linux.git
16 months agoperf test: Simplify metric value validation test final report
Weilin Wang [Tue, 30 Jan 2024 18:09:07 +0000 (10:09 -0800)]
perf test: Simplify metric value validation test final report

The original test report was too complicated to read with information
that not really useful. This new update simplify the report which should
largely improve the readibility.

Signed-off-by: Weilin Wang <weilin.wang@intel.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Perry Taylor <perry.taylor@intel.com>
Cc: Samantha Alt <samantha.alt@intel.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240130180907.639729-1-weilin.wang@intel.com
16 months agoperf report: Prevent segfault with --no-parent
Andi Kleen [Tue, 30 Jan 2024 18:55:52 +0000 (10:55 -0800)]
perf report: Prevent segfault with --no-parent

Prevent a perf report segfault with the (non sensical) --no-parent
option

Signed-off-By: Andi Kleen <ak@linux.intel.com>
Acked-by: Namhyung <namhyung@kernel.org>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240130185552.150578-1-ak@linux.intel.com
16 months agoperf evsel: Fix duplicate initialization of data->id in evsel__parse_sample()
Yang Jihong [Sat, 27 Jan 2024 02:57:56 +0000 (02:57 +0000)]
perf evsel: Fix duplicate initialization of data->id in evsel__parse_sample()

data->id has been initialized at line 2362, remove duplicate initialization.

Fixes: 3ad31d8a0df2 ("perf evsel: Centralize perf_sample initialization")
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Reviewed-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240127025756.4041808-1-yangjihong1@huawei.com
16 months agoperf evsel: Rename get_states() to parse_task_states() and make it public
Ze Gao [Tue, 23 Jan 2024 07:02:11 +0000 (02:02 -0500)]
perf evsel: Rename get_states() to parse_task_states() and make it public

Since get_states() assumes the existence of libtraceevent, so move
to where it should belong, i.e, util/trace-event-parse.c, and also
rename it to parse_task_states().

Leave evsel_getstate() untouched as it fits well in the evsel
category.

Also make some necessary tweaks for python support, and get it
verified with: perf test python.

Signed-off-by: Ze Gao <zegao@tencent.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240123070210.1669843-2-zegao@tencent.com
16 months agoperf tools: Add -H short option for --hierarchy
Namhyung Kim [Thu, 25 Jan 2024 05:51:24 +0000 (21:51 -0800)]
perf tools: Add -H short option for --hierarchy

I found the hierarchy mode useful, but it's easy to make a typo when
using it.  Let's add a short option for that.

Also update the documentation. :)

Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Link: https://lore.kernel.org/r/20240125055124.1579617-1-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf pmu: Treat the msr pmu as software
Ian Rogers [Wed, 24 Jan 2024 23:42:00 +0000 (15:42 -0800)]
perf pmu: Treat the msr pmu as software

The msr PMU is a software one, meaning msr events may be grouped
with events in a hardware context. As the msr PMU isn't marked as a
software PMU by perf_pmu__is_software, groups with the msr PMU in
are broken and the msr events placed in a different group. This
may lead to multiplexing errors where a hardware event isn't
counted while the msr event, such as tsc, is. Fix all of this by
marking the msr PMU as software, which agrees with the driver.

Before:
```
$ perf stat -e '{slots,tsc}' -a true
WARNING: events were regrouped to match PMUs

 Performance counter stats for 'system wide':

         1,750,335      slots
         4,243,557      tsc

       0.001456717 seconds time elapsed
```

After:
```
$ perf stat -e '{slots,tsc}' -a true
 Performance counter stats for 'system wide':

        12,526,380      slots
         3,415,163      tsc

       0.001488360 seconds time elapsed
```

Fixes: 251aa040244a ("perf parse-events: Wildcard most "numeric" events")
Signed-off-by: Ian Rogers <irogers@google.com>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Cc: James Clark <james.clark@arm.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Edward Baker <edward.baker@intel.com>
Cc: Perry Taylor <perry.taylor@intel.com>
Cc: Samantha Alt <samantha.alt@intel.com>
Cc: Weilin Wang <weilin.wang@intel.com>
Link: https://lore.kernel.org/r/20240124234200.1510417-1-irogers@google.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf test: Skip test_arm_callgraph_fp.sh if unwinding isn't built in
James Clark [Tue, 23 Jan 2024 16:39:02 +0000 (16:39 +0000)]
perf test: Skip test_arm_callgraph_fp.sh if unwinding isn't built in

Even though this is a frame pointer unwind test, it's testing that a
frame pointer stack can be augmented correctly with a partial
Dwarf unwind. So add a feature check so that this test skips instead of
fails if Dwarf unwinding isn't present.

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Spoorthy S <spoorts2@in.ibm.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Link: https://lore.kernel.org/r/20240123163903.350306-3-james.clark@arm.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf version: Display availability of HAVE_DWARF_UNWIND_SUPPORT
James Clark [Tue, 23 Jan 2024 16:39:01 +0000 (16:39 +0000)]
perf version: Display availability of HAVE_DWARF_UNWIND_SUPPORT

Even though unwinding depends on either HAVE_DWARF_SUPPORT or
HAVE_LIBUNWIND, scripts testing unwinding can't just look for the
existence of either of those flags. This is because
NO_LIBDW_DWARF_UNWIND=1 can disable unwinding with libdw, but libdw will
still be linked leaving HAVE_DWARF_SUPPORT turned on. Presumably because
it is used for things other than unwinding, so I don't think this needs
to be fixed.

HAVE_DWARF_UNWIND_SUPPORT already takes the combination of all those
things into account, and is used to gate the built in tests like "Test
dwarf unwind", so add it to the feature list output so that it can be
used by the script tests too.

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Spoorthy S <spoorts2@in.ibm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Link: https://lore.kernel.org/r/20240123163903.350306-2-james.clark@arm.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf evlist: Fix evlist__new_default() for > 1 core PMU
James Clark [Wed, 24 Jan 2024 09:43:57 +0000 (09:43 +0000)]
perf evlist: Fix evlist__new_default() for > 1 core PMU

The 'Session topology' test currently fails with this message when
evlist__new_default() opens more than one event:

  32: Session topology                                                :
  --- start ---
  templ file: /tmp/perf-test-vv5YzZ
  Using CPUID 0x00000000410fd070
  Opening: unknown-hardware:HG
  ------------------------------------------------------------
  perf_event_attr:
    type                             0 (PERF_TYPE_HARDWARE)
    config                           0xb00000000
    disabled                         1
  ------------------------------------------------------------
  sys_perf_event_open: pid 0  cpu -1  group_fd -1  flags 0x8 = 4
  Opening: unknown-hardware:HG
  ------------------------------------------------------------
  perf_event_attr:
    type                             0 (PERF_TYPE_HARDWARE)
    config                           0xa00000000
    disabled                         1
  ------------------------------------------------------------
  sys_perf_event_open: pid 0  cpu -1  group_fd -1  flags 0x8 = 5
  non matching sample_type
  FAILED tests/topology.c:73 can't get session
  ---- end ----
  Session topology: FAILED!

This is because when re-opening the file and parsing the header, Perf
expects that any file that has more than one event has the sample ID
flag set. Perf record already sets the flag in a similar way when there
is more than one event, so add the same logic to evlist__new_default().

evlist__new_default() is only currently used in tests, so I don't
expect this change to have any other side effects. The other tests that
use it don't save and re-open the file so don't hit this issue.

The session topology test has been failing on Arm big.LITTLE platforms
since commit 251aa040244a ("perf parse-events: Wildcard most
"numeric" events") when evlist__new_default() started opening multiple
events for 'cycles'.

Fixes: 251aa040244a ("perf parse-events: Wildcard most "numeric" events")
Closes: https://lore.kernel.org/lkml/CAP-5=fWVQ-7ijjK3-w1q+k2WYVNHbAcejb-xY0ptbjRw476VKA@mail.gmail.com/
Tested-by: Ian Rogers <irogers@google.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Tested-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: James Clark <james.clark@arm.com>
Cc: Changbin Du <changbin.du@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Link: https://lore.kernel.org/r/20240124094358.489372-1-james.clark@arm.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up perf_pmus__num_mem_pmus()
Kan Liang [Tue, 23 Jan 2024 18:50:36 +0000 (10:50 -0800)]
perf mem: Clean up perf_pmus__num_mem_pmus()

The number of mem PMUs can be calculated by searching the
perf_pmus__scan_mem().

Remove the ARCH specific perf_pmus__num_mem_pmus()

Tested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: ravi.bangoria@amd.com
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-8-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up perf_mem_events__record_args()
Kan Liang [Tue, 23 Jan 2024 18:50:35 +0000 (10:50 -0800)]
perf mem: Clean up perf_mem_events__record_args()

The current code iterates all memory PMUs. It doesn't matter if the
system has only one memory PMU or multiple PMUs. The check of
perf_pmus__num_mem_pmus() is not required anymore.

The rec_tmp is not used in c2c and mem. Removing them as well.

Suggested-by: Leo Yan <leo.yan@linaro.org>
Tested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: ravi.bangoria@amd.com
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-7-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up is_mem_loads_aux_event()
Kan Liang [Tue, 23 Jan 2024 18:50:34 +0000 (10:50 -0800)]
perf mem: Clean up is_mem_loads_aux_event()

The aux_event can be retrieved from the perf_pmu now. Implement a
generic support.

Reviewed-by: Ian Rogers <irogers@google.com>
Tested-by: Ravi Bangoria <ravi.bangoria@amd.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-6-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up perf_mem_event__supported()
Kan Liang [Tue, 23 Jan 2024 18:50:33 +0000 (10:50 -0800)]
perf mem: Clean up perf_mem_event__supported()

For some ARCHs, e.g., ARM and AMD, to get the availability of the
mem-events, perf checks the existence of a specific PMU. For the other
ARCHs, e.g., Intel and Power, perf has to check the existence of some
specific events.

The current perf only iterates the mem-events-supported PMUs. It's not
required to check the existence of a specific PMU anymore.

Rename sysfs_name to event_name, which stores the specific mem-events.
Perf only needs to check those events for the availability of the
mem-events.

Rename perf_mem_event__supported to perf_pmu__mem_events_supported.

Reviewed-by: Ian Rogers <irogers@google.com>
Tested-by: Ravi Bangoria <ravi.bangoria@amd.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-5-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up perf_mem_events__name()
Kan Liang [Tue, 23 Jan 2024 18:50:32 +0000 (10:50 -0800)]
perf mem: Clean up perf_mem_events__name()

Introduce a generic perf_mem_events__name(). Remove the ARCH-specific
one.

The mem_load events may have a different format. Add ldlat and aux_event
in the struct perf_mem_event to indicate the format and the extra aux
event.

Add perf_mem_events_intel_aux[] to support the extra mem_load_aux event.

Rename perf_mem_events__name to perf_pmu__mem_events_name.

Reviewed-by: Ian Rogers <irogers@google.com>
Tested-by: Ravi Bangoria <ravi.bangoria@amd.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-4-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Clean up perf_mem_events__ptr()
Kan Liang [Tue, 23 Jan 2024 18:50:31 +0000 (10:50 -0800)]
perf mem: Clean up perf_mem_events__ptr()

The mem_events can be retrieved from the struct perf_pmu now. An ARCH
specific perf_mem_events__ptr() is not required anymore. Remove all of
them.

The Intel hybrid has multiple mem-events-supported PMUs. But they share
the same mem_events. Other ARCHs only support one mem-events-supported
PMU. In the configuration, it's good enough to only configure the
mem_events for one PMU. Add perf_mem_events_find_pmu() which returns the
first mem-events-supported PMU.

In the perf_mem_events__init(), the perf_pmus__scan() is not required
anymore. It avoids checking the sysfs for every PMU on the system.

Make the perf_mem_events__record_args() more generic. Remove the
perf_mem_events__print_unsupport_hybrid().

Since pmu is added as a new parameter, rename perf_mem_events__ptr() to
perf_pmu__mem_events_ptr(). Several other functions also do a similar
rename.

Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: Kajol Jain <kjain@linux.ibm.com>
Tested-by: Ravi Bangoria <ravi.bangoria@amd.com>
Tested-by: Kajol jain <kjain@linux.ibm.com>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: james.clark@arm.com
Cc: will@kernel.org
Cc: leo.yan@linaro.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-3-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf mem: Add mem_events into the supported perf_pmu
Kan Liang [Tue, 23 Jan 2024 18:50:30 +0000 (10:50 -0800)]
perf mem: Add mem_events into the supported perf_pmu

With the mem_events, perf doesn't need to read sysfs for each PMU to
find the mem-events-supported PMU. The patch also makes it possible to
clean up the related __weak functions later.

The patch is only to add the mem_events into the perf_pmu for all ARCHs.
It will be used in the later cleanup patches.

Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: Kajol Jain <kjain@linux.ibm.com>
Tested-by: Ravi Bangoria <ravi.bangoria@amd.com>
Tested-by: Leo Yan <leo.yan@linaro.org>
Tested-by: Kajol Jain <kjain@linux.ibm.com>
Suggested-by: Leo Yan <leo.yan@linaro.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: will@kernel.org
Cc: mike.leach@linaro.org
Cc: renyu.zj@linux.alibaba.com
Cc: yuhaixin.yhx@linux.alibaba.com
Cc: tmricht@linux.ibm.com
Cc: atrajeev@linux.vnet.ibm.com
Cc: linux-arm-kernel@lists.infradead.org
Cc: john.g.garry@oracle.com
Link: https://lore.kernel.org/r/20240123185036.3461837-2-kan.liang@linux.intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf sched: Commit to evsel__taskstate() to parse task state info
Ze Gao [Tue, 23 Jan 2024 02:24:25 +0000 (21:24 -0500)]
perf sched: Commit to evsel__taskstate() to parse task state info

Now that we have evsel__taskstate() which no longer relies on the
hardcoded task state string and has good backward compatibility,
we have a good reason to use it.

Note TASK_STATE_TO_CHAR_STR and task bitmasks are useless now so
we remove them for good. And now we pass the state info back and
forth in a symbolic char which explains itself well instead.

Signed-off-by: Ze Gao <zegao@tencent.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: https://lore.kernel.org/r/20240123022425.1611483-1-zegao@tencent.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf util: Add evsel__taskstate() to parse the task state info instead
Ze Gao [Mon, 22 Jan 2024 07:08:58 +0000 (02:08 -0500)]
perf util: Add evsel__taskstate() to parse the task state info instead

Now that we have the __prinf_flags() parsing routines, we add a new
helper evsel__taskstate() to extract the task state info from the
recorded data.

Signed-off-by: Ze Gao <zegao@tencent.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: https://lore.kernel.org/r/20240122070859.1394479-5-zegao@tencent.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf util: Add helpers to parse task state string from libtraceevent
Ze Gao [Mon, 22 Jan 2024 07:08:57 +0000 (02:08 -0500)]
perf util: Add helpers to parse task state string from libtraceevent

Perf uses a hard coded string "RSDTtXZPI" to index the sched_switch
prev_state field raw bitmask value. This works well except for when
the kernel changes this string, in which case this will break again.

Instead we add a new way to parse task state string from tracepoint
print format already recorded by perf, which eliminates the further
dependencies with this hardcode and unmaintainable macro, and this
is exactly what libtraceevent[1] does for now.

So we borrow the print flags parsing logic from libtraceevent[1].
And in get_states(), we walk the print arguments until the
__print_flags() for the target state field is found, and use that to
build the states string for future parsing.

[1]: https://lore.kernel.org/linux-trace-devel/20231224140732.7d41698d@rorschach.local.home/

Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Ze Gao <zegao@tencent.com>
Link: https://lore.kernel.org/r/20240122070859.1394479-4-zegao@tencent.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf sched: Sync state char array with the kernel
Ze Gao [Mon, 22 Jan 2024 07:08:56 +0000 (02:08 -0500)]
perf sched: Sync state char array with the kernel

Update state char array to match the latest kernel definitions and
remove unused state mapping macros.

Note this is the preparing patch for get rid of the way to parse
process state from raw bitmask value. Instead we are going to
parse it from the recorded tracepoint print format, and this change
marks why we're doing it.

Signed-off-by: Ze Gao <zegao@tencent.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: https://lore.kernel.org/r/20240122070859.1394479-3-zegao@tencent.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf data: Minor code style alignment cleanup
Yang Jihong [Fri, 19 Jan 2024 04:03:04 +0000 (04:03 +0000)]
perf data: Minor code style alignment cleanup

Minor code style alignment cleanup for perf_data__switch() and
perf_data__write().

No functional change.

Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240119040304.3708522-4-yangjihong1@huawei.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf record: Check conflict between '--timestamp-filename' option and pipe mode befor...
Yang Jihong [Fri, 19 Jan 2024 04:03:03 +0000 (04:03 +0000)]
perf record: Check conflict between '--timestamp-filename' option and pipe mode before recording

In pipe mode, no need to switch perf data output, therefore,
'--timestamp-filename' option should not take effect.
Check the conflict before recording and output WARNING.
In this case, the check pipe mode in perf_data__switch() can be removed.

Before:

  # perf record --timestamp-filename -o- perf test -w noploop | perf report -i- --percent-limit=1
  # To display the perf.data header info, please use --header/--header-only options.
  #
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Dump -.2024011812110182 ]
  #
  # Total Lost Samples: 0
  #
  # Samples: 4K of event 'cycles:P'
  # Event count (approx.): 2176784359
  #
  # Overhead  Command  Shared Object         Symbol
  # ........  .......  ....................  ......................................
  #
      97.83%  perf     perf                  [.] noploop

  #
  # (Tip: Print event counts in CSV format with: perf stat -x,)
  #

After:

  # perf record --timestamp-filename -o- perf test -w noploop | perf report -i- --percent-limit=1
  WARNING: --timestamp-filename option is not available in pipe mode.
  # To display the perf.data header info, please use --header/--header-only options.
  #
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.000 MB - ]
  #
  # Total Lost Samples: 0
  #
  # Samples: 4K of event 'cycles:P'
  # Event count (approx.): 2185575421
  #
  # Overhead  Command  Shared Object          Symbol
  # ........  .......  .....................  .............................................
  #
      97.75%  perf     perf                   [.] noploop

  #
  # (Tip: Profiling branch (mis)predictions with: perf record -b / perf report)
  #

Fixes: ecfd7a9c044e ("perf record: Add '--timestamp-filename' option to append timestamp to output file name")
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240119040304.3708522-3-yangjihong1@huawei.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf record: Fix possible incorrect free in record__switch_output()
Yang Jihong [Fri, 19 Jan 2024 04:03:02 +0000 (04:03 +0000)]
perf record: Fix possible incorrect free in record__switch_output()

perf_data__switch() may not assign a legal value to 'new_filename'.
In this case, 'new_filename' uses the on-stack value, which may cause a
incorrect free and unexpected result.

Fixes: 03724b2e9c45 ("perf record: Allow to limit number of reported perf.data files")
Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240119040304.3708522-2-yangjihong1@huawei.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf dwarf-aux: Check allowed DWARF Ops
Namhyung Kim [Wed, 17 Jan 2024 06:26:57 +0000 (22:26 -0800)]
perf dwarf-aux: Check allowed DWARF Ops

The DWARF location expression can be fairly complex and it'd be hard
to match it with the condition correctly.  So let's be conservative
and only allow simple expressions.  For now it just checks the first
operation in the list.  The following operations looks ok:

 * DW_OP_stack_value
 * DW_OP_deref_size
 * DW_OP_deref
 * DW_OP_piece

To refuse complex (and unsupported) location expressions, add
check_allowed_ops() to compare the rest of the list.  It seems earlier
result contained those unsupported expressions.  For example, I found
some local struct variable is placed like below.

 <2><43d1517>: Abbrev Number: 62 (DW_TAG_variable)
    <43d1518>   DW_AT_location    : 15 byte block: 91 50 93 8 91 78 93 4 93 84 8 91 68 93 4
        (DW_OP_fbreg: -48; DW_OP_piece: 8;
         DW_OP_fbreg: -8; DW_OP_piece: 4;
         DW_OP_piece: 1028;
         DW_OP_fbreg: -24; DW_OP_piece: 4)

Another example is something like this.

    0057c8be ffffffffffffffff ffffffff812109f0 (base address)
    0057c8ce ffffffff812112b5 ffffffff812112c8 (DW_OP_breg3 (rbx): 0;
                                                DW_OP_constu: 18446744073709551612;
                                                DW_OP_and;
                                                DW_OP_stack_value)

It should refuse them.  After the change, the stat shows:

  Annotate data type stats:
  total 294, ok 158 (53.7%), bad 136 (46.3%)
  -----------------------------------------------------------
          30 : no_sym
          32 : no_mem_ops
          53 : no_var
          14 : no_typeinfo
           7 : bad_offset

Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20240117062657.985479-10-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Support stack variables
Namhyung Kim [Wed, 17 Jan 2024 06:26:56 +0000 (22:26 -0800)]
perf annotate-data: Support stack variables

Local variables are allocated in the stack and the location list
should look like base register(s) and an offset.  Extend the
die_find_variable_by_reg() to handle the following expressions

 * DW_OP_breg{0..31}
 * DW_OP_bregx
 * DW_OP_fbreg

Ususally DWARF subprogram entries have frame base information and
use it to locate stack variable like below:

 <2><43d1575>: Abbrev Number: 62 (DW_TAG_variable)
    <43d1576>   DW_AT_location    : 2 byte block: 91 7c         (DW_OP_fbreg: -4)  <--- here
    <43d1579>   DW_AT_name        : (indirect string, offset: 0x2c00c9): i
    <43d157d>   DW_AT_decl_file   : 1
    <43d157e>   DW_AT_decl_line   : 78
    <43d157f>   DW_AT_type        : <0x43d19d7>

I found some differences on saving the frame base between gcc and clang.
The gcc uses the CFA to get the base so it needs to check the current
frame's CFI info.  In this case, stack offset needs to be adjusted from
the start of the CFA.

 <1><1bb8d>: Abbrev Number: 102 (DW_TAG_subprogram)
    <1bb8e>   DW_AT_name        : (indirect string, offset: 0x74d41): kernel_init
    <1bb92>   DW_AT_decl_file   : 2
    <1bb92>   DW_AT_decl_line   : 1440
    <1bb94>   DW_AT_decl_column : 18
    <1bb95>   DW_AT_prototyped  : 1
    <1bb95>   DW_AT_type        : <0xcc>
    <1bb99>   DW_AT_low_pc      : 0xffffffff81bab9e0
    <1bba1>   DW_AT_high_pc     : 0x1b2
    <1bba9>   DW_AT_frame_base  : 1 byte block: 9c      (DW_OP_call_frame_cfa)  <------ here
    <1bbab>   DW_AT_call_all_calls: 1
    <1bbab>   DW_AT_sibling     : <0x1bf5a>

While clang sets it to a register directly and it can check the register
and offset in the instruction directly.

 <1><43d1542>: Abbrev Number: 60 (DW_TAG_subprogram)
    <43d1543>   DW_AT_low_pc      : 0xffffffff816a7c60
    <43d154b>   DW_AT_high_pc     : 0x98
    <43d154f>   DW_AT_frame_base  : 1 byte block: 56    (DW_OP_reg6 (rbp))  <---------- here
    <43d1551>   DW_AT_GNU_all_call_sites: 1
    <43d1551>   DW_AT_name        : (indirect string, offset: 0x3bce91): foo
    <43d1555>   DW_AT_decl_file   : 1
    <43d1556>   DW_AT_decl_line   : 75
    <43d1557>   DW_AT_prototyped  : 1
    <43d1557>   DW_AT_type        : <0x43c7332>
    <43d155b>   DW_AT_external    : 1

Also it needs to update the offset after finding the type like global
variables since the offset was from the frame base.  Factor out
match_var_offset() to check global and local variables in the same way.

