- 05 Nov, 2018 7 commits
-
-
Adrian Hunter authored
Fetching data from the database can be slow. Add a report that provides the ability to select a subset of branches. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/r/20181104151238.15947-3-adrian.hunter@intel.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Adrian Hunter authored
Fall back to /usr/local/lib/libxed.so to cater for distributions that do not have /usr/local/lib in the library path by default. Signed-off-by:
-
Jin Yao authored
'perf report' has supported the displaying of LBR stats (such as cycles, predicted%) in callchain entry. For example: $ perf report --branch-history --stdio --1.01%--intel_idle mwait.h:29 intel_idle cpufeature.h:164 (cycles:5) intel_idle cpufeature.h:164 (predicted:76.4%) intel_idle mwait.h:102 (cycles:41) intel_idle current.h:15 While 'perf top' doesn't support that. For example: $ perf top -a -b --call-graph branch - 13.86% 0.23% [kernel] [k] __x86_indirect_thunk_rax - 13.65% __x86_indirect_thunk_rax + 1.69% do_syscall_64 + 1.68% do_select + 1.41% ktime_get + 0.70% __schedule + 0.62% do_sys_poll 0.58% __x86_indirect_thunk_rax Actually it's very easy to enable this feature in 'perf top'. With this patch, the result is: $ perf top -a -b --call-graph branch $ - 13.58% 0.00% [kernel] [k] __x86_indirect_thunk_rax $ - 13.57% __x86_indirect_thunk_rax (predicted:93.9%) $ + 1.78% do_select (cycles:2) $ + 1.68% perf_pmu_disable.part.99 (cycles:1) $ + 1.45% ___sys_recvmsg (cycles:25) $ + 0.81% unix_stream_sendmsg (cycles:18) $ + 0.80% ktime_get (cycles:400) $ 0.58% pick_next_task_fair (cycles:47) $ + 0.56% i915_request_retire (cycles:2) $ + 0.52% do_sys_poll (cycles:4) Signed-off-by:Jin Yao <yao.jin@linux.intel.com> Tested-by:
-
Thomas Richter authored
On s390 the CPU Measurement Facility for counters now supports 2 PMUs named cpum_cf (CPU Measurement Facility for counters) and cpum_cf_diag (CPU Measurement Facility for diagnostic counters) for one and the same CPU. Running command [root@s35lp76 perf]# ./perf stat -e tx_c_tend \ -- ~/mytests/cf-tx-events 1 Measuring transactions TX_C_TABORT_NO_SPECIAL: 0 expected:0 TX_C_TABORT_SPECIAL: 0 expected:0 TX_C_TEND: 1 expected:1 TX_NC_TABORT: 11 expected:11 TX_NC_TEND: 1 expected:1 Performance counter stats for '/root/mytests/cf-tx-events 1': 2 tx_c_tend 0.002120091 seconds time elapsed 0.000121000 seconds user 0.002127000 seconds sys [root@s35lp76 perf]# displays output which is unexpected (and wrong): 2 tx_c_tend The test program definitely triggers only one transaction, as shown in line 'TX_C_TEND: 1 expected:1'. This is caused by the following call sequence: pmu_lookup() scans and installs a PMU. +--> pmu_aliases() parses all aliases in directory .../<pmu-name>/events/* which are file names. +--> pmu_aliases_parse() Read each file in directory and create an new alias entry. This is done with +--> perf_pmu__new_alias() and +--> __perf_pmu__new_alias() which also check for identical alias names. After pmu_aliases() returns, a complete list of event names for this pmu has been created. Now function pmu_add_cpu_aliases() is called to add the events listed in the json | files to the alias list of the cpu. +--> perf_pmu__find_map() Returns a pointer to the json events. Now function pmu_add_cpu_aliases() scans through all events listed