perf record: adapt NUMA awareness to machines with #CPUs > 1K
Status from this message: The following commit has been merged into the perf/core branch of tip:
Get series with b4
b4 am [email protected]Versions newest first · linked by author and title
- v5 Latest3 patches 6 repliesApplied
- v43 patches 4 repliesSuperseded
- v3 posting 2/23 patches 4 repliesSuperseded
Earlier versions · 3
| Version | Posted | Patches | Replies | Status |
|---|---|---|---|---|
| v3 posting 1/2 | 3 patches | 0 replies | Superseded | |
| v2 | 3 patches | 5 replies | Superseded | |
| v1 | 3 patches | 7 replies | Superseded |
Changes in v5 from the cover letter
- avoided allocation of mmap affinity masks in case of rec->opts.affinity == PERF_AFFINITY_SYS
Patches in v5 Applied
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/3 | tools bitmap: implement bitmap_equal() operation at bitmap APItools/ tools/ | +45 | 0 replies | |
| 2/3 | perf mmap: declare type for cpu mask of arbitrary lengthtools/ tools/ | +23 | 2 replies | |
| 3/3 | perf record: adapt affinity to machines with #CPUs > 1Ktools/ tools/ tools/ | +45 | 2 replies |
Reviews trailers per patch
None of the 3 patches has a Reviewed-by or Acked-by of their own; the series has one for all of them.
| Patch | Jiri Olsa |
|---|---|
| Series | A-b |
| 1/3 | |
| 2/3 | |
| 3/3 |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.
Cover letter [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K
Current implementation of cpu_set_t type by glibc has internal cpu mask size limitation of no more than 1024 CPUs. This limitation confines NUMA awareness of Perf tool in record mode, thru --affinity option, to the first 1024 CPUs on machines with larger amount of CPUs.
This patch set enables Perf tool to overcome 1024 CPUs limitation by using a dedicated struct mmap_cpu_mask type and applying tool's bitmap API operations to manipulate affinity masks of the tool's thread and the mmaped data buffers.
tools bitmap API has been extended with bitmap_free() function and bitmap_equal() operation whose implementation is derived from the kernel one.
---
Changes in v5:
- avoided allocation of mmap affinity masks in case of rec->opts.affinity == PERF_AFFINITY_SYS
Changes in v4:
- renamed perf_mmap__print_cpu_mask() to mmap_cpu_mask__scnprintf()
- avoided checking mask bits for NULL prior calling bitmask_free()
- avoided thread affinity mask allocation for case of rec->opts.affinity == PERF_AFFINITY_SYS
Changes in v3:
- implemented perf_mmap__print_cpu_mask() function
- use perf_mmap__print_cpu_mask() to log thread and mmap cpus masks when verbose level is equal to 2
Changes in v2:
- implemented bitmap_free() for symmetry with bitmap_alloc()
- capitalized MMAP_CPU_MASK_BYTES() macro
- returned -1 from perf_mmap__setup_affinity_mask()
- implemented releasing of masks using bitmap_free()
- moved debug printing under -vv option
---
Alexey Budankov (3):
tools bitmap: implement bitmap_equal() operation at bitmap API perf mmap: declare type for cpu mask of arbitrary length perf record: adapt affinity to machines with #CPUs > 1K
tools/include/linux/bitmap.h | 30 +++++++++++++++++++++++++++ tools/lib/bitmap.c | 15 ++++++++++++++ tools/perf/builtin-record.c | 28 +++++++++++++++++++------ tools/perf/util/mmap.c | 40 ++++++++++++++++++++++++++++++------ tools/perf/util/mmap.h | 13 +++++++++++- 5 files changed, 113 insertions(+), 13 deletions(-)
---
Validation:
# tools/perf/perf record -vv -- ls
Using CPUID GenuineIntel-6-5E-3
intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branch
nr_cblocks: 0
affinity: SYS
mmap flush: 1
comp level: 0
------------------------------------------------------------
perf_event_attr:
size 120
{ sample_period, sample_freq } 4000
sample_type IP|TID|TIME|PERIOD
read_format ID
disabled 1
inherit 1
mmap 1
comm 1
freq 1
enable_on_exec 1
task 1
precise_ip 3
sample_id_all 1
exclude_guest 1
mmap2 1
comm_exec 1
ksymbol 1
bpf_event 1
------------------------------------------------------------
sys_perf_event_open: pid 23718 cpu 0 group_fd -1 flags 0x8 = 4
sys_perf_event_open: pid 23718 cpu 1 group_fd -1 flags 0x8 = 5
sys_perf_event_open: pid 23718 cpu 2 group_fd -1 flags 0x8 = 6
sys_perf_event_open: pid 23718 cpu 3 group_fd -1 flags 0x8 = 9
sys_perf_event_open: pid 23718 cpu 4 group_fd -1 flags 0x8 = 10
sys_perf_event_open: pid 23718 cpu 5 group_fd -1 flags 0x8 = 11
sys_perf_event_open: pid 23718 cpu 6 group_fd -1 flags 0x8 = 12