The type stats are improved too:

  Annotate data type stats:
  total 294, ok 160 (54.4%), bad 134 (45.6%)
  -----------------------------------------------------------
          30 : no_sym
          32 : no_mem_ops
          51 : no_var
          14 : no_typeinfo
           7 : bad_offset

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-9-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf dwarf-aux: Add die_get_cfa()
Namhyung Kim [Wed, 17 Jan 2024 06:26:55 +0000 (22:26 -0800)]
perf dwarf-aux: Add die_get_cfa()

The die_get_cfa() is to get frame base register and offset at the given
instruction address (pc).  This info will be used to locate stack
variables which have location expression using DW_OP_fbreg.

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-8-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Support global variables
Namhyung Kim [Wed, 17 Jan 2024 06:26:54 +0000 (22:26 -0800)]
perf annotate-data: Support global variables

Global variables are accessed using PC-relative address so it needs to
be handled separately.  The PC-rel addressing is detected by using
DWARF_REG_PC.  On x86, %rip register would be used.

The address can be calculated using the ip and offset in the
instruction.  But it should start from the next instruction so add
calculate_pcrel_addr() to do it properly.

But global variables defined in a different file would only have a
declaration which doesn't include a location list.  So it first tries
to get the type info using the address, and then looks up the variable
declarations using name.  The name of global variables should be get
from the symbol table.  The declaration would have the type info.

So extend find_var_type() to take both address and name for global
variables.

The stat is now looks like:

  Annotate data type stats:
  total 294, ok 153 (52.0%), bad 141 (48.0%)
  -----------------------------------------------------------
          30 : no_sym
          32 : no_mem_ops
          61 : no_var
          10 : no_typeinfo
           8 : bad_offset

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-7-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Handle PC-relative addressing
Namhyung Kim [Wed, 17 Jan 2024 06:26:53 +0000 (22:26 -0800)]
perf annotate-data: Handle PC-relative addressing

Extend find_data_type_die() to find data type from PC-relative address
using die_find_variable_by_addr().  Users need to pass the address for
the (global) variable.

The offset for the variable should be updated after finding the type
because the offset in the instruction is just to calcuate the address
for the variable.  So it changed to pass a pointer to offset and renamed
it to 'poffset'.

First it searches variables in the CU DIE as it's likely that the global
variables are defined in the file level.  And then it iterates the scope
DIEs to find a local (static) variable.

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-6-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Add stack operation pseudo type
Namhyung Kim [Wed, 17 Jan 2024 06:26:52 +0000 (22:26 -0800)]
perf annotate-data: Add stack operation pseudo type

A typical function prologue and epilogue include multiple stack
operations to save and restore the current value of registers.
On x86, it looks like below:

  push  r15
  push  r14
  push  r13
  push  r12

  ...

  pop   r12
  pop   r13
  pop   r14
  pop   r15
  ret

As these all touches the stack memory region, chances are high that they
appear in a memory profile data.  But these are not used for any real
purpose yet so it'd return no types.

One of my profile type shows that non neglible portion of data came from
the stack operations.  It also seems GCC generates more stack operations
than clang.

Annotate Instruction stats
total 264, ok 169 (64.0%), bad 95 (36.0%)

    Name      :  Good   Bad
  -----------------------------------------------------------
    movq      :    49    27
    movl      :    24     9
    popq      :     0    19   <-- here
    cmpl      :    17     2
    addq      :    14     1
    cmpq      :    12     2
    cmpxchgl  :     3     7

Instead of dealing them as unknown, let's create a seperate pseudo type
to represent those stack operations separately.

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-5-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Handle array style accesses
Namhyung Kim [Wed, 17 Jan 2024 06:26:51 +0000 (22:26 -0800)]
perf annotate-data: Handle array style accesses

On x86, instructions for array access often looks like below.

  mov  0x1234(%rax,%rbx,8), %rcx

Usually the first register holds the type information and the second one
has the index.  And the current code only looks up a variable for the
first register.  But it's possible to be in the other way around so it
needs to check the second register if the first one failed.

The stat changed like this.

  Annotate data type stats:
  total 294, ok 148 (50.3%), bad 146 (49.7%)
  -----------------------------------------------------------
          30 : no_sym
          32 : no_mem_ops
          66 : no_var
          10 : no_typeinfo
           8 : bad_offset

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-4-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Handle macro fusion on x86
Namhyung Kim [Wed, 17 Jan 2024 06:26:50 +0000 (22:26 -0800)]
perf annotate-data: Handle macro fusion on x86

When a sample was come from a conditional branch without a memory
operand, it could be due to a macro fusion with a previous instruction.
So it needs to check the memory operand in the previous one.

This improves the stat like below:

  Annotate data type stats:
  total 294, ok 147 (50.0%), bad 147 (50.0%)
  -----------------------------------------------------------
          30 : no_sym
          32 : no_mem_ops
          71 : no_var
           6 : no_typeinfo
           8 : bad_offset

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-3-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf annotate-data: Parse 'lock' prefix from llvm-objdump
Namhyung Kim [Wed, 17 Jan 2024 06:26:49 +0000 (22:26 -0800)]
perf annotate-data: Parse 'lock' prefix from llvm-objdump

For the performance reason, I prefer llvm-objdump over GNU's.  But I
found that llvm-objdump puts x86 lock prefix in a separate line like
below.

  ffffffff81000695: f0                    lock
  ffffffff81000696: ff 83 54 0b 00 00     incl    2900(%rbx)

This should be parsed properly, but I just changed to find the insn
with next offset for now.

This improves the statistics as it can process more instructions.

  Annotate data type stats:
  total 294, ok 144 (49.0%), bad 150 (51.0%)
  -----------------------------------------------------------
          30 : no_sym
          35 : no_mem_ops
          71 : no_var
           6 : no_typeinfo
           8 : bad_offset

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lore.kernel.org/r/20240117062657.985479-2-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf build: Check whether pkg-config is installed when libtraceevent is linked
Yang Jihong [Fri, 12 Jan 2024 03:40:19 +0000 (03:40 +0000)]
perf build: Check whether pkg-config is installed when libtraceevent is linked

If pkg-config is not installed when libtraceevent is linked, the build fails.

The error information is as follows:

  $ make
  <SNIP>
  In file included from /home/yjh/projects_linux/perf-tool-next/linux/tools/perf/util/evsel.c:43:
  /home/yjh/projects_linux/perf-tool-next/linux/tools/perf/util/trace-event.h:149:62: error: operator '&&' has no right operand
    149 | #if defined(LIBTRACEEVENT_VERSION) &&  LIBTRACEEVENT_VERSION >= MAKE_LIBTRACEEVENT_VERSION(1, 5, 0)
        |                                                              ^~
  error: command '/usr/bin/gcc' failed with exit code 1
  cp: cannot stat 'python_ext_build/lib/perf*.so': No such file or directory
  make[2]: *** [Makefile.perf:668: python/perf.cpython-310-x86_64-linux-gnu.so] Error 1
  make[2]: *** Waiting for unfinished jobs....

Because pkg-config is not installed, fail to get libtraceevent version in
Makefile.config file. As a result, LIBTRACEEVENT_VERSION is empty.
However, the preceding error information is not user-friendly.

Identify errors in advance by checking that pkg-config is installed at
compile time.

The build results of various scenarios are as follows:

1. build successful when libtraceevent is not linked and pkg-config is not installed

  $ pkg-config --version
  -bash: /usr/bin/pkg-config: No such file or directory
  $ make clean >/dev/null
  $ make NO_LIBTRACEEVENT=1 >/dev/null
  Makefile.config:1133: No alternatives command found, you need to set JDIR= to point to the root of your Java directory
    PERF_VERSION = 6.7.rc6.gd988c9f511af
  $ echo $?
  0

2. dummy pkg-config is missing when libtraceevent is linked

  $ pkg-config --version
  -bash: /usr/bin/pkg-config: No such file or directory
  $ make clean >/dev/null
  $ make >/dev/null
  Makefile.config:221: *** Error: pkg-config needed by libtraceevent is missing on this system, please install it.  Stop.
  make[1]: *** [Makefile.perf:251: sub-make] Error 2
  make: *** [Makefile:70: all] Error 2
  $ echo $?
  2

3. build successful when libtraceevent is linked and pkg-config is installed

  $ pkg-config --version
  0.29.2
  $ make clean >/dev/null
  $ make >/dev/null
  Makefile.config:1133: No alternatives command found, you need to set JDIR= to point to the root of your Java directory
    PERF_VERSION = 6.7.rc6.gd988c9f511af
  $ echo $?
  0

Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240112034019.3558584-1-yangjihong1@huawei.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoperf test: raise limit to 20 percent for perf_stat_--bpf-counters_test
Thomas Richter [Mon, 8 Jan 2024 08:40:09 +0000 (09:40 +0100)]
perf test: raise limit to 20 percent for perf_stat_--bpf-counters_test

This test case often fails on s390 (about 2 out of 10) because the
10% percent limit on the difference between --bpf-counters event counting
and s390 hardware counting is more than 10% in all failure cases.
Raise the limit to 20% on s390 and the test case succeeds.

Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: gor@linux.ibm.com
Cc: hca@linux.ibm.com
Cc: sumanthk@linux.ibm.com
Cc: svens@linux.ibm.com
Link: https://lore.kernel.org/r/20240108084009.3959211-1-tmricht@linux.ibm.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
16 months agoLinux 6.8-rc1
Linus Torvalds [Sun, 21 Jan 2024 22:11:32 +0000 (14:11 -0800)]
Linux 6.8-rc1

16 months agoMerge tag 'bcachefs-2024-01-21' of https://evilpiepirate.org/git/bcachefs
Linus Torvalds [Sun, 21 Jan 2024 22:01:12 +0000 (14:01 -0800)]
Merge tag 'bcachefs-2024-01-21' of https://evilpiepirate.org/git/bcachefs

Pull more bcachefs updates from Kent Overstreet:
 "Some fixes, Some refactoring, some minor features:

   - Assorted prep work for disk space accounting rewrite

   - BTREE_TRIGGER_ATOMIC: after combining our trigger callbacks, this
     makes our trigger context more explicit

   - A few fixes to avoid excessive transaction restarts on
     multithreaded workloads: fstests (in addition to ktest tests) are
     now checking slowpath counters, and that's shaking out a few bugs

   - Assorted tracepoint improvements

   - Starting to break up bcachefs_format.h and move on disk types so
     they're with the code they belong to; this will make room to start
     documenting the on disk format better.

   - A few minor fixes"

* tag 'bcachefs-2024-01-21' of https://evilpiepirate.org/git/bcachefs: (46 commits)
  bcachefs: Improve inode_to_text()
  bcachefs: logged_ops_format.h
  bcachefs: reflink_format.h
  bcachefs; extents_format.h
  bcachefs: ec_format.h
  bcachefs: subvolume_format.h
  bcachefs: snapshot_format.h
  bcachefs: alloc_background_format.h
  bcachefs: xattr_format.h
  bcachefs: dirent_format.h
  bcachefs: inode_format.h
  bcachefs; quota_format.h
  bcachefs: sb-counters_format.h
  bcachefs: counters.c -> sb-counters.c
  bcachefs: comment bch_subvolume
  bcachefs: bch_snapshot::btime
  bcachefs: add missing __GFP_NOWARN
  bcachefs: opts->compression can now also be applied in the background
  bcachefs: Prep work for variable size btree node buffers
  bcachefs: grab s_umount only if snapshotting
  ...

16 months agoMerge tag 'timers-core-2024-01-21' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 21 Jan 2024 19:14:40 +0000 (11:14 -0800)]
Merge tag 'timers-core-2024-01-21' of git://git./linux/kernel/git/tip/tip

Pull timer updates from Thomas Gleixner:
 "Updates for time and clocksources:

   - A fix for the idle and iowait time accounting vs CPU hotplug.

     The time is reset on CPU hotplug which makes the accumulated
     systemwide time jump backwards.

