From: Muhammad Usama Anjum <usama.anjum@arm.com>
To: Catalin Marinas <catalin.marinas@arm.com>,
linux-arm-kernel@lists.infradead.org,
Mark Rutland <mark.rutland@arm.com>,
Ryan Roberts <ryan.roberts@arm.com>
Cc: vladimir.murzin@arm.com, ryan.roberts@arm.com,
usama.anjum@arm.com, peterz@infradead.org, maz@kernel.org,
david.laight.linux@gmail.com, stable@vger.kernel.org,
james.morse@arm.com, yang@os.amperecomputing.com, cl@gentwo.org,
ruanjinjie@huawei.com, david@kernel.org, ljs@kernel.org,
Will Deacon <will@kernel.org>,
ardb@kernel.org
Subject: Re: (subset) [PATCH v4 00/21] Preemptible this_cpu_*() operations
Date: Wed, 30 Sep 2026 14:14:22 +0100 [thread overview]
Message-ID: <28978498-348e-4cfd-af72-1146a106e96b@arm.com> (raw)
In-Reply-To: <179026582920.3572476.8424936957104740430.b4-ty@b4>
On 24/09/2026 5:03 pm, Catalin Marinas wrote:
> On Tue, 08 Sep 2026 16:17:20 +0100, Mark Rutland wrote:
>> This series reworks arm64's this_cpu_*() operations such that they do
>> not need to disable preemption. The scheme is described in detail in patch 13,
>> and is similar to the approach Heiko Carstens applied to s390 [1], which
>> inspired this series.
>>
>> This scheme should build and function in all kernel configurations (e.g.
>> regardless of KPTI, SW PAN, CNP, VA BITS), and has no dependency on new
>> architectural features. The fixup logic should "just work" with kprobes,
>> etc, and I don't expect that this will need to become more complicated.
>>
>> [...]
>
> Applied to arm64 (for-next/preemptible-this-cpu), thanks!
>
> Patches 1-3 are already upstream as fixes (in -rc4, used as a base for
> for-next/core). I fixed some minor comment typos in a couple of patches,
> I haven't touched the code.
>
> On v2, Usama reported performance regression in perf/sched and
> perf/futex benchmarks that have not been addressed yet. It would be good
> to run the tests again on this version, in case they were just noise. I
> don't see why these would regress, maybe because of pt_regs being larger
> and we get some more cache misses?
I've just performed testing on arm64 for-next/core as it is and
by reverting this entire series:
aws-m7g.metal:
21 improvements
1 regression
AmpereOne:
35 improvements
2 regression
Results for SUT Class aws-m7g.metal:
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| Benchmark | Result Class | arm64-core-no-mark21-20260929-m7g (base) | arm64-core-mark21-20260929-m7g |
+=========================+==========================================================+============================================+==================================+
| lmbench/lat-mem-rd | size:4MB P:1 stride:128B rnd_read_lat (ns) | 25.36 | -3.89% |
| | size:4MB P:1 stride:128B seq_read_lat (ns) | 3.00 | 0.61% |
| | size:16MB P:1 stride:128B rnd_read_lat (ns) | 28.17 | -4.00% |
| | size:16MB P:1 stride:128B seq_read_lat (ns) | 2.89 | -1.95% |
| | size:64MB P:1 stride:128B rnd_read_lat (ns) | 102.19 | 0.29% |
| | size:64MB P:1 stride:128B seq_read_lat (ns) | 5.42 | -1.17% |
| | size:256MB P:1 stride:128B rnd_read_lat (ns) | 113.45 | -0.66% |
| | size:256MB P:1 stride:128B seq_read_lat (ns) | 5.56 | -0.59% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| micromm/fork | fork: p:1, d:10 (seconds) | 0.0076 | -2.29% |
| | fork: p:512, d:10 (seconds) | 0.0055 | 0.32% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| micromm/munmap | munmap: p:1, d:10 (seconds) | 0.061 | (I) 40.50% |
| | munmap: p:512, d:10 (seconds) | 0.0046 | (I) 2.64% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| micromm/vmalloc | fix_align_alloc_test: p:1, h:0, l:500000 (usec) | 635669.17 | (I) 8.06% |
| | fix_size_alloc_test: p:1, h:0, l:500000 (usec) | 374395.50 | (I) 4.24% |
| | fix_size_alloc_test: p:4, h:0, l:500000 (usec) | 454288.83 | (I) 7.98% |
| | fix_size_alloc_test: p:16, h:0, l:500000 (usec) | 870184.17 | (I) 19.28% |
| | fix_size_alloc_test: p:16, h:1, l:500000 (usec) | 813688.83 | (I) 15.82% |
| | fix_size_alloc_test: p:64, h:0, l:100000 (usec) | 500721.33 | (I) 28.32% |
