpu_device *adev = psp->adev;
+
+ if (amdgpu_sriov_vf(adev))
+ data = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102);
+ else
+ data = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_67);
+
+ return data;
+}
+
+static void psp_v14_0_ring_set_wptr(struct psp_context *psp, uint32_t value)
+{
+ struct amdgpu_device *adev = psp->adev;
+
+ if (amdgpu_sriov_vf(adev)) {
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_102, value);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_101,
+ GFX_CTRL_CMD_ID_CONSUME_CMD);
+ } else
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_67, value);
+}
+
+static int psp_v14_0_memory_training_send_msg(struct psp_context *psp, int msg)
+{
+ int ret;
+ int i;
+ uint32_t data_32;
+ int max_wait;
+ struct amdgpu_device *adev = psp->adev;
+
+ data_32 = (psp->mem_train_ctx.c2p_train_data_offset >> 20);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36, data_32);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, msg);
+
+ max_wait = MEM_TRAIN_SEND_MSG_TIMEOUT_US / adev->usec_timeout;
+ for (i = 0; i < max_wait; i++) {
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+ 0x80000000, 0x80000000, false);
+ if (ret == 0)
+ break;
+ }
+ if (i < max_wait)
+ ret = 0;
+ else
+ ret = -ETIME;
+
+ dev_dbg(adev->dev, "training %s %s, cost %d @ %d ms\n",
+ (msg == PSP_BL__DRAM_SHORT_TRAIN) ? "short" : "long",
+ (ret == 0) ? "succeed" : "failed",
+ i, adev->usec_timeout/1000);
+ return ret;
+}
+
+
+static int psp_v14_0_memory_training(struct psp_context *psp, uint32_t ops)
+{
+ struct psp_memory_training_context *ctx = &psp->mem_train_ctx;
+ uint32_t *pcache = (uint32_t *)ctx->sys_cache;
+ struct amdgpu_device *adev = psp->adev;
+ uint32_t p2c_header[4];
+ uint32_t sz;
+ void *buf;
+ int ret, idx;
+
+ if (ctx->init == PSP_MEM_TRAIN_NOT_SUPPORT) {
+ dev_dbg(adev->dev, "Memory training is not supported.\n");
+ return 0;
+ } else if (ctx->init != PSP_MEM_TRAIN_INIT_SUCCESS) {
+ dev_err(adev->dev, "Memory training initialization failure.\n");
+ return -EINVAL;
+ }
+
+ if (psp_v14_0_is_sos_alive(psp)) {
+ dev_dbg(adev->dev, "SOS is alive, skip memory training.\n");
+ return 0;
+ }
+
+ amdgpu_device_vram_access(adev, ctx->p2c_train_data_offset, p2c_header, sizeof(p2c_header), false);
+ dev_dbg(adev->dev, "sys_cache[%08x,%08x,%08x,%08x] p2c_header[%08x,%08x,%08x,%08x]\n",
+ pcache[0], pcache[1], pcache[2], pcache[3],
+ p2c_header[0], p2c_header[1], p2c_header[2], p2c_header[3]);
+
+ if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+ dev_dbg(adev->dev, "Short training depends on restore.\n");
+ ops |= PSP_MEM_TRAIN_RESTORE;
+ }
+
+ if ((ops & PSP_MEM_TRAIN_RESTORE) &&
+ pcache[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+ dev_dbg(adev->dev, "sys_cache[0] is invalid, restore depends on save.\n");
+ ops |= PSP_MEM_TRAIN_SAVE;
+ }
+
+ if (p2c_header[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+ !(pcache[0] == MEM_TRAIN_SYSTEM_SIGNATURE &&
+ pcache[3] == p2c_header[3])) {
+ dev_dbg(adev->dev, "sys_cache is invalid or out-of-date, need save training data to sys_cache.\n");
+ ops |= PSP_MEM_TRAIN_SAVE;
+ }
+
+ if ((ops & PSP_MEM_TRAIN_SAVE) &&
+ p2c_header[0] != MEM_TRAIN_SYSTEM_SIGNATURE) {
+ dev_dbg(adev->dev, "p2c_header[0] is invalid, save depends on long training.\n");
+ ops |= PSP_MEM_TRAIN_SEND_LONG_MSG;
+ }
+
+ if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+ ops &= ~PSP_MEM_TRAIN_SEND_SHORT_MSG;
+ ops |= PSP_MEM_TRAIN_SAVE;
+ }
+
+ dev_dbg(adev->dev, "Memory training ops:%x.\n", ops);
+
+ if (ops & PSP_MEM_TRAIN_SEND_LONG_MSG) {
+ /*
+ * Long training will encroach a certain amount on the bottom of VRAM;
+ * save the content from the bottom of VRAM to system memory
+ * before training, and restore it after training to avoid
+ * VRAM corruption.
+ */
+ sz = BIST_MEM_TRAINING_ENCROACHED_SIZE;
+
+ if (adev->gmc.visible_vram_size < sz || !adev->mman.aper_base_kaddr) {
+ dev_err(adev->dev, "visible_vram_size %llx or aper_base_kaddr %p is not initialized.\n",
+ adev->gmc.visible_vram_size,
+ adev->mman.aper_base_kaddr);
+ return -EINVAL;
+ }
+
+ buf = vmalloc(sz);
+ if (!buf) {
+ dev_err(adev->dev, "failed to allocate system memory.\n");
+ return -ENOMEM;
+ }
+
+ if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+ memcpy_fromio(buf, adev->mman.aper_base_kaddr, sz);
+ ret = psp_v14_0_memory_training_send_msg(psp, PSP_BL__DRAM_LONG_TRAIN);
+ if (ret) {
+ DRM_ERROR("Send long training msg failed.\n");
+ vfree(buf);
+ drm_dev_exit(idx);
+ return ret;
+ }
+
+ memcpy_toio(adev->mman.aper_base_kaddr, buf, sz);
+ adev->hdp.funcs->flush_hdp(adev, NULL);
+ vfree(buf);
+ drm_dev_exit(idx);
+ } else {
+ vfree(buf);
+ return -ENODEV;
+ }
+ }
+
+ if (ops & PSP_MEM_TRAIN_SAVE) {
+ amdgpu_device_vram_access(psp->adev, ctx->p2c_train_data_offset, ctx->sys_cache, ctx->train_data_size, false);
+ }
+
+ if (ops & PSP_MEM_TRAIN_RESTORE) {
+ amdgpu_device_vram_access(psp->adev, ctx->c2p_train_data_offset, ctx->sys_cache, ctx->train_data_size, true);
+ }
+
+ if (ops & PSP_MEM_TRAIN_SEND_SHORT_MSG) {
+ ret = psp_v14_0_memory_training_send_msg(psp, (amdgpu_force_long_training > 0) ?
