- May 01, 2014
-
-
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: * Wire up perf_regs and unwind support for ARM64 (Jean Pihet) * Move u64_swap union to its single user's header, evsel.h (Borislav Petkov) * Fix for s390 to properly parse tracepoints plus test code (Alexander Yarygin) * Handle EINTR error for readn/writen (Namhyung Kim) Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Ingo Molnar authored
Conflicts: tools/perf/arch/x86/tests/dwarf-unwind.c Signed-off-by: -
Ingo Molnar authored
Merge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf into perf/urgent Pull perf/urgent fixes from Jiri Olsa: * Fix perf to use non-executable stack, again (Mathias Krause) * Remove extra '/' character in events file path (Xia Kaixu) * Search for modules in %s/lib/modules/%s (Richard Yao) * Build related fixies plus static build test (Jiri Olsa) * Fix stack map lookup in dwarf unwind test (Jiri Olsa) Signed-off-by:
-
- Apr 30, 2014
-
-
Jiri Olsa authored
Previous commit 'perf x86: Fix perf to use non-executable stack, again' moved stack map into MAP__VARIABLE map type again. Fixing the dwarf unwind test stack map lookup appropriately. 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: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnaldo Carvalho de Melo <acme@kernel.org> Cc: Jean Pihet <jean.pihet@linaro.org> Link: http://lkml.kernel.org/n/tip-ttzyhbe4zls24z7ednkmhvxl@git.kernel.org Signed-off-by:
-
Mathias Krause authored
arch/x86/tests/regs_load.S is missing the linker note about the stack requirements, therefore making the linker fall back to an executable stack. As this object gets linked against the final perf binary, it'll needlessly end up with an executable stack. Fix this by adding the appropriate linker note. Also add a global linker flag to prevent future regressions, as suggested by Jiri. This way perf won't get an executable stack even if we fail to add the .GNU-stack linker note to future assembler files. Though, doing so might create regressions the other way around, when (statically) linking against libraries needing an executable stack. But, apparently, regressing in that direction is wanted as it is an indicator of poor code quality -- or just missing linker notes. Fixes: 3c8b06f9 ("perf tests x86: Introduce perf_regs_load function") Signed-off-by:
Mathias Krause <minipli@googlemail.com> Acked-by:
-
Xia Kaixu authored
The array debugfs_known_mountpoints[] will cause extra '/' character output. Remove it. pre: $ perf probe -l /sys/kernel/debug//tracing/uprobe_events file does not exist - please rebuild kernel with CONFIG_UPROBE_EVENTS. post: $ perf probe -l /sys/kernel/debug/tracing/uprobe_events file does not exist - please rebuild kernel with CONFIG_UPROBE_EVENTS. Signed-off-by:
Xia Kaixu <xiakaixu@huawei.com> Acked-by:
Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/r/535B6660.2060001@huawei.com Signed-off-by:
-
Richard Yao authored
Modules installed outside of the kernel's build system should go into "%s/lib/modules/%s/extra", but at present, perf will only look at them when they are in "%s/lib/modules/%s/kernel". Lets encourage good citizenship by relaxing this requirement to "%s/lib/modules/%s". This way open source modules that are out-of-tree have no incentive to start populating a directory reserved for in-kernel modules and I can stop hex-editing my system's perf binary when profiling OSS out-of-tree modules. Feedback from Namhyung Kim correctly revealed that the hex-edits that I had been doing meant that perf was also traversing the build and source symlinks in %s/lib/modules/%s. That is undesireable, so we explicitly exclude them from traversal with a minor tweak to the traversal routine. Signed-off-by:
Richard Yao <ryao@gentoo.org> Acked-by:
-
Jiri Olsa authored
Adding test for building static perf build into the automated suite. Also available via following commands: $ make -f tests/make make_static - make_static: cd . && make -f Makefile DESTDIR=/tmp/tmp.7u5MlB4njo LDFLAGS=-static $ make -f tests/make make_static_O - make_static_O: cd . && make -f Makefile O=/tmp/tmp.Ay6r3wEmtX DESTDIR=/tmp/tmp.vK0KQwO0Vi LDFLAGS=-static Acked-by:
David Ahern <dsahern@gmail.com> Cc: Arnaldo Carvalho de Melo <acme@kernel.org> 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: 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/1398760413-7574-1-git-send-email-jolsa@kernel.org Signed-off-by:
-
Jiri Olsa authored
There's false assumption in the library detection code assuming -liberty and -lz are always present once bfd is detected. The fails on Ubuntu (14.04) as reported by Ingo. Forcing the bdf dependency libraries detection any time bfd library is detected. Reported-by:
-
Jiri Olsa authored
We no longer use ALL_LDFLAGS, Replacing with LDFLAGS. Acked-by:
-
- Apr 29, 2014
-
-
Jean Pihet authored