| | fix_size_alloc_test: p:64, h:1, l:100000 (usec) | 465855.33 | (I) 22.94% |
| | fix_size_alloc_test: p:256, h:0, l:100000 (usec) | 1820499.17 | (I) 31.59% |
| | fix_size_alloc_test: p:256, h:1, l:100000 (usec) | 1706893.17 | (I) 25.98% |
| | fix_size_alloc_test: p:512, h:0, l:100000 (usec) | 3566528.17 | (I) 32.81% |
| | fix_size_alloc_test: p:512, h:1, l:100000 (usec) | 2626046.67 | (I) 36.14% |
| | full_fit_alloc_test: p:1, h:0, l:500000 (usec) | 558588.00 | -0.07% |
| | kvfree_rcu_1_arg_vmalloc_test: p:1, h:0, l:500000 (usec) | 359429.83 | 0.20% |
| | kvfree_rcu_2_arg_vmalloc_test: p:1, h:0, l:500000 (usec) | 359256.33 | 1.15% |
| | long_busy_list_alloc_test: p:1, h:0, l:500000 (usec) | 4388243.50 | (I) 25.27% |
| | pcpu_alloc_test: p:1, h:0, l:500000 (usec) | 197834.17 | 0.40% |
| | random_size_align_alloc_test: p:1, h:0, l:500000 (usec) | 758328.83 | (I) 8.63% |
| | random_size_alloc_test: p:1, h:0, l:500000 (usec) | 2306188.50 | (I) 24.32% |
| | vm_map_ram_test: p:1, h:0, l:500000 (usec) | 81248.50 | 0.80% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| mmtests/hackbench | hackbench-process-pipes-1 (seconds) | 0.15 | -1.56% |
| | hackbench-process-pipes-4 (seconds) | 0.39 | 1.85% |
| | hackbench-process-pipes-7 (seconds) | 0.66 | -3.13% |
| | hackbench-process-pipes-12 (seconds) | 1.11 | -0.37% |
| | hackbench-process-pipes-21 (seconds) | 1.48 | -0.28% |
| | hackbench-process-pipes-30 (seconds) | 1.87 | 0.62% |
| | hackbench-process-pipes-48 (seconds) | 2.57 | 0.81% |
| | hackbench-process-pipes-79 (seconds) | 3.62 | 1.44% |
| | hackbench-process-pipes-110 (seconds) | 4.70 | 1.42% |
| | hackbench-process-pipes-141 (seconds) | 5.81 | 1.59% |
| | hackbench-process-pipes-172 (seconds) | 6.97 | 1.65% |
| | hackbench-process-pipes-203 (seconds) | 8.14 | 1.63% |
| | hackbench-process-pipes-234 (seconds) | 9.34 | 1.85% |
| | hackbench-process-pipes-256 (seconds) | 10.16 | 1.88% |
| | hackbench-process-sockets-1 (seconds) | 0.28 | 1.02% |
| | hackbench-process-sockets-4 (seconds) | 0.62 | 0.05% |
| | hackbench-process-sockets-7 (seconds) | 0.99 | 0.32% |
| | hackbench-process-sockets-12 (seconds) | 1.60 | 0.46% |
| | hackbench-process-sockets-21 (seconds) | 2.76 | 0.47% |
| | hackbench-process-sockets-30 (seconds) | 3.94 | 0.67% |
| | hackbench-process-sockets-48 (seconds) | 6.29 | 0.65% |
| | hackbench-process-sockets-79 (seconds) | 9.78 | 0.26% |
| | hackbench-process-sockets-110 (seconds) | 13.60 | -0.33% |
| | hackbench-process-sockets-141 (seconds) | 17.53 | -0.59% |
| | hackbench-process-sockets-172 (seconds) | 21.74 | -0.84% |
| | hackbench-process-sockets-203 (seconds) | 25.95 | -1.06% |
| | hackbench-process-sockets-234 (seconds) | 30.33 | -1.22% |
| | hackbench-process-sockets-256 (seconds) | 33.43 | -1.23% |
| | hackbench-thread-pipes-1 (seconds) | 0.18 | 0.28% |
| | hackbench-thread-pipes-4 (seconds) | 0.39 | -0.50% |
| | hackbench-thread-pipes-7 (seconds) | 0.68 | 0.00% |
| | hackbench-thread-pipes-12 (seconds) | 1.14 | -0.88% |
| | hackbench-thread-pipes-21 (seconds) | 1.52 | -0.28% |
| | hackbench-thread-pipes-30 (seconds) | 1.94 | -0.08% |
| | hackbench-thread-pipes-48 (seconds) | 2.68 | -0.33% |
| | hackbench-thread-pipes-79 (seconds) | 3.68 | 0.01% |
| | hackbench-thread-pipes-110 (seconds) | 4.82 | 0.41% |
| | hackbench-thread-pipes-141 (seconds) | 5.94 | -0.26% |
| | hackbench-thread-pipes-172 (seconds) | 7.17 | 0.01% |
| | hackbench-thread-pipes-203 (seconds) | 8.39 | 0.06% |
| | hackbench-thread-pipes-234 (seconds) | 9.64 | 0.09% |
| | hackbench-thread-pipes-256 (seconds) | 10.51 | -0.11% |
| | hackbench-thread-sockets-1 (seconds) | 0.30 | 1.57% |
| | hackbench-thread-sockets-4 (seconds) | 0.64 | 0.68% |
| | hackbench-thread-sockets-7 (seconds) | 1.00 | 1.13% |
| | hackbench-thread-sockets-12 (seconds) | 1.60 | 0.69% |
| | hackbench-thread-sockets-21 (seconds) | 2.72 | 0.44% |
| | hackbench-thread-sockets-30 (seconds) | 3.86 | 0.37% |
| | hackbench-thread-sockets-48 (seconds) | 6.13 | 0.26% |
| | hackbench-thread-sockets-79 (seconds) | 9.51 | -0.17% |