in the JSON files for this CPU. Each json event pmu name is compared with the current PMU being built up and if they mismatch, the json event is added to the current PMUs alias list. To avoid duplicate entries the following comparison is done: if (!is_arm_pmu_core(name)) { pname = pe->pmu ? pe->pmu : "cpu"; if (strncmp(pname, name, strlen(pname))) continue; } The culprit is the strncmp() function. Using current s390 PMU naming, the first PMU is 'cpum_cf' and a long list of events is added, among them 'tx_c_tend' When the second PMU named 'cpum_cf_diag' is added, only one event named 'CF_DIAG' is added by the pmu_aliases() function. Now function pmu_add_cpu_aliases() is invoked for PMU 'cpum_cf_diag'. Since the CPUID string is the same for both PMUs, json file events for PMU named 'cpum_cf' are added to the PMU 'cpm_cf_diag' This happens because the strncmp() actually compares: strncmp("cpum_cf", "cpum_cf_diag", 6); The first parameter is the pmu name taken from the event in the json file. The second parameter is the pmu name of the PMU currently being built. They are different, but the length of the compare only tests the common prefix and this returns 0(true) when it should return false. Now all events for PMU cpum_cf are added to the alias list for pmu cpum_cf_diag. Later on in function parse_events_add_pmu() the event 'tx_c_end' is searched in all available PMUs and found twice, adding it two times to the evsel_list global variable which is the root of all events. This results in a counter value of 2 instead of 1. Output with this patch: [root@s35lp76 perf]# ./perf stat -e tx_c_tend \ -- ~/mytests/cf-tx-events 1 Measuring transactions TX_C_TABORT_NO_SPECIAL: 0 expected:0 TX_C_TABORT_SPECIAL: 0 expected:0 TX_C_TEND: 1 expected:1 TX_NC_TABORT: 11 expected:11 TX_NC_TEND: 1 expected:1 Performance counter stats for '/root/mytests/cf-tx-events 1': 1 tx_c_tend 0.001815365 seconds time elapsed 0.000123000 seconds user 0.001756000 seconds sys [root@s35lp76 perf]# Signed-off-by:Thomas Richter <tmricht@linux.ibm.com> Reviewed-by:
Hendrik Brueckner <brueckner@linux.ibm.com> Reviewed-by:
Sebastien Boisvert <sboisvert@gydle.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: stable@vger.kernel.org Fixes: 292c34c1 ("perf pmu: Fix core PMU alias list for X86 platform") Link: http://lkml.kernel.org/r/20181023151616.78193-1-tmricht@linux.ibm.comSigned-off-by:
-
Andi Kleen authored
Implement a weak group fallback for 'perf record', similar to the existing 'perf stat' support. This allows to use groups that might be longer than the available counters without failing. Before: $ perf record -e '{cycles,cache-misses,cache-references,cpu_clk_unhalted.thread,cycles,cycles,cycles}' -a sleep 1 Error: The sys_perf_event_open() syscall returned with 22 (Invalid argument) for event (cycles). /bin/dmesg | grep -i perf may provide additional information. After: $ ./perf record -e '{cycles,cache-misses,cache-references,cpu_clk_unhalted.thread,cycles,cycles,cycles}:W' -a sleep 1 WARNING: No sample_id_all support, falling back to unordered processing [ perf record: Woken up 3 times to write data ] [ perf record: Captured and wrote 8.136 MB perf.data (134069 samples) ] Signed-off-by:Andi Kleen <ak@linux.intel.com> Acked-by:
Jiri Olsa <jolsa@kernel.org> Link: http://lkml.kernel.org/r/20181001195927.14211-2-andi@firstfloor.orgSigned-off-by:
-
Andi Kleen authored
- Move the function from builtin-stat to evlist for reuse - Rename to evlist to match purpose better - Pass the evlist as first argument. - No functional changes Signed-off-by:
-
Arnaldo Carvalho de Melo authored
This is the start of having the raw_syscalls:sys_enter BPF handler collecting pointer arguments, namely pathnames, and with two syscalls that have that pointer in different arguments, "open" as it as its first argument, "openat" as the second. With this in place the existing beautifiers in 'perf trace' works, those args are shown instead of just the pointer that comes with the syscalls tracepoints. This also serves to show and document pitfalls in the process of using just that place in the kernel (raw_syscalls:sys_enter) plus tables provided by userspace to collect syscall pointer arguments. One is the need to use a barrier, as suggested by Edward, to avoid clang optimizations that make the kernel BPF verifier to refuse loading our pointer contents collector. The end result should be a generic eBPF program that works in all architectures, with the differences amongst archs resolved by the userspace component, 'perf trace', that should get all its tables created automatically from the kernel components where they are defined, via string table constructors for things not expressed in BTF/DWARF (enums, structs, etc), and otherwise using those observability files (BTF). Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexei Starovoitov <ast@kernel.org> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: David Ahern <dsahern@gmail.com> Cc: Edward Cree <ecree@solarflare.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Martin KaFai Lau <kafai@fb.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Cc: Yonghong Song <yhs@fb.com> Link: https://lkml.kernel.org/n/tip-37dz54pmotgpnwg9tb6zuk9j@git.kernel.orgSigned-off-by:
-
- 03 Nov, 2018 1 commit
-
-
Arnaldo Carvalho de Melo authored
For now with BPF raw_augmented we hook into raw_syscalls:sys_enter and there we get all 6 syscall args plus the tracepoint common fields (sizeof(long)) and the syscall_nr (another long). So we check if that is the case and if so don't look after the sc->args_size, but always after the full raw_syscalls:sys_enter payload, which is fixed. We'll revisit this later to pass s->args_size to the BPF augmenter (now tools/perf/examples/bpf/augmented_raw_syscalls.c, so that it copies only what we need for each syscall, like what happens when we use syscalls:sys_enter_NAME, so that we reduce the kernel/userspace traffic to just what is needed for each syscall. 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: https://lkml.kernel.org/n/tip-nlslrg8apxdsobt4pwl3n7ur@git.kernel.orgSigned-off-by:
-
- 01 Nov, 2018 3 commits
-
-
Arnaldo Carvalho de Melo authored