+ PSP_BL__DRAM_LONG_TRAIN : PSP_BL__DRAM_SHORT_TRAIN);
+ if (ret) {
+ dev_err(adev->dev, "send training msg failed.\n");
+ return ret;
+ }
+ }
+ ctx->training_cnt++;
+ return 0;
+}
+
+static int psp_v14_0_load_usbc_pd_fw(struct psp_context *psp, uint64_t fw_pri_mc_addr)
+{
+ struct amdgpu_device *adev = psp->adev;
+ uint32_t reg_status;
+ int ret, i = 0;
+
+ /*
+ * LFB address which is aligned to 1MB address and has to be
+ * right-shifted by 20 so that LFB address can be passed on a 32-bit C2P
+ * register
+ */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36, (fw_pri_mc_addr >> 20));
+
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+ 0x80000000, 0x80000000, false);
+ if (ret)
+ return ret;
+
+ /* Fireup interrupt so PSP can pick up the address */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, (GFX_CMD_USB_PD_USE_LFB << 16));
+
+ /* FW load takes very long time */
+ do {
+ msleep(1000);
+ reg_status = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35);
+
+ if (reg_status & 0x80000000)
+ goto done;
+
+ } while (++i < USBC_PD_POLLING_LIMIT_S);
+
+ return -ETIME;
+done:
+
+ if ((reg_status & 0xFFFF) != 0) {
+ DRM_ERROR("Address load failed - MP0_SMN_C2PMSG_35.Bits [15:0] = %04x\n",
+ reg_status & 0xFFFF);
+ return -EIO;
+ }
+
+ return 0;
+}
+
+static int psp_v14_0_read_usbc_pd_fw(struct psp_context *psp, uint32_t *fw_ver)
+{
+ struct amdgpu_device *adev = psp->adev;
+ int ret;
+
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_35, C2PMSG_CMD_GFX_USB_PD_FW_VER);
+
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_35),
+ 0x80000000, 0x80000000, false);
+ if (!ret)
+ *fw_ver = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_36);
+
+ return ret;
+}
+
+static int psp_v14_0_exec_spi_cmd(struct psp_context *psp, int cmd)
+{
+ uint32_t reg_status = 0, reg_val = 0;
+ struct amdgpu_device *adev = psp->adev;
+ int ret;
+
+ /* clear MBX ready (MBOX_READY_MASK bit is 0) and set update command */
+ reg_val |= (cmd << 16);
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115, reg_val);
+
+ /* Ring the doorbell */
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_73, 1);
+
+ if (cmd == C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE)
+ ret = psp_wait_for_spirom_update(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+ MBOX_READY_FLAG, MBOX_READY_MASK, PSP_SPIROM_UPDATE_TIMEOUT);
+ else
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+ MBOX_READY_FLAG, MBOX_READY_MASK, false);
+
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+ MBOX_READY_FLAG, MBOX_READY_MASK, false);
+ if (ret) {
+ dev_err(adev->dev, "SPI cmd %x timed out, ret = %d", cmd, ret);
+ return ret;
+ }
+
+ reg_status = RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115);
+ if ((reg_status & 0xFFFF) != 0) {
+ dev_err(adev->dev, "SPI cmd %x failed, fail status = %04x\n",
+ cmd, reg_status & 0xFFFF);
+ return -EIO;
+ }
+
+ return 0;
+}
+
+static int psp_v14_0_update_spirom(struct psp_context *psp,
+ uint64_t fw_pri_mc_addr)
+{
+ struct amdgpu_device *adev = psp->adev;
+ int ret;
+
+ /* Confirm PSP is ready to start */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMPASP_SMN_C2PMSG_115),
+ MBOX_READY_FLAG, MBOX_READY_MASK, false);
+ if (ret) {
+ dev_err(adev->dev, "PSP Not ready to start processing, ret = %d", ret);
+ return ret;
+ }
+
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_116, lower_32_bits(fw_pri_mc_addr));
+
+ ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_LO);
+ if (ret)
+ return ret;
+
+ WREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_116, upper_32_bits(fw_pri_mc_addr));
+
+ ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_ROM_IMAGE_ADDR_HI);
+ if (ret)
+ return ret;
+
+ psp->vbflash_done = true;
+
+ ret = psp_v14_0_exec_spi_cmd(psp, C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
+static int psp_v14_0_vbflash_status(struct psp_context *psp)
+{
+ struct amdgpu_device *adev = psp->adev;
+
+ return RREG32_SOC15(MP0, 0, regMPASP_SMN_C2PMSG_115);
+}
+
+static const struct psp_funcs psp_v14_0_funcs = {
+ .init_microcode = psp_v14_0_init_microcode,
+ .bootloader_load_kdb = psp_v14_0_bootloader_load_kdb,
+ .bootloader_load_spl = psp_v14_0_bootloader_load_spl,
+ .bootloader_load_sysdrv = psp_v14_0_bootloader_load_sysdrv,
+ .bootloader_load_soc_drv = psp_v14_0_bootloader_load_soc_drv,
+ .bootloader_load_intf_drv = psp_v14_0_bootloader_load_intf_drv,
+ .bootloader_load_dbg_drv = psp_v14_0_bootloader_load_dbg_drv,
+ .bootloader_load_ras_drv = psp_v14_0_bootloader_load_ras_drv,
+ .bootloader_load_sos = psp_v14_0_bootloader_load_sos,
+ .ring_create = psp_v14_0_ring_create,
+ .ring_stop = psp_v14_0_ring_stop,
+ .ring_destroy = psp_v14_0_ring_destroy,
+ .ring_get_wptr = psp_v14_0_ring_get_wptr,
+ .ring_set_wptr = psp_v14_0_ring_set_wptr,
+ .mem_training = psp_v14_0_memory_training,
+ .load_usbc_pd_fw = psp_v14_0_load_usbc_pd_fw,
+ .read_usbc_pd_fw = psp_v14_0_read_usbc_pd_fw,
+ .update_spirom = psp_v14_0_update_spirom,
+ .vbflash_stat = psp_v14_0_vbflash_status
+};
+
+void psp_v14_0_set_psp_funcs(struct psp_context *psp)
+{
+ psp->funcs = &psp_v14_0_funcs;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h
new file mode 100644
index 000000000000..dd18ba2cfad5
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+#ifndef __PSP_V14_0_H__
+#define __PSP_V14_0_H__
+
+#include "amdgpu_psp.h"
+
+#define PSP_SPIROM_UPDATE_TIMEOUT 60000 /* 60s */
+
+void psp_v14_0_set_psp_funcs(struct psp_context *psp);
+
+#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index 8d5d86675a7f..07e19caf2bc1 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -57,22 +57,19 @@ static void sdma_v2_4_set_irq_funcs(struct amdgpu_device *adev);
MODULE_FIRMWARE("amdgpu/topaz_sdma.bin");
MODULE_FIRMWARE("amdgpu/topaz_sdma1.bin");
-static const u32 sdma_offsets[SDMA_MAX_INSTANCE] =
-{
+static const u32 sdma_offsets[SDMA_MAX_INSTANCE] = {
SDMA0_REGISTER_OFFSET,
SDMA1_REGISTER_OFFSET
};
-static const u32 golden_settings_iceland_a11[] =
-{
+static const u32 golden_settings_iceland_a11[] = {
mmSDMA0_CHICKEN_BITS, 0xfc910007, 0x00810007,
mmSDMA0_CLK_CTRL, 0xff000fff, 0x00000000,
mmSDMA1_CHICKEN_BITS, 0xfc910007, 0x00810007,
mmSDMA1_CLK_CTRL, 0xff000fff, 0x00000000,
};
-static const u32 iceland_mgcg_cgcg_init[] =
-{
+static const u32 iceland_mgcg_cgcg_init[] = {
mmSDMA0_CLK_CTRL, 0xff000ff0, 0x00000100,
mmSDMA1_CLK_CTRL, 0xff000ff0, 0x00000100
};
@@ -142,7 +139,8 @@ static int sdma_v2_4_init_microcode(struct amdgpu_device *adev)
case CHIP_TOPAZ:
chip_name = "topaz";
break;
- default: BUG();
+ default:
+ BUG();
}
for (i = 0; i < adev->sdma.num_instances; i++) {
@@ -1258,8 +1256,7 @@ static void sdma_v2_4_set_vm_pte_funcs(struct amdgpu_device *adev)
adev->vm_manager.vm_pte_num_scheds = adev->sdma.num_instances;
}
-const struct amdgpu_ip_block_version sdma_v2_4_ip_block =
-{
+const struct amdgpu_ip_block_version sdma_v2_4_ip_block = {
.type = AMD_IP_BLOCK_TYPE_SDMA,
.major = 2,
.minor = 4,
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
index 2d688dca26be..fec5a3d1c4bc 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -45,6 +45,8 @@
MODULE_FIRMWARE("amdgpu/sdma_4_4_2.bin");
+#define mmSMNAID_AID0_MCA_SMU 0x03b30400
+
#define WREG32_SDMA(instance, offset, value) \
WREG32(sdma_v4_4_2_get_reg_offset(adev, (instance), (offset)), value)
#define RREG32_SDMA(instance, offset) \
@@ -2204,9 +2206,79 @@ static const struct amdgpu_ras_block_hw_ops sdma_v4_4_2_ras_hw_ops = {
.reset_ras_error_count = sdma_v4_4_2_reset_ras_error_count,
};
+static int sdma_v4_4_2_aca_bank_generate_report(struct aca_handle *handle,
+ struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data)
+{
+ u64 status, misc0;
+ int ret;
+
+ status = bank->regs[ACA_REG_IDX_STATUS];
+ if ((type == ACA_ERROR_TYPE_UE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) ||
+ (type == ACA_ERROR_TYPE_CE &&
+ ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_CE)) {
+
+ ret = aca_bank_info_decode(bank, &report->info);
+ if (ret)
+ return ret;
+
+ misc0 = bank->regs[ACA_REG_IDX_MISC0];
+ report->count[type] = ACA_REG__MISC0__ERRCNT(misc0);
+ }
+
+ return 0;
+}
+
+/* CODE_SDMA0 - CODE_SDMA4, reference to smu driver if header file */
+static int sdma_v4_4_2_err_codes[] = { 33, 34, 35, 36 };
+
+static bool sdma_v4_4_2_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+ enum aca_error_type type, void *data)
+{
+ u32 instlo;
+
+ instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+ instlo &= GENMASK(31, 1);
+
+ if (instlo != mmSMNAID_AID0_MCA_SMU)
+ return false;
+
+ if (aca_bank_check_error_codes(handle->adev, bank,
+ sdma_v4_4_2_err_codes,
+ ARRAY_SIZE(sdma_v4_4_2_err_codes)))
+ return false;
+
+ return true;
+}
+
+static const struct aca_bank_ops sdma_v4_4_2_aca_bank_ops = {
+ .aca_bank_generate_report = sdma_v4_4_2_aca_bank_generate_report,
+ .aca_bank_is_valid = sdma_v4_4_2_aca_bank_is_valid,
+};
+
+static const struct aca_info sdma_v4_4_2_aca_info = {
+ .hwip = ACA_HWIP_TYPE_SMU,
+ .mask = ACA_ERROR_UE_MASK,
+ .bank_ops = &sdma_v4_4_2_aca_bank_ops,
+};
+
+static int sdma_v4_4_2_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+ int r;
+
+ r = amdgpu_sdma_ras_late_init(adev, ras_block);
+ if (r)
+ return r;
+
+ return amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__SDMA,
+ &sdma_v4_4_2_aca_info, NULL);
+}
+
static struct amdgpu_sdma_ras sdma_v4_4_2_ras = {
.ras_block = {
.hw_ops = &sdma_v4_4_2_ras_hw_ops,
+ .ras_late_init = sdma_v4_4_2_ras_late_init,
},
};
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
index 3c7ddd219de8..4874ded45653 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c
@@ -49,6 +49,7 @@ MODULE_FIRMWARE("amdgpu/sdma_6_0_1.bin");
MODULE_FIRMWARE("amdgpu/sdma_6_0_2.bin");
MODULE_FIRMWARE("amdgpu/sdma_6_0_3.bin");
MODULE_FIRMWARE("amdgpu/sdma_6_1_0.bin");
+MODULE_FIRMWARE("amdgpu/sdma_6_1_1.bin");
#define SDMA1_REG_OFFSET 0x600
#define SDMA0_HYP_DEC_REG_START 0x5880
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 15033efec2ba..c64c01e2944a 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -1298,10 +1298,32 @@ static int soc15_common_suspend(void *handle)
return soc15_common_hw_fini(adev);
}
+static bool soc15_need_reset_on_resume(struct amdgpu_device *adev)
+{
+ u32 sol_reg;
+
+ sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+
+ /* Will reset for the following suspend abort cases.