| | hackbench-thread-sockets-110 (seconds) | 13.22 | 0.04% |
| | hackbench-thread-sockets-141 (seconds) | 17.02 | -0.13% |
| | hackbench-thread-sockets-172 (seconds) | 21.15 | -0.13% |
| | hackbench-thread-sockets-203 (seconds) | 25.32 | -0.16% |
| | hackbench-thread-sockets-234 (seconds) | 29.59 | -0.08% |
| | hackbench-thread-sockets-256 (seconds) | 32.62 | -0.15% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| mmtests/kernbench | kernbench-elsp-64 (seconds) | 359.85 | 0.24% |
| | kernbench-syst-64 (seconds) | 1130.19 | (I) 4.44% |
| | kernbench-user-64 (seconds) | 19739.56 | 0.10% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| mmtests/sysbench-cpu | sysbenchcpu-1 (ops/second) | 834.34 | -0.01% |
| | sysbenchcpu-4 (ops/second) | 3337.18 | -0.01% |
| | sysbenchcpu-7 (ops/second) | 5839.81 | -0.01% |
| | sysbenchcpu-12 (ops/second) | 10010.66 | -0.01% |
| | sysbenchcpu-21 (ops/second) | 17517.47 | -0.01% |
| | sysbenchcpu-30 (ops/second) | 25023.23 | -0.01% |
| | sysbenchcpu-48 (ops/second) | 40032.51 | -0.01% |
| | sysbenchcpu-79 (ops/second) | 53376.78 | -0.02% |
| | sysbenchcpu-110 (ops/second) | 53377.41 | -0.02% |
| | sysbenchcpu-128 (ops/second) | 53379.94 | -0.03% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| mmtests/sysbench-mutex | sysbenchmutex-1 (usec) | 194.30 | -0.01% |
| | sysbenchmutex-4 (usec) | 200.58 | 0.01% |
| | sysbenchmutex-7 (usec) | 200.95 | -0.14% |
| | sysbenchmutex-12 (usec) | 201.54 | -0.18% |
| | sysbenchmutex-21 (usec) | 202.49 | 0.22% |
| | sysbenchmutex-30 (usec) | 203.24 | 0.22% |
| | sysbenchmutex-48 (usec) | 242.40 | 3.64% |
| | sysbenchmutex-79 (usec) | 360.42 | 0.84% |
| | sysbenchmutex-110 (usec) | 509.17 | 0.52% |
| | sysbenchmutex-128 (usec) | 582.54 | 0.44% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| mmtests/sysbench-thread | sysbenchthread-1 (usec) | 0.37 | (I) 6.16% |
| | sysbenchthread-4 (usec) | 0.55 | (I) 7.19% |
| | sysbenchthread-7 (usec) | 2.68 | -19.77% |
| | sysbenchthread-12 (usec) | 5.89 | 11.27% |
| | sysbenchthread-21 (usec) | 11.84 | -0.66% |
| | sysbenchthread-30 (usec) | 19.81 | 24.87% |
| | sysbenchthread-48 (usec) | 33.30 | 21.51% |
| | sysbenchthread-79 (usec) | 57.47 | -7.01% |
| | sysbenchthread-110 (usec) | 92.38 | -3.64% |
| | sysbenchthread-128 (usec) | 115.79 | 7.20% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| perf/futex | hash (ops/sec) | 2655172.83 | (I) 2.83% |
| | lock-pi (ops/sec) | 281.00 | 0.00% |
| | requeue (ms) | 0.027 | 3.30% |
| | wake (ms) | 0.071 | -2.87% |
| | wake-parallel (ms) | 0.0042 | 0.08% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| perf/mem | memset -k 1GB -f default -s 16GB (bytes/sec) | 65213616658.81 | -0.28% |
| | mmap -p 1GB -f demand -s 32GB -l 5 (bytes/sec) | 64303069991.32 | -0.35% |
| | mmap -p 4KB -f demand -s 32GB -l 5 (bytes/sec) | 49764611027.28 | 0.10% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| perf/sched | messaging -l 10000 (sec) | 1.32 | 0.67% |
| | pipe (ops/sec) | 543170.30 | 7.68% |
| | seccomp-notify (ops/sec) | 243035.63 | (R) -4.65% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| perf/syscall | basic (ops/sec) | 5162993.33 | 0.47% |
| | execve (ops/sec) | 2329.13 | 1.72% |
| | fork (ops/sec) | 4050.90 | 1.43% |
| | getpgid (ops/sec) | 5174345.87 | (I) 2.15% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| sockperf/echo-lat-tcp | ping-pong m:64B t:10s, avg_latency (usec) | 4.86 | -0.10% |
| | ping-pong m:64B t:10s, latency_p99 (usec) | 5.14 | -0.20% |
| | ping-pong m:256B t:10s, avg_latency (usec) | 4.97 | -0.37% |
| | ping-pong m:256B t:10s, latency_p99 (usec) | 5.31 | -0.15% |
| | ping-pong m:1024B t:10s, avg_latency (usec) | 5.09 | 0.00% |
| | ping-pong m:1024B t:10s, latency_p99 (usec) | 5.40 | -0.16% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| sockperf/echo-lat-udp | ping-pong m:64B t:10s, avg_latency (usec) | 4.27 | 0.77% |
| | ping-pong m:64B t:10s, latency_p99 (usec) | 5.03 | -0.74% |
| | ping-pong m:256B t:10s, avg_latency (usec) | 4.34 | 0.99% |
