aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf (follow)
AgeCommit message (Collapse)AuthorFilesLines
2017-08-22perf tools: Expression parser enhancements for metricsAndi Kleen2-6/+60
Enhance the expression parser for more complex metric formulas. - Support python style IF ELSE operators - Add an #SMT_On magic variable for formulas that depend on the SMT status. Example: 4 *( CPU_CLK_UNHALTED.THREAD_ANY / 2 ) if #SMT_on else cycles - Support MIN/MAX operations Example: min(1 , IDQ.MITE_UOPS / ( UPI * 16 * ( ICACHE.HIT + ICACHE.MISSES ) / 4.0 ) ) This is useful to fix up problems caused by multiplexing. - Support | & ^ operators - Minor cleanups and fixes - Support an \ escape for operators. This allows to specify event names like c2-residency - Support @ as an alternative for / to be able to specify pmus without conflicts with operators (like msr/tsc/ as msr@tsc@) Example: (cstate_core@c3\\-residency@ / msr@tsc@) * 100 Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170811232634.30465-8-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-22perf tools: Add utility function to detect SMT statusAndi Kleen3-0/+51
Add an smt_on() function to return if SMT is enabled or disabled. Used in the next patch. Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170811232634.30465-7-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-22perf bpf: Tighten detection of BPF eventsAndi Kleen1-4/+19
perf stat -e cpu/uops_executed.core,cmask=1/ would be detected as a BPF source event because the .c matches the .c source BPF pattern. v2: Originally I tried to use lex lookahead, but it doesn't seem to work. This now extends the BPF pattern to match longer events, but then does an extra check in the C code to reject BPF matches that do not end with .c/.o/.obj This uses REJECT, which makes the flex scanner slower, but that shouldn't be a big problem for the perf events. Committer testing: # perf trace -e write -e /home/acme/bpf/tracepoint.c cat /etc/passwd > /dev/null 0.000 ( 0.006 ms): cat/18485 write(fd: 1, buf: 0x7f59eebe1000, count: 3494 ) ... 0.006 ( ): raw_syscalls:sys_enter:NR 1 (1, 7f59eebe1000, da6, 22, 7f59eebe0010, 0)) 0.008 ( ): perf_bpf_probe:_write:(ffffffff9626b2c0)) 0.000 ( 0.010 ms): cat/18485 ... [continued]: write()) = 3494 # It continues doing what was expected, i.e. identifying /home/acme/bpf/tracepoint.c as a BPF event and activates the clang machinery to build an eBPF object and then uses sys_bpf() to hook it up to the raw_syscalls:sys_enter tracepoint, etc. Andi forgot to add Wang to the CC list, fix it. Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/r/20170811232634.30465-4-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-22perf evsel: Fix buffer overflow while freeing eventsAndi Kleen5-40/+20
Fix buffer overflow for: % perf stat -e msr/tsc/,cstate_core/c7-residency/ true that causes glibc free list corruption. For some reason it doesn't trigger in valgrind, but it is visible in AS: ================================================================= ==32681==ERROR: AddressSanitizer: heap-buffer-overflow on address 0x603000003f5c at pc 0x0000005671ef bp 0x7ffdaaac9ac0 sp 0x7ffdaaac9ab0 READ of size 4 at 0x603000003f5c thread T0 #0 0x5671ee in perf_evsel__close_fd util/evsel.c:1196 #1 0x56c57a in perf_evsel__close util/evsel.c:1717 #2 0x55ed5f in perf_evlist__close util/evlist.c:1631 #3 0x4647e1 in __run_perf_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:749 #4 0x4648e3 in run_perf_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:767 #5 0x46e1bc in cmd_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:2785 #6 0x52f83d in run_builtin /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:296 #7 0x52fd49 in handle_internal_command /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:348 #8 0x5300de in run_argv /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:392 #9 0x5308f3 in main /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:530 #10 0x7f0672d13400 in __libc_start_main (/lib64/libc.so.6+0x20400) #11 0x428419 in _start (/home/ak/hle/obj-perf/perf+0x428419) 0x603000003f5c is located 0 bytes to the right of 28-byte region [0x603000003f40,0x603000003f5c) allocated by thread T0 here: #0 0x7f0675139020 in calloc (/lib64/libasan.so.3+0xc7020) #1 0x648a2d in zalloc util/util.h:23 #2 0x648a88 in xyarray__new util/xyarray.c:9 #3 0x566419 in perf_evsel__alloc_fd util/evsel.c:1039 #4 0x56b427 in perf_evsel__open util/evsel.c:1529 #5 0x56c620 in perf_evsel__open_per_thread util/evsel.c:1730 #6 0x461dea in create_perf_stat_counter /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:263 #7 0x4637d7 in __run_perf_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:600 #8 0x4648e3 in run_perf_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:767 #9 0x46e1bc in cmd_stat /home/ak/hle/linux-hle-2.6/tools/perf/builtin-stat.c:2785 #10 0x52f83d in run_builtin /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:296 #11 0x52fd49 in handle_internal_command /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:348 #12 0x5300de in run_argv /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:392 #13 0x5308f3 in main /home/ak/hle/linux-hle-2.6/tools/perf/perf.c:530 #14 0x7f0672d13400 in __libc_start_main (/lib64/libc.so.6+0x20400) The event is allocated with cpus == 1, but freed with cpus == real number When the evsel close function walks the file descriptors it exceeds the fd xyarray boundaries and reads random memory. v2: Now that xyarrays save their original dimensions we can use these to iterate the two dimensional fd arrays. Fix some users (close, ioctl) in evsel.c to use these fields directly. This allows simplifying the code and dropping quite a few function arguments. Adjust all callers by removing the unneeded arguments. The actual perf event reading still uses the original values from the evsel list. Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170811232634.30465-2-andi@firstfloor.org [ Fix up xy_max_[xy]() -> xyarray__max_[xy]() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-22perf xyarray: Save max_x, max_yAndi Kleen2-0/+14