+ * 1) Only reset limit on APU side, dGPU hasn't checked yet.
+ * 2) S3 suspend abort and TOS already launched.
+ */
+ if (adev->flags & AMD_IS_APU && adev->in_s3 &&
+ !adev->suspend_complete &&
+ sol_reg)
+ return true;
+
+ return false;
+}
+
static int soc15_common_resume(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ if (soc15_need_reset_on_resume(adev)) {
+ dev_info(adev->dev, "S3 suspend abort case, let's reset ASIC.\n");
+ soc15_asic_reset(adev);
+ }
return soc15_common_hw_init(adev);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c
index 48c6efcdeac9..5f81c264e310 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc21.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc21.c
@@ -50,13 +50,13 @@ static const struct amd_ip_funcs soc21_common_ip_funcs;
/* SOC21 */
static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn0[] = {
{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
- {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+ {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)},
};
static const struct amdgpu_video_codec_info vcn_4_0_0_video_codecs_encode_array_vcn1[] = {
{codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 2304, 0)},
- {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 4096, 2304, 0)},
+ {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 0)},
};
static const struct amdgpu_video_codecs vcn_4_0_0_video_codecs_encode_vcn0 = {
@@ -711,6 +711,7 @@ static int soc21_common_early_init(void *handle)
AMD_CG_SUPPORT_BIF_MGCG |
AMD_CG_SUPPORT_BIF_LS;
adev->pg_flags = AMD_PG_SUPPORT_VCN_DPG |
+ AMD_PG_SUPPORT_JPEG_DPG |
AMD_PG_SUPPORT_VCN |
AMD_PG_SUPPORT_JPEG |
AMD_PG_SUPPORT_GFX_PG;
@@ -865,6 +866,7 @@ static int soc21_common_set_clockgating_state(void *handle,
case IP_VERSION(7, 7, 0):
case IP_VERSION(7, 7, 1):
case IP_VERSION(7, 11, 0):
+ case IP_VERSION(7, 11, 1):
adev->nbio.funcs->update_medium_grain_clock_gating(adev,
state == AMD_CG_STATE_GATE);
adev->nbio.funcs->update_medium_grain_light_sleep(adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
index 879bb7af297c..056d4df8fa1f 100644
--- a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
@@ -36,6 +36,9 @@ enum ras_command {
TA_RAS_COMMAND__ENABLE_FEATURES = 0,
TA_RAS_COMMAND__DISABLE_FEATURES,
TA_RAS_COMMAND__TRIGGER_ERROR,
+ TA_RAS_COMMAND__QUERY_BLOCK_INFO,
+ TA_RAS_COMMAND__QUERY_SUB_BLOCK_INFO,
+ TA_RAS_COMMAND__QUERY_ADDRESS,
};
enum ta_ras_status {
@@ -105,6 +108,11 @@ enum ta_ras_error_type {
TA_RAS_ERROR__POISON = 8,
};
+enum ta_ras_address_type {
+ TA_RAS_MCA_TO_PA,
+ TA_RAS_PA_TO_MCA,
+};
+
/* Input/output structures for RAS commands */
/**********************************************************/
@@ -133,12 +141,38 @@ struct ta_ras_init_flags {
uint8_t channel_dis_num;
};
+struct ta_ras_mca_addr {
+ uint64_t err_addr;
+ uint32_t ch_inst;
+ uint32_t umc_inst;
+ uint32_t node_inst;
+};
+
+struct ta_ras_phy_addr {
+ uint64_t pa;
+ uint32_t bank;
+ uint32_t channel_idx;
+};
+
+struct ta_ras_query_address_input {
+ enum ta_ras_address_type addr_type;
+ struct ta_ras_mca_addr ma;
+ struct ta_ras_phy_addr pa;
+};
+
struct ta_ras_output_flags {
uint8_t ras_init_success_flag;
uint8_t err_inject_switch_disable_flag;
uint8_t reg_access_failure_flag;
};
+struct ta_ras_query_address_output {
+ /* don't use the flags here */
+ struct ta_ras_output_flags flags;
+ struct ta_ras_mca_addr ma;
+ struct ta_ras_phy_addr pa;
+};
+
/* Common input structure for RAS callbacks */
/**********************************************************/
union ta_ras_cmd_input {
@@ -146,12 +180,14 @@ union ta_ras_cmd_input {
struct ta_ras_enable_features_input enable_features;
struct ta_ras_disable_features_input disable_features;
struct ta_ras_trigger_error_input trigger_error;
+ struct ta_ras_query_address_input address;
uint32_t reserve_pad[256];
};
union ta_ras_cmd_output {
struct ta_ras_output_flags flags;
+ struct ta_ras_query_address_output address;
uint32_t reserve_pad[256];
};
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
index 7458a218e89d..14ef7a24be7b 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c
@@ -89,12 +89,28 @@ static void umc_v12_0_reset_error_count(struct amdgpu_device *adev)
umc_v12_0_reset_error_count_per_channel, NULL);
}
+bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
+{
+ dev_info(adev->dev,
+ "MCA_UMC_STATUS(0x%llx): Val:%llu, Poison:%llu, Deferred:%llu, PCC:%llu, UC:%llu, TCC:%llu\n",
+ mc_umc_status,
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val),
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Poison),
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred),
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC),
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC),
+ REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, TCC)
+ );
+
+ return (amdgpu_ras_is_poison_mode_supported(adev) &&
+ (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
+ (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1));
+}
+
bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
{
- if (amdgpu_ras_is_poison_mode_supported(adev) &&
- (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
- (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1))
- return true;
+ if (umc_v12_0_is_deferred_error(adev, mc_umc_status))
+ return false;
return ((REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
(REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, PCC) == 1 ||
@@ -104,9 +120,7 @@ bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_um
bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status)
{
- if (amdgpu_ras_is_poison_mode_supported(adev) &&
- (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1) &&
- (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Deferred) == 1))
+ if (umc_v12_0_is_deferred_error(adev, mc_umc_status))
return false;
return (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
@@ -119,9 +133,10 @@ bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_
!(umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)))));
}
-static void umc_v12_0_query_correctable_error_count(struct amdgpu_device *adev,
+static void umc_v12_0_query_error_count_per_type(struct amdgpu_device *adev,
uint64_t umc_reg_offset,
- unsigned long *error_count)
+ unsigned long *error_count,
+ check_error_type_func error_type_func)
{
uint64_t mc_umc_status;
uint64_t mc_umc_status_addr;
@@ -129,31 +144,11 @@ static void umc_v12_0_query_correctable_error_count(struct amdgpu_device *adev,
mc_umc_status_addr =
SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
- /* Rely on MCUMC_STATUS for correctable error counter
- * MCUMC_STATUS is a 64 bit register
- */
+ /* Check MCUMC_STATUS */
mc_umc_status =
RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4);
- if (umc_v12_0_is_correctable_error(adev, mc_umc_status))
- *error_count += 1;
-}
-
-static void umc_v12_0_query_uncorrectable_error_count(struct amdgpu_device *adev,
- uint64_t umc_reg_offset,
- unsigned long *error_count)
-{
- uint64_t mc_umc_status;
- uint64_t mc_umc_status_addr;
-
- mc_umc_status_addr =
- SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_STATUST0);
-
- /* Check the MCUMC_STATUS. */
- mc_umc_status =
- RREG64_PCIE_EXT((mc_umc_status_addr + umc_reg_offset) * 4);
-
- if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status))
+ if (error_type_func(adev, mc_umc_status))
*error_count += 1;
}
@@ -162,7 +157,7 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev,
uint32_t ch_inst, void *data)
{
struct ras_err_data *err_data = (struct ras_err_data *)data;
- unsigned long ue_count = 0, ce_count = 0;
+ unsigned long ue_count = 0, ce_count = 0, de_count = 0;
/* NOTE: node_inst is converted by adev->umc.active_mask and the range is [0-3],
* which can be used as die ID directly */
@@ -174,11 +169,16 @@ static int umc_v12_0_query_error_count(struct amdgpu_device *adev,
uint64_t umc_reg_offset =
get_umc_v12_0_reg_offset(adev, node_inst, umc_inst, ch_inst);
- umc_v12_0_query_correctable_error_count(adev, umc_reg_offset, &ce_count);
- umc_v12_0_query_uncorrectable_error_count(adev, umc_reg_offset, &ue_count);
+ umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+ &ce_count, umc_v12_0_is_correctable_error);
+ umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+ &ue_count, umc_v12_0_is_uncorrectable_error);
+ umc_v12_0_query_error_count_per_type(adev, umc_reg_offset,
+ &de_count, umc_v12_0_is_deferred_error);
amdgpu_ras_error_statistic_ue_count(err_data, &mcm_info, NULL, ue_count);
amdgpu_ras_error_statistic_ce_count(err_data, &mcm_info, NULL, ce_count);
+ amdgpu_ras_error_statistic_de_count(err_data, &mcm_info, NULL, de_count);
return 0;
}
@@ -203,14 +203,14 @@ static bool umc_v12_0_bit_wise_xor(uint32_t val)
return result;
}
-static void umc_v12_0_convert_error_address(struct amdgpu_device *adev,
