- 29 Sep, 2014 1 commit
-
-
Matt Fleming authored
Passing pointers to alias modifiers 'unit' and 'scale' isn't very future-proof since if we add more modifiers to the list we'll end up passing more arguments. Instead wrap everything up in a struct perf_pmu_info, which can easily be expanded when additional alias modifiers are necessary in the future. Signed-off-by:
Matt Fleming <matt.fleming@intel.com> Acked-by:
Jiri Olsa <jolsa@kernel.org> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1411567455-31264-3-git-send-email-matt@console-pimps.orgSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
- 27 Sep, 2014 1 commit
-
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo: User visible changes: o Restore "--callchain graph" output, broken in recent cset to end up being the same as "fractal" (Namhyung Kim) o Allow profiling when kptr_restrict == 1 for non root users, kernel samples will just remain unresolved (Andi Kleen) o Allow configuring default options for callchains in config file (Namhyung Kim) o Fix line number in the config file error message (Jiri Olsa) o Fix --per-core on multi socket systems (Andi Kleen) Cleanups: o Use ACCESS_ONCE() instead of volatile cast. (Pranith Kumar) o Modify error code for when perf_session__new() fails (Taeung Song) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 26 Sep, 2014 11 commits
-
-
Jiri Olsa authored
If we fail to parse the config file within the callback function, the line number counter 'could be' already on the next line. This results in wrong line number report like: $ cat ~/.perfconfig [call-graph] sort-key = krava $ perf record ls Fatal: bad config file line 3 in /home/jolsa/.perfconfig Fixing this by saving the current line number for this case. Signed-off-by:Namhyung Kim <namhyung@kernel.org> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Milian Wolff <mail@milianw.de> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20140923115656.GC2979@krava.brq.redhat.comSigned-off-by:
-
Namhyung Kim authored
So that it'll be passed to perf_callchain_config(). Reviewed-by:
David Ahern <dsahern@gmail.com> Signed-off-by:
Jiri Olsa <jolsa@redhat.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Milian Wolff <mail@milianw.de> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1411434104-5307-6-git-send-email-namhyung@kernel.orgSigned-off-by:
-
Namhyung Kim authored
This patch adds support for following config options to ~/.perfconfig file. [call-graph] record-mode = dwarf dump-size = 8192 print-type = fractal order = callee threshold = 0.5 print-limit = 128 sort-key = function Reviewed-by: -
Namhyung Kim authored
And rename record_callchain_parse() to parse_callchain_record_opt() in accordance to parse_callchain_report_opt(). Reviewed-by:
-
Namhyung Kim authored
So that all callchain config parameters can be read/written to a single place. It's a preparation to consolidate handling of all callchain options. Reviewed-by:
-
Namhyung Kim authored
Currently perf report -g graph option doesn't work as expected and always work as same as -g fractal. This was a bug during recent callchain print code cleanup. Before: $ perf report -g graph Children Self Command Shared Object Symbol ================================================================ - 56.19% 35.41% sleep [kernel.kallsyms] [k] page_fault - page_fault + 63.02% _dl_relocate_object + 36.98% clear_user After: Children Self Command Shared Object Symbol ================================================================ - 56.19% 35.41% sleep [kernel.kallsyms] [k] page_fault - page_fault + 35.41% _dl_relocate_object + 20.78% clear_user Reviewed-by: -
Pranith Kumar authored
Use ACCESS_ONCE() instead of the cast to volatile and read. This is just a style change which is reader friendly. Signed-off-by:
Pranith Kumar <bobby.prani@gmail.com> Acked-by:
-
Taeung Song authored
Because perf_session__new() can fail for more reasons than just ENOMEM, modify error code(ENOMEM or EINVAL) to -1. Signed-off-by:
Taeung Song <treeze.taeung@gmail.com> Acked-by:
-
Andi Kleen authored
Currently perf record always errors out when you run it as non-root with kptr_restrict == 1, which is often the default. Make it only warn instead and fix the kernel resolve code to not segfault later. Profiling works still fine, except kernel symbols are not resolved. Signed-off-by:
Andi Kleen <ak@linux.intel.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1411594794-7229-1-git-send-email-andi@firstfloor.orgSigned-off-by:
-
Andi Kleen authored
On systems with more than one socket perf stat --per-core would either segfault or stop before outputting all cores. The problem was that the output code referenced the id including the socket number in the higher bits, which is far beyond any per cpu array. Mask out the socket number before referencing cpus in abs_printout. I also renamed the variable in nsec_printout to be clear what it is, even though it doesn't reference cpus. Signed-off-by:
Stephane Eranian <eranian@google.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1411591846-32736-1-git-send-email-andi@firstfloor.orgSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-fdarray-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core Pull perf tooling updates from Arnaldo Carvalho de Melo. Infrastructure changes: * We were not handling POLLHUP notifications for event file descriptors. Fix it by filtering entries in the events file descriptor array after poll() returns, refcounting mmaps so that when the last fd pointing to a perf mmap goes away we do the unmap. (Arnaldo Carvalho de Melo) User visible changes: * Now 'record' and 'trace' properly exit when a target thread exits. (Arnaldo Carvalho de Melo) Signed-off-by:
-
- 25 Sep, 2014 14 commits
-
-
Arnaldo Carvalho de Melo authored
So that we don't continue polling on vanished file descriptors, i.e. file descriptors for events monitoring threads that exited. I.e. the following 'trace' command now exits as expected, instead of staying in an eternal loop: $ sleep 5s & $ trace -p `pidof sleep` Reported-by: -
Arnaldo Carvalho de Melo authored
So that we don't continue polling on vanished file descriptors, i.e. file descriptors for events monitoring threads that exited. I.e. the following 'perf record' command now exits as expected, instead of staying in an eternal loop: $ sleep 5s & $ perf record -p `pidof sleep` Reported-by:
-
Arnaldo Carvalho de Melo authored
As noticed by receiving a POLLHUP for all its pollfd entries. That will remove the refcount taken in perf_evlist__mmap_per_evsel(), and when all events are consumed via perf_evlist__mmap_read() + perf_evlist__mmap_consume(), the ring buffer will be unmap'ed. Thanks to Jiri Olsa for pointing out that we must wait till all events are consumed, not being ok to unmmap just when receiving all the POLLHUPs. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/n/tip-t10w1xk4myp7ca7m9fvip6a0@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
We will use this in perf's evlist class so that it can, at fdarray__filter() time, to unmap the associated ring buffer. We may need to have further info associated with each fdarray entry, in that case we'll make that int array a 'union fdarray_priv' one and put a pointer there so that users can stash whatever they want there. For now, an int is enough tho. v2: Add clarification to the per array entry priv area, as well as make it a union, which makes usage a bit longer, but if/when we make it use more space by allowing per entry pointers existing users source code will not have to be changed, just rebuilt. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: -
Arnaldo Carvalho de Melo authored
We need to know how many fds are using a perf mmap via PERF_EVENT_IOC_SET_OUTPUT, so that we can know when to ditch an mmap, refcount it. v2: Automatically unmap it when the refcount hits one, which will happen when all fds are filtered by perf_evlist__filter_pollfd(), in later patches. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20140908153824.GG2773@kernel.org Link: http://lkml.kernel.org/n/tip-cpv7v2lw0g74ucmxa39xdpms@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
The extensible file description array that grew in the perf_evlist class can be useful for other tools, as it is not something that only evlists need, so move it to tools/lib/api/fd to ease sharing it. v2: Don't use {} like in: libapi_dirs: $(QUIET_MKDIR)mkdir -p $(OUTPUT){fs,fd}/ in Makefiles, as it will not work in some systems, as in ubuntu13.10. v3: Add fd/*.[ch] to LIBAPIKFS_SOURCES (Fix from Jiri Olsa) v4: Leave the fcntl(fd, O_NONBLOCK) in the evlist layer, remains to be checked if it is really needed there, but has no place in the fdarray class (Fix from Jiri Olsa) v5: Remove evlist details from fdarray grow/filter tests. Improve it a bit doing more tests about expected internal state. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/n/tip-kleuni3hckbc3s0lu6yb9x40@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
Since we have access two evlist members in all these poll calls, provide a helper. This will also help to make the patch introducing the pollfd class more clear, as the evlist specific uses will be hiden away perf_evlist__poll(). Acked-by:
-
Arnaldo Carvalho de Melo authored
Since we can add file descriptors to the evlist pollfd and it will autogrow, no need to copy all events to a local pollfd array, just add the timer and stdin file descriptors. Reviewed-by:
-
Arnaldo Carvalho de Melo authored