| | ping-pong m:256B t:10s, latency_p99 (usec) | 5.10 | 0.77% |
| | ping-pong m:1024B t:10s, avg_latency (usec) | 4.43 | -0.16% |
| | ping-pong m:1024B t:10s, latency_p99 (usec) | 5.26 | 0.66% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| sockperf/packet-tp-tcp | throughput m:64B t:10s, msg_rate (msg/sec) | 1339599.17 | 2.00% |
| | throughput m:64B t:10s, throughput_mbps (Mbps) | 654.10 | 2.00% |
| | throughput m:256B t:10s, msg_rate (msg/sec) | 2248084.00 | 0.76% |
| | throughput m:256B t:10s, throughput_mbps (Mbps) | 4390.79 | 0.76% |
| | throughput m:1024B t:10s, msg_rate (msg/sec) | 1587001.83 | -1.56% |
| | throughput m:1024B t:10s, throughput_mbps (Mbps) | 12398.45 | -1.56% |
| | throughput m:1472B t:10s, msg_rate (msg/sec) | 1398038.00 | -0.50% |
| | throughput m:1472B t:10s, throughput_mbps (Mbps) | 15700.62 | -0.50% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| sockperf/packet-tp-udp | throughput m:64B t:10s, msg_rate (msg/sec) | 401108.00 | -0.79% |
| | throughput m:64B t:10s, throughput_mbps (Mbps) | 195.85 | -0.79% |
| | throughput m:256B t:10s, msg_rate (msg/sec) | 403048.17 | -2.33% |
| | throughput m:256B t:10s, throughput_mbps (Mbps) | 787.20 | -2.33% |
| | throughput m:1024B t:10s, msg_rate (msg/sec) | 390376.50 | -2.26% |
| | throughput m:1024B t:10s, throughput_mbps (Mbps) | 3049.82 | -2.26% |
| | throughput m:1472B t:10s, msg_rate (msg/sec) | 387575.83 | -1.26% |
| | throughput m:1472B t:10s, throughput_mbps (Mbps) | 4352.66 | -1.26% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| speedometer/v2.0 | score (runs/min) | 158.00 | 0.00% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| speedometer/v2.1 | score (runs/min) | 165.00 | 0.91% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| syscall/getpid | mean (ns) | 195.17 | 1.04% |
| | p99 (ns) | 199.67 | -1.16% |
| | p99.9 (ns) | 201.00 | -1.47% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| syscall/getppid | mean (ns) | 197.00 | 1.37% |
| | p99 (ns) | 203.67 | 2.60% |
| | p99.9 (ns) | 206.50 | 3.08% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
| syscall/invalid | mean (ns) | 187.33 | 0.81% |
| | p99 (ns) | 191.00 | 0.26% |
| | p99.9 (ns) | 191.50 | 0.00% |
+-------------------------+----------------------------------------------------------+--------------------------------------------+----------------------------------+
Results for SUT Class cesw-aarch64-ampereone-1s-a192-32x:
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| Benchmark | Result Class | arm64-core-no-mark21-20260929-ampereone (base) | arm64-core-mark21-20260929-ampereone |
+=========================+==========================================================+==================================================+========================================+
| lmbench/lat-mem-rd | size:4MB P:1 stride:128B rnd_read_lat (ns) | 41.64 | 0.85% |
| | size:4MB P:1 stride:128B seq_read_lat (ns) | 4.77 | 11.97% |
| | size:16MB P:1 stride:128B rnd_read_lat (ns) | 42.74 | 1.73% |
| | size:16MB P:1 stride:128B seq_read_lat (ns) | 2.95 | -0.42% |
| | size:64MB P:1 stride:128B rnd_read_lat (ns) | 54.51 | 5.85% |
| | size:64MB P:1 stride:128B seq_read_lat (ns) | 3.29 | -4.91% |
| | size:256MB P:1 stride:128B rnd_read_lat (ns) | 150.17 | -0.30% |
| | size:256MB P:1 stride:128B seq_read_lat (ns) | 6.21 | -0.58% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| micromm/fork | fork: p:1, d:10 (seconds) | 0.0084 | 1.11% |
| | fork: p:512, d:10 (seconds) | 0.0048 | 0.88% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| micromm/munmap | munmap: p:1, d:10 (seconds) | 0.077 | (I) 30.07% |
| | munmap: p:512, d:10 (seconds) | 0.022 | 0.68% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| micromm/vmalloc | fix_align_alloc_test: p:1, h:0, l:500000 (usec) | 725723.67 | (I) 18.02% |
| | fix_size_alloc_test: p:1, h:0, l:500000 (usec) | 393933.50 | (I) 9.95% |
| | fix_size_alloc_test: p:4, h:0, l:500000 (usec) | 516939.50 | (I) 23.67% |