- struct ras_err_data *err_data, uint64_t err_addr,
- uint32_t ch_inst, uint32_t umc_inst,
- uint32_t node_inst)
+static void umc_v12_0_mca_addr_to_pa(struct amdgpu_device *adev,
+ uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst,
+ uint32_t node_inst,
+ struct ta_ras_query_address_output *addr_out)
{
uint32_t channel_index, i;
- uint64_t soc_pa, na, retired_page, column;
- uint32_t bank_hash0, bank_hash1, bank_hash2, bank_hash3, col, row, row_xor;
+ uint64_t na, soc_pa;
+ uint32_t bank_hash0, bank_hash1, bank_hash2, bank_hash3, col, row;
uint32_t bank0, bank1, bank2, bank3, bank;
bank_hash0 = (err_addr >> UMC_V12_0_MCA_B0_BIT) & 0x1ULL;
@@ -260,12 +260,44 @@ static void umc_v12_0_convert_error_address(struct amdgpu_device *adev,
/* the umc channel bits are not original values, they are hashed */
UMC_V12_0_SET_CHANNEL_HASH(channel_index, soc_pa);
+ addr_out->pa.pa = soc_pa;
+ addr_out->pa.bank = bank;
+ addr_out->pa.channel_idx = channel_index;
+}
+
+static void umc_v12_0_convert_error_address(struct amdgpu_device *adev,
+ struct ras_err_data *err_data, uint64_t err_addr,
+ uint32_t ch_inst, uint32_t umc_inst,
+ uint32_t node_inst)
+{
+ uint32_t col, row, row_xor, bank, channel_index;
+ uint64_t soc_pa, retired_page, column;
+ struct ta_ras_query_address_input addr_in;
+ struct ta_ras_query_address_output addr_out;
+
+ addr_in.addr_type = TA_RAS_MCA_TO_PA;
+ addr_in.ma.err_addr = err_addr;
+ addr_in.ma.ch_inst = ch_inst;
+ addr_in.ma.umc_inst = umc_inst;
+ addr_in.ma.node_inst = node_inst;
+
+ if (psp_ras_query_address(&adev->psp, &addr_in, &addr_out))
+ /* fallback to old path if fail to get pa from psp */
+ umc_v12_0_mca_addr_to_pa(adev, err_addr, ch_inst, umc_inst,
+ node_inst, &addr_out);
+
+ soc_pa = addr_out.pa.pa;
+ bank = addr_out.pa.bank;
+ channel_index = addr_out.pa.channel_idx;
+
+ col = (err_addr >> 1) & 0x1fULL;
+ row = (err_addr >> 10) & 0x3fffULL;
+ row_xor = row ^ (0x1ULL << 13);
/* clear [C3 C2] in soc physical address */
soc_pa &= ~(0x3ULL << UMC_V12_0_PA_C2_BIT);
/* clear [C4] in soc physical address */
soc_pa &= ~(0x1ULL << UMC_V12_0_PA_C4_BIT);
- row_xor = row ^ (0x1ULL << 13);
/* loop for all possibilities of [C4 C3 C2] */
for (column = 0; column < UMC_V12_0_NA_MAP_PA_NUM; column++) {
retired_page = soc_pa | ((column & 0x3) << UMC_V12_0_PA_C2_BIT);
@@ -316,10 +348,7 @@ static int umc_v12_0_query_error_address(struct amdgpu_device *adev,
}
/* calculate error address if ue error is detected */
- if (REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, Val) == 1 &&
- REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, AddrV) == 1 &&
- REG_GET_FIELD(mc_umc_status, MCA_UMC_UMC0_MCUMC_STATUST0, UC) == 1) {
-
+ if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)) {
mc_umc_addrt0 =
SOC15_REG_OFFSET(UMC, 0, regMCA_UMC_UMC0_MCUMC_ADDRT0);
@@ -385,45 +414,69 @@ static void umc_v12_0_ecc_info_query_ras_error_address(struct amdgpu_device *ade
{
struct ras_err_node *err_node;
uint64_t mc_umc_status;
+ struct ras_err_info *err_info;
+ struct ras_err_addr *mca_err_addr, *tmp;
struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
for_each_ras_error(err_node, err_data) {
- mc_umc_status = err_node->err_info.err_addr.err_status;
- if (!mc_umc_status)
+ err_info = &err_node->err_info;
+ if (list_empty(&err_info->err_addr_list))
continue;
- if (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status)) {
- uint64_t mca_addr, err_addr, mca_ipid;
- uint32_t InstanceIdLo;
- struct amdgpu_smuio_mcm_config_info *mcm_info;
-
- mcm_info = &err_node->err_info.mcm_info;
- mca_addr = err_node->err_info.err_addr.err_addr;
- mca_ipid = err_node->err_info.err_addr.err_ipid;
-
- err_addr = REG_GET_FIELD(mca_addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
- InstanceIdLo = REG_GET_FIELD(mca_ipid, MCMP1_IPIDT0, InstanceIdLo);
-
- dev_info(adev->dev, "UMC:IPID:0x%llx, aid:%d, inst:%d, ch:%d, err_addr:0x%llx\n",
- mca_ipid,
- mcm_info->die_id,
- MCA_IPID_LO_2_UMC_INST(InstanceIdLo),
- MCA_IPID_LO_2_UMC_CH(InstanceIdLo),
- err_addr);
-
- umc_v12_0_convert_error_address(adev,
- err_data, err_addr,
- MCA_IPID_LO_2_UMC_CH(InstanceIdLo),
- MCA_IPID_LO_2_UMC_INST(InstanceIdLo),
- mcm_info->die_id);
-
- /* Clear umc error address content */
- memset(&err_node->err_info.err_addr,
- 0, sizeof(err_node->err_info.err_addr));
+ list_for_each_entry_safe(mca_err_addr, tmp, &err_info->err_addr_list, node) {
+ mc_umc_status = mca_err_addr->err_status;
+ if (mc_umc_status &&
+ (umc_v12_0_is_uncorrectable_error(adev, mc_umc_status) ||
+ umc_v12_0_is_deferred_error(adev, mc_umc_status))) {
+ uint64_t mca_addr, err_addr, mca_ipid;
+ uint32_t InstanceIdLo;
+
+ mca_addr = mca_err_addr->err_addr;
+ mca_ipid = mca_err_addr->err_ipid;
+
+ err_addr = REG_GET_FIELD(mca_addr,
+ MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr);
+ InstanceIdLo = REG_GET_FIELD(mca_ipid, MCMP1_IPIDT0, InstanceIdLo);
+
+ dev_info(adev->dev, "UMC:IPID:0x%llx, aid:%d, inst:%d, ch:%d, err_addr:0x%llx\n",
+ mca_ipid,
+ err_info->mcm_info.die_id,
+ MCA_IPID_LO_2_UMC_INST(InstanceIdLo),
+ MCA_IPID_LO_2_UMC_CH(InstanceIdLo),
+ err_addr);
+
+ umc_v12_0_convert_error_address(adev,
+ err_data, err_addr,
+ MCA_IPID_LO_2_UMC_CH(InstanceIdLo),
+ MCA_IPID_LO_2_UMC_INST(InstanceIdLo),
+ err_info->mcm_info.die_id);
+ }
+
+ /* Delete error address node from list and free memory */
+ amdgpu_ras_del_mca_err_addr(err_info, mca_err_addr);
}
}
}
+static bool umc_v12_0_check_ecc_err_status(struct amdgpu_device *adev,
+ enum amdgpu_mca_error_type type, void *ras_error_status)
+{
+ uint64_t mc_umc_status = *(uint64_t *)ras_error_status;
+
+ switch (type) {
+ case AMDGPU_MCA_ERROR_TYPE_UE:
+ return umc_v12_0_is_uncorrectable_error(adev, mc_umc_status);
+ case AMDGPU_MCA_ERROR_TYPE_CE:
+ return umc_v12_0_is_correctable_error(adev, mc_umc_status);
+ case AMDGPU_MCA_ERROR_TYPE_DE:
+ return umc_v12_0_is_deferred_error(adev, mc_umc_status);
+ default:
+ return false;
+ }
+
+ return false;
+}
+
static void umc_v12_0_err_cnt_init(struct amdgpu_device *adev)
{
amdgpu_umc_loop_channels(adev,
@@ -444,12 +497,71 @@ const struct amdgpu_ras_block_hw_ops umc_v12_0_ras_hw_ops = {
.query_ras_error_address = umc_v12_0_query_ras_error_address,
};
+static int umc_v12_0_aca_bank_generate_report(struct aca_handle *handle, struct aca_bank *bank, enum aca_error_type type,
+ struct aca_bank_report *report, void *data)
+{
+ struct amdgpu_device *adev = handle->adev;
+ u64 status;
+ int ret;
+
+ ret = aca_bank_info_decode(bank, &report->info);
+ if (ret)
+ return ret;
+
+ status = bank->regs[ACA_REG_IDX_STATUS];
+ switch (type) {
+ case ACA_ERROR_TYPE_UE:
+ if (umc_v12_0_is_uncorrectable_error(adev, status)) {
+ report->count[type] = 1;
+ }
+ break;
+ case ACA_ERROR_TYPE_CE:
+ if (umc_v12_0_is_correctable_error(adev, status)) {
+ report->count[type] = 1;
+ }
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static const struct aca_bank_ops umc_v12_0_aca_bank_ops = {
+ .aca_bank_generate_report = umc_v12_0_aca_bank_generate_report,
+};
+
+const struct aca_info umc_v12_0_aca_info = {
+ .hwip = ACA_HWIP_TYPE_UMC,
+ .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+ .bank_ops = &umc_v12_0_aca_bank_ops,
+};
+
+static int umc_v12_0_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
+{
+ int ret;
+
+ ret = amdgpu_umc_ras_late_init(adev, ras_block);
+ if (ret)
+ return ret;
+
+ ret = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__UMC,
+ &umc_v12_0_aca_info, NULL);
+ if (ret)
+ return ret;
+
+ return 0;
+}
+
struct amdgpu_umc_ras umc_v12_0_ras = {
.ras_block = {
.hw_ops = &umc_v12_0_ras_hw_ops,
+ .ras_late_init = umc_v12_0_ras_late_init,
},
.err_cnt_init = umc_v12_0_err_cnt_init,
.query_ras_poison_mode = umc_v12_0_query_ras_poison_mode,
.ecc_info_query_ras_error_count = umc_v12_0_ecc_info_query_ras_error_count,
.ecc_info_query_ras_error_address = umc_v12_0_ecc_info_query_ras_error_address,
+ .check_ecc_err_status = umc_v12_0_check_ecc_err_status,
};
+
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h
index e8de3a92251a..5973bfb14fce 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.h
@@ -121,9 +121,12 @@
(((_ipid_lo) >> 12) & 0xF))
#define MCA_IPID_LO_2_UMC_INST(_ipid_lo) (((_ipid_lo) >> 21) & 0x7)
+bool umc_v12_0_is_deferred_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
bool umc_v12_0_is_uncorrectable_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
bool umc_v12_0_is_correctable_error(struct amdgpu_device *adev, uint64_t mc_umc_status);
+typedef bool (*check_error_type_func)(struct amdgpu_device *adev, uint64_t mc_umc_status);
+
extern const uint32_t
umc_v12_0_channel_idx_tbl[]
[UMC_V12_0_UMC_INSTANCE_NUM]
diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c
index 0d6b50528d76..97fa88ed770c 100644
--- a/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_0.c
