- 17 May, 2016 10 commits
-
-
Arnaldo Carvalho de Melo authored
We need have different helpers to account how many contexts we have in the sample and for real addresses, so do it now as a prep patch, to ease review. Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/n/tip-q964tnyuqrxw5gld18vizs3c@git.kernel.orgSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Arnaldo Carvalho de Melo authored
We will use it to count how many addresses are in the entry->ip[] array, excluding PERF_CONTEXT_{KERNEL,USER,etc} entries, so that we can really return the number of entries specified by the user via the relevant sysctl, kernel.perf_event_max_contexts, or via the per event perf_event_attr.sample_max_stack knob. This way we keep the perf_sample->ip_callchain->nr meaning, that is the number of entries, be it real addresses or PERF_CONTEXT_ entries, while honouring the max_stack knobs, i.e. the end result will be max_stack entries if we have at least that many entries in a given stack trace. Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/n/tip-s8teto51tdqvlfhefndtat9r@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
This makes perf_callchain_{user,kernel}() receive the max stack as context for the perf_callchain_entry, instead of accessing the global sysctl_perf_event_max_stack. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexei Starovoitov <ast@kernel.org> Cc: Brendan Gregg <brendan.d.gregg@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: He Kuang <hekuang@huawei.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Milian Wolff <milian.wolff@kdab.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vince Weaver <vincent.weaver@maine.edu> Cc: Wang Nan <wangnan0@huawei.com> Cc: Zefan Li <lizefan@huawei.com> Link: http://lkml.kernel.org/n/tip-kolmn1yo40p7jhswxwrc7rrd@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
So that it can be used for other stack related knobs, such as the upcoming one to tweak the max number of of contexts per stack sample. In all those cases we can only change the value if there are no perf sessions collecting stacks, so they need to grab that mutex, etc. Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/n/tip-8t3fk94wuzp8m2z1n4gc0s17@git.kernel.orgSigned-off-by:
-
Masami Hiramatsu authored
Instead of using a raw string, use DSO__NAME_KALLSYMS and DSO__NAME_KCORE macros for kallsyms and kcore. Signed-off-by:
Masami Hiramatsu <mhiramat@kernel.org> Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com> Cc: Brendan Gregg <brendan.d.gregg@gmail.com> Cc: Hemant Kumar <hemant@linux.vnet.ibm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20160515031935.4017.50971.stgit@devboxSigned-off-by:
-
Namhyung Kim authored
Currently 'perf stat' always counts task-clock event by default. But it's somewhat confusing for system-wide targets (especially with 'sleep N' as the 'sleep' task just sleeps and doesn't use cputime). Changing to cpu-clock event instead for that case makes more sense IMHO. Before: # perf stat -a sleep 0.1 Performance counter stats for 'system wide': 403.038603 task-clock (msec) # 4.001 CPUs utilized 150 context-switches # 0.372 K/sec 7 cpu-migrations # 0.017 K/sec 71 page-faults # 0.176 K/sec 23,705,169 cycles # 0.059 GHz 15,888,166 instructions # 0.67 insn per cycle 3,326,078 branches # 8.253 M/sec 87,643 branch-misses # 2.64% of all branches 0.100737009 seconds time elapsed # After: # perf stat -a sleep 0.1 Performance counter stats for 'system wide': 404.271182 cpu-clock (msec) # 4.000 CPUs utilized 143 context-switches # 0.354 K/sec 13 cpu-migrations # 0.032 K/sec 73 page-faults # 0.181 K/sec 22,119,220 cycles # 0.055 GHz 13,622,065 instructions # 0.62 insn per cycle 2,918,769 branches # 7.220 M/sec 85,033 branch-misses # 2.91% of all branches 0.101073089 seconds time elapsed # Signed-off-by:Namhyung Kim <namhyung@kernel.org> Tested-by:
-
Namhyung Kim authored
Currently only the task-clock event updates the runtime_nsec so it cannot show the metric when using cpu-clock events. However cpu clock works basically same as task-clock, so no need to not update the runtime IMHO. Before: # perf stat -a -e cpu-clock,context-switches,page-faults,cycles sleep 0.1 Performance counter stats for 'system wide': 1217.759506 cpu-clock (msec) 93 context-switches 61 page-faults 18,958,022 cycles 0.101393794 seconds time elapsed After: Performance counter stats for 'system wide': 1220.471884 cpu-clock (msec) # 12.013 CPUs utilized 118 context-switches # 0.097 K/sec 59 page-faults # 0.048 K/sec 17,941,247 cycles # 0.015 GHz 0.101594777 seconds time elapsed Signed-off-by: -
Namhyung Kim authored