[acme@ssdandy linux]$ perf test "Add fd" 34: Add fd to pollfd array, making it autogrow : Ok [acme@ssdandy linux]$ perf test -v "Add fd" 34: Add fd to pollfd array, making it autogrow : --- start --- test child forked, pid 19817 before growing array: 2 [ 1, 2 ] after 3rd add_pollfd: 3 [ 1, 2, 35 ] after 4th add_pollfd: 4 [ 1, 2, 35, 88 ] test child finished with 0 ---- end ---- Add fd to pollfd array, making it autogrow: Ok [acme@ssdandy linux]$ Acked-by:
-
Arnaldo Carvalho de Melo authored
This way we will be able to add more file descriptors to be polled, like stdin or some timer fd. At this point we might as well yank the pollfd class from evlist so that it can be used in other places. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/n/tip-o2mzsjl7taumsoc35ryol00i@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Because we want to notice when they get POLLHUP'ed, so that we can figure out when all threads exited in a workload being monitored. We can't just monitor the fds that were mmaped, we need to notice when all the fds that were PERF_EVENT_IOC_SET_OUTPUT'ed too, because the mmap stays even after the fd that originally was used to do the mmap call went away, its only when all the set-output fds for a mmap are gone that the mmap is. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Don Zickus <dzickus@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/20140908151016.GH17728@krava.brq.redhat.com Link: http://lkml.kernel.org/n/tip-24omlq5asrfg4uo3muuzn2bl@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
We want to know when the fd went away, like when a monitored thread exits. If we do not monitor such events, then the tools will wait forever on events from a vanished thread, like when running: $ sleep 5s & $ perf record -p `pidof sleep` This builds upon the kernel patch by Jiri Olsa that actually makes a poll on those file descriptors to return POLLHUP. It is also needed to change the tools to use perf_evlist__filter_pollfd() to check if there are remainings fds to monitor or if all are gone, in which case they will exit the poll/mmap/read loop. Acked-by:
-
Arnaldo Carvalho de Melo authored
That will use a synthetic evlist with just what is touched by this new method to check that it works as expected. Output in verbose mode: $ perf test -v pollfd 33: Filter fds with revents mask in a pollfd array : --- start --- filtering all but pollfd[2]: before: 5 [ 5, 4, 3, 2, 1 ] after: 1 [ 3 ] filtering all but (pollfd[0], pollfd[3]): before: 5 [ 5, 4, 3, 2, 1 ] after: 2 [ 5, 2 ] test child finished with 0 ---- end ---- Filter fds with revents mask in a pollfd array: Ok $ Acked-by:
-
Arnaldo Carvalho de Melo authored
To remove all entries in evlist->pollfd[] that have revents matching at least one of the bits in the specified mask. It'll adjust evlist->nr_fds to the number of unfiltered fds and will return this value, as a convenience and to avoid requiring direct access to internal state of perf_evlist objects. This will be used after polling the evlist fds so that we remove fds that were closed by the kernel. Acked-by:
-
- 24 Sep, 2014 13 commits
-
-
Stephane Eranian authored
This patch restructures the memory controller (IMC) uncore PMU support for client SNB/IVB/HSW processors. The main change is that it can now cope with more than one PCI device ID per processor model. There are many flavors of memory controllers for each processor. They have different PCI device ID, yet they behave the same w.r.t. the memory controller PMU that we are interested in. The patch now supports two distinct memory controllers for IVB processors: one for mobile, one for desktop. Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Link: http://lkml.kernel.org/r/20140917090616.GA11281@quad Cc: ak@linux.intel.com Cc: kan.liang@intel.com Cc: Arnaldo Carvalho de Melo <acme@kernel.org> Cc: Bjorn Helgaas <bhelgaas@google.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by:
-
Andi Kleen authored
The PCU frequency band filters use 8 bit each in a register. When setting up the value the shift value was not correctly scaled, which resulted in all filters except for band 0 to be zero. Fix the scaling. This allows to correctly monitor multiple uncore frequency bands. Signed-off-by:
-
Andi Kleen authored
The IvyBridge-EP uncore driver was missing three filter flags: NC, ISOC, C6 which are useful in some cases. Support them in the same way as the Haswell EP driver, by allowing to set them and exposing them in the sysfs formats. Also fix a typo in a define. Relies on the Haswell EP driver to be applied earlier. Signed-off-by:
-
Yan, Zheng authored
Current code registers PMUs for all possible uncore pci devices. This is not good because, on some machines, one or more uncore pci devices can be missing. The missing pci device make corresponding PMU unusable. Register the PMU only if the uncore device exists. Signed-off-by:
Yan, Zheng <zheng.z.yan@intel.com> Signed-off-by:
-
Yan, Zheng authored