| | fix_size_alloc_test: p:16, h:0, l:500000 (usec) | 971323.17 | (I) 30.43% |
| | fix_size_alloc_test: p:16, h:1, l:500000 (usec) | 915152.17 | (I) 25.96% |
| | fix_size_alloc_test: p:64, h:0, l:100000 (usec) | 591172.17 | (I) 41.62% |
| | fix_size_alloc_test: p:64, h:1, l:100000 (usec) | 544844.17 | (I) 34.58% |
| | fix_size_alloc_test: p:256, h:0, l:100000 (usec) | 2166037.67 | (I) 46.80% |
| | fix_size_alloc_test: p:256, h:1, l:100000 (usec) | 1994903.83 | (I) 43.54% |
| | fix_size_alloc_test: p:512, h:0, l:100000 (usec) | 4203198.00 | (I) 48.54% |
| | fix_size_alloc_test: p:512, h:1, l:100000 (usec) | 3003280.00 | (I) 66.11% |
| | full_fit_alloc_test: p:1, h:0, l:500000 (usec) | 610786.67 | (I) 10.06% |
| | kvfree_rcu_1_arg_vmalloc_test: p:1, h:0, l:500000 (usec) | 539508.33 | (I) 5.40% |
| | kvfree_rcu_2_arg_vmalloc_test: p:1, h:0, l:500000 (usec) | 516781.83 | 4.94% |
| | long_busy_list_alloc_test: p:1, h:0, l:500000 (usec) | 4941369.00 | (I) 39.67% |
| | pcpu_alloc_test: p:1, h:0, l:500000 (usec) | 480082.17 | 0.12% |
| | random_size_align_alloc_test: p:1, h:0, l:500000 (usec) | 887208.00 | (I) 8.63% |
| | random_size_alloc_test: p:1, h:0, l:500000 (usec) | 2626017.00 | (I) 37.28% |
| | vm_map_ram_test: p:1, h:0, l:500000 (usec) | 106035.00 | 2.38% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| mmtests/hackbench | hackbench-process-pipes-1 (seconds) | 0.26 | 0.20% |
| | hackbench-process-pipes-4 (seconds) | 0.44 | -2.97% |
| | hackbench-process-pipes-7 (seconds) | 0.53 | -0.47% |
| | hackbench-process-pipes-12 (seconds) | 0.73 | 0.75% |
| | hackbench-process-pipes-21 (seconds) | 1.12 | -0.82% |
| | hackbench-process-pipes-30 (seconds) | 1.58 | -1.54% |
| | hackbench-process-pipes-48 (seconds) | 2.81 | -0.05% |
| | hackbench-process-pipes-79 (seconds) | 3.67 | 2.64% |
| | hackbench-process-pipes-110 (seconds) | 3.53 | -1.03% |
| | hackbench-process-pipes-141 (seconds) | 3.68 | 1.43% |
| | hackbench-process-pipes-172 (seconds) | 4.00 | 0.44% |
| | hackbench-process-pipes-203 (seconds) | 4.34 | -0.27% |
| | hackbench-process-pipes-234 (seconds) | 4.65 | 0.83% |
| | hackbench-process-pipes-265 (seconds) | 5.04 | -0.27% |
| | hackbench-process-pipes-296 (seconds) | 5.37 | 0.54% |
| | hackbench-process-sockets-1 (seconds) | 0.56 | 5.32% |
| | hackbench-process-sockets-4 (seconds) | 0.66 | 3.91% |
| | hackbench-process-sockets-7 (seconds) | 0.79 | -0.15% |
| | hackbench-process-sockets-12 (seconds) | 1.22 | -0.79% |
| | hackbench-process-sockets-21 (seconds) | 2.03 | -1.28% |
| | hackbench-process-sockets-30 (seconds) | 2.88 | -1.69% |
| | hackbench-process-sockets-48 (seconds) | 4.55 | -1.84% |
| | hackbench-process-sockets-79 (seconds) | 7.62 | -1.81% |
| | hackbench-process-sockets-110 (seconds) | 10.72 | -1.84% |
| | hackbench-process-sockets-141 (seconds) | 13.88 | -1.87% |
| | hackbench-process-sockets-172 (seconds) | 17.07 | -1.78% |
| | hackbench-process-sockets-203 (seconds) | 20.28 | -1.63% |
| | hackbench-process-sockets-234 (seconds) | 23.49 | -1.68% |
| | hackbench-process-sockets-265 (seconds) | 26.77 | -1.66% |
| | hackbench-process-sockets-296 (seconds) | 30.07 | -1.70% |
| | hackbench-thread-pipes-1 (seconds) | 0.28 | 1.26% |
| | hackbench-thread-pipes-4 (seconds) | 0.54 | -0.40% |
| | hackbench-thread-pipes-7 (seconds) | 0.65 | 1.73% |
| | hackbench-thread-pipes-12 (seconds) | 0.82 | -0.37% |
| | hackbench-thread-pipes-21 (seconds) | 1.12 | 0.75% |
| | hackbench-thread-pipes-30 (seconds) | 1.60 | 0.64% |
| | hackbench-thread-pipes-48 (seconds) | 2.75 | 0.33% |
| | hackbench-thread-pipes-79 (seconds) | 3.94 | -0.39% |
| | hackbench-thread-pipes-110 (seconds) | 3.88 | -2.09% |
| | hackbench-thread-pipes-141 (seconds) | 4.00 | -0.31% |
| | hackbench-thread-pipes-172 (seconds) | 4.31 | -0.52% |
| | hackbench-thread-pipes-203 (seconds) | 4.72 | 0.20% |
| | hackbench-thread-pipes-234 (seconds) | 5.13 | 1.08% |
| | hackbench-thread-pipes-265 (seconds) | 5.48 | -2.56% |
| | hackbench-thread-pipes-296 (seconds) | 5.87 | -0.40% |
| | hackbench-thread-sockets-1 (seconds) | 0.57 | 1.12% |