The commit 140aeadc ("perf stat: Abstract stat metrics printing") changed how shadow metrics are printed, but it missed to update the width of the stalled backend cycles event to 7.2% like others. This resulted in misaligned output like below: Performance counter stats for 'pwd': 0.638313 task-clock (msec) # 0.567 CPUs utilized 0 context-switches # 0.000 K/sec 0 cpu-migrations # 0.000 K/sec 54 page-faults # 0.085 M/sec 885,600 cycles # 1.387 GHz 558,438 stalled-cycles-frontend # 63.06% frontend cycles idle 431,355 stalled-cycles-backend # 48.71% backend cycles idle 674,956 instructions # 0.76 insn per cycle # 0.83 stalled cycles per insn 130,380 branches # 204.257 M/sec <not counted> branch-misses 0.001125426 seconds time elapsed Signed-off-by:
-
He Kuang authored
When unwinding callchains on a different machine, vdso info should be available so the unwind process won't be interrupted if address falls into vdso region. But in most cases, the addresses of sample events are not in vdso range, the buildid of a zero hit vdso won't be stored into perf.data. This patch stores vdso buildid regardless of whether the vdso is hit or not. Signed-off-by:
He Kuang <hekuang@huawei.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ekaterina Tumanova <tumanova@linux.vnet.ibm.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Kan Liang <kan.liang@intel.com> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/r/1463042596-61703-3-git-send-email-hekuang@huawei.comSigned-off-by:
-
Andi Kleen authored
When the scaling factor is a full integer don't display fractional digits. This avoids unnecessary .00 output for topdown metrics with scale factors. v2: Remove redundant check. Signed-off-by:
Andi Kleen <ak@linux.intel.com> Acked-by:
Jiri Olsa <jolsa@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1462489447-31832-7-git-send-email-andi@firstfloor.org [ Rename 'round' to 'stat_round' as 'round' is defined in math.h, included by this patch, and this breaks the build on ubuntu 12.04 ] Signed-off-by:
-
- 12 May, 2016 7 commits
-
-
Alexander Shishkin authored
When the PMU driver reports a truncated AUX record, it effectively means that there is no more usable room in the event's AUX buffer (even though there may still be some room, so that perf_aux_output_begin() doesn't take action). At this point the consumer still has to be woken up and the event has to be disabled, otherwise the event will just keep spinning between perf_aux_output_begin() and perf_aux_output_end() until its context gets unscheduled. Again, for cpu-wide events this means never, so once in this condition, they will be forever losing data. Fix this by disabling the event and waking up the consumer in case of a truncated AUX record. Reported-by:
Markus Metzger <markus.t.metzger@intel.com> Signed-off-by:
Alexander Shishkin <alexander.shishkin@linux.intel.com> Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@infradead.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vince Weaver <vincent.weaver@maine.edu> Cc: vince@deater.net Link: http://lkml.kernel.org/r/1462886313-13660-3-git-send-email-alexander.shishkin@linux.intel.comSigned-off-by:
Ingo Molnar <mingo@kernel.org>
-
Alexander Shishkin authored
Currently, the PT driver always sets the PMI bit one region (page) before the STOP region so that we can wake up the consumer before we run out of room in the buffer and have to disable the event. However, we also need an interrupt in the last output region, so that we actually get to disable the event (if no more room from new data is available at that point), otherwise hardware just quietly refuses to start, but the event is scheduled in and we end up losing trace data till the event gets removed. For a cpu-wide event it is even worse since there may not be any re-scheduling at all and no chance for the ring buffer code to notice that its buffer is filled up and the event needs to be disabled (so that the consumer can re-enable it when it finishes reading the data out). In other words, all the trace data will be lost after the buffer gets filled up. This patch makes PT also generate a PMI when the last output region is full. Reported-by:
-
Ingo Molnar authored
Signed-off-by: -
Andrey Ryabinin authored
Jim reported: UBSAN: Undefined behaviour in arch/x86/events/intel/core.c:3708:12 shift exponent 35 is too large for 32-bit type 'long unsigned int' The use of 'unsigned long' type obviously is not correct here, make it 'unsigned long long' instead. Reported-by:
Jim Cromie <jim.cromie@gmail.com> Signed-off-by:
Andrey Ryabinin <aryabinin@virtuozzo.com> Signed-off-by:
-
Peter Zijlstra authored
We compute 'delta' and properly sign extend it and then ignore it and recompute the raw value, loosing the sign extention. Signed-off-by:
-
hchrzani authored