@@ -25,7 +25,7 @@
static void umc_v6_0_init_registers(struct amdgpu_device *adev)
{
- unsigned i,j;
+ unsigned i, j;
for (i = 0; i < 4; i++)
for (j = 0; j < 4; j++)
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c
new file mode 100644
index 000000000000..d6ee9958ba5f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c
@@ -0,0 +1,1339 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_vcn.h"
+#include "amdgpu_pm.h"
+#include "soc15.h"
+#include "soc15d.h"
+#include "soc15_hw_ip.h"
+#include "vcn_v2_0.h"
+
+#include "vcn/vcn_5_0_0_offset.h"
+#include "vcn/vcn_5_0_0_sh_mask.h"
+#include "ivsrcid/vcn/irqsrcs_vcn_4_0.h"
+#include "vcn_v5_0_0.h"
+
+#include <drm/drm_drv.h>
+
+static int amdgpu_ih_clientid_vcns[] = {
+ SOC15_IH_CLIENTID_VCN,
+ SOC15_IH_CLIENTID_VCN1
+};
+
+static void vcn_v5_0_0_set_unified_ring_funcs(struct amdgpu_device *adev);
+static void vcn_v5_0_0_set_irq_funcs(struct amdgpu_device *adev);
+static int vcn_v5_0_0_set_powergating_state(void *handle,
+ enum amd_powergating_state state);
+static int vcn_v5_0_0_pause_dpg_mode(struct amdgpu_device *adev,
+ int inst_idx, struct dpg_pause_state *new_state);
+static void vcn_v5_0_0_unified_ring_set_wptr(struct amdgpu_ring *ring);
+
+/**
+ * vcn_v5_0_0_early_init - set function pointers and load microcode
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Set ring and irq function pointers
+ * Load microcode from filesystem
+ */
+static int vcn_v5_0_0_early_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ /* re-use enc ring as unified ring */
+ adev->vcn.num_enc_rings = 1;
+
+ vcn_v5_0_0_set_unified_ring_funcs(adev);
+ vcn_v5_0_0_set_irq_funcs(adev);
+
+ return amdgpu_vcn_early_init(adev);
+}
+
+/**
+ * vcn_v5_0_0_sw_init - sw init for VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Load firmware and sw initialization
+ */
+static int vcn_v5_0_0_sw_init(void *handle)
+{
+ struct amdgpu_ring *ring;
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int i, r;
+
+ r = amdgpu_vcn_sw_init(adev);
+ if (r)
+ return r;
+
+ amdgpu_vcn_setup_ucode(adev);
+
+ r = amdgpu_vcn_resume(adev);
+ if (r)
+ return r;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+ volatile struct amdgpu_vcn4_fw_shared *fw_shared;
+
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ atomic_set(&adev->vcn.inst[i].sched_score, 0);
+
+ /* VCN UNIFIED TRAP */
+ r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+ VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE, &adev->vcn.inst[i].irq);
+ if (r)
+ return r;
+
+ /* VCN POISON TRAP */
+ r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
+ VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst[i].irq);
+ if (r)
+ return r;
+
+ ring = &adev->vcn.inst[i].ring_enc[0];
+ ring->use_doorbell = true;
+ ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + 8 * i;
+
+ ring->vm_hub = AMDGPU_MMHUB0(0);
+ sprintf(ring->name, "vcn_unified_%d", i);
+
+ r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst[i].irq, 0,
+ AMDGPU_RING_PRIO_0, &adev->vcn.inst[i].sched_score);
+ if (r)
+ return r;
+
+ fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+ fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE);
+ fw_shared->sq.is_enabled = 1;
+
+ if (amdgpu_vcnfw_log)
+ amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]);
+ }
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+ adev->vcn.pause_dpg_mode = vcn_v5_0_0_pause_dpg_mode;
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_sw_fini - sw fini for VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * VCN suspend and free up sw allocation
+ */
+static int vcn_v5_0_0_sw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int i, r, idx;
+
+ if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+ for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
+ volatile struct amdgpu_vcn4_fw_shared *fw_shared;
+
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+ fw_shared->present_flag_0 = 0;
+ fw_shared->sq.is_enabled = 0;
+ }
+
+ drm_dev_exit(idx);
+ }
+
+ r = amdgpu_vcn_suspend(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_vcn_sw_fini(adev);
+
+ return r;
+}
+
+/**
+ * vcn_v5_0_0_hw_init - start and test VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Initialize the hardware, boot up the VCPU and do some testing
+ */
+static int vcn_v5_0_0_hw_init(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ struct amdgpu_ring *ring;
+ int i, r;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ ring = &adev->vcn.inst[i].ring_enc[0];
+
+ adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
+ ((adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 8 * i), i);
+
+ r = amdgpu_ring_test_helper(ring);
+ if (r)
+ goto done;
+ }
+
+done:
+ if (!r)
+ DRM_INFO("VCN decode and encode initialized successfully(under %s).\n",
+ (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)?"DPG Mode":"SPG Mode");
+
+ return r;
+}
+
+/**
+ * vcn_v5_0_0_hw_fini - stop the hardware block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Stop the VCN block, mark ring as not ready any more
+ */
+static int vcn_v5_0_0_hw_fini(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int i;
+
+ cancel_delayed_work_sync(&adev->vcn.idle_work);
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ amdgpu_irq_put(adev, &adev->vcn.inst[i].irq, 0);
+ }
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_suspend - suspend VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * HW fini and suspend VCN block
+ */
+static int vcn_v5_0_0_suspend(void *handle)
+{
+ int r;
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ r = vcn_v5_0_0_hw_fini(adev);
+ if (r)
+ return r;
+
+ r = amdgpu_vcn_suspend(adev);
+
+ return r;
+}
+
+/**
+ * vcn_v5_0_0_resume - resume VCN block
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Resume firmware and hw init VCN block
+ */
+static int vcn_v5_0_0_resume(void *handle)
+{
+ int r;
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+ r = amdgpu_vcn_resume(adev);
+ if (r)
+ return r;
+
+ r = vcn_v5_0_0_hw_init(adev);
+
+ return r;
+}
+
+/**
+ * vcn_v5_0_0_mc_resume - memory controller programming
+ *
+ * @adev: amdgpu_device pointer
+ * @inst: instance number
+ *
+ * Let the VCN memory controller know it's offsets
+ */
+static void vcn_v5_0_0_mc_resume(struct amdgpu_device *adev, int inst)
+{
+ uint32_t offset, size;
+ const struct common_firmware_header *hdr;
+
+ hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
+ size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+ /* cache window 0: fw */
+ if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+ (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_lo));
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+ (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst].tmr_mc_addr_hi));
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, 0);
+ offset = 0;
+ } else {
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
+ lower_32_bits(adev->vcn.inst[inst].gpu_addr));
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH,
+ upper_32_bits(adev->vcn.inst[inst].gpu_addr));
+ offset = size;
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET0, AMDGPU_UVD_FIRMWARE_OFFSET >> 3);
+ }
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE0, size);
+
+ /* cache window 1: stack */
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW,
+ lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH,
+ upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset));
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET1, 0);
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE1, AMDGPU_VCN_STACK_SIZE);
+
+ /* cache window 2: context */
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW,
+ lower_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH,
+ upper_32_bits(adev->vcn.inst[inst].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE));
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_OFFSET2, 0);
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+
+ /* non-cache window */
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+ lower_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+ WREG32_SOC15(VCN, inst, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+ upper_32_bits(adev->vcn.inst[inst].fw_shared.gpu_addr));
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_OFFSET0, 0);
+ WREG32_SOC15(VCN, inst, regUVD_VCPU_NONCACHE_SIZE0,
+ AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)));
+}
+
+/**
+ * vcn_v5_0_0_mc_resume_dpg_mode - memory controller programming for dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
+ *
+ * Let the VCN memory controller know it's offsets with dpg mode
+ */
+static void vcn_v5_0_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect)
+{
+ uint32_t offset, size;
+ const struct common_firmware_header *hdr;
+
+ hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
