From 5278ca048d93eac74e9a81b3e672da2b2264bce4 Mon Sep 17 00:00:00 2001 From: Thomas Hellström Date: Tue, 23 Apr 2024 14:11:14 +0200 Subject: drm/xe: Fix unexpected backmerge results MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The recent backmerge from drm-next to drm-xe-next brought with it some silent unexpected results. One code snippet was added twice and a partial revert had merge errors. Fix that up to reinstate the affected code as it was before the backmerge. v2: - Commit log message rewording (Lucas DeMarchi) Fixes: 79790b6818e9 ("Merge drm/drm-next into drm-xe-next") Signed-off-by: Thomas Hellström Reviewed-by: Matthew Brost Reviewed-by: Lucas De Marchi Link: https://patchwork.freedesktop.org/patch/msgid/20240423121114.39325-1-thomas.hellstrom@linux.intel.com (cherry picked from commit 06e7139a034f26804904368fe4af2ceb70724756) Signed-off-by: Thomas Hellström --- drivers/gpu/drm/xe/xe_vm.c | 13 +++++++------ drivers/gpu/drm/xe/xe_vm_types.h | 4 ++++ 2 files changed, 11 insertions(+), 6 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 85d6f359142d..7ae2b0300db6 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -863,11 +863,6 @@ static void xe_vma_destroy_late(struct xe_vma *vma) vma->ufence = NULL; } - if (vma->ufence) { - xe_sync_ufence_put(vma->ufence); - vma->ufence = NULL; - } - if (xe_vma_is_userptr(vma)) { struct xe_userptr_vma *uvma = to_userptr_vma(vma); struct xe_userptr *userptr = &uvma->userptr; @@ -2100,6 +2095,10 @@ vm_bind_ioctl_ops_create(struct xe_vm *vm, struct xe_bo *bo, struct xe_vma_op *op = gpuva_op_to_vma_op(__op); if (__op->op == DRM_GPUVA_OP_MAP) { + op->map.immediate = + flags & DRM_XE_VM_BIND_FLAG_IMMEDIATE; + op->map.read_only = + flags & DRM_XE_VM_BIND_FLAG_READONLY; op->map.is_null = flags & DRM_XE_VM_BIND_FLAG_NULL; op->map.dumpable = flags & DRM_XE_VM_BIND_FLAG_DUMPABLE; op->map.pat_index = pat_index; @@ -2294,6 +2293,8 @@ static int vm_bind_ioctl_ops_parse(struct xe_vm *vm, struct xe_exec_queue *q, switch (op->base.op) { case DRM_GPUVA_OP_MAP: { + flags |= op->map.read_only ? + VMA_CREATE_FLAG_READ_ONLY : 0; flags |= op->map.is_null ? VMA_CREATE_FLAG_IS_NULL : 0; flags |= op->map.dumpable ? @@ -2438,7 +2439,7 @@ static int op_execute(struct drm_exec *exec, struct xe_vm *vm, case DRM_GPUVA_OP_MAP: err = xe_vm_bind(vm, vma, op->q, xe_vma_bo(vma), op->syncs, op->num_syncs, - !xe_vm_in_fault_mode(vm), + op->map.immediate || !xe_vm_in_fault_mode(vm), op->flags & XE_VMA_OP_FIRST, op->flags & XE_VMA_OP_LAST); break; diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h index 7570c2c6c463..72a100671e5d 100644 --- a/drivers/gpu/drm/xe/xe_vm_types.h +++ b/drivers/gpu/drm/xe/xe_vm_types.h @@ -269,6 +269,10 @@ struct xe_vm { struct xe_vma_op_map { /** @vma: VMA to map */ struct xe_vma *vma; + /** @immediate: Immediate bind */ + bool immediate; + /** @read_only: Read only */ + bool read_only; /** @is_null: is NULL binding */ bool is_null; /** @dumpable: whether BO is dumped on GPU hang */ -- cgit v1.2.3-58-ga151 From 28d21e3e66c5c33aae9f32d189af0508bb918fd0 Mon Sep 17 00:00:00 2001 From: Matthew Auld Date: Tue, 23 Apr 2024 08:47:23 +0100 Subject: drm/xe/vm: prevent UAF in rebind_work_func() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We flush the rebind worker during the vm close phase, however in places like preempt_fence_work_func() we seem to queue the rebind worker without first checking if the vm has already been closed. The concern here is the vm being closed with the worker flushed, but then being rearmed later, which looks like potential uaf, since there is no actual refcounting to track the queued worker. We can't take the vm->lock here in preempt_rebind_work_func() to first check if the vm is closed since that will deadlock, so instead flush the worker again when the vm refcount reaches zero. v2: - Grabbing vm->lock in the preempt worker creates a deadlock, so checking the closed state is tricky. Instead flush the worker when the refcount reaches zero. It should be impossible to queue the preempt worker without already holding vm ref. Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1676 Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1591 Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1364 Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1304 Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1249 Signed-off-by: Matthew Auld Cc: Matthew Brost Cc: # v6.8+ Reviewed-by: Matthew Brost Link: https://patchwork.freedesktop.org/patch/msgid/20240423074721.119633-4-matthew.auld@intel.com (cherry picked from commit 3d44d67c441a9fe6f81a1d705f7de009a32a5b35) Signed-off-by: Thomas Hellström --- drivers/gpu/drm/xe/xe_vm.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 7ae2b0300db6..4aa3943e6f29 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -1499,6 +1499,9 @@ static void xe_vm_free(struct drm_gpuvm *gpuvm) /* xe_vm_close_and_put was not called? */ xe_assert(xe, !vm->size); + if (xe_vm_in_preempt_fence_mode(vm)) + flush_work(&vm->preempt.rebind_work); + mutex_destroy(&vm->snap_mutex); if (!(vm->flags & XE_VM_FLAG_MIGRATION)) -- cgit v1.2.3-58-ga151 From 3bc8848bb7f7478e6806e4522b06b63f40a53e1e Mon Sep 17 00:00:00 2001 From: Lucas De Marchi Date: Sat, 27 Apr 2024 06:53:39 -0700 Subject: drm/xe: Merge 16021540221 and 18034896535 WAs MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit In order to detect duplicate implementations for the same workaround, early in the implementation of RTP it was decided to error out even if the values set are exactly the same. With the introduction of 18034896535 in commit 74671d23ca18 ("drm/xe/xe2: Add workaround 18034896535"), LNL stepping with graphics stepping A1 now gives the following error on module load: xe 0000:00:02.0: [drm] *ERROR* GT0: [GT OTHER] \ discarding save-restore reg e48c (clear: 00000200, set: 00000200,\ masked: yes, mcr: yes): ret=-22 RTP may be improved in the future, but for now simply join the entries like done with e.g. "1607297627, 1607030317, 1607186500". Fixes: 74671d23ca18 ("drm/xe/xe2: Add workaround 18034896535") Cc: Bommu Krishnaiah Cc: Tejas Upadhyay Cc: Matt Roper Reviewed-by: Himal Prasad Ghimiray Link: https://patchwork.freedesktop.org/patch/msgid/20240427135339.3485559-1-lucas.demarchi@intel.com Signed-off-by: Lucas De Marchi (cherry picked from commit 4caf410766add8cf376a3afc910b17dd0961dd75) Signed-off-by: Thomas Hellström --- drivers/gpu/drm/xe/xe_wa.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) (limited to 'drivers/gpu') diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c index dcf7ed51757c..dd214d95e4b6 100644 --- a/drivers/gpu/drm/xe/xe_wa.c +++ b/drivers/gpu/drm/xe/xe_wa.c @@ -449,12 +449,7 @@ static const struct xe_rtp_entry_sr engine_was[] = { XE_RTP_RULES(GRAPHICS_VERSION(2004), FUNC(xe_rtp_match_first_render_or_compute)), XE_RTP_ACTIONS(SET(ROW_CHICKEN3, XE2_EUPEND_CHK_FLUSH_DIS)) }, - { XE_RTP_NAME("16021540221"), - XE_RTP_RULES(GRAPHICS_VERSION(2004), GRAPHICS_STEP(A0, B0), - FUNC(xe_rtp_match_first_render_or_compute)), - XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH)) - }, - { XE_RTP_NAME("18034896535"), + { XE_RTP_NAME("18034896535, 16021540221"), /* 16021540221: GRAPHICS_STEP(A0, B0) */ XE_RTP_RULES(GRAPHICS_VERSION_RANGE(2001, 2004), FUNC(xe_rtp_match_first_render_or_compute)), XE_RTP_ACTIONS(SET(ROW_CHICKEN4, DISABLE_TDL_PUSH)) -- cgit v1.2.3-58-ga151