From c42e2f076769c9c1bc5f3f0aa1c2032558e76647 Mon Sep 17 00:00:00 2001 From: Evan Green Date: Fri, 9 Aug 2024 14:44:43 -0700 Subject: RISC-V: hwprobe: Add MISALIGNED_PERF key RISCV_HWPROBE_KEY_CPUPERF_0 was mistakenly flagged as a bitmask in hwprobe_key_is_bitmask(), when in reality it was an enum value. This causes problems when used in conjunction with RISCV_HWPROBE_WHICH_CPUS, since SLOW, FAST, and EMULATED have values whose bits overlap with each other. If the caller asked for the set of CPUs that was SLOW or EMULATED, the returned set would also include CPUs that were FAST. Introduce a new hwprobe key, RISCV_HWPROBE_KEY_MISALIGNED_PERF, which returns the same values in response to a direct query (with no flags), but is properly handled as an enumerated value. As a result, SLOW, FAST, and EMULATED are all correctly treated as distinct values under the new key when queried with the WHICH_CPUS flag. Leave the old key in place to avoid disturbing applications which may have already come to rely on the key, with or without its broken behavior with respect to the WHICH_CPUS flag. Fixes: e178bf146e4b ("RISC-V: hwprobe: Introduce which-cpus flag") Signed-off-by: Evan Green Reviewed-by: Charlie Jenkins Reviewed-by: Andrew Jones Link: https://lore.kernel.org/r/20240809214444.3257596-2-evan@rivosinc.com Signed-off-by: Palmer Dabbelt --- Documentation/arch/riscv/hwprobe.rst | 20 +++++++++++++------- arch/riscv/include/asm/hwprobe.h | 2 +- arch/riscv/include/uapi/asm/hwprobe.h | 1 + arch/riscv/kernel/sys_hwprobe.c | 1 + 4 files changed, 16 insertions(+), 8 deletions(-) diff --git a/Documentation/arch/riscv/hwprobe.rst b/Documentation/arch/riscv/hwprobe.rst index 3db60a0911df..a994eed75bde 100644 --- a/Documentation/arch/riscv/hwprobe.rst +++ b/Documentation/arch/riscv/hwprobe.rst @@ -239,8 +239,13 @@ The following keys are defined: ratified in commit 98918c844281 ("Merge pull request #1217 from riscv/zawrs") of riscv-isa-manual. -* :c:macro:`RISCV_HWPROBE_KEY_CPUPERF_0`: A bitmask that contains performance - information about the selected set of processors. +* :c:macro:`RISCV_HWPROBE_KEY_CPUPERF_0`: Deprecated. Returns similar values to + :c:macro:`RISCV_HWPROBE_KEY_MISALIGNED_SCALAR_PERF`, but the key was + mistakenly classified as a bitmask rather than a value. + +* :c:macro:`RISCV_HWPROBE_KEY_MISALIGNED_SCALAR_PERF`: An enum value describing + the performance of misaligned scalar native word accesses on the selected set + of processors. * :c:macro:`RISCV_HWPROBE_MISALIGNED_UNKNOWN`: The performance of misaligned accesses is unknown. @@ -249,12 +254,13 @@ The following keys are defined: emulated via software, either in or below the kernel. These accesses are always extremely slow. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_SLOW`: Misaligned accesses are slower - than equivalent byte accesses. Misaligned accesses may be supported - directly in hardware, or trapped and emulated by software. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SLOW`: Misaligned native word + sized accesses are slower than the equivalent quantity of byte accesses. + Misaligned accesses may be supported directly in hardware, or trapped and + emulated by software. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_FAST`: Misaligned accesses are faster - than equivalent byte accesses. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_FAST`: Misaligned native word + sized accesses are faster than the equivalent quantity of byte accesses. * :c:macro:`RISCV_HWPROBE_MISALIGNED_UNSUPPORTED`: Misaligned accesses are not supported at all and will generate a misaligned address fault. diff --git a/arch/riscv/include/asm/hwprobe.h b/arch/riscv/include/asm/hwprobe.h index ef01c182af2b..ffb9484531af 100644 --- a/arch/riscv/include/asm/hwprobe.h +++ b/arch/riscv/include/asm/hwprobe.h @@ -8,7 +8,7 @@ #include -#define RISCV_HWPROBE_MAX_KEY 8 +#define RISCV_HWPROBE_MAX_KEY 9 static inline bool riscv_hwprobe_key_is_valid(__s64 key) { diff --git a/arch/riscv/include/uapi/asm/hwprobe.h b/arch/riscv/include/uapi/asm/hwprobe.h index b706c8e47b02..635753084275 100644 --- a/arch/riscv/include/uapi/asm/hwprobe.h +++ b/arch/riscv/include/uapi/asm/hwprobe.h @@ -82,6 +82,7 @@ struct riscv_hwprobe { #define RISCV_HWPROBE_KEY_ZICBOZ_BLOCK_SIZE 6 #define RISCV_HWPROBE_KEY_HIGHEST_VIRT_ADDRESS 7 #define RISCV_HWPROBE_KEY_TIME_CSR_FREQ 8 +#define RISCV_HWPROBE_KEY_MISALIGNED_SCALAR_PERF 9 /* Increase RISCV_HWPROBE_MAX_KEY when adding items. */ /* Flags */ diff --git a/arch/riscv/kernel/sys_hwprobe.c b/arch/riscv/kernel/sys_hwprobe.c index 8d1b5c35d2a7..2d0f4f6a32c3 100644 --- a/arch/riscv/kernel/sys_hwprobe.c +++ b/arch/riscv/kernel/sys_hwprobe.c @@ -225,6 +225,7 @@ static void hwprobe_one_pair(struct riscv_hwprobe *pair, break; case RISCV_HWPROBE_KEY_CPUPERF_0: + case RISCV_HWPROBE_KEY_MISALIGNED_SCALAR_PERF: pair->value = hwprobe_misaligned(cpus); break; -- cgit v1.2.3-58-ga151 From 1f5288874de776412041022607513ffac74ae1a6 Mon Sep 17 00:00:00 2001 From: Evan Green Date: Fri, 9 Aug 2024 14:44:44 -0700 Subject: RISC-V: hwprobe: Add SCALAR to misaligned perf defines In preparation for misaligned vector performance hwprobe keys, rename the hwprobe key values associated with misaligned scalar accesses to include the term SCALAR. Leave the old defines in place to maintain source compatibility. This change is intended to be a functional no-op. Signed-off-by: Evan Green Reviewed-by: Charlie Jenkins Link: https://lore.kernel.org/r/20240809214444.3257596-3-evan@rivosinc.com Signed-off-by: Palmer Dabbelt --- Documentation/arch/riscv/hwprobe.rst | 28 +++++++++++++++------------- arch/riscv/include/uapi/asm/hwprobe.h | 5 +++++ arch/riscv/kernel/sys_hwprobe.c | 10 +++++----- arch/riscv/kernel/traps_misaligned.c | 6 +++--- arch/riscv/kernel/unaligned_access_speed.c | 12 ++++++------ 5 files changed, 34 insertions(+), 27 deletions(-) diff --git a/Documentation/arch/riscv/hwprobe.rst b/Documentation/arch/riscv/hwprobe.rst index a994eed75bde..85b709257918 100644 --- a/Documentation/arch/riscv/hwprobe.rst +++ b/Documentation/arch/riscv/hwprobe.rst @@ -247,23 +247,25 @@ The following keys are defined: the performance of misaligned scalar native word accesses on the selected set of processors. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_UNKNOWN`: The performance of misaligned - accesses is unknown. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN`: The performance of + misaligned scalar accesses is unknown. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_EMULATED`: Misaligned accesses are - emulated via software, either in or below the kernel. These accesses are - always extremely slow. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED`: Misaligned scalar + accesses are emulated via software, either in or below the kernel. These + accesses are always extremely slow. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_SLOW`: Misaligned native word - sized accesses are slower than the equivalent quantity of byte accesses. - Misaligned accesses may be supported directly in hardware, or trapped and - emulated by software. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW`: Misaligned scalar native + word sized accesses are slower than the equivalent quantity of byte + accesses. Misaligned accesses may be supported directly in hardware, or + trapped and emulated by software. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_FAST`: Misaligned native word - sized accesses are faster than the equivalent quantity of byte accesses. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SCALAR_FAST`: Misaligned scalar native + word sized accesses are faster than the equivalent quantity of byte + accesses. - * :c:macro:`RISCV_HWPROBE_MISALIGNED_UNSUPPORTED`: Misaligned accesses are - not supported at all and will generate a misaligned address fault. + * :c:macro:`RISCV_HWPROBE_MISALIGNED_SCALAR_UNSUPPORTED`: Misaligned scalar + accesses are not supported at all and will generate a misaligned address + fault. * :c:macro:`RISCV_HWPROBE_KEY_ZICBOZ_BLOCK_SIZE`: An unsigned int which represents the size of the Zicboz block in bytes. diff --git a/arch/riscv/include/uapi/asm/hwprobe.h b/arch/riscv/include/uapi/asm/hwprobe.h index 635753084275..1e153cda57db 100644 --- a/arch/riscv/include/uapi/asm/hwprobe.h +++ b/arch/riscv/include/uapi/asm/hwprobe.h @@ -83,6 +83,11 @@ struct riscv_hwprobe { #define RISCV_HWPROBE_KEY_HIGHEST_VIRT_ADDRESS 7 #define RISCV_HWPROBE_KEY_TIME_CSR_FREQ 8 #define RISCV_HWPROBE_KEY_MISALIGNED_SCALAR_PERF 9 +#define RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN 0 +#define RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED 1 +#define RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW 2 +#define RISCV_HWPROBE_MISALIGNED_SCALAR_FAST 3 +#define RISCV_HWPROBE_MISALIGNED_SCALAR_UNSUPPORTED 4 /* Increase RISCV_HWPROBE_MAX_KEY when adding items. */ /* Flags */ diff --git a/arch/riscv/kernel/sys_hwprobe.c b/arch/riscv/kernel/sys_hwprobe.c index 2d0f4f6a32c3..cea0ca2bf2a2 100644 --- a/arch/riscv/kernel/sys_hwprobe.c +++ b/arch/riscv/kernel/sys_hwprobe.c @@ -178,13 +178,13 @@ static u64 hwprobe_misaligned(const struct cpumask *cpus) perf = this_perf; if (perf != this_perf) { - perf = RISCV_HWPROBE_MISALIGNED_UNKNOWN; + perf = RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN; break; } } if (perf == -1ULL) - return RISCV_HWPROBE_MISALIGNED_UNKNOWN; + return RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN; return perf; } @@ -192,12 +192,12 @@ static u64 hwprobe_misaligned(const struct cpumask *cpus) static u64 hwprobe_misaligned(const struct cpumask *cpus) { if (IS_ENABLED(CONFIG_RISCV_EFFICIENT_UNALIGNED_ACCESS)) - return RISCV_HWPROBE_MISALIGNED_FAST; + return RISCV_HWPROBE_MISALIGNED_SCALAR_FAST; if (IS_ENABLED(CONFIG_RISCV_EMULATED_UNALIGNED_ACCESS) && unaligned_ctl_available()) - return RISCV_HWPROBE_MISALIGNED_EMULATED; + return RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED; - return