CHA events in Knights Landing platform require programming filter registers properly. Remote node, local node and NonNearMemCachable bits should be set to 1 at all times. Signed-off-by:
Hubert Chrzaniuk <hubert.chrzaniuk@intel.com> Signed-off-by:
Lawrence F Meadows <lawrence.f.meadows@intel.com> Signed-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-20160511' 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: - Fix symbol insertion and callchain behavior in db-export (Chris Phlipot) Infrastructure changes: - Add libunwind build test (feature query), working towards supporting cross-platform DWARF callchains, starting with arm/arm64 (He Kuang) - Use lsdir() more extensively (Masami Hiramatsu) - Use SBUILD_ID_SIZE in places where the equivalent expression was being used (Masami Hiramatsu) - Split some more 'perf trace' syscall arg beautifiers (Arnaldo Carvalho de Melo) Signed-off-by:
-
- 11 May, 2016 16 commits
-
-
Masami Hiramatsu authored
Use new lsdir() for looking up buildid caches. This changes logic a bit to ignore all dot files, since the build-id cache must not start with dot. Signed-off-by:
Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com> Cc: Brendan Gregg <brendan.d.gregg@gmail.com> Cc: Hemant Kumar <hemant@linux.vnet.ibm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20160511135217.23943.94596.stgit@devboxSigned-off-by:
-
Masami Hiramatsu authored
Use lsdir() to search in kcore cache directory. This also avoids checking hidden dot directory entries, because kcore cache directories must always have the name from timestamps when taking the kcore snapshots, and it never start with dot. Signed-off-by:
-
Masami Hiramatsu authored
Use the existing SBUILD_ID_SIZE macro instead of the equivalent BUILD_ID_SIZE * 2 + 1 expression for allocating a buffer for build-id strings. Signed-off-by:
-
Masami Hiramatsu authored
Fix lsdir() to set correct positive error number (ENOMEM). Since "errno" must have a positive error number instead of negative number, fix lsdir to set it correctly. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
To reduce the size of builtin-trace.c. Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/n/tip-ovxifncj34ynrjjseg33lil3@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To reduce the size of builtin-trace.c. Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/n/tip-c4c47w2a2jx13terl2p2hros@git.kernel.orgSigned-off-by:
-
He Kuang authored
Debug-frame for remote platforms is not related to the host platform, so we should test each platform separately. Signed-off-by:
-
He Kuang authored
Currently only test for local libunwind. We should check all supported platforms so we can use them to parse perf.data with callchain info on different machines. Signed-off-by:
-
Chris Phlipot authored
When an IP with an unresolved symbol occurs in the callchain more than once (ie. recursion), then duplicate symbols can be created because the callchain nodes are never updated after they are first created. To fix this issue we call dso__find_symbol whenever we encounter a NULL symbol, in case we already added a symbol at that IP since we started traversing the callchain. This change prevents duplicate symbols from being exported when duplicate IPs are present in the callchain. Signed-off-by:
Chris Phlipot <cphlipot0@gmail.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1462937209-6032-5-git-send-email-cphlipot0@gmail.comSigned-off-by:
-
Chris Phlipot authored
Remove the call to map_ip() to adjust al.addr, because it has already been called when assembling the callchain, in: thread__resolve_callchain_sample(perf_sample) add_callchain_ip(ip = perf_sample->callchain->ips[j]) thread__find_addr_location(addr = ip) thread__find_addr_map(addr) { al->addr = addr if (al->map) al->addr = al->map->map_ip(al->map, al->addr); } Calling it a second time can result in incorrect addresses being used. This can have effects such as duplicate symbols being created and exported. Signed-off-by: -
Chris Phlipot authored
Use the dso__insert_symbol function instead of symbols__insert() in order to properly update the dso symbol cache. If the cache is not updated, then duplicate symbols can be unintentionally created, inserted, and exported. This change prevents duplicate symbols from being exported due to dso__find_symbol() using a stale symbol cache. Signed-off-by:
-
Chris Phlipot authored