   - Assorted fixes and improvements for clocksource/event drivers"

* tag 'timers-core-2024-01-21' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  tick-sched: Fix idle and iowait sleeptime accounting vs CPU hotplug
  clocksource/drivers/ep93xx: Fix error handling during probe
  clocksource/drivers/cadence-ttc: Fix some kernel-doc warnings
  clocksource/drivers/timer-ti-dm: Fix make W=n kerneldoc warnings
  clocksource/timer-riscv: Add riscv_clock_shutdown callback
  dt-bindings: timer: Add StarFive JH8100 clint
  dt-bindings: timer: thead,c900-aclint-mtimer: separate mtime and mtimecmp regs

16 months agoMerge tag 'powerpc-6.8-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sun, 21 Jan 2024 19:04:29 +0000 (11:04 -0800)]
Merge tag 'powerpc-6.8-2' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Aneesh Kumar:

 - Increase default stack size to 32KB for Book3S

Thanks to Michael Ellerman.

* tag 'powerpc-6.8-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/64s: Increase default stack size to 32KB

16 months agobcachefs: Improve inode_to_text()
Kent Overstreet [Sun, 21 Jan 2024 17:19:01 +0000 (12:19 -0500)]
bcachefs: Improve inode_to_text()

Add line breaks - inode_to_text() is now much easier to read.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: logged_ops_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:57:45 +0000 (02:57 -0500)]
bcachefs: logged_ops_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: reflink_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:54:47 +0000 (02:54 -0500)]
bcachefs: reflink_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs; extents_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:51:56 +0000 (02:51 -0500)]
bcachefs; extents_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: ec_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:47:14 +0000 (02:47 -0500)]
bcachefs: ec_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: subvolume_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:42:53 +0000 (02:42 -0500)]
bcachefs: subvolume_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: snapshot_format.h
Kent Overstreet [Sun, 21 Jan 2024 07:41:06 +0000 (02:41 -0500)]
bcachefs: snapshot_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: alloc_background_format.h
Kent Overstreet [Sun, 21 Jan 2024 05:01:52 +0000 (00:01 -0500)]
bcachefs: alloc_background_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: xattr_format.h
Kent Overstreet [Sun, 21 Jan 2024 04:59:15 +0000 (23:59 -0500)]
bcachefs: xattr_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: dirent_format.h
Kent Overstreet [Sun, 21 Jan 2024 04:57:10 +0000 (23:57 -0500)]
bcachefs: dirent_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: inode_format.h
Kent Overstreet [Sun, 21 Jan 2024 04:55:39 +0000 (23:55 -0500)]
bcachefs: inode_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs; quota_format.h
Kent Overstreet [Sun, 21 Jan 2024 04:53:52 +0000 (23:53 -0500)]
bcachefs; quota_format.h

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: sb-counters_format.h
Kent Overstreet [Sun, 21 Jan 2024 04:50:56 +0000 (23:50 -0500)]
bcachefs: sb-counters_format.h

bcachefs_format.h has gotten too big; let's do some organizing.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: counters.c -> sb-counters.c
Kent Overstreet [Sun, 21 Jan 2024 04:46:35 +0000 (23:46 -0500)]
bcachefs: counters.c -> sb-counters.c

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: comment bch_subvolume
Kent Overstreet [Sun, 21 Jan 2024 04:44:17 +0000 (23:44 -0500)]
bcachefs: comment bch_subvolume

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bch_snapshot::btime
Kent Overstreet [Sun, 21 Jan 2024 04:35:41 +0000 (23:35 -0500)]
bcachefs: bch_snapshot::btime

Add a field to bch_snapshot for creation time; this will be important
when we start exposing the snapshot tree to userspace.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: add missing __GFP_NOWARN
Kent Overstreet [Wed, 17 Jan 2024 22:16:07 +0000 (17:16 -0500)]
bcachefs: add missing __GFP_NOWARN

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: opts->compression can now also be applied in the background
Kent Overstreet [Tue, 16 Jan 2024 21:20:21 +0000 (16:20 -0500)]
bcachefs: opts->compression can now also be applied in the background

The "apply this compression method in the background" paths now use the
compression option if background_compression is not set; this means that
setting or changing the compression option will cause existing data to
be compressed accordingly in the background.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Prep work for variable size btree node buffers
Kent Overstreet [Tue, 16 Jan 2024 18:29:59 +0000 (13:29 -0500)]
bcachefs: Prep work for variable size btree node buffers

bcachefs btree nodes are big - typically 256k - and btree roots are
pinned in memory. As we're now up to 18 btrees, we now have significant
memory overhead in mostly empty btree roots.

And in the future we're going to start enforcing that certain btree node
boundaries exist, to solve lock contention issues - analagous to XFS's
AGIs.

Thus, we need to start allocating smaller btree node buffers when we
can. This patch changes code that refers to the filesystem constant
c->opts.btree_node_size to refer to the btree node buffer size -
btree_buf_bytes() - where appropriate.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: grab s_umount only if snapshotting
Su Yue [Mon, 15 Jan 2024 02:21:25 +0000 (10:21 +0800)]
bcachefs: grab s_umount only if snapshotting

When I was testing mongodb over bcachefs with compression,
there is a lockdep warning when snapshotting mongodb data volume.

$ cat test.sh
prog=bcachefs

$prog subvolume create /mnt/data
$prog subvolume create /mnt/data/snapshots

while true;do
    $prog subvolume snapshot /mnt/data /mnt/data/snapshots/$(date +%s)
    sleep 1s
done

$ cat /etc/mongodb.conf
systemLog:
  destination: file
  logAppend: true
  path: /mnt/data/mongod.log

storage:
  dbPath: /mnt/data/

lockdep reports:
[ 3437.452330] ======================================================
[ 3437.452750] WARNING: possible circular locking dependency detected
[ 3437.453168] 6.7.0-rc7-custom+ #85 Tainted: G            E
[ 3437.453562] ------------------------------------------------------
[ 3437.453981] bcachefs/35533 is trying to acquire lock:
[ 3437.454325] ffffa0a02b2b1418 (sb_writers#10){.+.+}-{0:0}, at: filename_create+0x62/0x190
[ 3437.454875]
               but task is already holding lock:
[ 3437.455268] ffffa0a02b2b10e0 (&type->s_umount_key#48){.+.+}-{3:3}, at: bch2_fs_file_ioctl+0x232/0xc90 [bcachefs]
[ 3437.456009]
               which lock already depends on the new lock.

[ 3437.456553]
               the existing dependency chain (in reverse order) is:
[ 3437.457054]
               -> #3 (&type->s_umount_key#48){.+.+}-{3:3}:
[ 3437.457507]        down_read+0x3e/0x170
[ 3437.457772]        bch2_fs_file_ioctl+0x232/0xc90 [bcachefs]
[ 3437.458206]        __x64_sys_ioctl+0x93/0xd0
[ 3437.458498]        do_syscall_64+0x42/0xf0
[ 3437.458779]        entry_SYSCALL_64_after_hwframe+0x6e/0x76
[ 3437.459155]
               -> #2 (&c->snapshot_create_lock){++++}-{3:3}:
[ 3437.459615]        down_read+0x3e/0x170
[ 3437.459878]        bch2_truncate+0x82/0x110 [bcachefs]
[ 3437.460276]        bchfs_truncate+0x254/0x3c0 [bcachefs]
[ 3437.460686]        notify_change+0x1f1/0x4a0
[ 3437.461283]        do_truncate+0x7f/0xd0
[ 3437.461555]        path_openat+0xa57/0xce0
[ 3437.461836]        do_filp_open+0xb4/0x160
[ 3437.462116]        do_sys_openat2+0x91/0xc0
[ 3437.462402]        __x64_sys_openat+0x53/0xa0
[ 3437.462701]        do_syscall_64+0x42/0xf0
[ 3437.462982]        entry_SYSCALL_64_after_hwframe+0x6e/0x76
[ 3437.463359]
               -> #1 (&sb->s_type->i_mutex_key#15){+.+.}-{3:3}:
[ 3437.463843]        down_write+0x3b/0xc0
[ 3437.464223]        bch2_write_iter+0x5b/0xcc0 [bcachefs]
[ 3437.464493]        vfs_write+0x21b/0x4c0
[ 3437.464653]        ksys_write+0x69/0xf0
[ 3437.464839]        do_syscall_64+0x42/0xf0
[ 3437.465009]        entry_SYSCALL_64_after_hwframe+0x6e/0x76
[ 3437.465231]
               -> #0 (sb_writers#10){.+.+}-{0:0}:
[ 3437.465471]        __lock_acquire+0x1455/0x21b0
[ 3437.465656]        lock_acquire+0xc6/0x2b0
[ 3437.465822]        mnt_want_write+0x46/0x1a0
[ 3437.465996]        filename_create+0x62/0x190
[ 3437.466175]        user_path_create+0x2d/0x50
[ 3437.466352]        bch2_fs_file_ioctl+0x2ec/0xc90 [bcachefs]
[ 3437.466617]        __x64_sys_ioctl+0x93/0xd0
[ 3437.466791]        do_syscall_64+0x42/0xf0
[ 3437.466957]        entry_SYSCALL_64_after_hwframe+0x6e/0x76
[ 3437.467180]
               other info that might help us debug this:

[ 3437.469670] 2 locks held by bcachefs/35533:
               other info that might help us debug this:

[ 3437.467507] Chain exists of:
                 sb_writers#10 --> &c->snapshot_create_lock --> &type->s_umount_key#48

[ 3437.467979]  Possible unsafe locking scenario:

[ 3437.468223]        CPU0                    CPU1
[ 3437.468405]        ----                    ----
[ 3437.468585]   rlock(&type->s_umount_key#48);
[ 3437.468758]                                lock(&c->snapshot_create_lock);
[ 3437.469030]                                lock(&type->s_umount_key#48);
[ 3437.469291]   rlock(sb_writers#10);
[ 3437.469434]
                *** DEADLOCK ***

[ 3437.469670] 2 locks held by bcachefs/35533:
[ 3437.469838]  #0: ffffa0a02ce00a88 (&c->snapshot_create_lock){++++}-{3:3}, at: bch2_fs_file_ioctl+0x1e3/0xc90 [bcachefs]
[ 3437.470294]  #1: ffffa0a02b2b10e0 (&type->s_umount_key#48){.+.+}-{3:3}, at: bch2_fs_file_ioctl+0x232/0xc90 [bcachefs]
[ 3437.470744]
               stack backtrace:
[ 3437.470922] CPU: 7 PID: 35533 Comm: bcachefs Kdump: loaded Tainted: G            E      6.7.0-rc7-custom+ #85
[ 3437.471313] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Arch Linux 1.16.3-1-1 04/01/2014
[ 3437.471694] Call Trace:
[ 3437.471795]  <TASK>
[ 3437.471884]  dump_stack_lvl+0x57/0x90
[ 3437.472035]  check_noncircular+0x132/0x150
[ 3437.472202]  __lock_acquire+0x1455/0x21b0
[ 3437.472369]  lock_acquire+0xc6/0x2b0
[ 3437.472518]  ? filename_create+0x62/0x190
[ 3437.472683]  ? lock_is_held_type+0x97/0x110
[ 3437.472856]  mnt_want_write+0x46/0x1a0
[ 3437.473025]  ? filename_create+0x62/0x190
[ 3437.473204]  filename_create+0x62/0x190
[ 3437.473380]  user_path_create+0x2d/0x50
[ 3437.473555]  bch2_fs_file_ioctl+0x2ec/0xc90 [bcachefs]
[ 3437.473819]  ? lock_acquire+0xc6/0x2b0
[ 3437.474002]  ? __fget_files+0x2a/0x190
[ 3437.474195]  ? __fget_files+0xbc/0x190
[ 3437.474380]  ? lock_release+0xc5/0x270
[ 3437.474567]  ? __x64_sys_ioctl+0x93/0xd0
[ 3437.474764]  ? __pfx_bch2_fs_file_ioctl+0x10/0x10 [bcachefs]
[ 3437.475090]  __x64_sys_ioctl+0x93/0xd0
[ 3437.475277]  do_syscall_64+0x42/0xf0
[ 3437.475454]  entry_SYSCALL_64_after_hwframe+0x6e/0x76
[ 3437.475691] RIP: 0033:0x7f2743c313af
======================================================

In __bch2_ioctl_subvolume_create(), we grab s_umount unconditionally
and unlock it at the end of the function. There is a comment
"why do we need this lock?" about the lock coming from
commit 42d237320e98 ("bcachefs: Snapshot creation, deletion")
The reason is that __bch2_ioctl_subvolume_create() calls
sync_inodes_sb() which enforce locked s_umount to writeback all dirty
nodes before doing snapshot works.