+ size = AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8);
+
+ /* cache window 0: fw */
+ if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+ if (!indirect) {
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+ (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_lo), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+ (adev->firmware.ucode[AMDGPU_UCODE_ID_VCN + inst_idx].tmr_mc_addr_hi), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+ } else {
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
+ }
+ offset = 0;
+ } else {
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr), 0, indirect);
+ offset = size;
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET0),
+ AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
+ }
+
+ if (!indirect)
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
+ else
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
+
+ /* cache window 1: stack */
+ if (!indirect) {
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+ } else {
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
+ }
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
+
+ /* cache window 2: context */
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst[inst_idx].gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
+
+ /* non-cache window */
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst[inst_idx].fw_shared.gpu_addr), 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_NONCACHE_SIZE0),
+ AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_vcn4_fw_shared)), 0, indirect);
+
+ /* VCN global tiling registers */
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, 0, regUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
+
+ return;
+}
+
+/**
+ * vcn_v5_0_0_disable_static_power_gating - disable VCN static power gating
+ *
+ * @adev: amdgpu_device pointer
+ * @inst: instance number
+ *
+ * Disable static power gating for VCN block
+ */
+static void vcn_v5_0_0_disable_static_power_gating(struct amdgpu_device *adev, int inst)
+{
+ uint32_t data = 0;
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+ } else {
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+ data = 1 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS, 0,
+ UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+ }
+
+ data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+ data &= ~0x103;
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN)
+ data |= UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON |
+ UVD_POWER_STATUS__UVD_PG_EN_MASK;
+
+ WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+ return;
+}
+
+/**
+ * vcn_v5_0_0_enable_static_power_gating - enable VCN static power gating
+ *
+ * @adev: amdgpu_device pointer
+ * @inst: instance number
+ *
+ * Enable static power gating for VCN block
+ */
+static void vcn_v5_0_0_enable_static_power_gating(struct amdgpu_device *adev, int inst)
+{
+ uint32_t data;
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
+ /* Before power off, this indicator has to be turned on */
+ data = RREG32_SOC15(VCN, inst, regUVD_POWER_STATUS);
+ data &= ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK;
+ data |= UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF;
+ WREG32_SOC15(VCN, inst, regUVD_POWER_STATUS, data);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO5_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO5_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO4_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO4_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO3_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO3_PWR_STATUS_MASK);
+
+ data = 2 << UVD_IPX_DLDO_CONFIG__ONO2_PWR_CONFIG__SHIFT;
+ WREG32_SOC15(VCN, inst, regUVD_IPX_DLDO_CONFIG, data);
+ SOC15_WAIT_ON_RREG(VCN, inst, regUVD_IPX_DLDO_STATUS,
+ 1 << UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS__SHIFT,
+ UVD_IPX_DLDO_STATUS__ONO2_PWR_STATUS_MASK);
+ }
+ return;
+}
+
+/**
+ * vcn_v5_0_0_disable_clock_gating - disable VCN clock gating
+ *
+ * @adev: amdgpu_device pointer
+ * @inst: instance number
+ *
+ * Disable clock gating for VCN block
+ */
+static void vcn_v5_0_0_disable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+ return;
+}
+
+#if 0
+/**
+ * vcn_v5_0_0_disable_clock_gating_dpg_mode - disable VCN clock gating dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @sram_sel: sram select
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
+ *
+ * Disable clock gating for VCN block with dpg mode
+ */
+static void vcn_v5_0_0_disable_clock_gating_dpg_mode(struct amdgpu_device *adev, uint8_t sram_sel,
+ int inst_idx, uint8_t indirect)
+{
+ return;
+}
+#endif
+
+/**
+ * vcn_v5_0_0_enable_clock_gating - enable VCN clock gating
+ *
+ * @adev: amdgpu_device pointer
+ * @inst: instance number
+ *
+ * Enable clock gating for VCN block
+ */
+static void vcn_v5_0_0_enable_clock_gating(struct amdgpu_device *adev, int inst)
+{
+ return;
+}
+
+/**
+ * vcn_v5_0_0_start_dpg_mode - VCN start with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @indirect: indirectly write sram
+ *
+ * Start VCN block with dpg mode
+ */
+static int vcn_v5_0_0_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, bool indirect)
+{
+ volatile struct amdgpu_vcn4_fw_shared *fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr;
+ struct amdgpu_ring *ring;
+ uint32_t tmp;
+
+ /* disable register anti-hang mechanism */
+ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 1,
+ ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+ /* enable dynamic power gating mode */
+ tmp = RREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS);
+ tmp |= UVD_POWER_STATUS__UVD_PG_MODE_MASK;
+ tmp |= UVD_POWER_STATUS__UVD_PG_EN_MASK;
+ WREG32_SOC15(VCN, inst_idx, regUVD_POWER_STATUS, tmp);
+
+ if (indirect)
+ adev->vcn.inst[inst_idx].dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst[inst_idx].dpg_sram_cpu_addr;
+
+ /* enable VCPU clock */
+ tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+ tmp |= UVD_VCPU_CNTL__CLK_EN_MASK | UVD_VCPU_CNTL__BLK_RST_MASK;
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+ /* disable master interrupt */
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_MASTINT_EN), 0, 0, indirect);
+
+ /* setup regUVD_LMI_CTRL */
+ tmp = (UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+ UVD_LMI_CTRL__REQ_MODE_MASK |
+ UVD_LMI_CTRL__CRC_RESET_MASK |
+ UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+ UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+ UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
+ (8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
+ 0x00100000L);
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_CTRL), tmp, 0, indirect);
+
+ vcn_v5_0_0_mc_resume_dpg_mode(adev, inst_idx, indirect);
+
+ tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
+ tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_VCPU_CNTL), tmp, 0, indirect);
+
+ /* enable LMI MC and UMC channels */
+ tmp = 0x1f << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT;
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_LMI_CTRL2), tmp, 0, indirect);
+
+ /* enable master interrupt */
+ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET(
+ VCN, inst_idx, regUVD_MASTINT_EN),
+ UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
+
+ if (indirect)
+ amdgpu_vcn_psp_update_sram(adev, inst_idx, 0);
+
+ ring = &adev->vcn.inst[inst_idx].ring_enc[0];
+
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_LO, ring->gpu_addr);
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_SIZE, ring->ring_size / 4);
+
+ tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+ tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+ WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+ fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR, 0);
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, 0);
+
+ tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_RPTR);
+ WREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR, tmp);
+ ring->wptr = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+
+ tmp = RREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE);
+ tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+ WREG32_SOC15(VCN, inst_idx, regVCN_RB_ENABLE, tmp);
+ fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+
+ WREG32_SOC15(VCN, inst_idx, regVCN_RB1_DB_CTRL,
+ ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+ VCN_RB1_DB_CTRL__EN_MASK);
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_start - VCN start
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Start VCN block
+ */
+static int vcn_v5_0_0_start(struct amdgpu_device *adev)
+{
+ volatile struct amdgpu_vcn4_fw_shared *fw_shared;
+ struct amdgpu_ring *ring;
+ uint32_t tmp;
+ int i, j, k, r;
+
+ if (adev->pm.dpm_enabled)