With just this commit we get to support all syscalls via hooking raw_syscalls:sys_{enter,exit} to the trace__sys_{enter,exit} routines to combine, strace-like, those tracepoints. # trace -e tools/perf/examples/bpf/augmented_raw_syscalls.c sleep 1 ? ( ): sleep/31680 ... [continued]: execve()) = 0 0.043 ( 0.004 ms): sleep/31680 brk() = 0x55652a851000 0.070 ( 0.009 ms): sleep/31680 access(filename:, mode: R) = -1 ENOENT No such file or directory 0.087 ( 0.006 ms): sleep/31680 openat(dfd: CWD, filename: , flags: CLOEXEC) = 3 0.096 ( 0.003 ms): sleep/31680 fstat(fd: 3, statbuf: 0x7ffc5269e190) = 0 0.101 ( 0.005 ms): sleep/31680 mmap(len: 103334, prot: READ, flags: PRIVATE, fd: 3) = 0x7f709c239000 0.109 ( 0.002 ms): sleep/31680 close(fd: 3) = 0 0.126 ( 0.006 ms): sleep/31680 openat(dfd: CWD, filename: , flags: CLOEXEC) = 3 0.135 ( 0.003 ms): sleep/31680 read(fd: 3, buf: 0x7ffc5269e358, count: 832) = 832 0.141 ( 0.002 ms): sleep/31680 fstat(fd: 3, statbuf: 0x7ffc5269e1f0) = 0 0.146 ( 0.005 ms): sleep/31680 mmap(len: 8192, prot: READ|WRITE, flags: PRIVATE|ANONYMOUS) = 0x7f709c237000 0.159 ( 0.007 ms): sleep/31680 mmap(len: 3889792, prot: EXEC|READ, flags: PRIVATE|DENYWRITE, fd: 3) = 0x7f709bc79000 0.168 ( 0.009 ms): sleep/31680 mprotect(start: 0x7f709be26000, len: 2093056) = 0 0.179 ( 0.010 ms): sleep/31680 mmap(addr: 0x7f709c025000, len: 24576, prot: READ|WRITE, flags: PRIVATE|FIXED|DENYWRITE, fd: 3, off: 1753088) = 0x7f709c025000 0.196 ( 0.005 ms): sleep/31680 mmap(addr: 0x7f709c02b000, len: 14976, prot: READ|WRITE, flags: PRIVATE|FIXED|ANONYMOUS) = 0x7f709c02b000 0.210 ( 0.002 ms): sleep/31680 close(fd: 3) = 0 0.230 ( 0.002 ms): sleep/31680 arch_prctl(option: 4098, arg2: 140121632638208) = 0 0.306 ( 0.009 ms): sleep/31680 mprotect(start: 0x7f709c025000, len: 16384, prot: READ) = 0 0.338 ( 0.005 ms): sleep/31680 mprotect(start: 0x556529607000, len: 4096, prot: READ) = 0 0.348 ( 0.005 ms): sleep/31680 mprotect(start: 0x7f709c253000, len: 4096, prot: READ) = 0 0.356 ( 0.019 ms): sleep/31680 munmap(addr: 0x7f709c239000, len: 103334) = 0 0.463 ( 0.002 ms): sleep/31680 brk() = 0x55652a851000 0.468 ( 0.004 ms): sleep/31680 brk(brk: 0x55652a872000) = 0x55652a872000 0.474 ( 0.002 ms): sleep/31680 brk() = 0x55652a872000 0.484 ( 0.008 ms): sleep/31680 open(filename: , flags: CLOEXEC) = 3 0.497 ( 0.002 ms): sleep/31680 fstat(fd: 3, statbuf: 0x7f709c02aaa0) = 0 0.501 ( 0.006 ms): sleep/31680 mmap(len: 113045344, prot: READ, flags: PRIVATE, fd: 3) = 0x7f70950aa000 0.514 ( 0.002 ms): sleep/31680 close(fd: 3) = 0 0.554 (1000.140 ms): sleep/31680 nanosleep(rqtp: 0x7ffc5269eed0) = 0 1000.734 ( 0.007 ms): sleep/31680 close(fd: 1) = 0 1000.748 ( 0.004 ms): sleep/31680 close(fd: 2) = 0 1000.769 ( ): sleep/31680 exit_group() # Now to allow selecting which syscalls should be traced, using a map. 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: https://lkml.kernel.org/n/tip-votqqmqhag8e1i9mgyzfez3o@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
The previous approach of attaching to each syscall showed how it is possible to augment tracepoints and use that augmentation, pointer payloads, in the existing beautifiers in 'perf trace', but for a more general solution we now will try to augment the main raw_syscalls:sys_{enter,exit} syscalls, and then pass instructions in maps so that it knows which syscalls and which pointer contents, and how many bytes for each of the arguments should be copied. Start with just the bare minimum to collect what is provided by those two tracepoints via the __augmented_syscalls__ map + bpf-output perf event, which results in perf trace showing them without connecting enter+exit: # perf trace -e tools/perf/examples/bpf/augmented_raw_syscalls.c sleep 1 0.000 sleep/11563 raw_syscalls:sys_exit:NR 59 = 0 0.019 ( ): sleep/11563 brk() ... 