Save the original array dimensions in xyarrays, so that users can retrieve them later. Add some inline functions to access these fields. Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170811232634.30465-1-andi@firstfloor.org [ As noticed by Jiri, fix up namespacing: xy__method() -> xyarray__method() ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-18perf annotate browser: Circulate percent, total-period and nr-samples viewTaeung Song1-3/+8
Using the existing 't' hotkey, support the three views: percent, total period and number of samples on the annotate TUI browser, circulating them like below: Percent -> Total Period -> Nr Samples -> Percent ... Committer notes: Removed new 'e' hotkey, should be resubmitted as a separate patch, with proper justification for its inclusion. Suggested-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Milian Wolff <milian.wolff@kdab.com> Link: http://lkml.kernel.org/r/1503046028-5691-1-git-send-email-treeze.taeung@gmail.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-18perf annotate browser: Support --show-nr-samples optionTaeung Song2-5/+13
Support the --show-nr-samples in the TUI browser. Committer notes: Lift the restriction about --tui but leave it for --gtk: $ export LD_LIBRARY_PATH=~/lib64 $ perf annotate --gtk --show-nr-samples --show-nr-samples is not available in --gtk mode at this time $ Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/r/1503046023-5646-1-git-send-email-treeze.taeung@gmail.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-18perf annotate: Document --show-total-period optionTaeung Song1-0/+2
When the --show-total-period option was introduced we forgot to add an entry in the man page, fix it. Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Martin Liška <mliska@suse.cz> Fixes: 0c4a5bcea460 ("perf annotate: Display total number of samples with --show-total-period") Link: http://lkml.kernel.org/r/1503046013-5555-1-git-send-email-treeze.taeung@gmail.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-18perf annotate stdio: Support --show-nr-samples optionTaeung Song3-3/+23
Add --show-nr-samples option to "perf annotate" so that it matches "perf report". Committer note: Note that it can't be used together with --show-total-period, which seems like a silly limitation, that can be lifted at some point. Made it bail out if not on --stdio. Signed-off-by: Taeung Song <treeze.taeung@gmail.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Milian Wolff <milian.wolff@kdab.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lkml.kernel.org/r/1503046008-5511-1-git-send-email-treeze.taeung@gmail.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf tools: Use default CPUINFO_PROC where it fitsArnaldo Carvalho de Melo2-20/+2
Several architectures don't need to define it since the string is the same as the default one, so nuke them. 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-v1e1jr1u474w9xcelpaoxamu@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf tools: Remove unused cpu_relax() macrosArnaldo Carvalho de Melo1-9/+0
Since 195564390210 ("perf_counter: kerneltop: simplify data_head read") we do not use it, and this was way back in 2009, remove it before some other arch maintainer adds its implementation, like so many did, needlessly :-) 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: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-3l2su9c58eaq4twjzrf9uu08@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf events parse: Rename parse_events_parse argumentsArnaldo Carvalho de Melo4-104/+104
Calling them just "data" is too vague, call it 'perf_state', to make it clearer, for instance, when looking at patch hunks. 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-rnhk5yb05wem77rjpclrh7so@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf events parse: Use just one parse events state structArnaldo Carvalho de Melo3-6/+3
Andi reported problems when parse errors were detected with vendor events (json), because in the yyparse/parse_events_parse function we dereferenced the _data parameter to two different structs, with different layouts, which ended up making parse_events_evlist->error to point to random stack addresses. Fix it by making _data to always be struct parse_events_state, changing the only place where 'struct parse_events_term' was used in parse_events.y. Reported-by: Andi Kleen <ak@linux.intel.com> 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-bc27lshz823hxl8n9nkelcgh@git.kernel.org Fixes: 90e2b22dee90 ("perf/tool: Add support to reuse event grammar to parse out terms") Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf events parse: Rename parsing state struct to clearer nameArnaldo Carvalho de Melo4-31/+31
Rename it from 'parse_events_evlist' to 'parse_events_state' to better state that this is parsing state that has to be passed around. 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-dursqtg2h2w98ztaa297u43x@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf events parse: Remove some needless local variablesArnaldo Carvalho de Melo1-13/+7
Those are just casting a void pointer to a struct to then pass them to functions, i.e. remove the local variables and pass the void pointer directly, the casting will be done and the code will be shorter. 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-bzfodzr3mb46gy7u7v0mqad6@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf trace: Fix off by one string allocation problemArnaldo Carvalho de Melo1-1/+1
We need to consider the null terminator, oops, fix it. 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> Fixes: 017037ff3d0b ("perf trace: Allow specifying list of syscalls and events in -e/--expr/--event") Link: http://lkml.kernel.org/n/tip-j79jpqqe91gvxqmsgxgfn2ni@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-17perf jevents: Support FCMask and PortMaskAndi Kleen1-0/+2
Skylake server uncore IIO events need new FCMask/PortMask fields. Support those in the json parser and pass it through as a filter. Signed-off-by: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170816220201.19182-2-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-16perf test shell: Replace '|&' with '2>&1 |' to work with more shellsKim Phillips3-4/+4
