diff options
author | Linus Torvalds | 2019-12-01 18:49:57 -0800 |
---|---|---|
committer | Linus Torvalds | 2019-12-01 18:49:57 -0800 |
commit | b7fcf31f7036895ca8fc3a30eefffab0e82f75f6 (patch) | |
tree | de2480af1cb8213dd32f56ab596759a76ec7bd57 /arch | |
parent | 72c0870e3a05d9cd5466d08c3d2a3069ed0a2f9f (diff) | |
parent | e680a41fcaf07ccac8817c589fc4824988b48eac (diff) |
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf fixes from Ingo Molnar:
- Make /sys/devices/cpu/rdpmc based RDPMC enforcement more
instantaneous
- decoder: Update the Intel opcode map
- Various tooling fixes, including a few late optimizations and
cleanups.
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (23 commits)
perf script: Fix invalid LBR/binary mismatch error
perf script: Fix brstackinsn for AUXTRACE
perf affinity: Add infrastructure to save/restore affinity
perf pmu: Use file system cache to optimize sysfs access
perf regs: Make perf_reg_name() return "unknown" instead of NULL
perf diff: Use llabs() with 64-bit values
perf diff: Use llabs() with 64-bit values
perf/x86: Implement immediate enforcement of /sys/devices/cpu/rdpmc value of 0
perf tools: Allow to link with libbpf dynamicaly
perf tests: Rename tests/map_groups.c to tests/maps.c
perf tests: Rename thread-mg-share to thread-maps-share
perf maps: Rename map_groups.h to maps.h
perf maps: Rename 'mg' variables to 'maps'
perf map_symbol: Rename ms->mg to ms->maps
perf addr_location: Rename al->mg to al->maps
perf thread: Rename thread->mg to thread->maps
perf maps: Merge 'struct maps' with 'struct map_groups'
x86/insn: perf tools: Add some more instructions to the new instructions test
x86/insn: Add some more Intel instructions to the opcode map
perf map: Remove unused functions
...
Diffstat (limited to 'arch')
-rw-r--r-- | arch/x86/events/core.c | 18 | ||||
-rw-r--r-- | arch/x86/include/asm/mmu_context.h | 4 | ||||
-rw-r--r-- | arch/x86/lib/x86-opcode-map.txt | 44 |
3 files changed, 47 insertions, 19 deletions
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c index 6e3f0c18908e..9a89d98c55bd 100644 --- a/arch/x86/events/core.c +++ b/arch/x86/events/core.c @@ -49,6 +49,7 @@ DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { .enabled = 1, }; +DEFINE_STATIC_KEY_FALSE(rdpmc_never_available_key); DEFINE_STATIC_KEY_FALSE(rdpmc_always_available_key); u64 __read_mostly hw_cache_event_ids @@ -2181,21 +2182,26 @@ static ssize_t set_attr_rdpmc(struct device *cdev, if (x86_pmu.attr_rdpmc_broken) return -ENOTSUPP; - if ((val == 2) != (x86_pmu.attr_rdpmc == 2)) { + if (val != x86_pmu.attr_rdpmc) { /* - * Changing into or out of always available, aka - * perf-event-bypassing mode. This path is extremely slow, + * Changing into or out of never available or always available, + * aka perf-event-bypassing mode. This path is extremely