The current method for inserting symbols is to use the symbols__insert() function. However symbols__insert() does not update the dso symbol cache. This causes problems in the following scenario: 1. symbol not found at addr using dso__find_symbol 2. symbol inserted at addr using the existing symbols__insert function 3. symbol still not found at addr using dso__find_symbol() because cache isn't updated. This is undesired behavior. The undesired behavior in (3) is addressed by creating a new function, dso__insert_symbol() to both insert the symbol and update the symbol cache if necessary. If dso__insert_symbol() is used in (2) instead of symbols__insert(), then the undesired behavior in (3) is avoided. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
It probably is equivalent, but that seems to be the "pythonic" way of dieing? Anyway, one less die() in the tools/perf codebase. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Cc: Chris Phlipot <cphlipot0@gmail.com> Link: http://lkml.kernel.org/n/tip-nlzgepdv2818zs4e7faif9tu@git.kernel.orgSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-20160510' 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: - Recording 'dwarf' callchains do not need DWARF unwinding support (He Kuang) - Print recently added perf_event_attr.write_backward bit flag in -vv verbose mode (Arnaldo Carvalho de Melo) - Fix incorrect python db-export error message in 'perf script' (Chris Phlipot) - Fix handling of zero-length symbols (Chris Phlipot) - perf stat: Scale values by unit before metrics (Andi Kleen) Infrastructure changes: - Rewrite strbuf not to die(), making tools using it to check its return value instead (Masami Hiramatsu) - Support reading from backward ring buffer, add a 'perf test' entry for it (Wang Nan) Signed-off-by:
-
Ingo Molnar authored
Signed-off-by: -
Namhyung Kim authored
The commit b97511c5 ("perf tools: Add overhead/overhead_children keys defaults via string") moved initialization of column headers but it missed to check the sort__mode. As 'perf diff' doesn't call perf_hpp__init(), the setup_overhead() also should not be called. Before: # Baseline Delta Children Overhead Shared Object Symbol # ........ ....... ........ ........ ................... ....................... # 28.48% -28.47% 28.48% 28.48% [kernel.vmlinux ] [k] intel_idle 11.51% -11.47% 11.51% 11.51% libxul.so [.] 0x0000000001a360f7 3.49% -3.49% 3.49% 3.49% [kernel.vmlinux] [k] generic_exec_single 2.91% -2.89% 2.91% 2.91% libdbus-1.so.3.8.11 [.] 0x000000000000cdc2 2.86% -2.85% 2.86% 2.86% libxcb.so.1.1.0 [.] 0x000000000000c890 2.44% -2.39% 2.44% 2.44% [kernel.vmlinux] [k] perf_event_aux_ctx After: # Baseline Delta Shared Object Symbol # ........ ....... ................... ....................... # 28.48% -28.47% [kernel.vmlinux] [k] intel_idle 11.51% -11.47% libxul.so [.] 0x0000000001a360f7 3.49% -3.49% [kernel.vmlinux] [k] generic_exec_single 2.91% -2.89% libdbus-1.so.3.8.11 [.] 0x000000000000cdc2 2.86% -2.85% libxcb.so.1.1.0 [.] 0x000000000000c890 2.44% -2.39% [kernel.vmlinux] [k] perf_event_aux_ctx Signed-off-by:
-
- 10 May, 2016 7 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pciLinus Torvalds authored
Pull PCI fixes from Bjorn Helgaas: "Since v4.5, we've WARNed during resume if a PCI device, including a Thunderbolt device, was added while we were suspended. A change we merged for v4.6-rc1 turned that warning into a system hang. These enumeration patches from Lukas Wunner fix this issue: - Fix BUG on device attach failure - Do not treat EPROBE_DEFER as device attach failure" * tag 'pci-v4.6-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: PCI: Do not treat EPROBE_DEFER as device attach failure PCI: Fix BUG on device attach failure
-
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tipLinus Torvalds authored
Pull x86 fixes from Ingo Molnar: "Two topology corner case fixes, and a MAINTAINERS file update for mmiotrace maintenance" * 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: x86/topology: Set x86_max_cores to 1 for CONFIG_SMP=n MAINTAINERS: Add mmiotrace entry x86/topology: Handle CPUID bogosity gracefully
-
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tipLinus Torvalds authored
Pull scheduler fixes from Ingo Molnar: "A UP kernel cpufreq fix and a rt/dl scheduler corner case fix" * 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: sched/rt, sched/dl: Don't push if task's scheduling class was changed sched/fair: Fix !CONFIG_SMP kernel cpufreq governor breakage
-
Masami Hiramatsu authored
Remove unused xrealloc() and ALLOC_GROW() from libperf. Signed-off-by:
-
Masami Hiramatsu authored
Replace ALLOC_GROW with normal realloc code in add_cmd_list() so that it can handle errors directly. Signed-off-by:
-
Masami Hiramatsu authored
Make pmu_formats_string() to check return value of strbuf APIs so that it can detect errors in it. Signed-off-by:
-
Masami Hiramatsu authored
Make topology checkers to check the return value of strbuf APIs so that it can detect errors in it. Signed-off-by:
-