Since we do not specify bash (and/or zsh) as a requirement, use the standard error redirection that is more widely supported. Signed-off-by: Kim Phillips <kim.phillips@arm.com> Link: http://lkml.kernel.org/n/tip-ji5mhn3iilgch3eaay6csr6z@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-16perf bpf: Fix endianness problem when loading parameters in prologueWang Nan2-3/+50
Perf's BPF prologue generator unconditionally fetches 8 bytes for function parameters, which causes problems on big endian machines. Thomas gives a detailed analysis for this problem: http://lkml.kernel.org/r/968ebda5-abe4-8830-8d69-49f62529d151@linux.vnet.ibm.com ---- 8< ---- I investigated perf test BPF for s390x and have a question regarding the 38.3 subtest (bpf-prologue test) which fails on s390x. When I turn on trace_printk in tests/bpf-script-test-prologue.c I see this output in /sys/kernel/debug/tracing/trace: [root@s8360047 perf]# cat /sys/kernel/debug/tracing/trace perf-30229 [000] d..2 170161.535791: : f_mode 2001d00000000 offset:0 orig:0 perf-30229 [000] d..2 170161.535809: : f_mode 6001f00000000 offset:0 orig:0 perf-30229 [000] d..2 170161.535815: : f_mode 6001f00000000 offset:1 orig:0 perf-30229 [000] d..2 170161.535819: : f_mode 2001d00000000 offset:1 orig:0 perf-30229 [000] d..2 170161.535822: : f_mode 2001d00000000 offset:2 orig:1 perf-30229 [000] d..2 170161.535825: : f_mode 6001f00000000 offset:2 orig:1 perf-30229 [000] d..2 170161.535828: : f_mode 6001f00000000 offset:3 orig:1 perf-30229 [000] d..2 170161.535832: : f_mode 2001d00000000 offset:3 orig:1 perf-30229 [000] d..2 170161.535835: : f_mode 2001d00000000 offset:4 orig:0 perf-30229 [000] d..2 170161.535841: : f_mode 6001f00000000 offset:4 orig:0 [...] There are 3 parameters the eBPF program tests/bpf-script-test-prologue.c accesses: f_mode (member of struct file at offset 140) offset and orig. They are parameters of the lseek() system call triggered in this test case in function llseek_loop(). What is really strange is the value of f_mode. It is an 8 byte value, whereas in the probe event it is defined as a 4 byte value. The lower 4 bytes are all zero and do not belong to member f_mode. The correct value should be 2001d for read-only and 6001f for read-write open mode. Here is the output of the 'perf test -vv bpf' trace: Try to find probe point from debuginfo. Matched function: null_lseek [2d9310d] Probe point found: null_lseek+0 Searching 'file' variable in context. Converting variable file into trace event. converting f_mode in file f_mode type is unsigned int. Opening /sys/kernel/debug/tracing//README write=0 Searching 'offset' variable in context. Converting variable offset into trace event. offset type is long long int. Searching 'orig' variable in context. Converting variable orig into trace event. orig type is int. Found 1 probe_trace_events. Opening /sys/kernel/debug/tracing//kprobe_events write=1 Writing event: p:perf_bpf_probe/func _text+8794224 f_mode=+140(%r2):x32 ---- 8< ---- This patch parses the type of each argument and converts data from memory to expected type. Now the test runs successfully on 4.13.0-rc5: [root@s8360046 perf]# ./perf test bpf 38: BPF filter : 38.1: Basic BPF filtering : Ok 38.2: BPF pinning : Ok 38.3: BPF prologue generation : Ok 38.4: BPF relocation checker : Ok [root@s8360046 perf]# Signed-off-by: Wang Nan <wangnan0@huawei.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Link: http://lkml.kernel.org/r/20170815092159.31912-1-tmricht@linux.vnet.ibm.com Signed-off-by: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf script python: Add support for sqlite3 to call-graph-from-sql.pyAdrian Hunter1-24/+36
Add support for SQLite 3 to the call-graph-from-sql.py script. The SQL statements work as is, so just detect the database type by checking if the SQLite 3 file exists. Committer notes: Tested collecting the PT data on a RHEL7.4, generating the SQLite3 database there and then moving it to a Fedora 26 system where the call-graph-from-sql.py script was run, using python-pyside version 1.2.2-7fc26 to see the callgraphs using Qt4. Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: http://lkml.kernel.org/r/1501749090-20357-6-git-send-email-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf script python: Rename call-graph-from-postgresql.py to call-graph-from-sql.pyAdrian Hunter3-7/+7
Rename call-graph-from-postgresql.py to call-graph-from-sql.py in preparation for adding support to it for SQLite 3. Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Link: http://lkml.kernel.org/r/1501749090-20357-5-git-send-email-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf script python: Add support for exporting to sqlite3Adrian Hunter4-3/+491
Add support for exporting to SQLite 3 the same data as the PostgreSQL export. Committer note: Tested on RHEL 7.4 using the 1.2.2-4el python-pyside packages from EPEL. Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: http://lkml.kernel.org/r/1501749090-20357-4-git-send-email-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf scripts python: Fix query in call-graph-from-postgresql.pyAdrian Hunter1-1/+1
Add a missing space which seemed not to affect PostgreSQL but upsets SQLite. Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Link: http://lkml.kernel.org/r/1501749090-20357-3-git-send-email-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf scripts python: Fix missing call_path_id in export-to-postgresql scriptAdrian Hunter1-1/+2
The export does not work if only branches are exported because of a missing column in the samples table. Fix by adding the missing call_path_id. Fixes: 3521f3bc9dae ("perf script: Update export-to-postgresql to support callchain export") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Link: http://lkml.kernel.org/r/1501749090-20357-2-git-send-email-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf test shell vfs_getname: Skip for tools built with NO_LIBDWARF=1Arnaldo Carvalho de Melo1-3/+3