This patch hooks in the perf_regs and libunwind code for ARM64. The tools/perf/arch/arm64 is created; it contains the arch specific code for DWARF unwinding. Signed-off-by:
Jean Pihet <jean.pihet@linaro.org> Acked-by:
Will Deacon <will.deacon@arm.com> Link: http://lkml.kernel.org/r/1398688353-3737-1-git-send-email-jean.pihet@linaro.org Signed-off-by:
-
Borislav Petkov authored
... to its single user's header, evsel.h. Signed-off-by:
Borislav Petkov <bp@suse.de> Link: http://lkml.kernel.org/r/tip-9os1chjyz12upubfsjc71d99@git.kernel.org Signed-off-by:
-
Alexander Yarygin authored
Add a s390 specific test of a hardcoded trace event with '-' in the name. Signed-off-by:
Alexander Yarygin <yarygin@linux.vnet.ibm.com> Link: http://lkml.kernel.org/r/1398440047-6641-4-git-send-email-yarygin@linux.vnet.ibm.com Signed-off-by:
-
Alexander Yarygin authored
In tests/parse-events.c test cases are declared in evlist_test[] arrays. Elements of arrays are initialized in following pattern: [i] = { .name = ..., .check = ..., }, When perf-test is running with '-v' option, 'i' variable will be printed for every existing test. However, we can't add any arch specific tests inside #ifdefs, because it will create collision between the element number inside #ifdef and the next one outside. This patch adds 'id' field in evlist_test, uses it as a test identifier and removes explicit numbering of array elements. This helps to number tests with gaps. Signed-off-by: -
Alexander Yarygin authored
Trace events potentially can have a '-' in their trace system name, e.g. kvm on s390 defines kvm-s390:* tracepoints. We could not parse them, because there was no rule for this: $ sudo ./perf top -e "kvm-s390:*" invalid or unsupported event: 'kvm-s390:*' This patch adds an extra rule to event_legacy_tracepoint which handles those cases. Without the patch, perf will not accept such tracepoints in the -e option. Signed-off-by:
Christian Borntraeger <borntraeger@de.ibm.com> Acked-by:
-
Namhyung Kim authored
Those readn/writen functions are to ensure read/write does I/O for a given size exactly. But ion() - its implementation - does not handle in case it returns prematurely due to a signal. As it's not an error itself so just retry the operation. 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 a test case for hists filtering (Namhyung Kim) * Share map_groups among threads of the same group (Arnaldo Carvalho de Melo, Jiri Olsa) Signed-off-by:
-
- Apr 28, 2014
-
-
Jiri Olsa authored
This test create 2 processes abstractions, with several threads and checks they properly share and maintain map groups info. Acked-by:
-
Jiri Olsa authored
Sharing map groups within all process threads. This way there's only one copy of mmap info and it's reachable from any thread within the process. Original-patch-by:
Arnaldo Carvalho de Melo <acme@kernel.org> Acked-by:
-
Arnaldo Carvalho de Melo authored
We will share it among threads in the same process. Adding map_groups__get/map_groups__put interface for that. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
Moving towards sharing map groups within a process threads. Because of this we need the map groups to be dynamically allocated. No other functional change is intended in here. Based on a patch by Jiri Olsa, but this time _just_ making the conversion from statically allocating thread->mg to turning it into a pointer and instead of initializing it at thread's constructor, introduce a constructor/destructor for the map_groups class and call at thread creation time. Later we will introduce the get/put methods when we move to sharing those map_groups, when the get/put refcounting semantics will be needed. Signed-off-by:
-
Jiri Olsa authored
Adding automated test for memory maps lookup within multiple machines threads. The test creates 4 threads and separated memory maps. It checks that we could use thread__find_addr_map function with thread object based on TID to find memory maps. Acked-by:
-
Namhyung Kim authored
Now we have changed how hists stats are accounted especially when filter(s) applied. So add a test case to verify it. Signed-off-by:
-
Namhyung Kim authored
The fake_setup_machine() is for setting up a environment for testing various hists operations. As it'll be used for other test cases it'd better factoring it out. Signed-off-by:
-
- Apr 25, 2014
-
-
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: * Factor hists statistics counts processing which in turn also fixes several bugs in TUI report command (Namhyung Kim) Signed-off-by:
-
Ingo Molnar authored
Signed-off-by:
-
- Apr 24, 2014
-
-
Namhyung Kim authored
When TUI hist browser expands/collapses callchains it accounted number of callchain nodes into total entries to show. However this code ignores filtering so that it can make the cursor go to out of screen. Thanks to Jiri Olsa for pointing out a bug (and a fix) in the code. Signed-off-by:
-
