Re: [PATCH V4 0/6] Add the page size in the perf record (user tools)

[Athira Rajeev]

> On 06-Jan-2021, at 1:27 AM, kan.liang@xxxxxxxxxxxxxxx wrote:
> 
> From: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
> 
> Changes since V3:
> - Rebase on top of acme's perf/core branch
>  commit c07b45a355ee ("perf record: Tweak "Lowering..." warning in record_opts__config_freq")
> 
> Changes since V2:
> - Rebase on top of acme perf/core branch
>  commit eec7b53d5916 ("perf test: Make sample-parsing test aware of PERF_SAMPLE_{CODE,DATA}_PAGE_SIZE")
> - Use unit_number__scnprintf() in get_page_size_name()
> - Emit warning about kernel not supporting the code page size sample_type bit
> 
> Changes since V1:
> - Fix the compile warning with GCC 10
> - Add Acked-by from Namhyung Kim
> 
> Current perf can report both virtual addresses and physical addresses,
> but not the page size. Without the page size information of the utilized
> page, users cannot decide whether to promote/demote large pages to
> optimize memory usage.
> 
> The kernel patches have been merged into tip perf/core branch,
> commit 8d97e71811aa ("perf/core: Add PERF_SAMPLE_DATA_PAGE_SIZE")
> commit 76a5433f95f3 ("perf/x86/intel: Support PERF_SAMPLE_DATA_PAGE_SIZE")
> commit 4cb6a42e4c4b ("powerpc/perf: Support PERF_SAMPLE_DATA_PAGE_SIZE")
> commit 995f088efebe ("perf/core: Add support for PERF_SAMPLE_CODE_PAGE_SIZE")
> commit 51b646b2d9f8 ("perf,mm: Handle non-page-table-aligned hugetlbfs")
> 
> and Peter's perf/core branch
> commit 524680ce47a1 ("mm/gup: Provide gup_get_pte() more generic")
> commit 44a35d6937d2 ("mm: Introduce pXX_leaf_size()")
> commit 2f1e2f091ad0 ("perf/core: Fix arch_perf_get_page_size()")
> commit 7649e44aacdd ("arm64/mm: Implement pXX_leaf_size() support")
> commit 1df1ae7e262c ("sparc64/mm: Implement pXX_leaf_size() support")
> 
> This patch set is to enable the page size support in user tools.

Hi Kan Liang,

I am trying to check this series on powerpc. 

# perf mem --phys-data --data-page-size record <workload>

To my observation, some of the samples returned zero size and comes as ’N/A’ in the perf report 

# perf mem --phys-data --data-page-size report 

For fetching the page size, though initially there was a weak function added ( as arch_perf_get_page_size ) here: 

https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/commit/?h=perf/core&id=51b646b2d9f84d6ff6300e3c1d09f2be4329a424

later I see it got removed here: 

https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/commit/?h=perf/core&id=8af26be062721e52eba1550caf50b712f774c5fd 

I picked kernel changes from git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git , or I am missing something ?

Thanks
Athira

> 
> Kan Liang (3):
>  perf mem: Clean up output format
>  perf mem: Support data page size
>  perf tools: Add support for PERF_SAMPLE_CODE_PAGE_SIZE
> 
> Stephane Eranian (3):
>  perf script: Add support for PERF_SAMPLE_CODE_PAGE_SIZE
>  perf report: Add support for PERF_SAMPLE_CODE_PAGE_SIZE
>  perf test: Add test case for PERF_SAMPLE_CODE_PAGE_SIZE
> 
> tools/perf/Documentation/perf-mem.txt     |   3 +
> tools/perf/Documentation/perf-record.txt  |   3 +
> tools/perf/Documentation/perf-report.txt  |   1 +
> tools/perf/Documentation/perf-script.txt  |   2 +-
> tools/perf/builtin-mem.c                  | 111 +++++++++++-----------
> tools/perf/builtin-record.c               |   2 +
> tools/perf/builtin-script.c               |  13 ++-
> tools/perf/tests/sample-parsing.c         |   4 +
> tools/perf/util/event.h                   |   1 +
> tools/perf/util/evsel.c                   |  18 +++-
> tools/perf/util/evsel.h                   |   1 +
> tools/perf/util/hist.c                    |   2 +
> tools/perf/util/hist.h                    |   1 +
> tools/perf/util/perf_event_attr_fprintf.c |   2 +-
> tools/perf/util/record.h                  |   1 +
> tools/perf/util/session.c                 |   3 +
> tools/perf/util/sort.c                    |  26 +++++
> tools/perf/util/sort.h                    |   2 +
> tools/perf/util/synthetic-events.c        |   8 ++
> 19 files changed, 144 insertions(+), 60 deletions(-)
> 
> -- 
> 2.25.1
> 
> 
>