If that is the case, or if the required lib is not present, e.g. elfutils-devel in Fedora systems, then just skip the tests requiring DWARF analysis. Before: # rpm -e elfutils-devel # perf test ping vfs_getname 60: Use vfs_getname probe to get syscall args filenames : FAILED! 61: probe libc's inet_pton & backtrace it with ping : Ok 62: Check open filename arg using perf trace + vfs_getname: FAILED! 63: Add vfs_getname probe to get syscall args filenames : FAILED! # After: # perf test vfs_getname 60: Use vfs_getname probe to get syscall args filenames : Skip 62: Check open filename arg using perf trace + vfs_getname: Skip 63: Add vfs_getname probe to get syscall args filenames : Skip # Then, reinstalling elfutils-devel, rebuilding the tool and running again: # perf test vfs_getname 60: Use vfs_getname probe to get syscall args filenames : Ok 62: Check open filename arg using perf trace + vfs_getname: Ok 63: Add vfs_getname probe to get syscall args filenames : Ok # Reported-by: Kim Phillips <kim.phillips@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-d67tvn401fxrwr97pu5ihfb1@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf test shell: Check if 'perf probe' is available, skip tests if notArnaldo Carvalho de Melo6-1/+22
Add a library function that checks if 'perf probe' is built into the tool being tested, skipping tests that need it. Testing it on a system after removing the library needed to build 'probe' as a perf subcommand: # perf test ping vfs_getname 59: Use vfs_getname probe to get syscall args filenames : Skip 60: probe libc's inet_pton & backtrace it with ping : Skip 61: Check open filename arg using perf trace + vfs_getname: Skip 62: Add vfs_getname probe to get syscall args filenames : Skip # perf probe perf: 'probe' is not a perf-command. See 'perf --help'. # Now reinstalling elfutils-libelf-devel on this Fedora 26 system to rebuild perf and then retest this: # perf test ping vfs_getname 60: Use vfs_getname probe to get syscall args filenames : Ok 61: probe libc's inet_pton & backtrace it with ping : Ok 62: Check open filename arg using perf trace + vfs_getname: Ok 63: Add vfs_getname probe to get syscall args filenames : Ok # Reported-by: Kim Phillips <kim.phillips@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-ctdck2gzsskqhjzu3ebb62zm@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-15perf tests shell: Remove duplicate skip_if_no_debuginfo() functionArnaldo Carvalho de Melo1-5/+0
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-3zxjswdbs2au3ih0rino0iy1@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Add uprobes + backtrace ping testArnaldo Carvalho de Melo1-0/+40
Installs a probe on libc's inet_pton function, that will use uprobes, then use 'perf trace' on a ping to localhost asking for just one packet with the a backtrace 3 levels deep, check that it is what we expect. This needs no debuginfo package, all is done using the libc ELF symtab and the CFI info in the binaries. Testing it: # perf test ping 61: probe libc's inet_pton & backtrace it with ping : Ok In verbose mode: # perf test -v ping 61: probe libc's inet_pton & backtrace it with ping : --- start --- test child forked, pid 1007 PING ::1(::1) 56 data bytes 64 bytes from ::1: icmp_seq=1 ttl=64 time=0.058 ms --- ::1 ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 0.058/0.058/0.058/0.000 ms 0.000 probe_libc:inet_pton:(7f75fce12a20)) __GI___inet_pton (/usr/lib64/libc-2.24.so) getaddrinfo (/usr/lib64/libc-2.24.so) _init (/usr/bin/ping) test child finished with 0 ---- end ---- probe libc's inet_pton & backtrace it with ping: Ok # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-idrntt4nbg15aafu8hjmv7sk@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf report: Fix module symbol adjustment for s390xThomas Richter3-1/+17
The 'perf report' tool does not display the addresses of kernel module symbols correctly. For example symbol qeth_send_ipa_cmd in kernel module qeth.ko has this relative address for function qeth_send_ipa_cmd(): [root@s8360047 linux]# nm -g drivers/s390/net/qeth.ko | fgrep send_ipa_cmd 0000000000013088 T qeth_send_ipa_cmd The module is loaded at address: [root@s8360047 linux]# cat /sys/module/qeth/sections/.text 0x000003ff80296d20 [root@s8360047 linux]# This should result in a start address of: 0x13088 + 0x3ff80296d20 = 0x3ff802a9da8 Using crash to verify the address on a live system: [root@s8360046 linux]# crash vmlinux crash 7.1.9++ Copyright (C) 2002-2016 Red Hat, Inc. Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation [...] crash> mod -s qeth drivers/s390/net/qeth.ko MODULE NAME SIZE OBJECT FILE 3ff8028d700 qeth 151552 drivers/s390/net/qeth.ko crash> sym qeth_send_ipa_cmd 3ff802a9da8 (T) qeth_send_ipa_cmd [qeth] /root/linux/drivers/s390/net/qeth_core_main.c: 2944 crash> Now perf report displays the address of symbol qeth_send_ipa_cmd: symbol__new: qeth_send_ipa_cmd 0x130f0-0x132ce There is a difference of 0x68 between the entry in the symbol table (see nm command above) and perf. The difference is from the offset the .text segment of qeth.ko: [root@s8360047 perf]# readelf -a drivers/s390/net/qeth.ko Section Headers: [Nr] Name Type Address Offset Size EntSize Flags Link Info Align [ 0] NULL 0000000000000000 00000000 0000000000000000 0000000000000000 0 0 0 [ 1] .note.gnu.build-i NOTE 0000000000000000 00000040 0000000000000024 0000000000000000 A 0 0 4 [ 2] .text PROGBITS 0000000000000000 00000068 000000000001c8a0 0000000000000000 AX 0 0 8 As seen the .text segment has an offset of 0x68 with start address 0x0. Therefore 0x68 is added to the address of qeth_send_ipa_cmd and thus 0x13088 + 0x68 = 0x130f0 is displayed. This is wrong, perf report needs to display the start address of symbol qeth_send_ipa_cmd at 0x13088 + qeth.ko.text section start address. The qeth.ko module .text start address is available in the qeth.ko DSO map. Just identify the kernel module symbols and correct the addresses. With the fix I see this correct address for symbol: symbol__new: qeth_send_ipa_cmd 0x3ff802a9da8-0x3ff802a9f86 Signed-off-by: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com> Reviewed-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com> Cc: Zvonko Kosic <zvonko.kosic@de.ibm.com> LPU-Reference: 20170803134902.47207-1-tmricht@linux.vnet.ibm.com Link: http://lkml.kernel.org/n/tip-q8lktlpoxb5e3dj52u1s1rw4@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf record: Fix wrong size in perf_record_mmap for last kernel moduleThomas Richter4-10/+19
