- Jun 05, 2014
-
-
Anshuman Khandual authored
This patch adds conditional branch filtering support, enabling it for PERF_SAMPLE_BRANCH_COND in perf branch stack sampling framework by utilizing an available software filter X86_BR_JCC. Signed-off-by:
Anshuman Khandual <khandual@linux.vnet.ibm.com> Reviewed-by:
Stephane Eranian <eranian@google.com> Reviewed-by:
Andi Kleen <ak@linux.intel.com> Signed-off-by:
Peter Zijlstra <peterz@infradead.org> Cc: mpe@ellerman.id.au Cc: benh@kernel.crashing.org Cc: Arnaldo Carvalho de Melo <acme@kernel.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Link: http://lkml.kernel.org/r/1400743210-32289-3-git-send-email-khandual@linux.vnet.ibm.com Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Anshuman Khandual authored
Adding perf record support for new branch stack filter criteria PERF_SAMPLE_BRANCH_COND. Signed-off-by:
-
Anshuman Khandual authored
This patch introduces new branch filter PERF_SAMPLE_BRANCH_COND which will extend the existing perf ABI. This will filter branches which are conditional. Various architectures can provide this functionality either with HW filtering support (if present) or with SW filtering of captured branch instructions. Signed-off-by:
-
Oleg Nesterov authored
tmpfs is widely used but as Denys reports shmem_aops doesn't have ->readpage() and thus you can't probe a binary on this filesystem. As Hugh suggested we can use shmem_read_mapping_page() in this case, just we need to check shmem_mapping() if ->readpage == NULL. Reported-by:
Denys Vlasenko <dvlasenk@redhat.com> Suggested-by:
Hugh Dickins <hughd@google.com> Signed-off-by:
Oleg Nesterov <oleg@redhat.com> Acked-by:
Srikar Dronamraju <srikar@linux.vnet.ibm.com> Signed-off-by:
-
Oleg Nesterov authored
copy_insn() fails with -EIO if ->readpage == NULL, but this error is not propagated unless uprobe_register() path finds ->mm which already mmaps this file. In this case (say) "perf record" does not actually install the probe, but the user can't know about this. Move this check into uprobe_register() so that this problem can be detected earlier and reported to user. Note: this is still not perfect, - copy_insn() and arch_uprobe_analyze_insn() should be called by uprobe_register() but this is not simple, we need vm_file for read_mapping_page() (although perhaps we can pass NULL), and we need ->mm for is_64bit_mm() (although this logic is broken anyway). - uprobe_register() should be called by create_trace_uprobe(), not by probe_event_enable(), so that an error can be detected at "perf probe -x" time. This also needs more changes in the core uprobe code, uprobe register/unregister interface was poorly designed from the very beginning. Reported-by:
-
Vince Weaver authored
Make the x86 perf code use the new common PMU interrupt disabled code. Typically most x86 machines have working PMU interrupts, although some older p6-class machines had this problem. Signed-off-by:
Vince Weaver <vincent.weaver@maine.edu> Signed-off-by:
-
Vince Weaver authored
Make the ARM perf code use the new common PMU interrupt disabled code. This allows perf to work on ARM machines without a working PMU interrupt (for example, raspberry pi). Acked-by:
Will Deacon <will.deacon@arm.com> Signed-off-by:
-
Vince Weaver authored
Add common code to generate -ENOTSUPP at event creation time if an architecture attempts to create a sampled event and PERF_PMU_NO_INTERRUPT is set. This adds a new pmu->capabilities flag. Initially we only support PERF_PMU_NO_INTERRUPT (to indicate a PMU has no support for generating hardware interrupts) but there are other capabilities that can be added later. Signed-off-by:
-
Peter Zijlstra authored
While that mutex should guard the elements, it doesn't guard against the use-after-free that's from list_for_each_entry_rcu(). __perf_event_exit_task() can actually free the event. And because list addition/deletion is guarded by both ctx->mutex and ctx->lock, holding ctx->mutex is sufficient for reading the list, so we don't actually need the rcu list iteration. Fixes: 3a497f48 ("perf: Simplify perf_event_exit_task_context()") Reported-by:
Sasha Levin <sasha.levin@oracle.com> Tested-by:
-
Ingo Molnar authored
Conflicts: arch/x86/kernel/traps.c The kprobes enhancements are fully cooked, ship them upstream. Signed-off-by: -
Ingo Molnar authored
These bits from Oleg are fully cooked, ship them to Linus. Signed-off-by: -
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf into perf/core Pull perf/core improvements and fixes from Jiri Olsa: * Warn the user when trace command is not available (Arnaldo Carvalho de Melo) * Add warning when disabling perl scripting support due to missing devel files (Arnaldo Carvalho de Melo) * Consider header files outside perf directory in tags target (Sebastian Andrzej Siewior) * Allow overriding sysfs and proc finding with env var (Cody P Schafer) * Fix "==" into "=" in ui_browser__warning assignment (zhangdianfang) * Factor elide bool handling in sort code (Jiri Olsa) * Fix poll return value propagation (Jiri Olsa) * Fix 'make help' message error (Jianyu Zhan) Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Signed-off-by:
-
- Jun 04, 2014
-
-
Jianyu Zhan authored
Currently 'make help' message has such hint: use "make prefix=<path> <install target>" to install to a particular path like make prefix=/usr/local install install-doc But this is misleading, when I specify "prefix=/usr/local", it has got no respect at all. This is because that, "DESTDIR" is considered first. In this case, "DESTDIR" has an empty value, so "prefix" is honored. However, "prefix" is unconditionally assigned to $HOME, regardless of what it is set to from command line. So our "prefix" setting got no respect and the actual destination falls back to $HOME. This patch fixes this issue and corrects the help message. Signed-off-by:Jianyu Zhan <nasa4836@gmail.com> Acked-by:
Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/r/1401727474-19370-1-git-send-email-nasa4836@gmail.com Signed-off-by:
-
Jiri Olsa authored
If the perf record command is interrupted in record__mmap_read_all function, the 'done' is set and err has the latest poll return value, which is most likely positive number (= number of pollfds ready to read). This 'positive err' is then propagated to the exit code, resulting in not finishing the perf.data header properly, causing following error in report: # perf record -F 50000 -a --- make the system real busy, so there's more chance to interrupt perf in event writing code --- ^C[ perf record: Woken up 16 times to write data ] [ perf record: Captured and wrote 30.292 MB perf.data (~1323468 samples) ] # perf report --stdio > /dev/null WARNING: The perf.data file's data size field is 0 which is unexpected. Was the 'perf record' command properly terminated? Fixing this by checking for positive poll return value and setting err to 0. Acked-by:
Arnaldo Carvalho de Melo <acme@redhat.com> Acked-by:
-
Jiri Olsa authored
After output/sort fields refactoring, it's expensive to check the elide bool in its current location inside the 'struct sort_entry'. The perf_hpp__should_skip function gets highly noticable in workloads with high number of output/sort fields, like for: $ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio Performance report: 9.70% perf [.] perf_hpp__should_skip Moving the elide bool into the 'struct perf_hpp_fmt', which makes the perf_hpp__should_skip just single struct read. Got speedup of around 22% for my test perf.data workload. The change should not harm any other workload types. Performance counter stats for (10 runs): before: 358,319,732,626 cycles ( +- 0.55% ) 467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% ) 150.943975206 seconds time elapsed ( +- 0.62% ) now: 278,785,972,990 cycles ( +- 0.12% ) 370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% ) 116.416670507 seconds time elapsed ( +- 0.31% ) Acked-by: -
Jiri Olsa authored
There's no need to setup elide of sort_dso sort entry again with symbol_conf.dso_list list. The only difference were list names of memory mode data, which does not make much sense to me. Acked-by:
-
zhangdianfang authored
Convert "==" into "=" in ui_browser__warning assignment. Bug description: https://bugzilla.kernel.org/show_bug.cgi?id=76751 Reported-by:
David Binderman <dcb314@hotmail.com> Signed-off-by:
Dianfang Zhang <zhangdianfang@huawei.com> Acked-by:
-
Cody P Schafer authored
SYSFS_PATH and PROC_PATH environment variables now let the user override the detection of sysfs and proc locations for testing purposes. Signed-off-by:
Cody P Schafer <dev@codyps.com> Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> Link: http://lkml.kernel.org/r/1401236684-10579-2-git-send-email-dev@codyps.com Signed-off-by:
-
Sebastian Andrzej Siewior authored
This fixes lookups like "vi -t event_format" Signed-off-by:
Sebastian Andrzej Siewior <bigeasy@linutronix.de> Link: http://lkml.kernel.org/r/20140528081918.GA28567@linutronix.de Signed-off-by:
-
Arnaldo Carvalho de Melo authored
We were just showing "libperl: OFF", unlike other features where we present the user with a message helping have a feature built in. Fix it by adding the following message: config/Makefile:450: Missing perl devel files. Disabling perl scripting support, consider installing perl-ExtUtils-Embed Signed-off-by:
-
Arnaldo Carvalho de Melo authored
When the audit-libs devel package is not found at build time we disable the 'trace' command, as we are not able to map syscall numbers to strings, but then the message the user is presented is cryptic: [root@zoo linux]# trace ls perf: 'ls' is not a perf-command. See 'perf --help'. Fix it by presenting a more helpful message: [root@zoo linux]# trace l trace command not available: missing audit-libs devel package at build time. Signed-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf into perf/core Pull perf/core improvements and fixes from Jiri Olsa: * Add support to accumulate hist periods (Namhyung Kim) Signed-off-by:
-
- Jun 01, 2014
-
-
Namhyung Kim authored