Fix it by read locking s_umount for snapshotting only and unlocking
s_umount after sync_inodes_sb().

Signed-off-by: Su Yue <glass.su@suse.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: kvfree bch_fs::snapshots in bch2_fs_snapshots_exit
Su Yue [Tue, 16 Jan 2024 11:05:37 +0000 (19:05 +0800)]
bcachefs: kvfree bch_fs::snapshots in bch2_fs_snapshots_exit

bch_fs::snapshots is allocated by kvzalloc in __snapshot_t_mut.
It should be freed by kvfree not kfree.
Or umount will triger:

[  406.829178 ] BUG: unable to handle page fault for address: ffffe7b487148008
[  406.830676 ] #PF: supervisor read access in kernel mode
[  406.831643 ] #PF: error_code(0x0000) - not-present page
[  406.832487 ] PGD 0 P4D 0
[  406.832898 ] Oops: 0000 [#1] PREEMPT SMP PTI
[  406.833512 ] CPU: 2 PID: 1754 Comm: umount Kdump: loaded Tainted: G           OE      6.7.0-rc7-custom+ #90
[  406.834746 ] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Arch Linux 1.16.3-1-1 04/01/2014
[  406.835796 ] RIP: 0010:kfree+0x62/0x140
[  406.836197 ] Code: 80 48 01 d8 0f 82 e9 00 00 00 48 c7 c2 00 00 00 80 48 2b 15 78 9f 1f 01 48 01 d0 48 c1 e8 0c 48 c1 e0 06 48 03 05 56 9f 1f 01 <48> 8b 50 08 48 89 c7 f6 c2 01 0f 85 b0 00 00 00 66 90 48 8b 07 f6
[  406.837810 ] RSP: 0018:ffffb9d641607e48 EFLAGS: 00010286
[  406.838213 ] RAX: ffffe7b487148000 RBX: ffffb9d645200000 RCX: ffffb9d641607dc4
[  406.838738 ] RDX: 000065bb00000000 RSI: ffffffffc0d88b84 RDI: ffffb9d645200000
[  406.839217 ] RBP: ffff9a4625d00068 R08: 0000000000000001 R09: 0000000000000001
[  406.839650 ] R10: 0000000000000001 R11: 000000000000001f R12: ffff9a4625d4da80
[  406.840055 ] R13: ffff9a4625d00000 R14: ffffffffc0e2eb20 R15: 0000000000000000
[  406.840451 ] FS:  00007f0a264ffb80(0000) GS:ffff9a4e2d500000(0000) knlGS:0000000000000000
[  406.840851 ] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  406.841125 ] CR2: ffffe7b487148008 CR3: 000000018c4d2000 CR4: 00000000000006f0
[  406.841464 ] Call Trace:
[  406.841583 ]  <TASK>
[  406.841682 ]  ? __die+0x1f/0x70
[  406.841828 ]  ? page_fault_oops+0x159/0x470
[  406.842014 ]  ? fixup_exception+0x22/0x310
[  406.842198 ]  ? exc_page_fault+0x1ed/0x200
[  406.842382 ]  ? asm_exc_page_fault+0x22/0x30
[  406.842574 ]  ? bch2_fs_release+0x54/0x280 [bcachefs]
[  406.842842 ]  ? kfree+0x62/0x140
[  406.842988 ]  ? kfree+0x104/0x140
[  406.843138 ]  bch2_fs_release+0x54/0x280 [bcachefs]
[  406.843390 ]  kobject_put+0xb7/0x170
[  406.843552 ]  deactivate_locked_super+0x2f/0xa0
[  406.843756 ]  cleanup_mnt+0xba/0x150
[  406.843917 ]  task_work_run+0x59/0xa0
[  406.844083 ]  exit_to_user_mode_prepare+0x197/0x1a0
[  406.844302 ]  syscall_exit_to_user_mode+0x16/0x40
[  406.844510 ]  do_syscall_64+0x4e/0xf0
[  406.844675 ]  entry_SYSCALL_64_after_hwframe+0x6e/0x76
[  406.844907 ] RIP: 0033:0x7f0a2664e4fb

Signed-off-by: Su Yue <glass.su@suse.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bios must be 512 byte algined
Kent Overstreet [Tue, 16 Jan 2024 16:38:04 +0000 (11:38 -0500)]
bcachefs: bios must be 512 byte algined

Fixes: 023f9ac9f70f bcachefs: Delete dio read alignment check
Reported-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: remove redundant variable tmp
Colin Ian King [Tue, 16 Jan 2024 11:07:23 +0000 (11:07 +0000)]
bcachefs: remove redundant variable tmp

The variable tmp is being assigned a value but it isn't being
read afterwards. The assignment is redundant and so tmp can be
removed.

Cleans up clang scan build warning:
warning: Although the value stored to 'ret' is used in the enclosing
expression, the value is never actually read from 'ret'
[deadcode.DeadStores]

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Improve trace_trans_restart_relock
Kent Overstreet [Tue, 16 Jan 2024 01:40:06 +0000 (20:40 -0500)]
bcachefs: Improve trace_trans_restart_relock

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Fix excess transaction restarts in __bchfs_fallocate()
Kent Overstreet [Tue, 16 Jan 2024 01:37:23 +0000 (20:37 -0500)]
bcachefs: Fix excess transaction restarts in __bchfs_fallocate()

drop_locks_do() should not be used in a fastpath without first trying
the do in nonblocking mode - the unlock and relock will cause excessive
transaction restarts and potentially livelocking with other threads that
are contending for the same locks.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: extents_to_bp_state
Kent Overstreet [Mon, 15 Jan 2024 23:19:52 +0000 (18:19 -0500)]
bcachefs: extents_to_bp_state

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bkey_and_val_eq()
Kent Overstreet [Mon, 15 Jan 2024 23:08:32 +0000 (18:08 -0500)]
bcachefs: bkey_and_val_eq()

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Better journal tracepoints
Kent Overstreet [Mon, 15 Jan 2024 22:59:51 +0000 (17:59 -0500)]
bcachefs: Better journal tracepoints

Factor out bch2_journal_bufs_to_text(), and use it in the
journal_entry_full() tracepoint; when we can't get a journal reservation
we need to know the outstanding journal entry sizes to know if the
problem is due to excessive flushing.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Print size of superblock with space allocated
Kent Overstreet [Mon, 15 Jan 2024 22:57:44 +0000 (17:57 -0500)]
bcachefs: Print size of superblock with space allocated

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Avoid flushing the journal in the discard path
Kent Overstreet [Mon, 15 Jan 2024 22:56:22 +0000 (17:56 -0500)]
bcachefs: Avoid flushing the journal in the discard path

When issuing discards, we may need to flush the journal if there's too
many buckets that can't be discarded until a journal flush.

But the heuristic was bad; we should be comparing the number of buckets
that need to flushes against the number of free buckets, not the number
of buckets we saw.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Improve move_extent tracepoint
Kent Overstreet [Mon, 15 Jan 2024 20:33:39 +0000 (15:33 -0500)]
bcachefs: Improve move_extent tracepoint

Also print out the data_opts, so that we can see what specifically is
being done to an extent.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Add missing bch2_moving_ctxt_flush_all()
Kent Overstreet [Mon, 15 Jan 2024 20:06:43 +0000 (15:06 -0500)]
bcachefs: Add missing bch2_moving_ctxt_flush_all()

This fixes a bug with rebalance IOs getting stuck with reads completed,
but writes never being issued.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Re-add move_extent_write tracepoint
Kent Overstreet [Mon, 15 Jan 2024 20:04:40 +0000 (15:04 -0500)]
bcachefs: Re-add move_extent_write tracepoint

It appears this was accidentally deleted at some point - also, do a bit
of cleanup.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bch2_kthread_io_clock_wait() no longer sleeps until full amount
Kent Overstreet [Mon, 15 Jan 2024 19:15:26 +0000 (14:15 -0500)]
bcachefs: bch2_kthread_io_clock_wait() no longer sleeps until full amount

Drop t he loop in bch2_kthread_io_clock_wait(): this allows the code
that uses it to be woken up for other reasons, and fixes a bug where
rebalance wouldn't wake up when a scan was requested.

This raises the possibility of spurious wakeups, but callers should
always be able to handle that reasonably well.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Add .val_to_text() for KEY_TYPE_cookie
Kent Overstreet [Mon, 15 Jan 2024 19:15:03 +0000 (14:15 -0500)]
bcachefs: Add .val_to_text() for KEY_TYPE_cookie

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Don't pass memcmp() as a pointer
Kent Overstreet [Mon, 15 Jan 2024 19:12:43 +0000 (14:12 -0500)]
bcachefs: Don't pass memcmp() as a pointer

Some (buggy!) compilers have issues with this.

Fixes: https://github.com/koverstreet/bcachefs/issues/625
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agoMerge tag 'header_cleanup-2024-01-20' of https://evilpiepirate.org/git/bcachefs
Linus Torvalds [Sun, 21 Jan 2024 18:21:43 +0000 (10:21 -0800)]
Merge tag 'header_cleanup-2024-01-20' of https://evilpiepirate.org/git/bcachefs

Pull header fix from Kent Overstreet:
 "Just one small fixup for the RT build"

* tag 'header_cleanup-2024-01-20' of https://evilpiepirate.org/git/bcachefs:
  spinlock: Fix failing build for PREEMPT_RT

16 months agobcachefs: Reduce would_deadlock restarts
Kent Overstreet [Thu, 11 Jan 2024 04:47:04 +0000 (23:47 -0500)]
bcachefs: Reduce would_deadlock restarts

We don't have to take locks in any particular ordering - we'll make
forward progress just fine - but if we try to stick to an ordering, it
can help to avoid excessive would_deadlock transaction restarts.

This tweaks the reflink path to take extents btree locks in the right
order.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bch2_trans_account_disk_usage_change()
Kent Overstreet [Sat, 11 Nov 2023 20:08:36 +0000 (15:08 -0500)]
bcachefs: bch2_trans_account_disk_usage_change()

The disk space accounting rewrite is splitting out accounting for each
replicas set - those are moving to btree keys, instead of percpu
counters.

This breaks bch2_trans_fs_usage_apply() up, splitting out the part we
will still need.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bch_fs_usage_base
Kent Overstreet [Fri, 17 Nov 2023 05:03:45 +0000 (00:03 -0500)]
bcachefs: bch_fs_usage_base

Split out base filesystem usage into its own type; prep work for
breaking up bch2_trans_fs_usage_apply().