During work on perf report for s390 I ran into the following issue: 0 0x318 [0x78]: PERF_RECORD_MMAP -1/0: [0x3ff804d6990(0xfffffc007fb2966f) @ 0]: x /lib/modules/4.12.0perf1+/kernel/drivers/s390/net/qeth_l2.ko This is a PERF_RECORD_MMAP entry of the perf.data file with an invalid module size for qeth_l2.ko (the s390 ethernet device driver). Even a mainframe does not have 0xfffffc007fb2966f bytes of main memory. It turned out that this wrong size is created by the perf record command. What happens is this function call sequence from __cmd_record(): perf_session__new(): perf_session__create_kernel_maps(): machine__create_kernel_maps(): machine__create_modules(): Creates map for all loaded kernel modules. modules__parse(): Reads /proc/modules and extracts module name and load address (1st and last column) machine__create_module(): Called for every module found in /proc/modules. Creates a new map for every module found and enters module name and start address into the map. Since the module end address is unknown it is set to zero. This ends up with a kernel module map list sorted by module start addresses. All module end addresses are zero. Last machine__create_kernel_maps() calls function map_groups__fixup_end(). This function iterates through the maps and assigns each map entry's end address the successor map entry start address. The last entry of the map group has no successor, so ~0 is used as end to consume the remaining memory. Later __cmd_record calls function record__synthesize() which in turn calls perf_event__synthesize_kernel_mmap() and perf_event__synthesize_modules() to create PERF_REPORT_MMAP entries into the perf.data file. On s390 this results in the last module qeth_l2.ko (which has highest start address, see module table: [root@s8360047 perf]# cat /proc/modules qeth_l2 86016 1 - Live 0x000003ff804d6000 qeth 266240 1 qeth_l2, Live 0x000003ff80296000 ccwgroup 24576 1 qeth, Live 0x000003ff80218000 vmur 36864 0 - Live 0x000003ff80182000 qdio 143360 2 qeth_l2,qeth, Live 0x000003ff80002000 [root@s8360047 perf]# ) to be the last entry and its map has an end address of ~0. When the PERF_RECORD_MMAP entry is created for kernel module qeth_l2.ko its start address and length is written. The length is calculated in line: event->mmap.len = pos->end - pos->start; and results in 0xffffffffffffffff - 0x3ff804d6990(*) = 0xfffffc007fb2966f (*) On s390 the module start address is actually determined by a __weak function named arch__fix_module_text_start() in machine__create_module(). I think this improvable. We can use the module size (2nd column of /proc/modules) to get each loaded kernel module size and calculate its end address. Only for map entries which do not have a valid end address (end is still zero) we can use the heuristic we have now, that is use successor start address or ~0. Signed-off-by: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com> Reviewed-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com> Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com> Cc: Zvonko Kosic <zvonko.kosic@de.ibm.com> LPU-Reference: 20170803134902.47207-2-tmricht@linux.vnet.ibm.com Link: http://lkml.kernel.org/n/tip-nmoqij5b5vxx7rq2ckwu8iaj@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf srcline: Do not consider empty files as valid srclinesMilian Wolff1-0/+6
Sometimes we get a non-null, but empty, string for the filename from bfd. This then results in srclines of the form ":0", which is different from the canonical SRCLINE_UNKNOWN in the form "??:0". Set the file to NULL if it is empty to fix this. Signed-off-by: Milian Wolff <milian.wolff@kdab.com> Cc: David Ahern <dsahern@gmail.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Yao Jin <yao.jin@linux.intel.com> Link: http://lkml.kernel.org/r/20170806212446.24925-14-milian.wolff@kdab.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf util: Take elf_name as const string in dso__demangle_symMilian Wolff3-3/+3
The input string is not modified and thus can be passed in as a pointer to const data. Signed-off-by: Milian Wolff <milian.wolff@kdab.com> Cc: David Ahern <dsahern@gmail.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Yao Jin <yao.jin@linux.intel.com> Link: http://lkml.kernel.org/r/20170806212446.24925-3-milian.wolff@kdab.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Add test using vfs_getname + 'perf trace'Arnaldo Carvalho de Melo1-0/+31
Uses the 'perf test shell' library to add probe:vfs_getname to the system then use it with 'perf trace' using 'touch' to write to a temp file, then checks that that was captured by the vfs_getname was used by 'perf trace', that already handles "probe:vfs_getname" if present, and used in the "open" syscall "filename" argument beautifier. Testing it: # perf test "trace + vfs_getname" 61: Check open filename arg using perf trace + vfs_getname: Ok # # perf test -v "trace + vfs_getname" 61: Check open filename arg using perf trace + vfs_getname: --- start --- test child forked, pid 30846 Added new event: probe:vfs_getname (on getname_flags:72 with pathname=result->name:string) You can now use it in all perf tools, such as: perf record -e probe:vfs_getname -aR sleep 1 2.237 ( 0.012 ms): touch/30855 open(filename: /tmp/temporary_file.kmoWQ, flags: CREAT|NOCTTY|NONBLOCK|WRONLY, mode: IRUGO|IWUGO) = 3 test child finished with 0 ---- end ---- Check open filename arg using perf trace + vfs_getname: Ok # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-j02nobfvvn9c7yrphdsnbqx0@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Add test using probe:vfs_getname and verifying resultsArnaldo Carvalho de Melo1-0/+37