Namhyung Kim authored
The hist_browser__reset() is only called right after a filter is applied so it needs to udpate browser->nr_entries properly. We cannot use hists->nr_non_filtered_entreis directly since it's possible that such entries are also filtered out by minimum percentage limit. In addition when a filter is used for perf top, hist browser's nr_entries field was not updated after applying the filter. But it needs to be updated as new samples are coming. Signed-off-by:
-
Namhyung Kim authored
Rename ->nr_pcnt_entries and hist_browser__update_pcnt_entries() to ->nr_non_filtered_entries and hist_browser__update_nr_entries() since it's now used for filtering as well. Signed-off-by:
-
Namhyung Kim authored
The nr_entries variable is increased inside the loop in the function but it always count the first entry regardless of it's filtered or not; caused an off-by-one error. It'd become a problem especially there's no entry at all - it'd get a segfault during referencing a NULL pointer. Signed-off-by:
-
Namhyung Kim authored
When a filter is used for perf top, its hists->nr_non_filtered_entries was not updated after it removed an entry in hists__decay_entries(). Also hists->stats.total_non_filtered_period was missed too. Signed-off-by:
-
Namhyung Kim authored
Currently, accounting each sample is done in multiple places - once when adding them to the input tree, other when adding them to the output tree. It's not only confusing but also can cause a subtle problem since concurrent processing like in perf top might see the updated stats before adding entries into the output tree - like seeing more (blank) lines at the end and/or slight inaccurate percentage. To fix this, only account the entries when it's moved into the output tree so that they cannot be seen prematurely. There're some exceptional cases here and there - they should be addressed separately with comments. Signed-off-by:
-
Namhyung Kim authored
When a filter is applied a hist entry checks whether its callchain was folded and account it to the output stat. But this is rather hacky and only TUI-specific. Simply fold the callchains for the entry looks like a simpler and more generic solution IMHO. Signed-off-by:
-
Namhyung Kim authored
Add hists__{reset,inc}_[filter_]stats() functions to cleanup accesses to hist stats (for output). Note that number of samples in the stat is not handled here since it belongs to the input stage. Signed-off-by: -
Namhyung Kim authored
It's not the part of logic of hists__inc_stats() so it'd be better to move it out of the function. Signed-off-by:
-
Namhyung Kim authored
The existing hists__inc_nr_entries() is a misnomer as it's not only increasing ->nr_entries but also other stats. So rename it to more general hists__inc_stats(). Signed-off-by:
-
Namhyung Kim authored
The hists->nr_entries is counted in multiple places so that they can confuse readers of the code. This is a preparation of later change and do not intend any functional difference. Note that report__collapse_hists() now changed to return nothing since its return value (nr_samples) is only for checking if there's any data in the input file and this can be acheived by checking ->nr_entries. Signed-off-by:
-
Stephane Eranian authored
This patch fixes a bug introduced by: 24223657 ("perf/x86/intel: Use rdmsrl_safe() when initializing RAPL PMU") The rdmsrl_safe() function returns 0 on success. The current code was failing to detect the RAPL PMU on real hardware (missing /sys/devices/power) because the return value of rdmsrl_safe() was misinterpreted. Signed-off-by:
Stephane Eranian <eranian@google.com> Acked-by:
Venkatesh Srinivas <venkateshs@google.com> Cc: peterz@infradead.org Cc: zheng.z.yan@intel.com Link: http://lkml.kernel.org/r/20140423170418.GA12767@quad Signed-off-by:
-
- Apr 23, 2014
-
-
Ingo Molnar authored
Merge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf into perf/urgent Pull perf/urgent fixes from Jiri Olsa: * Fix memory leak and backward compatibility macros for pevent filter enums in traceevent library (Steven Rostedt) * Disable libdw unwind for all but x86 arch (Jiri Olsa) * Fix memory leak in sample_ustack (Masanari Iida) Signed-off-by:
Jiri Olsa <jolsa@redhat.com> Signed-off-by:
-
Steven Rostedt authored
Commit 12e55569 "tools lib traceevent: Use helper trace-seq in print functions like kernel does" added a extra trace_seq helper to process string arguments like the kernel does it. But the difference between the kernel and the userspace library is that the kernel's trace_seq structure has a static allocated buffer. The userspace one has a dynamically allocated one. It requires a trace_seq_destroy(), otherwise it produces a nasty memory leak. Cc: stable@vger.kernel.org # 3.14+ Signed-off-by:
Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/r/20140422192330.6bb09bf8@gandalf.local.home Signed-off-by:
-