| | hackbench-thread-sockets-4 (seconds) | 0.73 | 2.84% |
| | hackbench-thread-sockets-7 (seconds) | 0.84 | -0.96% |
| | hackbench-thread-sockets-12 (seconds) | 1.27 | -0.08% |
| | hackbench-thread-sockets-21 (seconds) | 2.07 | -1.10% |
| | hackbench-thread-sockets-30 (seconds) | 2.91 | -1.53% |
| | hackbench-thread-sockets-48 (seconds) | 4.59 | -1.42% |
| | hackbench-thread-sockets-79 (seconds) | 7.66 | -1.45% |
| | hackbench-thread-sockets-110 (seconds) | 10.73 | -1.92% |
| | hackbench-thread-sockets-141 (seconds) | 13.89 | -1.81% |
| | hackbench-thread-sockets-172 (seconds) | 17.07 | -1.67% |
| | hackbench-thread-sockets-203 (seconds) | 20.24 | -1.70% |
| | hackbench-thread-sockets-234 (seconds) | 23.43 | -1.77% |
| | hackbench-thread-sockets-265 (seconds) | 26.63 | -1.87% |
| | hackbench-thread-sockets-296 (seconds) | 29.91 | -1.82% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| mmtests/kernbench | kernbench-elsp-192 (seconds) | 172.18 | 0.44% |
| | kernbench-syst-192 (seconds) | 1755.36 | (I) 6.41% |
| | kernbench-user-192 (seconds) | 22994.12 | 0.05% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| mmtests/sysbench-cpu | sysbenchcpu-1 (ops/second) | 232.13 | -0.01% |
| | sysbenchcpu-4 (ops/second) | 929.37 | 0.06% |
| | sysbenchcpu-7 (ops/second) | 1628.48 | -0.09% |
| | sysbenchcpu-12 (ops/second) | 2791.82 | -0.63% |
| | sysbenchcpu-21 (ops/second) | 4884.10 | 0.02% |
| | sysbenchcpu-30 (ops/second) | 6976.07 | -0.02% |
| | sysbenchcpu-48 (ops/second) | 11160.90 | -0.06% |
| | sysbenchcpu-79 (ops/second) | 18354.35 | -0.14% |
| | sysbenchcpu-110 (ops/second) | 25531.52 | -0.16% |
| | sysbenchcpu-141 (ops/second) | 32597.33 | 0.00% |
| | sysbenchcpu-172 (ops/second) | 39486.52 | -0.10% |
| | sysbenchcpu-203 (ops/second) | 43977.55 | -0.43% |
| | sysbenchcpu-234 (ops/second) | 44051.35 | -0.60% |
| | sysbenchcpu-265 (ops/second) | 44152.90 | -0.36% |
| | sysbenchcpu-296 (ops/second) | 44255.55 | -0.19% |
| | sysbenchcpu-327 (ops/second) | 44225.89 | -0.42% |
| | sysbenchcpu-358 (ops/second) | 44301.57 | -0.27% |
| | sysbenchcpu-384 (ops/second) | 44171.27 | -0.04% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| mmtests/sysbench-mutex | sysbenchmutex-1 (usec) | 167.90 | -0.27% |
| | sysbenchmutex-4 (usec) | 176.95 | 0.12% |
| | sysbenchmutex-7 (usec) | 178.62 | -0.08% |
| | sysbenchmutex-12 (usec) | 179.43 | -0.24% |
| | sysbenchmutex-21 (usec) | 183.11 | -0.14% |
| | sysbenchmutex-30 (usec) | 199.32 | -4.26% |
| | sysbenchmutex-48 (usec) | 328.61 | -1.71% |
| | sysbenchmutex-79 (usec) | 531.33 | -1.75% |
| | sysbenchmutex-110 (usec) | 651.77 | 3.22% |
| | sysbenchmutex-141 (usec) | 817.90 | -2.82% |
| | sysbenchmutex-172 (usec) | 993.03 | 1.08% |
| | sysbenchmutex-203 (usec) | 1140.15 | -1.22% |
| | sysbenchmutex-234 (usec) | 1261.51 | -1.21% |
| | sysbenchmutex-265 (usec) | 1458.96 | -2.62% |
| | sysbenchmutex-296 (usec) | 1643.47 | 0.26% |
| | sysbenchmutex-327 (usec) | 1770.20 | -0.67% |
| | sysbenchmutex-358 (usec) | 1930.64 | -5.61% |
| | sysbenchmutex-384 (usec) | 2135.30 | -2.11% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| mmtests/sysbench-thread | sysbenchthread-1 (usec) | 0.34 | (I) 6.77% |
| | sysbenchthread-4 (usec) | 0.49 | 7.30% |
| | sysbenchthread-7 (usec) | 3.89 | 12.19% |
| | sysbenchthread-12 (usec) | 6.70 | -15.50% |
| | sysbenchthread-21 (usec) | 18.05 | -9.78% |
| | sysbenchthread-30 (usec) | 29.92 | 5.75% |
| | sysbenchthread-48 (usec) | 48.52 | 5.21% |
| | sysbenchthread-79 (usec) | 80.01 | 3.30% |
| | sysbenchthread-110 (usec) | 115.83 | 5.85% |
| | sysbenchthread-141 (usec) | 142.43 | -1.78% |
| | sysbenchthread-172 (usec) | 171.79 | 2.44% |
| | sysbenchthread-203 (usec) | 269.94 | -0.28% |
| | sysbenchthread-234 (usec) | 294.78 | -7.66% |
| | sysbenchthread-265 (usec) | 345.62 | -1.59% |
| | sysbenchthread-296 (usec) | 419.84 | 7.43% |
| | sysbenchthread-327 (usec) | 452.17 | 18.71% |
| | sysbenchthread-358 (usec) | 515.08 | 12.55% |