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: bch2_prt_compression_type()
Kent Overstreet [Sun, 7 Jan 2024 02:01:47 +0000 (21:01 -0500)]
bcachefs: bch2_prt_compression_type()

bounds checking helper, since compression types are extensible

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: helpers for printing data types
Kent Overstreet [Sun, 7 Jan 2024 01:57:43 +0000 (20:57 -0500)]
bcachefs: helpers for printing data types

We need bounds checking since new versions may introduce new data types.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: BTREE_TRIGGER_ATOMIC
Kent Overstreet [Sun, 7 Jan 2024 22:14:46 +0000 (17:14 -0500)]
bcachefs: BTREE_TRIGGER_ATOMIC

Add a new flag to be explicit about when we're running atomic triggers.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: drop to_text code for obsolete bps in alloc keys
Kent Overstreet [Sun, 7 Jan 2024 00:47:09 +0000 (19:47 -0500)]
bcachefs: drop to_text code for obsolete bps in alloc keys

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: eytzinger_for_each() declares loop iter
Kent Overstreet [Sun, 7 Jan 2024 00:29:14 +0000 (19:29 -0500)]
bcachefs: eytzinger_for_each() declares loop iter

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: Don't log errors if BCH_WRITE_ALLOC_NOWAIT
Kent Overstreet [Thu, 11 Jan 2024 04:08:30 +0000 (23:08 -0500)]
bcachefs: Don't log errors if BCH_WRITE_ALLOC_NOWAIT

Previously, we added logging in the write path to ensure that any
unexpected errors getting reported to userspace have a log message; but
BCH_WRITE_ALLOC_NOWAIT is a special case, it's used for promotes where
errors are expected and not reported out to userspace - so we need to
silence those.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agobcachefs: fix memleak in bch2_split_devs
Su Yue [Mon, 8 Jan 2024 15:11:08 +0000 (23:11 +0800)]
bcachefs: fix memleak in bch2_split_devs

The pointer dev_name can be modified by strseq(),
then causes the memleak:

unreferenced object 0xffff9d08a2916c80 (size 32):
  comm "mount.bcachefs", pid 9090, jiffies 4295856224 (age 17.564s)
  hex dump (first 32 bytes):
    2f 64 65 76 2f 6d 61 70 70 65 72 2f 74 65 73 74  /dev/mapper/test
    2d 30 00 00 00 00 00 00 00 00 00 00 00 00 00 00  -0..............
  backtrace:
    [<00000000c5d3be7d>] __kmem_cache_alloc_node+0x1f3/0x2c0
    [<0000000052215d26>] __kmalloc_node_track_caller+0x51/0x150
    [<0000000069fea956>] kstrdup+0x32/0x60
    [<000000000877fcf1>] bch2_split_devs+0x3f/0x150 [bcachefs]
    [<000000007ee93204>] bch2_mount+0xcb/0x640 [bcachefs]
    [<000000002dd1e04b>] legacy_get_tree+0x30/0x60
    [<000000006afc31d3>] vfs_get_tree+0x28/0xf0
    [<000000007b0c538e>] path_mount+0x475/0xb60
    [<0000000092de5882>] __x64_sys_mount+0x105/0x140
    [<0000000054fc05d8>] do_syscall_64+0x42/0xf0
    [<00000000df584910>] entry_SYSCALL_64_after_hwframe+0x6e/0x76

Fix it by copy pointer dev_name at beginning and free the copied
pointer at end.

Signed-off-by: Su Yue <glass.su@suse.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
16 months agoMerge tag 'v6.8-rc-part2-smb-client' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 21 Jan 2024 00:48:07 +0000 (16:48 -0800)]
Merge tag 'v6.8-rc-part2-smb-client' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client updates from Steve French:
 "Various smb client fixes, including multichannel and for SMB3.1.1
  POSIX extensions:

   - debugging improvement (display start time for stats)

   - two reparse point handling fixes

   - various multichannel improvements and fixes

   - SMB3.1.1 POSIX extensions open/create parsing fix

   - retry (reconnect) improvement including new retrans mount parm, and
     handling of two additional return codes that need to be retried on

   - two minor cleanup patches and another to remove duplicate query
     info code

   - two documentation cleanup, and one reviewer email correction"

* tag 'v6.8-rc-part2-smb-client' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: update iface_last_update on each query-and-update
  cifs: handle servers that still advertise multichannel after disabling
  cifs: new mount option called retrans
  cifs: reschedule periodic query for server interfaces
  smb: client: don't clobber ->i_rdev from cached reparse points
  smb: client: get rid of smb311_posix_query_path_info()
  smb: client: parse owner/group when creating reparse points
  smb: client: fix parsing of SMB3.1.1 POSIX create context
  cifs: update known bugs mentioned in kernel docs for cifs
  cifs: new nt status codes from MS-SMB2
  cifs: pick channel for tcon and tdis
  cifs: open_cached_dir should not rely on primary channel
  smb3: minor documentation updates
  Update MAINTAINERS email address
  cifs: minor comment cleanup
  smb3: show beginning time for per share stats
  cifs: remove redundant variable tcon_exist

16 months agoMerge tag 'dmaengine-fix-6.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 20 Jan 2024 23:03:25 +0000 (15:03 -0800)]
Merge tag 'dmaengine-fix-6.8-rc1' of git://git./linux/kernel/git/vkoul/dmaengine

Pull dmaengine updates from Vinod Koul:
 "New support:
   - Loongson LS2X APB DMA controller
   - sf-pdma: mpfs-pdma support
   - Qualcomm X1E80100 GPI dma controller support

  Updates:
   - Xilinx XDMA updates to support interleaved DMA transfers
   - TI PSIL threads for AM62P and J722S and cfg register regions
     description
   - axi-dmac Improving the cyclic DMA transfers
   - Tegra Support dma-channel-mask property
   - Remaining platform remove callback returning void conversions

 Driver fixes for:
   - Xilinx xdma driver operator precedence and initialization fix
   - Excess kernel-doc warning fix in imx-sdma xilinx xdma drivers
   - format-overflow warning fix for rz-dmac, sh usb dmac drivers
   - 'output may be truncated' fix for shdma, fsl-qdma and dw-edma
     drivers"

* tag 'dmaengine-fix-6.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (58 commits)
  dmaengine: dw-edma: increase size of 'name' in debugfs code
  dmaengine: fsl-qdma: increase size of 'irq_name'
  dmaengine: shdma: increase size of 'dev_id'
  dmaengine: xilinx: xdma: Fix kernel-doc warnings
  dmaengine: usb-dmac: Avoid format-overflow warning
  dmaengine: sh: rz-dmac: Avoid format-overflow warning
  dmaengine: imx-sdma: fix Excess kernel-doc warnings
  dmaengine: xilinx: xdma: Fix initialization location of desc in xdma_channel_isr()
  dmaengine: xilinx: xdma: Fix operator precedence in xdma_prep_interleaved_dma()
  dmaengine: xilinx: xdma: statify xdma_prep_interleaved_dma
  dmaengine: xilinx: xdma: Workaround truncation compilation error
  dmaengine: pl330: issue_pending waits until WFP state
  dmaengine: xilinx: xdma: Implement interleaved DMA transfers
  dmaengine: xilinx: xdma: Prepare the introduction of interleaved DMA transfers
  dmaengine: xilinx: xdma: Add transfer error reporting
  dmaengine: xilinx: xdma: Add error checking in xdma_channel_isr()
  dmaengine: xilinx: xdma: Rework xdma_terminate_all()
  dmaengine: xilinx: xdma: Ease dma_pool alignment requirements
  dmaengine: xilinx: xdma: Add necessary macro definitions
  dmaengine: xilinx: xdma: Get rid of unused code
  ...

16 months agoMerge tag 'coccinelle-for-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/jlawa...
Linus Torvalds [Sat, 20 Jan 2024 22:20:34 +0000 (14:20 -0800)]
Merge tag 'coccinelle-for-6.8' of git://git./linux/kernel/git/jlawall/linux

Pull coccinelle updates from Julia Lawall:
 "Updates to the device_attr_show semantic patch to reflect the new
  guidelines of the Linux kernel documentation.

  The problem was identified by Li Zhijian <lizhijian@fujitsu.com>, who
  proposed an initial fix"

* tag 'coccinelle-for-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/jlawall/linux:
  coccinelle: device_attr_show: simplify patch case
  coccinelle: device_attr_show: Adapt to the latest Documentation/filesystems/sysfs.rst

16 months agomedia: solo6x10: replace max(a, min(b, c)) by clamp(b, a, c)
Aurelien Jarno [Sat, 13 Jan 2024 18:33:31 +0000 (19:33 +0100)]
media: solo6x10: replace max(a, min(b, c)) by clamp(b, a, c)

This patch replaces max(a, min(b, c)) by clamp(b, a, c) in the solo6x10
driver.  This improves the readability and more importantly, for the
solo6x10-p2m.c file, this reduces on my system (x86-64, gcc 13):

 - the preprocessed size from 121 MiB to 4.5 MiB;

 - the build CPU time from 46.8 s to 1.6 s;

 - the build memory from 2786 MiB to 98MiB.

In fine, this allows this relatively simple C file to be built on a
32-bit system.

Reported-by: Jiri Slaby <jirislaby@gmail.com>
Closes: https://lore.kernel.org/lkml/18c6df0d-45ed-450c-9eda-95160a2bbb8e@gmail.com/
Cc: <stable@vger.kernel.org> # v6.7+
Suggested-by: David Laight <David.Laight@ACULAB.COM>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Reviewed-by: David Laight <David.Laight@ACULAB.COM>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 months agococcinelle: device_attr_show: simplify patch case
Julia Lawall [Sat, 20 Jan 2024 20:56:11 +0000 (21:56 +0100)]
coccinelle: device_attr_show: simplify patch case

Replacing the final expression argument by ... allows the format
string to have multiple arguments.

It also has the advantage of allowing the change to be recognized as
a change in a single statement, thus avoiding adding unneeded braces.

Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
16 months agoexecve: open the executable file before doing anything else
Linus Torvalds [Tue, 9 Jan 2024 00:43:04 +0000 (16:43 -0800)]
execve: open the executable file before doing anything else

No point in allocating a new mm, counting arguments and environment
variables etc if we're just going to return ENOENT.

This patch does expose the fact that 'do_filp_open()' that execve() uses
is still unnecessarily expensive in the failure case, because it
allocates the 'struct file *' early, even if the path lookup (which is
heavily optimized) fails.

So that remains an unnecessary cost in the "no such executable" case,
but it's a separate issue.  Regardless, I do not want to do _both_ a
filename_lookup() and a later do_filp_open() like the origin patch by
Josh Triplett did in [1].

Reported-by: Josh Triplett <josh@joshtriplett.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Mateusz Guzik <mjguzik@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Link: https://lore.kernel.org/lkml/5c7333ea4bec2fad1b47a8fa2db7c31e4ffc4f14.1663334978.git.josh@joshtriplett.org/
Link: https://lore.kernel.org/lkml/202209161637.9EDAF6B18@keescook/
Link: https://lore.kernel.org/lkml/CAHk-=wgznerM-xs+x+krDfE7eVBiy_HOam35rbsFMMOwvYuEKQ@mail.gmail.com/
Link: https://lore.kernel.org/lkml/CAHk-=whf9qLO8ipps4QhmS0BkM8mtWJhvnuDSdtw5gFjhzvKNA@mail.gmail.com/
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 months agoMerge tag 'riscv-for-linus-6.8-mw4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 20 Jan 2024 19:06:04 +0000 (11:06 -0800)]
Merge tag 'riscv-for-linus-6.8-mw4' of git://git./linux/kernel/git/riscv/linux

Pull more RISC-V updates from Palmer Dabbelt:

 - Support for tuning for systems with fast misaligned accesses.

 - Support for SBI-based suspend.

 - Support for the new SBI debug console extension.

 - The T-Head CMOs now use PA-based flushes.

 - Support for enabling the V extension in kernel code.

 - Optimized IP checksum routines.

 - Various ftrace improvements.

 - Support for archrandom, which depends on the Zkr extension.

 - The build is no longer broken under NET=n, KUNIT=y for ports that
   don't define their own ipv6 checksum.

* tag 'riscv-for-linus-6.8-mw4' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux: (56 commits)
  lib: checksum: Fix build with CONFIG_NET=n
  riscv: lib: Check if output in asm goto supported
  riscv: Fix build error on rv32 + XIP
  riscv: optimize ELF relocation function in riscv
  RISC-V: Implement archrandom when Zkr is available
  riscv: Optimize hweight API with Zbb extension
  riscv: add dependency among Image(.gz), loader(.bin), and vmlinuz.efi
  samples: ftrace: Add RISC-V support for SAMPLE_FTRACE_DIRECT[_MULTI]
  riscv: ftrace: Add DYNAMIC_FTRACE_WITH_DIRECT_CALLS support
  riscv: ftrace: Make function graph use ftrace directly
  riscv: select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
  lib/Kconfig.debug: Update AS_HAS_NON_CONST_LEB128 comment and name
  riscv: Restrict DWARF5 when building with LLVM to known working versions
  riscv: Hoist linker relaxation disabling logic into Kconfig
  kunit: Add tests for csum_ipv6_magic and ip_fast_csum
  riscv: Add checksum library
  riscv: Add checksum header
  riscv: Add static key for misaligned accesses
  asm-generic: Improve csum_fold
  RISC-V: selftests: cbo: Ensure asm operands match constraints
  ...