+ amdgpu_dpm_enable_uvd(adev, true);
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+ r = vcn_v5_0_0_start_dpg_mode(adev, i, adev->vcn.indirect_sram);
+ continue;
+ }
+
+ /* disable VCN power gating */
+ vcn_v5_0_0_disable_static_power_gating(adev, i);
+
+ /* set VCN status busy */
+ tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY;
+ WREG32_SOC15(VCN, i, regUVD_STATUS, tmp);
+
+ /* enable VCPU clock */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+ UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK);
+
+ /* disable master interrupt */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0,
+ ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+ /* enable LMI MC and UMC channels */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0,
+ ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK);
+
+ tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+ tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+ tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+ WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+ /* setup regUVD_LMI_CTRL */
+ tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL);
+ WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp |
+ UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK |
+ UVD_LMI_CTRL__MASK_MC_URGENT_MASK |
+ UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK |
+ UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK);
+
+ vcn_v5_0_0_mc_resume(adev, i);
+
+ /* VCN global tiling registers */
+ WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG,
+ adev->gfx.config.gb_addr_config);
+
+ /* unblock VCPU register access */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0,
+ ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+ /* release VCPU reset to boot */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+ ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+ for (j = 0; j < 10; ++j) {
+ uint32_t status;
+
+ for (k = 0; k < 100; ++k) {
+ status = RREG32_SOC15(VCN, i, regUVD_STATUS);
+ if (status & 2)
+ break;
+ mdelay(10);
+ if (amdgpu_emu_mode == 1)
+ msleep(1);
+ }
+
+ if (amdgpu_emu_mode == 1) {
+ r = -1;
+ if (status & 2) {
+ r = 0;
+ break;
+ }
+ } else {
+ r = 0;
+ if (status & 2)
+ break;
+
+ dev_err(adev->dev,
+ "VCN[%d] is not responding, trying to reset the VCPU!!!\n", i);
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+ UVD_VCPU_CNTL__BLK_RST_MASK,
+ ~UVD_VCPU_CNTL__BLK_RST_MASK);
+ mdelay(10);
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+ ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+ mdelay(10);
+ r = -1;
+ }
+ }
+
+ if (r) {
+ dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i);
+ return r;
+ }
+
+ /* enable master interrupt */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN),
+ UVD_MASTINT_EN__VCPU_EN_MASK,
+ ~UVD_MASTINT_EN__VCPU_EN_MASK);
+
+ /* clear the busy bit of VCN_STATUS */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0,
+ ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT));
+
+ ring = &adev->vcn.inst[i].ring_enc[0];
+ WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL,
+ ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT |
+ VCN_RB1_DB_CTRL__EN_MASK);
+
+ WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr);
+ WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
+ WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4);
+
+ tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+ tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK);
+ WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+ fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET;
+ WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0);
+ WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0);
+
+ tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR);
+ WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp);
+ ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR);
+
+ tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE);
+ tmp |= VCN_RB_ENABLE__RB1_EN_MASK;
+ WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp);
+ fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF);
+ }
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_stop_dpg_mode - VCN stop with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ *
+ * Stop VCN block with dpg mode
+ */
+static void vcn_v5_0_0_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx)
+{
+ struct dpg_pause_state state = {.fw_based = VCN_DPG_STATE__UNPAUSE};
+ uint32_t tmp;
+
+ vcn_v5_0_0_pause_dpg_mode(adev, inst_idx, &state);
+
+ /* Wait for power status to be 1 */
+ SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 1,
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+ /* wait for read ptr to be equal to write ptr */
+ tmp = RREG32_SOC15(VCN, inst_idx, regUVD_RB_WPTR);
+ SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_RB_RPTR, tmp, 0xFFFFFFFF);
+
+ /* disable dynamic power gating mode */
+ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0,
+ ~UVD_POWER_STATUS__UVD_PG_MODE_MASK);
+
+ return;
+}
+
+/**
+ * vcn_v5_0_0_stop - VCN stop
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Stop VCN block
+ */
+static int vcn_v5_0_0_stop(struct amdgpu_device *adev)
+{
+ volatile struct amdgpu_vcn4_fw_shared *fw_shared;
+ uint32_t tmp;
+ int i, r = 0;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr;
+ fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF;
+
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
+ vcn_v5_0_0_stop_dpg_mode(adev, i);
+ continue;
+ }
+
+ /* wait for vcn idle */
+ r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7);
+ if (r)
+ return r;
+
+ tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK |
+ UVD_LMI_STATUS__READ_CLEAN_MASK |
+ UVD_LMI_STATUS__WRITE_CLEAN_MASK |
+ UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
+ r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+ if (r)
+ return r;
+
+ /* disable LMI UMC channel */
+ tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2);
+ tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK;
+ WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp);
+ tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK |
+ UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
+ r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp);
+ if (r)
+ return r;
+
+ /* block VCPU register access */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL),
+ UVD_RB_ARB_CTRL__VCPU_DIS_MASK,
+ ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK);
+
+ /* reset VCPU */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL),
+ UVD_VCPU_CNTL__BLK_RST_MASK,
+ ~UVD_VCPU_CNTL__BLK_RST_MASK);
+
+ /* disable VCPU clock */
+ WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0,
+ ~(UVD_VCPU_CNTL__CLK_EN_MASK));
+
+ /* apply soft reset */
+ tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+ tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK;
+ WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+ tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET);
+ tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK;
+ WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp);
+
+ /* clear status */
+ WREG32_SOC15(VCN, i, regUVD_STATUS, 0);
+
+ /* enable VCN power gating */
+ vcn_v5_0_0_enable_static_power_gating(adev, i);
+ }
+
+ if (adev->pm.dpm_enabled)
+ amdgpu_dpm_enable_uvd(adev, false);
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_pause_dpg_mode - VCN pause with dpg mode
+ *
+ * @adev: amdgpu_device pointer
+ * @inst_idx: instance number index
+ * @new_state: pause state
+ *
+ * Pause dpg mode for VCN block
+ */
+static int vcn_v5_0_0_pause_dpg_mode(struct amdgpu_device *adev, int inst_idx,
+ struct dpg_pause_state *new_state)
+{
+ uint32_t reg_data = 0;
+ int ret_code;
+
+ /* pause/unpause if state is changed */
+ if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
+ DRM_DEV_DEBUG(adev->dev, "dpg pause state changed %d -> %d",
+ adev->vcn.inst[inst_idx].pause_state.fw_based, new_state->fw_based);
+ reg_data = RREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE) &
+ (~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+
+ if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
+ ret_code = SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_POWER_STATUS, 0x1,
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
+
+ if (!ret_code) {
+ /* pause DPG */
+ reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+ WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+
+ /* wait for ACK */
+ SOC15_WAIT_ON_RREG(VCN, inst_idx, regUVD_DPG_PAUSE,
+ UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
+ UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+ }
+ } else {
+ /* unpause dpg, no need to wait */
+ reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
+ WREG32_SOC15(VCN, inst_idx, regUVD_DPG_PAUSE, reg_data);