This test uses the 'perf test shell' library to add probe:vfs_getname to the system then use it with 'perf record' using 'touch' to write to a temp file, then checks that that was captured by the vfs_getname probe in the generated perf.data file, with the temp file name as the pathname argument. Using it: # perf test "Use vfs_getname" 60: Use vfs_getname probe to get syscall args filenames: Ok # perf test -v "Use vfs_getname" 60: Use vfs_getname probe to get syscall args filenames: --- start --- test child forked, pid 16414 Added new event: probe:vfs_getname (on getname_flags:72 with pathname=result->name:string) You can now use it in all perf tools, such as: perf record -e probe:vfs_getname -aR sleep 1 Recording open file: [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.022 MB /tmp/vaca.perf.data.QZsn7 (13 samples) ] Looking at perf.data file for vfs_getname records for the file we touched: touch 16421 [002] 1255152.879561: probe:vfs_getname: (ffffffffa626e608) pathname="/tmp/vaca.l10SL" test child finished with 0 ---- end ---- Use vfs_getname probe to get syscall args filenames: Ok # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-t555fnhbcbxnukltk23dqxur@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Move vfs_getname probe function to libArnaldo Carvalho de Melo3-22/+32
Multiple tests will be able to reuse these functions, to test things like perf report, 'trace', etc, using this probe. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-48xagvozhouhyi8fjota6o2d@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Install shell testsArnaldo Carvalho de Melo1-1/+3
Now that we have shell tests, install them. Developers don't need this pass, as 'perf test' will look first at the in tree scripts at tools/perf/tests/shell/. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-j21u4v0jsehi0lpwqwjb4j45@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test shell: Add 'probe_vfs_getname' shell testArnaldo Carvalho de Melo1-0/+30
First perf shell test: # perf test vfs_getname 60: Add vfs_getname probe to get syscall args filenames: Ok # In verbose mode: # perf test -v vfs_getname 60: Add vfs_getname probe to get syscall args filenames: --- start --- test child forked, pid 19146 Added new event: probe:vfs_getname (on getname_flags:72 with pathname=result->name:string) You can now use it in all perf tools, such as: perf record -e probe:vfs_getname -aR sleep 1 test child finished with 0 ---- end ---- Add vfs_getname probe to get syscall args filenames: Ok # And if the vmlinux file is not found: # mv ../build/v4.12.0-rc6+/vmlinux ../build/v4.12.0-rc6+/vmlinux.hidden # perf test vfs_getname 60: Add vfs_getname probe to get syscall args filenames: Skip # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-8f3n22c1yn516ev30s603ow2@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test: Make 'list' use same filtering code as main 'perf test'Arnaldo Carvalho de Melo1-8/+10
Before: # perf test Synth 39: Synthesize thread map : Ok 41: Synthesize cpu map : Ok 42: Synthesize stat config : Ok 43: Synthesize stat : Ok 44: Synthesize stat round : Ok 45: Synthesize attr update : Ok # perf test list Synth # After: # perf test Synth 39: Synthesize thread map : Ok 41: Synthesize cpu map : Ok 42: Synthesize stat config : Ok 43: Synthesize stat : Ok 44: Synthesize stat round : Ok 45: Synthesize attr update : Ok # perf test list Synth 39: Synthesize thread map 41: Synthesize cpu map 42: Synthesize stat config 43: Synthesize stat 44: Synthesize stat round 45: Synthesize attr update # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-v95tqqzuwawsmds3zn2mosje@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test: Add infrastructure to run shell based testsArnaldo Carvalho de Melo2-1/+170
To allow testing by directly using perf tools in scripts, checking that the effects on the system are the ones expected and that the output produced is as well the desired one. For instance, adding a probe at a well known location with 'perf probe', then checking that the results from using that probe to record are the desired ones, etc. The next csets will introduce tests using this new testing infrastructure. The scripts should return 0 for Ok, 1 for FAIL and 2 for SKIP. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-swbpn7amrjqffh83lsr39s9p@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test: Add 'struct test *' to the test functionsArnaldo Carvalho de Melo54-124/+127
This way we'll be able to pass more test specific parameters without having to change this function signature. Will be used by the upcoming 'shell tests', shell scripts that will call perf tools and check if they work as expected, comparing its effects on the system (think 'perf probe foo') the output produced, etc. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-wq250w7j1opbzyiynozuajbl@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf test: Make 'list' subcommand match main 'perf test' numbering/matchingArnaldo Carvalho de Melo1-1/+3
Before: # perf test Synth 39: Synthesize thread map : Ok 41: Synthesize cpu map : Ok 42: Synthesize stat config : Ok 43: Synthesize stat : Ok 44: Synthesize stat round : Ok 45: Synthesize attr update : Ok # # perf test list Synth 1: Synthesize thread map 2: Synthesize cpu map 3: Synthesize stat config 4: Synthesize stat 5: Synthesize stat round 6: Synthesize attr update # After: # perf test Synth 39: Synthesize thread map : Ok 41: Synthesize cpu map : Ok 42: Synthesize stat config : Ok 43: Synthesize stat : Ok 44: Synthesize stat round : Ok 45: Synthesize attr update : Ok # # perf test list Synth 39: Synthesize thread map 41: Synthesize cpu map 42: Synthesize stat config 43: Synthesize stat 44: Synthesize stat round 45: Synthesize attr update # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Matt Fleming <matt.fleming@intel.com> Cc: Michael Petlan <mpetlan@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Thomas Richter <tmricht@linux.vnet.ibm.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-pjhuhkphs7o3tkbqrukfv6bz@git.kernel.org Fixes: e8210cefb7e1 ("perf tests: Introduce iterator function for tests") Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf tools: Add missing newline to expr parser error messagesAndi Kleen1-1/+1
Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170724234015.5165-6-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf stat: Fix saved values rbtree lookupAndi Kleen1-1/+5
