- 23 Jun, 2016 2 commits
-
-
He Kuang authored
There's a problem in machine__findnew_vdso(), vdso buildid generated by a 32-bit machine stores it with the name 'vdso', but when processing buildid on a 64-bit machine with the same 'perf.data', perf will search for vdso named as 'vdso32' and get failed. This patch tries to find the existing dsos in machine->dsos by thread dso_type. 64-bit thread tries to find vdso with name 'vdso', because all 64-bit vdso is named as that. 32-bit thread first tries to find vdso with name 'vdso32' if this thread was run on 64-bit machine, if failed, then it tries 'vdso' which indicates that the thread was run on 32-bit machine when recording. Committer note: Additional explanation by Adrian Hunter: We match maps to builds ids using the file name - consider machine__findnew_[v]dso() called in map__new(). So in the context of a perf data file, we consider the file name to be unique. A vdso map does not have a file name - all we know is that it is vdso. We look at the thread to tell if it is 32-bit, 64-bit or x32. Then we need to get the build id which has been recorded using short name "[vdso]" or "[vdso32]" or "[vdsox32]". The problem is that on a 32-bit machine, we use the name "[vdso]". If you take a 32-bit perf data file to a 64-bit machine, it gets hard to figure out if "[vdso]" is 32-bit or 64-bit. This patch solves that problem. ---- This also merges a followup patch fixing a problem introduced by the original submission of this patch, that would crash 'perf record' when recording samples for a 32-bit app on a 64-bit system. Signed-off-by:
He Kuang <hekuang@huawei.com> Acked-by:
Adrian Hunter <adrian.hunter@intel.com> Tested-by:
Arnaldo Carvalho de Melo <acme@redhat.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 <mhiramat@kernel.org> 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/1463475894-163531-1-git-send-email-hekuang@huawei.com Link: http://lkml.kernel.org/r/1466578626-92406-6-git-send-email-hekuang@huawei.comSigned-off-by:
-
Taeung Song authored
Lately util/config.h has been added but util/cache.h has declarations of functions and a global variable for config features. To manage codes about configuration at one spot, move them to util/config.h and let source files that need config features include config.h And if the source files that included previous cache.h need only config.h, remove including cache.h. Signed-off-by:
Taeung Song <treeze.taeung@gmail.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/r/1466672119-4852-2-git-send-email-treeze.taeung@gmail.comSigned-off-by:
-
- 22 Jun, 2016 14 commits
-
-
He Kuang authored
Currently, python uses host gcc instead of cross-compile gcc in the last step of compiling build_ext(remove '--quiet' to show verbose): cross-gcc ... cross-gcc ... creating ~/out/python_ext_build/lib gcc -pthread -shared -Wl,-z ... This is wrong but may not cause any errors unless the features detected by cross-compiler do not match those for host compiler, and causes the following errors: /usr/lib64/gcc/bin/ld: cannot find -lunwind-x86 collect2: error: ld returned 1 exit status error: command 'gcc' failed with exit status 1 cp: cannot stat ‘~/out/python_ext_build/lib/perf.so’: No such file or directory Makefile.perf:257: recipe for target '~/out/python/perf.so' failed make[1]: *** [~/out/python/perf.so] Error 1 Makefile:68: recipe for target 'all' failed make: *** [all] Error 2 This issue is also reported and anwsered on stackoverflow. Link: http://stackoverflow.com/questions/5986256/python-distutils-gcc-pathSigned-off-by:
-
Arnaldo Carvalho de Melo authored
And do nothing, just like free(), to avoid having to test it in callers, usually in error paths. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-q42gj3b3znhho9z1mrbo4jce@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Because at the destructor we will call close() and that will do the disable. And we destructors can accept NULL, just like free(), so no need to check it. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/n/tip-i98mcyfkkjh5qp62dle27ac1@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
And do nothing, just like free(), to avoid having to test it in callers, usually in error paths. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-dyuupcj0hnoyt96vma8b3anv@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
And do nothing, just like free(), to avoid having to test it in callers, usually in error paths. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-mexbavy0ft387j5w89t365eu@git.kernel.orgSigned-off-by:
-
Jiri Olsa authored
The pid sort entry currently aligns pids with 5 digits, which is not enough for current 4 million pids limit. This leads to unaligned ':' header-data output when we display 7 digits pid: # Children Self Symbol Pid:Command # ........ ........ ...................... ..................... # 0.12% 0.12% [.] 0x0000000000147e0f 2052894:krava ... Adding 2 more digit to properly align the pid limit: # Children Self Symbol Pid:Command # ........ ........ ...................... ....................... # 0.12% 0.12% [.] 0x0000000000147e0f 2052894:krava Signed-off-by:Jiri Olsa <jolsa@kernel.org> Cc: David Ahern <dsahern@gmail.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1466459899-1166-9-git-send-email-jolsa@kernel.orgSigned-off-by:
-
Jiri Olsa authored
Factoring out the hist_browser initialization code, so it could be used from other parts in following patches. Signed-off-by:
-
Jiri Olsa authored
So we could use hist_browser__new for generic hist browser in following patches. Signed-off-by:
-
Jiri Olsa authored
Moving horizontal scroll init to initialization function as already intended. Signed-off-by:
-
Jiri Olsa authored
We can now setup title callback for hist_browser, which will be useful in following changes to create customized hist_browsers. This also separates struct perf_evsel dependency out of hist_browser basic code. Signed-off-by:
-
Jiri Olsa authored
This way we can use it outside of ui/browsers/hists.c and extend it in following patches. Signed-off-by:
-
Jiri Olsa authored
This way we can use it outside of ui/browsers/hists.c and extend it in following patches. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
It is ignored and this is actually a python script, not a perl one. Reported-by:
Brendan Gregg <brendan.d.gregg@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Milian Wolff <milian.wolff@kdab.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paolo Bonzini <pbonzini@redhat.com> Link: http://lkml.kernel.org/n/tip-0w4bpbqd79v3sl34jvpr11v0@git.kernel.orgSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-20160621' 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: New features: - Add --dry-run option to 'perf record' to check if command line options can be parsed, but not doing any recording (Wang Nan) - Allow dumping the object files generated by llvm when processing eBPF scriptlet events (Wang Nan) - Add stackcollapse.py script to help generating flame graphs (Paolo Bonzini) Documentation changes: - Fix 'perf script' documentation of '-f' when it should be '-F' (Adrian Hunter) Infrastructure changes: - Fix write_backwards fallback when using a new tool on older kernels without support for this feature (Arnaldo Carvalho de Melo) - Remove some leftovers from the initial codebase copying from git (Arnaldo Carvalho de Melo) - List libelf-devel as an alternative, as this is how the libelf development package is called on OpenSuSE (Jean Delvare) - Rename __hists__add_entry to hists__add_entry (Jiri Olsa) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 21 Jun, 2016 9 commits
-
-
Paolo Bonzini authored
Add stackcollapse.py script as an example of parsing call chains, and also of using optparse to access command line options. The flame graph tools include a set of scripts that parse output from various tools (including "perf script"), remove the offsets in the function and collapse each stack to a single line. The website also says "perf report could have a report style [...] that output folded stacks directly, obviating the need for stackcollapse-perf.pl", so here it is. This script is a Python rewrite of stackcollapse-perf.pl, using the perf scripting interface to access the perf data directly from Python. Signed-off-by:
Paolo Bonzini <pbonzini@redhat.com> Acked-by:
-
Arnaldo Carvalho de Melo authored
Commit b90dc17a "perf evsel: Add overwrite attribute and check write_backward" misunderstood the 'order' should be obeyed in __perf_evsel__open. But the way this was done for attr.write_backwards was buggy, as we need to check features in the inverse order of their introduction to the kernel, so that a newer tool checks first the newest perf_event_attr fields, detecting that the older kernel doesn't have support for them. Also, we can avoid calling sys_perf_event_open() if we have already detected the missing of write_backward. Cc: He Kuang <hekuang@huawei.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Zefan Li <lizefan@huawei.com> Cc: pi3orama@163.com Fixes: b90dc17a ("perf evsel: Add overwrite attribute and check write_backward") Link: http://lkml.kernel.org/r/1466419645-75551-2-git-send-email-wangnan0@huawei.com Link: http://lkml.kernel.org/r/20160616214724.GI13337@kernel.orgSigned-off-by:
Wang Nan <wangnan0@huawei.com> Signed-off-by:
-
Wang Nan authored
With '--dry-run', 'perf record' doesn't do reall recording. Combine with llvm.dump-obj option, --dry-run can be used to help compile BPF objects for embedded platform. Signed-off-by:
-
Wang Nan authored
Add a 'llvm.dump-obj' config option to enable perf dump BPF object files compiled by LLVM. This option is useful when using BPF objects in embedded platforms. LLVM compiler won't be deployed in these platforms, and currently we don't support dynamic compiling library. Before this patch users have to explicitly issue llvm commands to compile BPF scripts, and can't use helpers (like include path detection and default macros) in perf. With this option, user is allowed to use perf to compile their BPF objects then copy them into their embedded platforms. Committer notice: Testing it: # cat ~/.perfconfig [llvm] dump-obj = true # # ls -la filter.o ls: cannot access filter.o: No such file or directory # cat filter.c #include <uapi/linux/bpf.h> #define SEC(NAME) __attribute__((section(NAME), used)) SEC("func=hrtimer_nanosleep rqtp->tv_nsec") int func(void *ctx, int err, long nsec) { return nsec > 1000; } char _license[] SEC("license") = "GPL"; int _version SEC("version") = LINUX_VERSION_CODE; # trace -e nanosleep --event filter.c usleep 6 LLVM: dumping filter.o 0.007 ( 0.007 ms): usleep/13976 nanosleep(rqtp: 0x7ffc5847f640 ) ... 0.007 ( ): perf_bpf_probe:func:(ffffffff811137d0) tv_nsec=6000) 0.070 ( 0.070 ms): usleep/13976 ... [continued]: nanosleep()) = 0 # ls -la filter.o -rw-r--r--. 1 root root 776 Jun 20 17:01 filter.o # readelf -SW filter.o There are 7 section headers, starting at offset 0x148: Section Headers: [Nr] Name Type Address Off Size ES Flg Lk Inf Al [ 0] NULL 0000000000000000 000000 000000 00 0 0 0 [ 1] .strtab STRTAB 0000000000000000 0000e8 00005a 00 0 0 1 [ 2] .text PROGBITS 0000000000000000 000040 000000 00 AX 0 0 4 [ 3] func=hrtimer_nanosleep rqtp->tv_nsec PROGBITS 0000000000000000 000040 000028 00 AX 0 0 8 [ 4] license PROGBITS 0000000000000000 000068 000004 00 WA 0 0 1 [ 5] version PROGBITS 0000000000000000 00006c 000004 00 WA 0 0 4 [ 6] .symtab SYMTAB 0000000000000000 000070 000078 18 1 2 8 Key to Flags: W (write), A (alloc), X (execute), M (merge), S (strings) I (info), L (link order), G (group), T (TLS), E (exclude), x (unknown) O (extra OS processing required) o (OS specific), p (processor specific) # Signed-off-by: -
Arnaldo Carvalho de Melo authored
Completely unused in perf, carried along all this time from the initial copy of git infrastructure, ditch'em. 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> Link: http://lkml.kernel.org/n/tip-wtiln26gyqndprmkl0kdswvi@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Probably are there since the beginning, taken from git but never used. 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> Link: http://lkml.kernel.org/n/tip-lr65jeefffjeaywoapps9a6i@git.kernel.orgSigned-off-by:
-
Jiri Olsa authored
There's no reason we should suffer the '__' prefix for the base global function. Signed-off-by:
-
Adrian Hunter authored
The documentation for perf script mixes up '-f' and '-F'. Fix it. Signed-off-by:
-
Jean Delvare authored
On openSUSE, the libelf development files are in package libelf-devel. Signed-off-by:
Jean Delvare <jdelvare@suse.de> Acked-by:
Josh Poimboeuf <jpoimboe@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Michal Marek <mmarek@suse.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: linux-kbuild@vger.kernel.org Link: http://lkml.kernel.org/n/tip-s8nyk3pyy2927sd7qp7u42oi@git.kernel.orgSigned-off-by:
-
- 16 Jun, 2016 1 commit
-
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-20160615' 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: - Add --ldlat option to 'perf mem' to specify load latency for loads event (e.g. cpu/mem-loads/ ) (Jiri Olsa) Build fixes: - Fix libunwind related compile error for static cross build (He Kuang) Infrastructure changes: - UI refactorings to support headers with multiple lines, non-evsel hists browsers, toggle showing callchains, etc (Jiri Olsa) - More prep work for caching probe definitions, paving the way for supporting SDT (Statically Defined Traces) userspace probes (Masami Hiramatsu) - Handle NULL at perf_config_set__delete() (Taeung Song) Signed-off-by:
-
- 15 Jun, 2016 13 commits
-
-
Masami Hiramatsu authored
Add --cache option to cache the probe definitions. This just saves the result of the dwarf analysis to probe cache. Signed-off-by:
Masami Hiramatsu <mhiramat@kernel.org> 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: Masami Hiramatsu <mhiramat@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20160615032840.31330.44412.stgit@devboxSigned-off-by:
-
Masami Hiramatsu authored
Introduce perf_cache object and interfaces to create, add entries, commit, and delete the object. perf_cache represents a file for the cached "perf probe" definitions on one binary file or vmlinux which has its own build id. The probe cache file is located under the build-id cache directory of the target binary, as below; <perf-debug-dir>/.build-id/<BU>/<ILDID>/probe Signed-off-by:
-
Jiri Olsa authored
Replacing perf_evsel arg perf_hpp_fmt's width callback with hists object. This will be helpful in future for non evsel related hist browsers. Signed-off-by:
-
Jiri Olsa authored
Replacing perf_evsel arg perf_hpp_fmt's header callback with hists object. None of the actual callbacks actually use evsel object, also this will be helpful in future for non evsel related hist browsers. Signed-off-by:
-
Jiri Olsa authored
It will be convenient in following patches to display hists entries without callchains even if they are defined. Signed-off-by:
-
Jiri Olsa authored
There's no need, we have the hists pointer in struct hist_entry. Signed-off-by:
-
Jiri Olsa authored
Introducing hists__fprintf_standard_headers function to separate standard headers display code. Signed-off-by:
-
Jiri Olsa authored
Introducing hists__fprintf_hierarchy_headers function to separate hierarchy headers display code. Signed-off-by:
-
Jiri Olsa authored
Introducing hists__fprintf_headers function to separate the code that displays headers. Signed-off-by:
-
Jiri Olsa authored
It will be useful for future changes that enhance headers with multiple lines and span columns, which don't affect hierarchy headers. Signed-off-by:
-
Jiri Olsa authored
Putting correct HISTC_MEM_DADDR_DSO index to Data Object sort entry. Signed-off-by:
-
Jiri Olsa authored
Adding --ldlat option to specify desired latency for loads event. Specify 50 as loads event latency: $ perf mem record -e ldlat-loads -v --ldlat 50 true calling: record -W -d -e cpu/mem-loads,ldlat=50/P true Signed-off-by:
-
He Kuang authored
Build failure for static cross-compiling on aarch64, with libunwind-x86 provided: $ file ./libunwind_for_x86_on_aarch64/lib/libunwind-x86.so.8.0.1 libunwind-x86.so.8.0.1: ELF 64-bit LSB shared object, ARM aarch64, version 1 (SYSV), dynamically linked, not stripped $ make LDFLAGS=-static LIBUNWIND_DIR=./libunwind_for_x86_on_aarch64 ARCH=aarch64 CROSS_COMPILE=aarch64-buildroot-linux-gnu- ~/libperf.a(libperf-in.o): In function `find_proc_info': :(.text+0xae4ac): undefined reference to `_Ux86_dwarf_search_unwind_table' ~/libperf.a(libperf-in.o): In function `_unwind__prepare_access': :(.text+0xaedd0): undefined reference to `_Ux86_create_addr_space' :(.text+0xaee24): undefined reference to `_Ux86_set_caching_policy' ~/libperf.a(libperf-in.o): In function `_unwind__flush_access': :(.text+0xaee98): undefined reference to `_Ux86_flush_cache' ~/libperf.a(libperf-in.o): In function `_unwind__finish_access': :(.text+0xaef08): undefined reference to `_Ux86_destroy_addr_space' ~/libperf.a(libperf-in.o): In function `get_entries': :(.text+0xaf148): undefined reference to `_Ux86_init_remote' :(.text+0xaf184): undefined reference to `_Ux86_get_reg' :(.text+0xaf1a4): undefined reference to `_Ux86_step' collect2: error: ld returned 1 exit status Makefile.perf:350: recipe for target '~/perf' failed make[1]: *** [~/perf] Error 1 Makefile:68: recipe for target 'all' failed make: *** [all] Error 2 This is because the remote libunwind library detected is not appended to EXTLIBS variable, which will be included between 'start-group' and 'end-group' when linking. The existing variable LIBUNWIND_LIBS is assigned to libs for local unwind, this patch introduces a new variable EXTLIBS_LIBUNWIND for storing remote libunwind libraries instead. Signed-off-by:
-
- 14 Jun, 2016 1 commit
-
-
Masami Hiramatsu authored
Uncomment and export synthesize_perf_probe_point() which had once introduced but has been disabled for a long time. This renews the code and re-enable it. Signed-off-by:
-