16 months agoMerge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 20 Jan 2024 17:42:32 +0000 (09:42 -0800)]
Merge tag 'scsi-misc' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI updates from James Bottomley:
 "Final round of fixes that came in too late to send in the first
  request.

  It's nine bug fixes and one version update (because of a bug fix) and
  one set of PCI ID additions. There's one bug fix in the core which is
  really a one liner (except that an additional sdev pointer was added
  for convenience) and the rest are in drivers"

* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: target: core: Add TMF to tmr_list handling
  scsi: core: Kick the requeue list after inserting when flushing
  scsi: fnic: unlock on error path in fnic_queuecommand()
  scsi: fcoe: Fix unsigned comparison with zero in store_ctlr_mode()
  scsi: mpi3mr: Fix mpi3mr_fw.c kernel-doc warnings
  scsi: smartpqi: Bump driver version to 2.1.26-030
  scsi: smartpqi: Fix logical volume rescan race condition
  scsi: smartpqi: Add new controller PCI IDs
  scsi: ufs: qcom: Remove unnecessary goto statement from ufs_qcom_config_esi()
  scsi: ufs: core: Remove the ufshcd_hba_exit() call from ufshcd_async_scan()
  scsi: ufs: core: Simplify power management during async scan

16 months agoMerge tag 'sh-for-v6.8-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/glaubit...
Linus Torvalds [Sat, 20 Jan 2024 17:24:06 +0000 (09:24 -0800)]
Merge tag 'sh-for-v6.8-tag1' of git://git./linux/kernel/git/glaubitz/sh-linux

Pull sh updates from John Paul Adrian Glaubitz:
 "Since the large patch series to convert arch/sh to device tree support
  has not been finalized yet due to various maintainers still asking for
  changes to the series, this ended up being rather small consisting of
  just two fixes.

  The first patch by Geert Uytterhoeven addresses a build failure in the
  EcoVec platform code. And the second patch by Masahiro Yamada removes
  an unnecessary $(foreach ...) found in a Makefile of the vsyscall
  code.

   - Rename missed backlight field from fbdev to dev

   - Remove unnecessary $(foreach ...)"

* tag 'sh-for-v6.8-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/glaubitz/sh-linux:
  sh: vsyscall: Remove unnecessary $(foreach ...)
  sh: ecovec24: Rename missed backlight field from fbdev to dev

16 months agoMerge tag 'fbdev-for-6.8-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 20 Jan 2024 17:14:04 +0000 (09:14 -0800)]
Merge tag 'fbdev-for-6.8-rc1-2' of git://git./linux/kernel/git/deller/linux-fbdev

Pull fbdev fix from Helge Deller:
 "There were various reports from people without any graphics output on
  the screen and it turns out one commit triggers the problem.

   - Revert 'firmware/sysfb: Clear screen_info state after consuming it'"

* tag 'fbdev-for-6.8-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/linux-fbdev:
  Revert "firmware/sysfb: Clear screen_info state after consuming it"

16 months agoMerge tag 'perf-tools-for-v6.8-1-2024-01-09' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Fri, 19 Jan 2024 22:25:23 +0000 (14:25 -0800)]
Merge tag 'perf-tools-for-v6.8-1-2024-01-09' of git://git./linux/kernel/git/perf/perf-tools

Pull perf tools updates from Arnaldo Carvalho de Melo:
 "Add Namhyung Kim as tools/perf/ co-maintainer, we're taking turns
  processing patches, switching roles from perf-tools to perf-tools-next
  at each Linux release.

  Data profiling:

   - Associate samples that identify loads and stores with data
     structures. This uses events available on Intel, AMD and others and
     DWARF info:

       # To get memory access samples in kernel for 1 second (on Intel)
       $ perf mem record -a -K --ldlat=4 -- sleep 1

       # Similar for the AMD (but it requires 6.3+ kernel for BPF filters)
       $ perf mem record -a --filter 'mem_op == load || mem_op == store, ip > 0x8000000000000000' -- sleep 1

     Then, amongst several modes of post processing, one can do things like:

       $ perf report -s type,typeoff --hierarchy --group --stdio
       ...
       #
       # Samples: 10K of events 'cpu/mem-loads,ldlat=4/P, cpu/mem-stores/P, dummy:u'
       # Event count (approx.): 602758064
       #
       #                    Overhead  Data Type / Data Type Offset
       # ...........................  ............................
       #
           26.09%   3.28%   0.00%     long unsigned int
              26.09%   3.28%   0.00%     long unsigned int +0 (no field)
           18.48%   0.73%   0.00%     struct page
              10.83%   0.02%   0.00%     struct page +8 (lru.next)
               3.90%   0.28%   0.00%     struct page +0 (flags)
               3.45%   0.06%   0.00%     struct page +24 (mapping)
               0.25%   0.28%   0.00%     struct page +48 (_mapcount.counter)
               0.02%   0.06%   0.00%     struct page +32 (index)
               0.02%   0.00%   0.00%     struct page +52 (_refcount.counter)
               0.02%   0.01%   0.00%     struct page +56 (memcg_data)
               0.00%   0.01%   0.00%     struct page +16 (lru.prev)
           15.37%  17.54%   0.00%     (stack operation)
              15.37%  17.54%   0.00%     (stack operation) +0 (no field)
           11.71%  50.27%   0.00%     (unknown)
              11.71%  50.27%   0.00%     (unknown) +0 (no field)

       $ perf annotate --data-type
       ...
       Annotate type: 'struct cfs_rq' in [kernel.kallsyms] (13 samples):
       ============================================================================
           samples     offset       size  field
                13          0        640  struct cfs_rq         {
                 2          0         16      struct load_weight       load {
                 2          0          8          unsigned long        weight;
                 0          8          4          u32  inv_weight;
                                              };
                 0         16          8      unsigned long    runnable_weight;
                 0         24          4      unsigned int     nr_running;
                 1         28          4      unsigned int     h_nr_running;
       ...

       $ perf annotate --data-type=page --group
       Annotate type: 'struct page' in [kernel.kallsyms] (480 samples):
        event[0] = cpu/mem-loads,ldlat=4/P
        event[1] = cpu/mem-stores/P
        event[2] = dummy:u
       ===================================================================================
                samples  offset  size  field
       447  33        0       0    64  struct page     {
       108   8        0       0     8  long unsigned int  flags;
       319  13        0       8    40  union       {
       319  13        0       8    40          struct          {
       236   2        0       8    16              union       {
       236   2        0       8    16                  struct list_head       lru {
       236   1        0       8     8                      struct list_head*  next;
         0   1        0      16     8                      struct list_head*  prev;
                                                       };
       236   2        0       8    16                  struct          {
       236   1        0       8     8                      void*      __filler;
         0   1        0      16     4                      unsigned int       mlock_count;
                                                       };
       236   2        0       8    16                  struct list_head       buddy_list {
       236   1        0       8     8                      struct list_head*  next;
         0   1        0      16     8                      struct list_head*  prev;
                                                       };
       236   2        0       8    16                  struct list_head       pcp_list {
       236   1        0       8     8                      struct list_head*  next;
         0   1        0      16     8                      struct list_head*  prev;
                                                       };
                                                   };
        82   4        0      24     8              struct address_space*      mapping;
         1   7        0      32     8              union       {
         1   7        0      32     8                  long unsigned int      index;
         1   7        0      32     8                  long unsigned int      share;
                                                   };
         0   0        0      40     8              long unsigned int  private;
                                                                 };

     This uses the existing annotate code, calling objdump to do the
     disassembly, with improvements to avoid having this take too long,
     but longer term a switch to a disassembler library, possibly
     reusing code in the kernel will be pursued.

     This is the initial implementation, please use it and report
     impressions and bugs. Make sure the kernel-debuginfo packages match
     the running kernel. The 'perf report' phase for non short perf.data
     files may take a while.

     There is a great article about it on LWN:

       https://lwn.net/Articles/955709/ - "Data-type profiling for perf"

     One last test I did while writing this text, on a AMD Ryzen 5950X,
     using a distro kernel, while doing a simple 'find /' on an
     otherwise idle system resulted in:

     # uname -r
     6.6.9-100.fc38.x86_64
     # perf -vv | grep BPF_
                      bpf: [ on  ]  # HAVE_LIBBPF_SUPPORT
            bpf_skeletons: [ on  ]  # HAVE_BPF_SKEL
     # rpm -qa | grep kernel-debuginfo
     kernel-debuginfo-common-x86_64-6.6.9-100.fc38.x86_64
     kernel-debuginfo-6.6.9-100.fc38.x86_64
     #
     # perf mem record -a --filter 'mem_op == load || mem_op == store, ip > 0x8000000000000000'
     ^C[ perf record: Woken up 1 times to write data ]
     [ perf record: Captured and wrote 2.199 MB perf.data (2913 samples) ]
     #
     # ls -la perf.data
     -rw-------. 1 root root 2346486 Jan  9 18:36 perf.data
     # perf evlist
     ibs_op//
     dummy:u
     # perf evlist -v
     ibs_op//: type: 11, size: 136, config: 0, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|ADDR|CPU|PERIOD|IDENTIFIER|DATA_SRC|WEIGHT, read_format: ID, disabled: 1, inherit: 1, freq: 1, sample_id_all: 1
     dummy:u: type: 1 (PERF_TYPE_SOFTWARE), size: 136, config: 0x9 (PERF_COUNT_SW_DUMMY), { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|ADDR|CPU|IDENTIFIER|DATA_SRC|WEIGHT, read_format: ID, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, task: 1, mmap_data: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1
     #
     # perf report -s type,typeoff --hierarchy --group --stdio
     # Total Lost Samples: 0
     #
     # Samples: 2K of events 'ibs_op//, dummy:u'
     # Event count (approx.): 1904553038
     #
     #            Overhead  Data Type / Data Type Offset
     # ...................  ............................
     #
         73.70%   0.00%     (unknown)
            73.70%   0.00%     (unknown) +0 (no field)
          3.01%   0.00%     long unsigned int
             3.00%   0.00%     long unsigned int +0 (no field)
             0.01%   0.00%     long unsigned int +2 (no field)
          2.73%   0.00%     struct task_struct
             1.71%   0.00%     struct task_struct +52 (on_cpu)
             0.38%   0.00%     struct task_struct +2104 (rcu_read_unlock_special.b.blocked)
             0.23%   0.00%     struct task_struct +2100 (rcu_read_lock_nesting)
             0.14%   0.00%     struct task_struct +2384 ()
             0.06%   0.00%     struct task_struct +3096 (signal)
             0.05%   0.00%     struct task_struct +3616 (cgroups)
             0.05%   0.00%     struct task_struct +2344 (active_mm)
             0.02%   0.00%     struct task_struct +46 (flags)
             0.02%   0.00%     struct task_struct +2096 (migration_disabled)
             0.01%   0.00%     struct task_struct +24 (__state)
             0.01%   0.00%     struct task_struct +3956 (mm_cid_active)
             0.01%   0.00%     struct task_struct +1048 (cpus_ptr)
             0.01%   0.00%     struct task_struct +184 (se.group_node.next)
             0.01%   0.00%     struct task_struct +20 (thread_info.cpu)
             0.00%   0.00%     struct task_struct +104 (on_rq)
             0.00%   0.00%     struct task_struct +2456 (pid)
          1.36%   0.00%     struct module
             0.59%   0.00%     struct module +952 (kallsyms)
             0.42%   0.00%     struct module +0 (state)
             0.23%   0.00%     struct module +8 (list.next)
             0.12%   0.00%     struct module +216 (syms)
          0.95%   0.00%     struct inode
             0.41%   0.00%     struct inode +40 (i_sb)
             0.22%   0.00%     struct inode +0 (i_mode)
             0.06%   0.00%     struct inode +76 (i_rdev)
             0.06%   0.00%     struct inode +56 (i_security)
     <SNIP>

  perf top/report:

   - Don't ignore job control, allowing control+Z + bg to work.