slow, * but only root can trigger it, so it's okay. */ + if (val == 0) + static_branch_inc(&rdpmc_never_available_key); + else if (x86_pmu.attr_rdpmc == 0) + static_branch_dec(&rdpmc_never_available_key); + if (val == 2) static_branch_inc(&rdpmc_always_available_key); - else + else if (x86_pmu.attr_rdpmc == 2) static_branch_dec(&rdpmc_always_available_key); + on_each_cpu(refresh_pce, NULL, 1); + x86_pmu.attr_rdpmc = val; } - x86_pmu.attr_rdpmc = val; - return count; } diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h index 16ae821483c8..5f33924e200f 100644 --- a/arch/x86/include/asm/mmu_context.h +++ b/arch/x86/include/asm/mmu_context.h @@ -26,12 +26,14 @@ static inline void paravirt_activate_mm(struct mm_struct *prev, #ifdef CONFIG_PERF_EVENTS +DECLARE_STATIC_KEY_FALSE(rdpmc_never_available_key); DECLARE_STATIC_KEY_FALSE(rdpmc_always_available_key); static inline void load_mm_cr4_irqsoff(struct mm_struct *mm) { if (static_branch_unlikely(&rdpmc_always_available_key) || - atomic_read(&mm->context.perf_rdpmc_allowed)) + (!static_branch_unlikely(&rdpmc_never_available_key) && + atomic_read(&mm->context.perf_rdpmc_allowed))) cr4_set_bits_irqsoff(X86_CR4_PCE); else cr4_clear_bits_irqsoff(X86_CR4_PCE); diff --git a/arch/x86/lib/x86-opcode-map.txt b/arch/x86/lib/x86-opcode-map.txt index 0a0e9112f284..8908c58bd6cd 100644 --- a/arch/x86/lib/x86-opcode-map.txt +++ b/arch/x86/lib/x86-opcode-map.txt @@ -695,16 +695,28 @@ AVXcode: 2 4d: vrcp14ss/d Vsd,Hpd,Wsd (66),(ev) 4e: vrsqrt14ps/d Vpd,Wpd (66),(ev) 4f: vrsqrt14ss/d Vsd,Hsd,Wsd (66),(ev) -# Skip 0x50-0x57 +50: vpdpbusd Vx,Hx,Wx (66),(ev) +51: vpdpbusds Vx,Hx,Wx (66),(ev) +52: vdpbf16ps Vx,Hx,Wx (F3),(ev) | vpdpwssd Vx,Hx,Wx (66),(ev) | vp4dpwssd Vdqq,Hdqq,Wdq (F2),(ev) +53: vpdpwssds Vx,Hx,Wx (66),(ev) | vp4dpwssds Vdqq,Hdqq,Wdq (F2),(ev) +54: vpopcntb/w Vx,Wx (66),(ev) +55: vpopcntd/q Vx,Wx (66),(ev) 58: vpbroadcastd Vx,Wx (66),(v) 59: vpbroadcastq Vx,Wx (66),(v) | vbroadcasti32x2 Vx,Wx (66),(evo) 5a: vbroadcasti128 Vqq,Mdq (66),(v) | vbroadcasti32x4/64x2 Vx,Wx (66),(evo) 5b: vbroadcasti32x8/64x4 Vqq,Mdq (66),(ev) -# Skip 0x5c-0x63 +# Skip 0x5c-0x61 +62: vpexpandb/w Vx,Wx (66),(ev) +63: vpcompressb/w Wx,Vx (66),(ev) 64: vpblendmd/q Vx,Hx,Wx (66),(ev) 65: vblendmps/d Vx,Hx,Wx (66),(ev) 66: vpblendmb/w Vx,Hx,Wx (66),(ev) -# Skip 0x67-0x74 +68: vp2intersectd/q Kx,Hx,Wx (F2),(ev) +# Skip 0x69-0x6f +70: vpshldvw Vx,Hx,Wx (66),(ev) +71: vpshldvd/q Vx,Hx,Wx (66),(ev) +72: vcvtne2ps2bf16 Vx,Hx,Wx (F2),(ev) | vcvtneps2bf16 Vx,Wx (F3),(ev) | vpshrdvw Vx,Hx,Wx (66),(ev) +73: vpshrdvd/q Vx,Hx,Wx (66),(ev) 75: vpermi2b/w Vx,Hx,Wx (66),(ev) 76: vpermi2d/q Vx,Hx,Wx (66),(ev) 77: vpermi2ps/d Vx,Hx,Wx (66),(ev) @@ -727,6 +739,7 @@ AVXcode: 2 8c: vpmaskmovd/q Vx,Hx,Mx (66),(v) 8d: vpermb/w Vx,Hx,Wx (66),(ev) 8e: vpmaskmovd/q Mx,Vx,Hx (66),(v) +8f: vpshufbitqmb Kx,Hx,Wx (66),(ev) # 0x0f 0x38 0x90-0xbf (FMA) 90: vgatherdd/q Vx,Hx,Wx (66),(v) | vpgatherdd/q Vx,Wx (66),(evo) 91: vgatherqd/q Vx,Hx,Wx (66),(v) | vpgatherqd/q Vx,Wx (66),(evo) @@ -738,8 +751,8 @@ AVXcode: 2 97: vfmsubadd132ps/d Vx,Hx,Wx (66),(v) 98: vfmadd132ps/d Vx,Hx,Wx (66),(v) 99: vfmadd132ss/d Vx,Hx,Wx (66),(v),(v1) -9a: vfmsub132ps/d Vx,Hx,Wx (66),(v) -9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1) +9a: vfmsub132ps/d Vx,Hx,Wx (66),(v) | v4fmaddps Vdqq,Hdqq,Wdq (F2),(ev) +9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1) | v4fmaddss Vdq,Hdq,Wdq (F2),(ev) 9c: vfnmadd132ps/d Vx,Hx,Wx (66),(v) 9d: vfnmadd132ss/d Vx,Hx,Wx (66),(v),(v1) 9e: vfnmsub132ps/d Vx,Hx,Wx (66),(v) @@ -752,8 +765,8 @@ a6: vfmaddsub213ps/d Vx,Hx,Wx (66),(v) a7: vfmsubadd213ps/d Vx,Hx,Wx (66),(v) a8: vfmadd213ps/d Vx,Hx,Wx (66),(v) a9: vfmadd213ss/d Vx,Hx,Wx (66),(v),(v1) -aa: vfmsub213ps/d Vx,Hx,Wx (66),(v) -ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1) +aa: vfmsub213ps/d Vx,Hx,Wx (66),(v) | v4fnmaddps Vdqq,Hdqq,Wdq (F2),(ev) +ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1) | v4fnmaddss Vdq,Hdq,Wdq (F2),(ev) ac: vfnmadd213ps/d Vx,Hx,Wx (66),(v) ad: vfnmadd213ss/d Vx,Hx,Wx (66),(v),(v1) ae: vfnmsub213ps/d Vx,Hx,Wx (66),(v) @@ -780,11 +793,12 @@ ca: sha1msg2 Vdq,Wdq | vrcp28ps/d Vx,Wx (66),(ev) cb: sha256rnds2 Vdq,Wdq | vrcp28ss/d Vx,Hx,Wx (66),(ev) cc: sha256msg1 Vdq,Wdq | vrsqrt28ps/d Vx,Wx (66),(ev) cd: sha256msg2 Vdq,Wdq | vrsqrt28ss/d Vx,Hx,Wx (66),(ev) +cf: vgf2p8mulb Vx,Wx (66) db: VAESIMC Vdq,Wdq (66),(v1) -dc: VAESENC Vdq,Hdq,Wdq (66),(v1) -dd: VAESENCLAST Vdq,Hdq,Wdq (66),(v1) -de: VAESDEC Vdq,Hdq,Wdq (66),(v1) -df: VAESDECLAST Vdq,Hdq,Wdq (66),(v1) +dc: vaesenc Vx,Hx,Wx (66) +dd: vaesenclast Vx,Hx,Wx (66) +de: vaesdec Vx,Hx,Wx (66) +df: vaesdeclast Vx,Hx,Wx (66) f0: MOVBE Gy,My | MOVBE Gw,Mw (66) | CRC32 Gd,Eb (F2) | CRC32 Gd,Eb (66&F2) f1: MOVBE My,Gy | MOVBE Mw,Gw (66) | CRC32 Gd,Ey (F2) | CRC32 Gd,Ew (66&F2) f2: ANDN Gy,By,Ey (v) @@ -848,7 +862,7 @@ AVXcode: 3 41: vdppd Vdq,Hdq,Wdq,Ib (66),(v1) 42: vmpsadbw Vx,Hx,Wx,Ib (66),(v1) | vdbpsadbw Vx,Hx,Wx,Ib (66),(evo) 43: vshufi32x4/64x2 Vx,Hx,Wx,Ib (66),(ev) -44: vpclmulqdq Vdq,Hdq,Wdq,Ib (66),(v1) +44: vpclmulqdq Vx,Hx,Wx,Ib (66) 46: vperm2i128 Vqq,Hqq,Wqq,Ib (66),(v) 4a: vblendvps Vx,Hx,Wx,Lx (66),(v) 4b: vblendvpd Vx,Hx,Wx,Lx (66),(v) @@ -865,7 +879,13 @@ AVXcode: 3 63: vpcmpistri Vdq,Wdq,Ib (66),(v1) 66: vfpclassps/d Vk,Wx,Ib (66),(ev) 67: vfpclassss/d Vk,Wx,Ib (66),(ev) +70: vpshldw Vx,Hx,Wx,Ib (66),(ev) +71: vpshldd/q Vx,Hx,Wx,Ib (66),(ev) +72: vpshrdw Vx,Hx,Wx,Ib (66),(ev) +73: vpshrdd/q Vx,Hx,Wx,Ib (66),(ev) cc: sha1rnds4 Vdq,Wdq,Ib +ce: vgf2p8affineqb Vx,Wx,Ib (66) +cf: vgf2p8affineinvqb Vx,Wx,Ib (66) df: VAESKEYGEN Vdq,Wdq,Ib (66),(v1) f0: RORX Gy,Ey,Ib (F2),(v) EndTable |