pmc: add PERF, RDPMC, row, sampling and unload tests
Needs ReviewPublic
Actions

Authored by ojanerif_amd.com on Fri, Oct 30, 8:12 PM.

Project Tags

None

Referenced Files

None

Subscribers

imp

Details

Reviewers

mhorne
netchild
ali_mashtizadeh.com
afscoelho_gmail.com
anascime_amd.com

Summary

pmc_perf_test: the PERF class (MPERF, APERF, IRPERF).
pmc_rdpmc_test: pmc_get_msr() and RDPMC.
pmc_row_test: row disposition per mode; system sampling needs a log.
pmc_ss_test: system sampling into a log, and its interrupt accounting.
pmc_unload_test: unloading hwpmc.ko leaks no memory.

Sponsored by: Advanced Micro Devices, Inc.
Signed-off-by: Osvaldo Janeri Filho <ojanerif@amd.com>
Signed-off-by: Davi C.A <davi.chavesazevedo@amd.com>

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Skipped

Unit

Tests Skipped

Event Timeline

ojanerif_amd.com created this revision.Fri, Oct 30, 8:12 PM

Herald added a subscriber: imp. · View Herald TranscriptFri, Oct 30, 8:12 PM

ojanerif_amd.com requested review of this revision.Fri, Oct 30, 8:12 PM

mhorne requested changes to this revision.Thu, Nov 5, 2:11 PM

mhorne added inline comments.

tests/sys/pmc/pmc_unload_test.c
23–43	This seems too fragile to me, building these tests around a kernel warning string. I am also confused how it would run at all, given `M_PMC` is not defined within the `hwpmc` module, but in the kernel itself. Are you able to trigger a failure by inserting a purposeful leak? How does it work? You should look into `libmemstat(3)` for a real interface for querying memory types and stats.

This revision now requires changes to proceed.Thu, Nov 5, 2:11 PM

Right, M_PMC is the kernel's, so it could never fail. Switched to libmemstat(3): record the pmc type with hwpmc unloaded, then load/use/unload three times and compare. A hwpmc.ko that leaks one M_PMC allocation on unload fails it (2 -> 3 allocations after the first unload).