+ }
+ adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
+ }
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_get_rptr - get unified read pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified read pointer
+ */
+static uint64_t vcn_v5_0_0_unified_ring_get_rptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+ DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+ return RREG32_SOC15(VCN, ring->me, regUVD_RB_RPTR);
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_get_wptr - get unified write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Returns the current hardware unified write pointer
+ */
+static uint64_t vcn_v5_0_0_unified_ring_get_wptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+ DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+ if (ring->use_doorbell)
+ return *ring->wptr_cpu_addr;
+ else
+ return RREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR);
+}
+
+/**
+ * vcn_v5_0_0_unified_ring_set_wptr - set enc write pointer
+ *
+ * @ring: amdgpu_ring pointer
+ *
+ * Commits the enc write pointer to the hardware
+ */
+static void vcn_v5_0_0_unified_ring_set_wptr(struct amdgpu_ring *ring)
+{
+ struct amdgpu_device *adev = ring->adev;
+
+ if (ring != &adev->vcn.inst[ring->me].ring_enc[0])
+ DRM_ERROR("wrong ring id is identified in %s", __func__);
+
+ if (ring->use_doorbell) {
+ *ring->wptr_cpu_addr = lower_32_bits(ring->wptr);
+ WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
+ } else {
+ WREG32_SOC15(VCN, ring->me, regUVD_RB_WPTR, lower_32_bits(ring->wptr));
+ }
+}
+
+static const struct amdgpu_ring_funcs vcn_v5_0_0_unified_ring_vm_funcs = {
+ .type = AMDGPU_RING_TYPE_VCN_ENC,
+ .align_mask = 0x3f,
+ .nop = VCN_ENC_CMD_NO_OP,
+ .get_rptr = vcn_v5_0_0_unified_ring_get_rptr,
+ .get_wptr = vcn_v5_0_0_unified_ring_get_wptr,
+ .set_wptr = vcn_v5_0_0_unified_ring_set_wptr,
+ .emit_frame_size =
+ SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 +
+ SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 +
+ 4 + /* vcn_v2_0_enc_ring_emit_vm_flush */
+ 5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */
+ 1, /* vcn_v2_0_enc_ring_insert_end */
+ .emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */
+ .emit_ib = vcn_v2_0_enc_ring_emit_ib,
+ .emit_fence = vcn_v2_0_enc_ring_emit_fence,
+ .emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush,
+ .test_ring = amdgpu_vcn_enc_ring_test_ring,
+ .test_ib = amdgpu_vcn_unified_ring_test_ib,
+ .insert_nop = amdgpu_ring_insert_nop,
+ .insert_end = vcn_v2_0_enc_ring_insert_end,
+ .pad_ib = amdgpu_ring_generic_pad_ib,
+ .begin_use = amdgpu_vcn_ring_begin_use,
+ .end_use = amdgpu_vcn_ring_end_use,
+ .emit_wreg = vcn_v2_0_enc_ring_emit_wreg,
+ .emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait,
+ .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
+};
+
+/**
+ * vcn_v5_0_0_set_unified_ring_funcs - set unified ring functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set unified ring functions
+ */
+static void vcn_v5_0_0_set_unified_ring_funcs(struct amdgpu_device *adev)
+{
+ int i;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ adev->vcn.inst[i].ring_enc[0].funcs = &vcn_v5_0_0_unified_ring_vm_funcs;
+ adev->vcn.inst[i].ring_enc[0].me = i;
+
+ DRM_INFO("VCN(%d) encode/decode are enabled in VM mode\n", i);
+ }
+}
+
+/**
+ * vcn_v5_0_0_is_idle - check VCN block is idle
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Check whether VCN block is idle
+ */
+static bool vcn_v5_0_0_is_idle(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int i, ret = 1;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ ret &= (RREG32_SOC15(VCN, i, regUVD_STATUS) == UVD_STATUS__IDLE);
+ }
+
+ return ret;
+}
+
+/**
+ * vcn_v5_0_0_wait_for_idle - wait for VCN block idle
+ *
+ * @handle: amdgpu_device pointer
+ *
+ * Wait for VCN block idle
+ */
+static int vcn_v5_0_0_wait_for_idle(void *handle)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int i, ret = 0;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ ret = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE,
+ UVD_STATUS__IDLE);
+ if (ret)
+ return ret;
+ }
+
+ return ret;
+}
+
+/**
+ * vcn_v5_0_0_set_clockgating_state - set VCN block clockgating state
+ *
+ * @handle: amdgpu_device pointer
+ * @state: clock gating state
+ *
+ * Set VCN block clockgating state
+ */
+static int vcn_v5_0_0_set_clockgating_state(void *handle, enum amd_clockgating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+ int i;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ if (enable) {
+ if (RREG32_SOC15(VCN, i, regUVD_STATUS) != UVD_STATUS__IDLE)
+ return -EBUSY;
+ vcn_v5_0_0_enable_clock_gating(adev, i);
+ } else {
+ vcn_v5_0_0_disable_clock_gating(adev, i);
+ }
+ }
+
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_set_powergating_state - set VCN block powergating state
+ *
+ * @handle: amdgpu_device pointer
+ * @state: power gating state
+ *
+ * Set VCN block powergating state
+ */
+static int vcn_v5_0_0_set_powergating_state(void *handle, enum amd_powergating_state state)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ int ret;
+
+ if (state == adev->vcn.cur_state)
+ return 0;
+
+ if (state == AMD_PG_STATE_GATE)
+ ret = vcn_v5_0_0_stop(adev);
+ else
+ ret = vcn_v5_0_0_start(adev);
+
+ if (!ret)
+ adev->vcn.cur_state = state;
+
+ return ret;
+}
+
+/**
+ * vcn_v5_0_0_set_interrupt_state - set VCN block interrupt state
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @type: interrupt types
+ * @state: interrupt states
+ *
+ * Set VCN block interrupt state
+ */
+static int vcn_v5_0_0_set_interrupt_state(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+ unsigned type, enum amdgpu_interrupt_state state)
+{
+ return 0;
+}
+
+/**
+ * vcn_v5_0_0_process_interrupt - process VCN block interrupt
+ *
+ * @adev: amdgpu_device pointer
+ * @source: interrupt sources
+ * @entry: interrupt entry from clients and sources
+ *
+ * Process VCN block interrupt
+ */
+static int vcn_v5_0_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source,
+ struct amdgpu_iv_entry *entry)
+{
+ uint32_t ip_instance;
+
+ switch (entry->client_id) {
+ case SOC15_IH_CLIENTID_VCN:
+ ip_instance = 0;
+ break;
+ case SOC15_IH_CLIENTID_VCN1:
+ ip_instance = 1;
+ break;
+ default:
+ DRM_ERROR("Unhandled client id: %d\n", entry->client_id);
+ return 0;
+ }
+
+ DRM_DEBUG("IH: VCN TRAP\n");
+
+ switch (entry->src_id) {
+ case VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE:
+ amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]);
+ break;
+ case VCN_4_0__SRCID_UVD_POISON:
+ amdgpu_vcn_process_poison_irq(adev, source, entry);
+ break;
+ default:
+ DRM_ERROR("Unhandled interrupt: %d %d\n",
+ entry->src_id, entry->src_data[0]);
+ break;
+ }
+
+ return 0;
+}
+
+static const struct amdgpu_irq_src_funcs vcn_v5_0_0_irq_funcs = {
+ .set = vcn_v5_0_0_set_interrupt_state,
+ .process = vcn_v5_0_0_process_interrupt,
+};
+
+/**
+ * vcn_v5_0_0_set_irq_funcs - set VCN block interrupt irq functions
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Set VCN block interrupt irq functions
+ */
+static void vcn_v5_0_0_set_irq_funcs(struct amdgpu_device *adev)
+{
+ int i;
+
+ for (i = 0; i < adev->vcn.num_vcn_inst; ++i) {
+ if (adev->vcn.harvest_config & (1 << i))
+ continue;
+
+ adev->vcn.inst[i].irq.num_types = adev->vcn.num_enc_rings + 1;
+ adev->vcn.inst[i].irq.funcs = &vcn_v5_0_0_irq_funcs;
+ }
+}
+
+static const struct amd_ip_funcs vcn_v5_0_0_ip_funcs = {
+ .name = "vcn_v5_0_0",
+ .early_init = vcn_v5_0_0_early_init,
+ .late_init = NULL,
+ .sw_init = vcn_v5_0_0_sw_init,
+ .sw_fini = vcn_v5_0_0_sw_fini,
+ .hw_init = vcn_v5_0_0_hw_init,
+ .hw_fini = vcn_v5_0_0_hw_fini,
+ .suspend = vcn_v5_0_0_suspend,
+ .resume = vcn_v5_0_0_resume,
+ .is_idle = vcn_v5_0_0_is_idle,
+ .wait_for_idle = vcn_v5_0_0_wait_for_idle,
+ .check_soft_reset = NULL,
+ .pre_soft_reset = NULL,
+ .soft_reset = NULL,
+ .post_soft_reset = NULL,
+ .set_clockgating_state = vcn_v5_0_0_set_clockgating_state,
+ .set_powergating_state = vcn_v5_0_0_set_powergating_state,
+};
+
+const struct amdgpu_ip_block_version vcn_v5_0_0_ip_block = {
+ .type = AMD_IP_BLOCK_TYPE_VCN,
+ .major = 5,
+ .minor = 0,
+ .rev = 0,
+ .funcs = &vcn_v5_0_0_ip_funcs,
+};
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h
new file mode 100644
index 000000000000..51bbccd4360f
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright 2023 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __VCN_V5_0_0_H__
+#define __VCN_V5_0_0_H__
+
+#define VCN_VID_SOC_ADDRESS 0x1FC00
+#define VCN_AON_SOC_ADDRESS 0x1F800
+#define VCN1_VID_SOC_ADDRESS 0x48300
+#define VCN1_AON_SOC_ADDRESS 0x48000
+
+#define VCN_VID_IP_ADDRESS 0x0
+#define VCN_AON_IP_ADDRESS 0x30000
+
+extern const struct amdgpu_ip_block_version vcn_v5_0_0_ip_block;
+
+#endif /* __VCN_V5_0_0_H__ */