The stat shadow saved values rbtree is indexed by a pointer. Fix the comparison function: - We cannot return a pointer delta as an int because that loses bits on 64bit. - Doing pointer arithmetic on the struct pointer only works if the objects are spaced by the multiple of the object size, which is not guaranteed for individual malloc'ed object Replace it with a proper comparison. This fixes various problems with values not being found. Signed-off-by: Andi Kleen <ak@linux.intel.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20170724234015.5165-4-andi@firstfloor.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf vendor events powerpc: Update POWER9 eventsSukadev Bhattiprolu9-2298/+3511
Update and cleanup POWER9 PMU events. Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> Cc: Anton Blanchard <anton@au1.ibm.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Link: http://lkml.kernel.org/r/20170802174617.GA32545@us.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf vendor events powerpc: remove suffix in mapfileSukadev Bhattiprolu1-10/+10
Drop the .json suffix for events directory in the mapfile.csv. Now that we have separate JSON files for each topic in a CPU (eg: see tools/perf/pmu-events/arch/powerpc/power8/*.json) the .json suffix in the mapfile is misleading and redundant. Reported-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> Cc: Anton Blanchard <anton@au1.ibm.com> Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/r/20170802174617.GA32545@us.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-11perf scripting python: Add ppc64le to audit uname listNaveen N. Rao1-0/+1
Before patch: $ uname -m ppc64le $ ./perf script -s ./scripts/python/syscall-counts.py Install the audit-libs-python package to get syscall names. For example: # apt-get install python-audit (Ubuntu) # yum install audit-libs-python (Fedora) etc. Press control+C to stop and show the summary ^CWarning: 4 out of order events recorded. syscall events: event count ---------------------------------------- ----------- 4 504638 54 1206 221 42 55 21 3 12 167 10 11 8 6 7 125 6 5 6 108 5 162 4 90 4 45 3 33 3 311 1 246 1 238 1 93 1 91 1 After patch: ./perf script -s ./scripts/python/syscall-counts.py Press control+C to stop and show the summary ^CWarning: 5 out of order events recorded. syscall events: event count ---------------------------------------- ----------- write 643411 ioctl 1206 futex 54 fcntl 27 poll 14 read 12 execve 8 close 7 mprotect 6 open 6 nanosleep 5 fstat 5 mmap 4 inotify_add_watch 3 brk 3 access 3 timerfd_settime 1 clock_gettime 1 epoll_wait 1 ftruncate 1 munmap 1 Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com> Acked-by: Paul Clarke <pc@us.ibm.com> Cc: linuxppc-dev@lists.ozlabs.org Link: http://lkml.kernel.org/n/tip-bnl67p1alkvx97pn9moxz3qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-01perf trace beautify ioctl: Beautify perf ioctl's 'cmd' argArnaldo Carvalho de Melo3-8/+39
Also trying a new approach, using the copy of uapi/linux/perf_event.h we auto generate the string tables, then include it in the ioctl cmd beautifier. This way either the perf developers will add the new commands to the tools/ copy, like is happening with other areas of tools/include/ (bpf.h comes to mind), or we'll be notified when building perf that our copy drifted. E.g., looking at some of the perf ioctls issued by the 'perf test' test cases: # (perf trace -e perf_event_open,ioctl perf test) 2>&1 | egrep "(cmd: PERF_|perf_event_open)" 4: Read samples using the mmap interface : 348.811 ( 0.062 ms): perf/23351 perf_event_open(attr_uptr: 0x414a5e8, pid: 23351 (perf), group_fd: -1, flags: FD_CLOEXEC) = 3 348.878 ( 0.039 ms): perf/23351 perf_event_open(attr_uptr: 0x414a5e8, pid: 23351 (perf), cpu: 1, group_fd: -1, flags: FD_CLOEXEC) = 4 348.919 ( 0.036 ms): perf/23351 perf_event_open(attr_uptr: 0x414a5e8, pid: 23351 (perf), cpu: 2, group_fd: -1, flags: FD_CLOEXEC) = 5 348.958 ( 0.036 ms): perf/23351 perf_event_open(attr_uptr: 0x414a5e8, pid: 23351 (perf), cpu: 3, group_fd: -1, flags: FD_CLOEXEC) = 6 349.070 ( 0.046 ms): perf/23351 perf_event_open(attr_uptr: 0x414aa38, pid: 23351 (perf), group_fd: -1, flags: FD_CLOEXEC) = 7 349.120 ( 0.037 ms): perf/23351 perf_event_open(attr_uptr: 0x414aa38, pid: 23351 (perf), cpu: 1, group_fd: -1, flags: FD_CLOEXEC) = 8 349.161 ( 0.036 ms): perf/23351 perf_event_open(attr_uptr: 0x414aa38, pid: 23351 (perf), cpu: 2, group_fd: -1, flags: FD_CLOEXEC) = 9 349.201 ( 0.035 ms): perf/23351 perf_event_open(attr_uptr: 0x414aa38, pid: 23351 (perf), cpu: 3, group_fd: -1, flags: FD_CLOEXEC) = 10 349.306 ( 0.041 ms): perf/23351 perf_event_open(attr_uptr: 0x414b2d8, pid: 23351 (perf), group_fd: -1, flags: FD_CLOEXEC) = 11 349.611 ( 0.005 ms): perf/23351 ioctl(fd: 3<anon_inode:[perf_event]>, cmd: PERF_ID, arg: 0x7fff025999b8) = 0 349.619 ( 0.002 ms): perf/23351 ioctl(fd: 7<anon_inode:[perf_event]>, cmd: PERF_SET_OUTPUT, arg: 0x3 ) = 0 349.623 ( 0.002 ms): perf/23351 ioctl(fd: 7<anon_inode:[perf_event]>, cmd: PERF_ID, arg: 0x7fff025999b8) = 0 349.627 ( 0.002 ms): perf/23351 ioctl(fd: 11<anon_inode:[perf_event]>, cmd: PERF_SET_OUTPUT, arg: 0x3 ) = 0 349.630 ( 0.001 ms): perf/23351 ioctl(fd: 11<anon_inode:[perf_event]>, cmd: PERF_ID, arg: 0x7fff025999b8) = 0 <SNIP> 7: PERF_RECORD_* events & perf_sample fields : 647.150 ( 0.014 ms): perf/23354 perf_event_open(attr_uptr: 0x7fff02599920, pid: -1, cpu: 2, group_fd: -1, flags: FD_CLOEXEC) = 3 647.197 ( 0.076 ms): perf/23354 perf_event_open(attr_uptr: 0x414b478, pid: -1, group_fd: -1, flags: FD_CLOEXEC) = 3 647.289 ( 0.040 ms): perf/23354 perf_event_open(attr_uptr: 0x414b478, pid: -1, group_fd: -1, flags: FD_CLOEXEC) = 3 647.368 ( 0.011 ms): perf/23354 perf_event_open(attr_uptr: 0x414a5e8, pid: 23355 (perf), group_fd: -1, flags: FD_CLOEXEC) = 3 647.381 ( 0.005 ms): perf/23354 perf_event_open(attr_uptr: 0x414a5e8, pid: 23355 (perf), cpu: 1, group_fd: -1, flags: FD_CLOEXEC) = 4 647.387 ( 0.005 ms): perf/23354 perf_event_open(attr_uptr: 0x414a5e8, pid: 23355 (perf), cpu: 2, group_fd: -1, flags: FD_CLOEXEC) = 5 647.393 ( 0.004 ms): perf/23354 perf_event_open(attr_uptr: 0x414a5e8, pid: 23355 (perf), cpu: 3, group_fd: -1, flags: FD_CLOEXEC) = 7 648.026 ( 0.011 ms): perf/23354 ioctl(fd: 3<anon_inode:[perf_event]>, cmd: PERF_ENABLE) = 0 648.038 ( 0.002 ms): perf/23354 