sys_perf_event_open: pid 23718 cpu 7 group_fd -1 flags 0x8 = 13
mmap size 528384B
0x7f3e06e060b8: mmap mask[8]:
0x7f3e06e16180: mmap mask[8]:
0x7f3e06e26248: mmap mask[8]:
0x7f3e06e36310: mmap mask[8]:
0x7f3e06e463d8: mmap mask[8]:
0x7f3e06e564a0: mmap mask[8]:
0x7f3e06e66568: mmap mask[8]:
0x7f3e06e76630: mmap mask[8]:
------------------------------------------------------------
perf_event_attr:
type 1
size 120
config 0x9
watermark 1
sample_id_all 1
bpf_event 1
{ wakeup_events, wakeup_watermark } 1
------------------------------------------------------------
sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14
sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15
sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16
sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17
sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18
sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19
sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20
sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21
mmap size 528384B
0x7f3e0697d0b8: mmap mask[8]:
0x7f3e0698d180: mmap mask[8]:
0x7f3e0699d248: mmap mask[8]:
0x7f3e069ad310: mmap mask[8]:
0x7f3e069bd3d8: mmap mask[8]:
0x7f3e069cd4a0: mmap mask[8]:
0x7f3e069dd568: mmap mask[8]:
0x7f3e069ed630: mmap mask[8]:
Synthesizing TSC conversion information
arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux
block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt
[ perf record: Woken up 1 times to write data ]
Looking at the vmlinux_path (8 entries long)
Using vmlinux for symbols
[ perf record: Captured and wrote 0.013 MB perf.data (8 samples) ]tools/perf/perf record -vv --affinity=cpu -- ls
thread mask[8]: empty
Using CPUID GenuineIntel-6-5E-3intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branchnr_cblocks: 0
affinity: CPU
mmap flush: 1
comp level: 0
------------------------------------------------------------perf_event_attr:
size 120
{ sample_period, sample_freq } 4000
sample_type IP|TID|TIME|PERIOD
read_format ID
disabled 1
inherit 1
mmap 1
comm 1
freq 1
enable_on_exec 1
task 1
precise_ip 3
sample_id_all 1
exclude_guest 1
mmap2 1
comm_exec 1
ksymbol 1
bpf_event 1------------------------------------------------------------sys_perf_event_open: pid 23713 cpu 0 group_fd -1 flags 0x8 = 4sys_perf_event_open: pid 23713 cpu 1 group_fd -1 flags 0x8 = 5sys_perf_event_open: pid 23713 cpu 2 group_fd -1 flags 0x8 = 6sys_perf_event_open: pid 23713 cpu 3 group_fd -1 flags 0x8 = 9sys_perf_event_open: pid 23713 cpu 4 group_fd -1 flags 0x8 = 10sys_perf_event_open: pid 23713 cpu 5 group_fd -1 flags 0x8 = 11sys_perf_event_open: pid 23713 cpu 6 group_fd -1 flags 0x8 = 12sys_perf_event_open: pid 23713 cpu 7 group_fd -1 flags 0x8 = 13 mmap size 528384B0x7f3e005bc0b8: mmap mask[8]: 00x7f3e005cc180: mmap mask[8]: 10x7f3e005dc248: mmap mask[8]: 20x7f3e005ec310: mmap mask[8]: 30x7f3e005fc3d8: mmap mask[8]: 40x7f3e0060c4a0: mmap mask[8]: 50x7f3e0061c568: mmap mask[8]: 60x7f3e0062c630: mmap mask[8]: 7
------------------------------------------------------------perf_event_attr:
type 1 size 120 config 0x9 watermark 1 sample_id_all 1 bpf_event 1
{ wakeup_events, wakeup_watermark } 1
------------------------------------------------------------sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21 mmap size 528384B0x7f3e001330b8: mmap mask[8]:0x7f3e00143180: mmap mask[8]:0x7f3e00153248: mmap mask[8]:0x7f3e00163310: mmap mask[8]:0x7f3e001733d8: mmap mask[8]:0x7f3e001834a0: mmap mask[8]:0x7f3e00193568: mmap mask[8]:0x7f3e001a3630: mmap mask[8]:
Synthesizing TSC conversion information0x9c9ff0: thread mask[8]: 00x9c9ff0: thread mask[8]: 10x9c9ff0: thread mask[8]: 20x9c9ff0: thread mask[8]: 30x9c9ff0: thread mask[8]: 4
arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt0x9c9ff0: thread mask[8]: 50x9c9ff0: thread mask[8]: 60x9c9ff0: thread mask[8]: 70x9c9ff0: thread mask[8]: 00x9c9ff0: thread mask[8]: 10x9c9ff0: thread mask[8]: 20x9c9ff0: thread mask[8]: 30x9c9ff0: thread mask[8]: 40x9c9ff0: thread mask[8]: 50x9c9ff0: thread mask[8]: 60x9c9ff0: thread mask[8]: 7
[ perf record: Woken up 0 times to write data ] 0x9c9ff0: thread mask[8]: 00x9c9ff0: thread mask[8]: 10x9c9ff0: thread mask[8]: 20x9c9ff0: thread mask[8]: 30x9c9ff0: thread mask[8]: 40x9c9ff0: thread mask[8]: 50x9c9ff0: thread mask[8]: 60x9c9ff0: thread mask[8]: 7
Looking at the vmlinux_path (8 entries long)
Using vmlinux for symbols
...
[ perf record: Captured and wrote 0.013 MB perf.data (10 samples) ]