The uncore subsystem in Haswell-EP is similar to Sandy/Ivy Bridge-EP. There are some differences in config register encoding and pci device IDs. The Haswell-EP uncore also supports a few new events. Add the Haswell-EP driver to the snbep split driver. Signed-off-by:
-
Andi Kleen authored
Use the newly added Broadwell cache event list for Haswell too. All Haswell and Broadwell events and offcore masks used in these lists are identical. However Haswell is very different from the Sandy Bridge list that was used previously. That fixes a wide range of mis-counting cache events. The node events are now only for retired memory events, so prefetching and speculative memory accesses are not included. They are PEBS capable now, which makes it much easier to sample for them, plus it's possible to create address maps with -d. The prefetch events are gone now. They way the hardware counts them is very misleading (some prefetches included, others not), so it seemed best to leave them out. Signed-off-by:
-
Andi Kleen authored
On Broadwell INST_RETIRED.ALL cannot be used with any period that doesn't have the lowest 6 bits cleared. And the period should not be smaller than 128. Add a new callback to enforce this, and set it for Broadwell. This is erratum BDM57 and BDM11. How does this handle the case when an app requests a specific period with some of the bottom bits set The apps thinks it is sampling at X occurences per sample, when it is in fact at X - 63 (worst case). Short answer: Any useful instruction sampling period needs to be 4-6 orders of magnitude larger than 128, as an PMI every 128 instructions would instantly overwhelm the system and be throttled. So the +-64 error from this is really small compared to the period, much smaller than normal system jitter. Long answer: <write up by Peter:> IFF we guarantee perf_event_attr::sample_period >= 128. Suppose we start out with sample_period=192; then we'll set period_left to 192, we'll end up with left = 128 (we truncate the lower bits). We get an interrupt, find that period_left = 64 (>0 so we return 0 and don't get an overflow handler), up that to 128. Then we trigger again, at n=256. Then we find period_left = -64 (<=0 so we return 1 and do get an overflow). We increment with sample_period so we get left = 128. We fire again, at n=384, period_left = 0 (<=0 so we return 1 and get an overflow). And on and on. So while the individual interrupts are 'wrong' we get then with interval=256,128 in exactly the right ratio to average out at 192. And this works for everything >=128. So the num_samples*fixed_period thing is still entirely correct +- 127, which is good enough I'd say, as you already have that error anyhow. So no need to 'fix' the tools, al we need to do is refuse to create INST_RETIRED:ALL events with sample_period < 128. Signed-off-by:
-
Andi Kleen authored
Add Broadwell support for Broadwell Client to perf. This is very similar to Haswell. It uses a new cache event table, because there were various changes there. The constraint list has one new event that needs to be handled over Haswell. The PEBS event list is the same, so we reuse Haswell's. [fengguang.wu: make intel_bdw_event_constraints[] static] Signed-off-by:
-
Andi Kleen authored
Add names for each Haswell model as requested by Peter. Signed-off-by:
-
Andi Kleen authored
71 is a Broadwell, not a Haswell. The model number was added by mistake earlier. Remove it for now, until it can be re-added later with real Broadwell support. In practice it does not cause a lot of issues because the Broadwell PMU is very similar to Haswell, but some details were wrong, and it's better to handle it correctly. Signed-off-by:
-
Jiri Olsa authored
This reverts commit 1f9a7268. With the fix of the initial state for the cloned event we now correctly handle the error described in: 1f9a7268 perf: Do not allow optimized switch for non-cloned events so we can revert it. I made an automated test for this, but its not suitable for automated perf tests framework. It needs to be customized for each machine (the more cpu the higher numbers for GROUPS/WORKERS/BYTES) and it could take longer time to hit the issue. Signed-off-by:
-
Jiri Olsa authored
Currently we initialize the child event based on the original parent state. This is wrong, because the original parent event (and its state) is not related to current fork and also could be already gone. We need to initialize the child state based on the immediate parent event state. Signed-off-by:
-
Jiri Olsa authored
Currently we return POLLHUP in event polling if the monitored process is done, but we didn't consider possible children, that might be still running and producing data. Before returning POLLHUP making sure that: 1) the monitored task has exited and that 2) we don't have any children to monitor Also adding parent wakeup when the child event is gone. Suggested-by:
Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by:
-