ioctl(fd: 4<anon_inode:[perf_event]>, cmd: PERF_ENABLE) = 0 648.042 ( 0.002 ms): perf/23354 ioctl(fd: 5<anon_inode:[perf_event]>, cmd: PERF_ENABLE) = 0 648.045 ( 0.002 ms): perf/23354 ioctl(fd: 7<anon_inode:[perf_event]>, cmd: PERF_ENABLE) = 0 <SNIP> 18: Breakpoint overflow signal handler : 2772.721 ( 0.017 ms): perf/23375 perf_event_open(attr_uptr: 0x7fff02599d20, pid: -1, cpu: 3, group_fd: -1, flags: FD_CLOEXEC) = 3 2772.748 ( 0.009 ms): perf/23375 perf_event_open(attr_uptr: 0x7fff02599e60, cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 3 2772.768 ( 0.002 ms): perf/23375 ioctl(fd: 3, cmd: PERF_RESET) = 0 2772.776 ( 0.008 ms): perf/23375 perf_event_open(attr_uptr: 0x7fff02599e60, cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 4 2772.788 ( 0.002 ms): perf/23375 ioctl(fd: 4, cmd: PERF_RESET) = 0 2772.791 ( 0.006 ms): perf/23375 perf_event_open(attr_uptr: 0x7fff02599e60, cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 5 2772.800 ( 0.001 ms): perf/23375 ioctl(fd: 5, cmd: PERF_RESET) = 0 2772.803 ( 0.005 ms): perf/23375 ioctl(fd: 3, cmd: PERF_ENABLE) = 0 2772.810 ( 0.004 ms): perf/23375 ioctl(fd: 4, cmd: PERF_ENABLE) = 0 2772.815 ( 0.004 ms): perf/23375 ioctl(fd: 5, cmd: PERF_ENABLE) = 0 <SNIP> 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-ahotwscqt080ae0ulu3zznh2@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-01perf trace beautify ioctl: Beautify vhost virtio ioctl's 'cmd' argArnaldo Carvalho de Melo3-2/+42
Also trying a new approach, using a copy of uapi/linux/vhost.h we auto generate the string tables, then include it in the ioctl cmd beautifier. This way either the KVM developers will add the new commands to the tools/ copy, like is happening with other areas of tools/include/ (bpf.h comes to mind), or we'll be notified when building perf that our copy drifted. E.g., doing syswide tracing grepping for the newly beautified VHOST ioctls: # perf trace -e ioctl 2>&1 | grep VHOST 3873.064 ( 0.099 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_NET_SET_BACKEND, arg: 0x7fff053dffe0) = 0 3873.168 ( 0.019 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_NET_SET_BACKEND, arg: 0x7fff053dffe0) = 0 3873.226 ( 0.006 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_GET_VRING_BASE, arg: 0x7fff053dff60) = 0 3873.244 ( 0.002 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_GET_VRING_BASE, arg: 0x7fff053dff60) = 0 3873.817 ( 0.014 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_CALL, arg: 0x7fff053dff20) = 0 3873.838 ( 0.004 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_CALL, arg: 0x7fff053dff20) = 0 4701.372 ( 0.006 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_CALL, arg: 0x7fff053dfe20) = 0 4701.417 ( 0.007 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_CALL, arg: 0x7fff053dfe20) = 0 4701.563 ( 0.004 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_FEATURES, arg: 0x7fff053dfe88) = 0 4701.571 ( 0.028 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_MEM_TABLE, arg: 0x563c7c906870) = 0 4701.604 ( 0.003 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_NUM, arg: 0x7fff053dff00) = 0 4701.609 ( 0.002 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_BASE, arg: 0x7fff053dff00) = 0 4701.615 ( 0.002 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_ADDR, arg: 0x7fff053dfe70) = 0 4701.619 ( 0.008 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_KICK, arg: 0x7fff053dfef0) = 0 4701.634 ( 0.004 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_NUM, arg: 0x7fff053dff00) = 0 4701.640 ( 0.002 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_BASE, arg: 0x7fff053dff00) = 0 4701.644 ( 0.002 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_ADDR, arg: 0x7fff053dfe70) = 0 4701.648 ( 0.009 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_SET_VRING_KICK, arg: 0x7fff053dfef0) = 0 4701.665 ( 0.005 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_NET_SET_BACKEND, arg: 0x7fff053dff80) = 0 4701.672 ( 0.004 ms): qemu-system-x8/21238 ioctl(fd: 27</dev/vhost-net>, cmd: VHOST_NET_SET_BACKEND, arg: 0x7fff053dff80) = 0 ^C '-e ioctl' uses tracepoint filters, in time this will be replaces by eBPF filters hooked at the syscall tracepoints and that "grep VHOST" will also be done with eBPF, right at the kernel, to reduce overhead. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-2gthnhpliunvakywjterrzz3@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-01tools include uapi: Grab a copy of linux/vhost.hArnaldo Carvalho de Melo2-0/+2
We will use it to generate tables for beautifying ioctl's 'cmd' arg. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-nxwpq34hu6te1m2ra5m7o8n9@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-08-01perf trace beauty ioctl: Pass _IOC_DIR to the per _IOC_TYPE scnprintfArnaldo Carvalho de Melo1-12/+12
Not all subsystems use the fact that we may have the same _IOC_NR for different _IOC_DIR, as in the end it'll result in a different ioctl number. So, for instance, vhost virtio has: #define VHOST_GET_FEATURES _IOR(VHOST_VIRTIO, 0x00, __u64) #define VHOST_SET_FEATURES _IOW(VHOST_VIRTIO, 0x00, __u64) So same _IOC_NR (0x00) but different _IOC_DIR (R versus W), but it also have: #define VHOST_SET_VRING_ENDIAN _IOW(VHOST_VIRTIO, 0x13, struct vhost_vring_state) #define VHOST_GET_VRING_ENDIAN _IOW(VHOST_VIRTIO, 0x14, struct vhost_vring_state) A "get" operation that uses a "W" _IOC_DIR, and its implementation, uses copy_to_user, it should've probably been _IOR(). Then: /* Base value where queue looks for available descriptors */ #define VHOST_SET_VRING_BASE _IOW(VHOST_VIRTIO, 0x12, struct vhost_vring_state) /* Get accessor: reads index, writes value in num */ #define VHOST_GET_VRING_BASE _IOWR(VHOST_VIRTIO, 0x12, struct vhost_vring_state) So we'll need to use _IOC_DIR() to disambiguate the VHOST_VIRTIO ioctl bautifier. 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: Paolo Bonzini <pbonzini@redhat.com> Cc: Radim Krčmář <rkrcmar@redhat.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-rq6q717ql7j2z7kuccafgq84@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Copyright © 1996 – 2023 Jason A. Donenfeld. All Rights Reverse Engineered.