Now it adds a new testcase to verify --children option working correctly. Signed-off-by:
-
Namhyung Kim authored
In various histogram test cases, fake symbols are used as raw numbers. Define macros for each pid, map, symbols so that it can increase readability somewhat. Signed-off-by:
-
Namhyung Kim authored
When reset_output_field() is called, also reset field/sort order to NULL so that it can have the default values. It's needed for testing. Signed-off-by:
-
Namhyung Kim authored
With current output field change, GTK browser cannot display callchain information correctly since it couldn't determine where the symbol column is. This is a problem - just for now I changed to use the last column since it'll work for most cases. Also it has a same problem of the percentage as stdio code. Signed-off-by:
-
Namhyung Kim authored
On stdio, there's a problem that it shows invalid values for callchains in cumulated hist entries. It's because it only cares about the self period. But with --children behavior, we always add callchain info to the cumulated entries so it should use the value in that case. Before: # Children Self Command Shared Object Symbol # ........ ........ ....... ................. ................ # 61.22% 0.32% swapper [kernel.kallsyms] [k] cpu_idle | --- cpu_idle | |--16530.76%-- start_secondary | |--2758.70%-- rest_init | start_kernel | x86_64_start_reservations | x86_64_start_kernel --6837850969203030.00%-- [...] After: # Children Self Command Shared Object Symbol # ........ ........ ....... ................. ................ # 61.22% 0.32% swapper [kernel.kallsyms] [k] cpu_idle | --- cpu_idle | |--85.70%-- start_secondary | --14.30%-- rest_init start_kernel x86_64_start_reservations x86_64_start_kernel Signed-off-by: -
Namhyung Kim authored
Now perf top and perf report will show children column by default if it has callchain information. Requested-by:
Rodrigo Campos <rodrigo@sdfg.com.ar> Tested-by:
Arun Sharma <asharma@fb.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Link: http://lkml.kernel.org/r/1401335910-16832-23-git-send-email-namhyung@kernel.org Signed-off-by:
-
Namhyung Kim authored
Add top.children config option for setting default value of callchain accumulation. It affects the output only if one of -g or --call-graph option is given as well. A user can write .perfconfig file like below to enable accumulation by default: $ cat ~/.perfconfig [top] children = true And it can be disabled through command line: $ perf top --no-children Signed-off-by:
-
Namhyung Kim authored
The --children option is for showing accumulated overhead (period) value as well as self overhead. It should be used with one of -g or --call-graph option. Signed-off-by:
-
Namhyung Kim authored
Reuse hist_entry_iter__add() function to share the similar code with perf report. Note that it needs to be called with hists.lock so tweak some internal functions not to deadlock or hold the lock too long. Signed-off-by:
-
Namhyung Kim authored
The new ->add_entry_cb() will be called after an entry was added to the histogram. It's used for code sharing between perf report and perf top. Note that ops->add_*_entry() should set iter->he properly in order to call the ->add_entry_cb. Also pass @arg to the callback function. It'll be used by perf top later. Signed-off-by:
-
Namhyung Kim authored
Depending on the configuration perf inserts/removes the Children column in the output automatically. But it might not be what user wants if [s]he give --fields option explicitly. Signed-off-by:
-
Namhyung Kim authored
Add report.children config option for setting default value of callchain accumulation. It affects the report output only if perf.data contains callchain info. A user can write .perfconfig file like below to enable accumulation by default: $ cat ~/.perfconfig [report] children = true And it can be disabled through command line: $ perf report --no-children Signed-off-by:
-
Namhyung Kim authored
The --children option is for showing accumulated overhead (period) value as well as self overhead. Signed-off-by:
-
Namhyung Kim authored
Sometimes it needs to disable some columns at runtime. Add help functions to support that. Signed-off-by:
-
Namhyung Kim authored
If -g cumulative option is given, it needs to show entries which don't have self overhead. So apply percent-limit to accumulated overhead percentage in this case. Signed-off-by:
-
Namhyung Kim authored
Print accumulated stat of a hist entry if requested. Signed-off-by:
-
Namhyung Kim authored
Print accumulated stat of a hist entry if requested. Signed-off-by:
-
Namhyung Kim authored
Print accumulated stat of a hist entry if requested. To do that, add new HPP_PERCENT_ACC_FNS macro and generate a perf_hpp_fmt using it. The __hpp__sort_acc() function sorts entries by accumulated period value. When accumulated periods of two entries are same (i.e. single path callchain) put the caller above since accumulation tends to put callers on higher position for obvious reason. Also add "overhead_children" output field to be selected by user. Signed-off-by:
-