| | sysbenchthread-384 (usec) | 497.78 | 6.22% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| perf/futex | hash (ops/sec) | 2653557.20 | 7.63% |
| | lock-pi (ops/sec) | 78.53 | 0.59% |
| | requeue (ms) | 0.12 | (I) 2.94% |
| | wake (ms) | 0.29 | -0.72% |
| | wake-parallel (ms) | 0.0027 | (R) -10.60% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| perf/mem | memset -k 1GB -f default -s 16GB (bytes/sec) | 33861254728.37 | 0.04% |
| | mmap -p 1GB -f demand -s 32GB -l 5 (bytes/sec) | 33677436964.25 | 0.44% |
| | mmap -p 4KB -f demand -s 32GB -l 5 (bytes/sec) | 25464266889.26 | -0.94% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| perf/sched | messaging -l 10000 (sec) | 0.99 | -0.55% |
| | pipe (ops/sec) | 319346.30 | -0.83% |
| | seccomp-notify (ops/sec) | 130238.67 | (R) -2.86% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| perf/syscall | basic (ops/sec) | 7377677.90 | 1.14% |
| | execve (ops/sec) | 1526.50 | (I) 4.25% |
| | fork (ops/sec) | 2393.27 | (I) 4.30% |
| | getpgid (ops/sec) | 7699746.87 | -0.93% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| sockperf/echo-lat-tcp | ping-pong m:64B t:10s, avg_latency (usec) | 8.09 | 1.82% |
| | ping-pong m:64B t:10s, latency_p99 (usec) | 9.20 | 1.92% |
| | ping-pong m:256B t:10s, avg_latency (usec) | 8.15 | 1.93% |
| | ping-pong m:256B t:10s, latency_p99 (usec) | 9.21 | 1.95% |
| | ping-pong m:1024B t:10s, avg_latency (usec) | 8.35 | 1.75% |
| | ping-pong m:1024B t:10s, latency_p99 (usec) | 9.39 | 1.79% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| sockperf/echo-lat-udp | ping-pong m:64B t:10s, avg_latency (usec) | 6.98 | (I) 4.78% |
| | ping-pong m:64B t:10s, latency_p99 (usec) | 9.15 | (I) 3.76% |
| | ping-pong m:256B t:10s, avg_latency (usec) | 7.04 | (I) 4.08% |
| | ping-pong m:256B t:10s, latency_p99 (usec) | 9.20 | (I) 3.37% |
| | ping-pong m:1024B t:10s, avg_latency (usec) | 7.29 | (I) 4.28% |
| | ping-pong m:1024B t:10s, latency_p99 (usec) | 9.47 | 3.08% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| sockperf/packet-tp-tcp | throughput m:64B t:10s, msg_rate (msg/sec) | 2741038.17 | -0.59% |
| | throughput m:64B t:10s, throughput_mbps (Mbps) | 1338.40 | -0.59% |
| | throughput m:256B t:10s, msg_rate (msg/sec) | 2287435.17 | 0.08% |
| | throughput m:256B t:10s, throughput_mbps (Mbps) | 4467.65 | 0.08% |
| | throughput m:1024B t:10s, msg_rate (msg/sec) | 1681400.50 | -0.11% |
| | throughput m:1024B t:10s, throughput_mbps (Mbps) | 13135.94 | -0.11% |
| | throughput m:1472B t:10s, msg_rate (msg/sec) | 1527520.00 | 0.14% |
| | throughput m:1472B t:10s, throughput_mbps (Mbps) | 17154.77 | 0.14% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| sockperf/packet-tp-udp | throughput m:64B t:10s, msg_rate (msg/sec) | 428854.17 | (I) 4.13% |
| | throughput m:64B t:10s, throughput_mbps (Mbps) | 209.40 | (I) 4.13% |
| | throughput m:256B t:10s, msg_rate (msg/sec) | 424411.50 | (I) 3.17% |
| | throughput m:256B t:10s, throughput_mbps (Mbps) | 828.93 | (I) 3.17% |
| | throughput m:1024B t:10s, msg_rate (msg/sec) | 399098.33 | (I) 5.30% |
| | throughput m:1024B t:10s, throughput_mbps (Mbps) | 3117.96 | (I) 5.30% |
| | throughput m:1472B t:10s, msg_rate (msg/sec) | 399116.00 | (I) 4.98% |
| | throughput m:1472B t:10s, throughput_mbps (Mbps) | 4482.26 | (I) 4.98% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| speedometer/v2.0 | score (runs/min) | 181.33 | -0.09% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| speedometer/v2.1 | score (runs/min) | 189.67 | -0.09% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| syscall/getpid | mean (ns) | 145.83 | 0.11% |
| | p99 (ns) | 160.00 | 0.00% |
| | p99.9 (ns) | 180.00 | 0.00% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| syscall/getppid | mean (ns) | 146.67 | -0.68% |
| | p99 (ns) | 160.00 | 0.00% |
| | p99.9 (ns) | 180.00 | -1.82% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
| syscall/invalid | mean (ns) | 136.00 | -0.49% |
| | p99 (ns) | 143.33 | -6.52% |
| | p99.9 (ns) | 180.00 | 0.00% |