0.021 sleep/11563 raw_syscalls:sys_exit:NR 12 = 94682642325504 0.033 ( ): sleep/11563 access(filename:, mode: R) ... 0.037 sleep/11563 raw_syscalls:sys_exit:NR 21 = -2 0.041 ( ): sleep/11563 openat(dfd: CWD, filename: , flags: CLOEXEC) ... 0.044 sleep/11563 raw_syscalls:sys_exit:NR 257 = 3 0.045 ( ): sleep/11563 fstat(fd: 3, statbuf: 0x7ffdbf7119b0) ... 0.046 sleep/11563 raw_syscalls:sys_exit:NR 5 = 0 0.047 ( ): sleep/11563 mmap(len: 103334, prot: READ, flags: PRIVATE, fd: 3) ... 0.049 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196285493248 0.050 ( ): sleep/11563 close(fd: 3) ... 0.051 sleep/11563 raw_syscalls:sys_exit:NR 3 = 0 0.059 ( ): sleep/11563 openat(dfd: CWD, filename: , flags: CLOEXEC) ... 0.062 sleep/11563 raw_syscalls:sys_exit:NR 257 = 3 0.063 ( ): sleep/11563 read(fd: 3, buf: 0x7ffdbf711b78, count: 832) ... 0.065 sleep/11563 raw_syscalls:sys_exit:NR 0 = 832 0.066 ( ): sleep/11563 fstat(fd: 3, statbuf: 0x7ffdbf711a10) ... 0.067 sleep/11563 raw_syscalls:sys_exit:NR 5 = 0 0.068 ( ): sleep/11563 mmap(len: 8192, prot: READ|WRITE, flags: PRIVATE|ANONYMOUS) ... 0.070 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196285485056 0.073 ( ): sleep/11563 mmap(len: 3889792, prot: EXEC|READ, flags: PRIVATE|DENYWRITE, fd: 3) ... 0.076 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196279463936 0.077 ( ): sleep/11563 mprotect(start: 0x7f81fd8a8000, len: 2093056) ... 0.083 sleep/11563 raw_syscalls:sys_exit:NR 10 = 0 0.084 ( ): sleep/11563 mmap(addr: 0x7f81fdaa7000, len: 24576, prot: READ|WRITE, flags: PRIVATE|FIXED|DENYWRITE, fd: 3, off: 1753088) ... 0.088 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196283314176 0.091 ( ): sleep/11563 mmap(addr: 0x7f81fdaad000, len: 14976, prot: READ|WRITE, flags: PRIVATE|FIXED|ANONYMOUS) ... 0.093 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196283338752 0.097 ( ): sleep/11563 close(fd: 3) ... 0.098 sleep/11563 raw_syscalls:sys_exit:NR 3 = 0 0.107 ( ): sleep/11563 arch_prctl(option: 4098, arg2: 140196285490432) ... 0.108 sleep/11563 raw_syscalls:sys_exit:NR 158 = 0 0.143 ( ): sleep/11563 mprotect(start: 0x7f81fdaa7000, len: 16384, prot: READ) ... 0.146 sleep/11563 raw_syscalls:sys_exit:NR 10 = 0 0.157 ( ): sleep/11563 mprotect(start: 0x561d037e7000, len: 4096, prot: READ) ... 0.160 sleep/11563 raw_syscalls:sys_exit:NR 10 = 0 0.163 ( ): sleep/11563 mprotect(start: 0x7f81fdcd5000, len: 4096, prot: READ) ... 0.165 sleep/11563 raw_syscalls:sys_exit:NR 10 = 0 0.166 ( ): sleep/11563 munmap(addr: 0x7f81fdcbb000, len: 103334) ... 0.174 sleep/11563 raw_syscalls:sys_exit:NR 11 = 0 0.216 ( ): sleep/11563 brk() ... 0.217 sleep/11563 raw_syscalls:sys_exit:NR 12 = 94682642325504 0.217 ( ): sleep/11563 brk(brk: 0x561d05453000) ... 0.219 sleep/11563 raw_syscalls:sys_exit:NR 12 = 94682642460672 0.220 ( ): sleep/11563 brk() ... 0.221 sleep/11563 raw_syscalls:sys_exit:NR 12 = 94682642460672 0.224 ( ): sleep/11563 open(filename: , flags: CLOEXEC) ... 0.228 sleep/11563 raw_syscalls:sys_exit:NR 2 = 3 0.229 ( ): sleep/11563 fstat(fd: 3, statbuf: 0x7f81fdaacaa0) ... 0.230 sleep/11563 raw_syscalls:sys_exit:NR 5 = 0 0.231 ( ): sleep/11563 mmap(len: 113045344, prot: READ, flags: PRIVATE, fd: 3) ... 0.234 sleep/11563 raw_syscalls:sys_exit:NR 9 = 140196166418432 0.237 ( ): sleep/11563 close(fd: 3) ... 