   - Add s390 raw data interpretation for PAI (Processor Activity
     Instrumentation) counters.

  perf archive:

   - Add new option '--all' to pack perf.data with DSOs.

   - Add new option '--unpack' to expand tarballs.

  Initialization speedups:

   - Lazily initialize zstd streams to save memory when not using it.

   - Lazily allocate/size mmap event copy.

   - Lazy load kernel symbols in 'perf record'.

   - Be lazier in allocating lost samples buffer in 'perf record'.

   - Don't synthesize BPF events when disabled via the command line
     (perf record --no-bpf-event).

  Assorted improvements:

   - Show note on AMD systems that the :p, :pp, :ppp and :P are all the
     same, as IBS (Instruction Based Sampling) is used and it is
     inherentely precise, not having levels of precision like in Intel
     systems.

   - When 'cycles' isn't available, fall back to the "task-clock" event
     when not system wide, not to 'cpu-clock'.

   - Add --debug-file option to redirect debug output, e.g.:

       $ perf --debug-file /tmp/perf.log record -v true

   - Shrink 'struct map' to under one cacheline by avoiding function
     pointers for selecting if addresses are identity or DSO relative,
     and using just a byte for some boolean struct members.

   - Resolve the arch specific strerrno just once to use in
     perf_env__arch_strerrno().

   - Reduce memory for recording PERF_RECORD_LOST_SAMPLES event.

  Assorted fixes:

   - Fix the default 'perf top' usage on Intel hybrid systems, now it
     starts with a browser showing the number of samples for Efficiency
     (cpu_atom/cycles/P) and Performance (cpu_core/cycles/P). This
     behaviour is similar on ARM64, with its respective set of
     big.LITTLE processors.

   - Fix segfault on build_mem_topology() error path.

   - Fix 'perf mem' error on hybrid related to availability of mem event
     in a PMU.

   - Fix missing reference count gets (map, maps) in the db-export code.

   - Avoid recursively taking env->bpf_progs.lock in the 'perf_env'
     code.

   - Use the newly introduced maps__for_each_map() to add missing
     locking around iteration of 'struct map' entries.

   - Parse NOTE segments until the build id is found, don't stop on the
     first one, ELF files may have several such NOTE segments.

   - Remove 'egrep' usage, its deprecated, use 'grep -E' instead.

   - Warn first about missing libelf, not libbpf, that depends on
     libelf.

   - Use alternative to 'find ... -printf' as this isn't supported in
     busybox.

   - Address python 3.6 DeprecationWarning for string scapes.

   - Fix memory leak in uniq() in libsubcmd.

   - Fix man page formatting for 'perf lock'

   - Fix some spelling mistakes.

  perf tests:

   - Fail shell tests that needs some symbol in perf itself if it is
     stripped. These tests check if a symbol is resolved, if some hot
     function is indeed detected by profiling, etc.

   - The 'perf test sigtrap' test is currently failing on PREEMPT_RT,
     skip it if sleeping spinlocks are detected (using BTF) and point to
     the mailing list discussion about it. This test is also being
     skipped on several architectures (powerpc, s390x, arm and aarch64)
     due to other pending issues with intruction breakpoints.

   - Adjust test case perf record offcpu profiling tests for s390.

   - Fix 'Setup struct perf_event_attr' fails on s390 on z/VM guest,
     addressing issues caused by the fallback from cycles to task-clock
     done in this release.

   - Fix mask for VG register in the user-regs test.

   - Use shellcheck on 'perf test' shell scripts automatically to make
     sure changes don't introduce things it flags as problematic.

   - Add option to change objdump binary and allow it to be set via
     'perf config'.

   - Add basic 'perf script', 'perf list --json" and 'perf diff' tests.

   - Basic branch counter support.

   - Make DSO tests a suite rather than individual.

   - Remove atomics from test_loop to avoid test failures.

   - Fix call chain match on powerpc for the record+probe_libc_inet_pton
     test.

   - Improve Intel hybrid tests.

  Vendor event files (JSON):

  powerpc:

   - Update datasource event name to fix duplicate events on IBM's
     Power10.

   - Add PVN for HX-C2000 CPU with Power8 Architecture.

  Intel:

   - Alderlake/rocketlake metric fixes.

   - Update emeraldrapids events to v1.02.

   - Update icelakex events to v1.23.

   - Update sapphirerapids events to v1.17.

   - Add skx, clx, icx and spr upi bandwidth metric.

  AMD:

   - Add Zen 4 memory controller events.

  RISC-V:

   - Add StarFive Dubhe-80 and Dubhe-90 JSON files.
       https://www.starfivetech.com/en/site/cpu-u

   - Add T-HEAD C9xx JSON file.
       https://github.com/riscv-software-src/opensbi/blob/master/docs/platform/thead-c9xx.md

  ARM64:

   - Remove UTF-8 characters from cmn.json, that were causing build
     failure in some distros.

   - Add core PMU events and metrics for Ampere One X.

   - Rename Ampere One's BPU_FLUSH_MEM_FAULT to GPC_FLUSH_MEM_FAULT

  libperf:

   - Rename several perf_cpu_map constructor names to clarify what they
     really do.

   - Ditto for some other methods, coping with some issues in their
     semantics, like perf_cpu_map__empty() ->
     perf_cpu_map__has_any_cpu_or_is_empty().

   - Document perf_cpu_map__nr()'s behavior

  perf stat:

   - Exit if parse groups fails.

   - Combine the -A/--no-aggr and --no-merge options.

   - Fix help message for --metric-no-threshold option.

  Hardware tracing:

  ARM64 CoreSight:

   - Bump minimum OpenCSD version to ensure a bugfix is present.

   - Add 'T' itrace option for timestamp trace

   - Set start vm addr of exectable file to 0 and don't ignore first
     sample on the arm-cs-trace-disasm.py 'perf script'"

* tag 'perf-tools-for-v6.8-1-2024-01-09' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools: (179 commits)
  MAINTAINERS: Add Namhyung as tools/perf/ co-maintainer
  perf test: test case 'Setup struct perf_event_attr' fails on s390 on z/vm
  perf db-export: Fix missing reference count get in call_path_from_sample()
  perf tests: Add perf script test
  libsubcmd: Fix memory leak in uniq()
  perf TUI: Don't ignore job control
  perf vendor events intel: Update sapphirerapids events to v1.17
  perf vendor events intel: Update icelakex events to v1.23
  perf vendor events intel: Update emeraldrapids events to v1.02
  perf vendor events intel: Alderlake/rocketlake metric fixes
  perf x86 test: Add hybrid test for conflicting legacy/sysfs event
  perf x86 test: Update hybrid expectations
  perf vendor events amd: Add Zen 4 memory controller events
  perf stat: Fix hard coded LL miss units
  perf record: Reduce memory for recording PERF_RECORD_LOST_SAMPLES event
  perf env: Avoid recursively taking env->bpf_progs.lock
  perf annotate: Add --insn-stat option for debugging
  perf annotate: Add --type-stat option for debugging
  perf annotate: Support event group display
  perf annotate: Add --data-type option
  ...

16 months agoMerge tag 'strlcpy-removal-v6.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 19 Jan 2024 21:49:16 +0000 (13:49 -0800)]
Merge tag 'strlcpy-removal-v6.8-rc1' of git://git./linux/kernel/git/kees/linux

Pull strlcpy removal from Kees Cook:
 "As promised, this is 'part 2' of the hardening tree, late in -rc1 now
  that all the other trees with strlcpy() removals have landed. One new
  user appeared (in bcachefs) but was a trivial refactor. The kernel is
  now free of the strlcpy() API!

   - Remove of the final (very recent) user of strlcpy() (in bcachefs)

   - Remove the strlcpy() API. Long live strscpy()"

* tag 'strlcpy-removal-v6.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  string: Remove strlcpy()
  bcachefs: Replace strlcpy() with strscpy()

16 months agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 19 Jan 2024 21:36:15 +0000 (13:36 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "I think the main one is fixing the dynamic SCS patching when full LTO
  is enabled (clang was silently getting this horribly wrong), but it's
  all good stuff.

  Rob just pointed out that the fix to the workaround for erratum
  #2966298 might not be necessary, but in the worst case it's harmless
  and since the official description leaves a little to be desired here,
  I've left it in.

  Summary:

   - Fix shadow call stack patching with LTO=full

   - Fix voluntary preemption of the FPSIMD registers from assembly code

   - Fix workaround for A520 CPU erratum #2966298 and extend to A510

   - Fix SME issues that resulted in corruption of the register state

   - Minor fixes (missing includes, formatting)"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Fix silcon-errata.rst formatting
  arm64/sme: Always exit sme_alloc() early with existing storage
  arm64/fpsimd: Remove spurious check for SVE support
  arm64/ptrace: Don't flush ZA/ZT storage when writing ZA via ptrace
  arm64: entry: simplify kernel_exit logic
  arm64: entry: fix ARM64_WORKAROUND_SPECULATIVE_UNPRIV_LOAD
  arm64: errata: Add Cortex-A510 speculative unprivileged load workaround
  arm64: Rename ARM64_WORKAROUND_2966298
  arm64: fpsimd: Bring cond_yield asm macro in line with new rules
  arm64: scs: Work around full LTO issue with dynamic SCS
  arm64: irq: include <linux/cpumask.h>

16 months agoMerge tag 'loongarch-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai...
Linus Torvalds [Fri, 19 Jan 2024 21:30:49 +0000 (13:30 -0800)]
Merge tag 'loongarch-6.8' of git://git./linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch updates from Huacai Chen:

 - Raise minimum clang version to 18.0.0

 - Enable initial Rust support for LoongArch

 - Add built-in dtb support for LoongArch

 - Use generic interface to support crashkernel=X,[high,low]

 - Some bug fixes and other small changes

 - Update the default config file.

* tag 'loongarch-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson: (22 commits)
  MAINTAINERS: Add BPF JIT for LOONGARCH entry
  LoongArch: Update Loongson-3 default config file
  LoongArch: BPF: Prevent out-of-bounds memory access
  LoongArch: BPF: Support 64-bit pointers to kfuncs
  LoongArch: Fix definition of ftrace_regs_set_instruction_pointer()
  LoongArch: Use generic interface to support crashkernel=X,[high,low]
  LoongArch: Fix and simplify fcsr initialization on execve()
  LoongArch: Let cores_io_master cover the largest NR_CPUS
  LoongArch: Change SHMLBA from SZ_64K to PAGE_SIZE
  LoongArch: Add a missing call to efi_esrt_init()
  LoongArch: Parsing CPU-related information from DTS
  LoongArch: dts: DeviceTree for Loongson-2K2000
  LoongArch: dts: DeviceTree for Loongson-2K1000
  LoongArch: dts: DeviceTree for Loongson-2K0500
  LoongArch: Allow device trees be built into the kernel
  dt-bindings: interrupt-controller: loongson,liointc: Fix dtbs_check warning for interrupt-names
  dt-bindings: interrupt-controller: loongson,liointc: Fix dtbs_check warning for reg-names
  dt-bindings: loongarch: Add Loongson SoC boards compatibles
  dt-bindings: loongarch: Add CPU bindings for LoongArch
  LoongArch: Enable initial Rust support
  ...

16 months agoRevert "firmware/sysfb: Clear screen_info state after consuming it"
Helge Deller [Fri, 19 Jan 2024 20:47:15 +0000 (21:47 +0100)]
Revert "firmware/sysfb: Clear screen_info state after consuming it"

This reverts commit df67699c9cb0ceb70f6cc60630ca938c06773eda.

Jens Axboe reported a regression that his machine is failing to show a
console, or in fact anything, on current -git. There's no output and no
console after:

Loading Linux 6.7.0+ ...
Loading initial ramdisk ...

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Jens Axboe <axboe@kernel.dk>