+-------------------------+----------------------------------------------------------+--------------------------------------------------+----------------------------------------+
--
Thanks,
Usama
prev parent reply other threads:[~2026-09-30 13:15 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-08 15:17 [PATCH v4 00/21] Preemptible this_cpu_*() operations Mark Rutland
2026-09-08 15:17 ` [PATCH v4 01/21] arm64: percpu: Fix this_cpu_write() casting Mark Rutland
2026-09-10 10:56 ` Lorenzo Stoakes (ARM)
2026-09-08 15:17 ` [PATCH v4 02/21] arm64: percpu: Fix this_cpu_and() mask generation Mark Rutland
2026-09-08 15:17 ` [PATCH v4 03/21] arm64: percpu: Fix LSE operations on {8,16}-bit types Mark Rutland
2026-09-10 15:18 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 04/21] arm64: cmpxchg: LL/SC: Avoid redundant extension Mark Rutland
2026-09-15 14:21 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 05/21] arm64: cmpxchg128: LSE: Remove redundant operands Mark Rutland
2026-09-10 10:06 ` Vladimir Murzin
2026-09-11 11:43 ` Mark Rutland
2026-09-08 15:17 ` [PATCH v4 06/21] arm64: preempt: Simplify and optimize __preempt_count_dec_and_test() Mark Rutland
2026-09-08 15:17 ` [PATCH v4 07/21] arm64: preempt: Treat should_resched() as unlikely Mark Rutland
2026-09-08 15:17 ` [PATCH v4 08/21] arm64: ptrace: Always inline pt_regs_[read,write}_reg() Mark Rutland
2026-09-11 10:20 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 09/21] arm64: percpu: Factor out percpu offset asm Mark Rutland
2026-09-11 12:39 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 10/21] arm64: gpr-num: Add wxN aliases for wN registers Mark Rutland
2026-09-11 12:43 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 11/21] arm64: gpr-num: add __GPR_NUM() helper Mark Rutland
2026-09-10 14:24 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 12/21] arm64: entry: sdei: Restore all clobberable GPRs Mark Rutland
2026-09-11 12:47 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 13/21] arm64: entry: sdei: Make 'tsk' available Mark Rutland
2026-09-10 13:06 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 14/21] arm64: percpu: Add infrastructure for preemptible this_cpu_*() ops Mark Rutland
2026-09-15 14:19 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 15/21] arm64: percpu: Implement preemptible read/write ops Mark Rutland
2026-09-15 11:58 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 16/21] arm64: percpu: Implement preemptible void RMW ops Mark Rutland
2026-09-15 12:01 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 17/21] arm64: percpu: Implement preemptible return " Mark Rutland
2026-09-15 12:03 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 18/21] arm64: percpu: Implement preemptible XCHG ops Mark Rutland
2026-09-15 12:07 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 19/21] arm64: percpu: Implement preemptible CMPXCHG ops Mark Rutland
2026-09-15 14:20 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 20/21] arm64: percpu: Implement preemptible CMPXCHG128 ops Mark Rutland
2026-09-15 14:20 ` Vladimir Murzin
2026-09-08 15:17 ` [PATCH v4 21/21] arm64: percpu: Remove _pcp_protect*() wrappers Mark Rutland
2026-09-15 14:21 ` Vladimir Murzin
2026-09-11 13:44 ` [PATCH v4 00/21] Preemptible this_cpu_*() operations Will Deacon
2026-09-24 16:03 ` (subset) " Catalin Marinas
2026-09-30 13:14 ` Muhammad Usama Anjum [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=28978498-348e-4cfd-af72-1146a106e96b@arm.com \
--to=usama.anjum@arm.com \
--cc=ardb@kernel.org \
--cc=catalin.marinas@arm.com \
--cc=cl@gentwo.org \
--cc=david.laight.linux@gmail.com \
--cc=david@kernel.org \
--cc=james.morse@arm.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=ljs@kernel.org \
--cc=mark.rutland@arm.com \
--cc=maz@kernel.org \
--cc=peterz@infradead.org \
--cc=ruanjinjie@huawei.com \
--cc=ryan.roberts@arm.com \
--cc=stable@vger.kernel.org \
--cc=vladimir.murzin@arm.com \
--cc=will@kernel.org \
--cc=yang@os.amperecomputing.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.