RISCV_HWPROBE_MISALIGNED_SLOW; + return RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW; } #endif diff --git a/arch/riscv/kernel/traps_misaligned.c b/arch/riscv/kernel/traps_misaligned.c index b62d5a2f4541..192cd5603e95 100644 --- a/arch/riscv/kernel/traps_misaligned.c +++ b/arch/riscv/kernel/traps_misaligned.c @@ -338,7 +338,7 @@ int handle_misaligned_load(struct pt_regs *regs) perf_sw_event(PERF_COUNT_SW_ALIGNMENT_FAULTS, 1, regs, addr); #ifdef CONFIG_RISCV_PROBE_UNALIGNED_ACCESS - *this_cpu_ptr(&misaligned_access_speed) = RISCV_HWPROBE_MISALIGNED_EMULATED; + *this_cpu_ptr(&misaligned_access_speed) = RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED; #endif if (!unaligned_enabled) @@ -532,13 +532,13 @@ static bool check_unaligned_access_emulated(int cpu) unsigned long tmp_var, tmp_val; bool misaligned_emu_detected; - *mas_ptr = RISCV_HWPROBE_MISALIGNED_UNKNOWN; + *mas_ptr = RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN; __asm__ __volatile__ ( " "REG_L" %[tmp], 1(%[ptr])\n" : [tmp] "=r" (tmp_val) : [ptr] "r" (&tmp_var) : "memory"); - misaligned_emu_detected = (*mas_ptr == RISCV_HWPROBE_MISALIGNED_EMULATED); + misaligned_emu_detected = (*mas_ptr == RISCV_HWPROBE_MISALIGNED_SCALAR_EMULATED); /* * If unaligned_ctl is already set, this means that we detected that all * CPUS uses emulated misaligned access at boot time. If that changed diff --git a/arch/riscv/kernel/unaligned_access_speed.c b/arch/riscv/kernel/unaligned_access_speed.c index a9a6bcb02acf..160628a2116d 100644 --- a/arch/riscv/kernel/unaligned_access_speed.c +++ b/arch/riscv/kernel/unaligned_access_speed.c @@ -34,9 +34,9 @@ static int check_unaligned_access(void *param) struct page *page = param; void *dst; void *src; - long speed = RISCV_HWPROBE_MISALIGNED_SLOW; + long speed = RISCV_HWPROBE_MISALIGNED_SCALAR_SLOW; - if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_UNKNOWN) + if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN) return 0; /* Make an unaligned destination buffer. */ @@ -95,14 +95,14 @@ static int check_unaligned_access(void *param) } if (word_cycles < byte_cycles) - speed = RISCV_HWPROBE_MISALIGNED_FAST; + speed = RISCV_HWPROBE_MISALIGNED_SCALAR_FAST; ratio = div_u64((byte_cycles * 100), word_cycles); pr_info("cpu%d: Ratio of byte access time to unaligned word access is %d.%02d, unaligned accesses are %s\n", cpu, ratio / 100, ratio % 100, - (speed == RISCV_HWPROBE_MISALIGNED_FAST) ? "fast" : "slow"); + (speed == RISCV_HWPROBE_MISALIGNED_SCALAR_FAST) ? "fast" : "slow"); per_cpu(misaligned_access_speed, cpu) = speed; @@ -110,7 +110,7 @@ static int check_unaligned_access(void *param) * Set the value of fast_misaligned_access of a CPU. These operations * are atomic to avoid race conditions. */ - if (speed == RISCV_HWPROBE_MISALIGNED_FAST) + if (speed == RISCV_HWPROBE_MISALIGNED_SCALAR_FAST) cpumask_set_cpu(cpu, &fast_misaligned_access); else cpumask_clear_cpu(cpu, &fast_misaligned_access); @@ -188,7 +188,7 @@ static int riscv_online_cpu(unsigned int cpu) static struct page *buf; /* We are already set since the last check */ - if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_UNKNOWN) + if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_SCALAR_UNKNOWN) goto exit; buf = alloc_pages(GFP_KERNEL, MISALIGNED_BUFFER_ORDER); -- cgit v1.2.3-58-ga151