0.238 sleep/11563 raw_syscalls:sys_exit:NR 3 = 0 0.262 ( ): sleep/11563 nanosleep(rqtp: 0x7ffdbf7126f0) ... 1000.399 sleep/11563 raw_syscalls:sys_exit:NR 35 = 0 1000.440 ( ): sleep/11563 close(fd: 1) ... 1000.447 sleep/11563 raw_syscalls:sys_exit:NR 3 = 0 1000.454 ( ): sleep/11563 close(fd: 2) ... 1000.468 ( ): sleep/11563 exit_group( ) # In the next csets we'll connect those events to the existing enter/exit raw_syscalls handlers in 'perf trace', just like we did with the syscalls:sys_{enter,exit}_* tracepoints. 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: https://lkml.kernel.org/n/tip-5nl8l4hx1tl9pqdx65nkp6pw@git.kernel.orgSigned-off-by: -
Will Deacon authored
Cheers for reporting this. I managed to reproduce the build failure with gcc version 6.3.0 20170516 (Debian 6.3.0-18+deb9u1). The code in question is the arm64 versions of smp_load_acquire() and smp_store_release(). Unlike other architectures, these are not built around READ_ONCE() and WRITE_ONCE() since we have instructions we can use instead of fences. Bringing our macros up-to-date with those (i.e. tweaking the union initialisation and using the special "uXX_alias_t" types) appears to fix the issue for me. Committer notes: Testing it in the systems previously failing: # time dm android-ndk:r12b-arm \ android-ndk:r15c-arm \ debian:experimental-x-arm64 \ ubuntu:14.04.4-x-linaro-arm64 \ ubuntu:16.04-x-arm \ ubuntu:16.04-x-arm64 \ ubuntu:18.04-x-arm \ ubuntu:18.04-x-arm64 1 android-ndk:r12b-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 2 android-ndk:r15c-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 3 debian:experimental-x-arm64 : Ok aarch64-linux-gnu-gcc (Debian 8.2.0-7) 8.2.0 4 ubuntu:14.04.4-x-linaro-arm64 : Ok aarch64-linux-gnu-gcc (Linaro GCC 5.5-2017.10) 5.5.0 5 ubuntu:16.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 6 ubuntu:16.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 7 ubuntu:18.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 7.3.0-27ubuntu1~18.04) 7.3.0 8 ubuntu:18.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 7.3.0-27ubuntu1~18.04) 7.3.0 Reported-by:Will Deacon <will.deacon@arm.com> Tested-by:
Daniel Borkmann <daniel@iogearbox.net> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20181031174408.GA27871@arm.comSigned-off-by:
-
- 31 Oct, 2018 19 commits
-
-
Ingo Molnar authored
Merge tag 'perf-urgent-for-mingo-4.20-20181031' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent Pull perf/urgent improvements and fixes from Arnaldo Carvalho de Melo: - Fixes dealing with the removal of the fallback to looking up samples marked as userspace in the kernel maps, done recently: - For intel-pt, that was setting the synthesized header misc field as PERF_RECORD_MISC_USER, depending thus on the fallback to take place, now it sets as USER or KERNEL according to x86 specific knowledge. Also now it inserts the PERF_CONTEXT_{USER,KERNEL} into the PERF_SAMPLE_CALLCHAINs it synthesizes from hw traces (Adrian Hunter) - Similar fixes for the cs-etm ARM HW trace code, that used the Intel PT model as a starting point (Leo Yan) - For the "caller" callchain order, where the callchain returned by the kernel was simply reversed without taking into account the PERF_CONTEXT_{USER,KERNEL,etc} markers from where to define if an entry was for kernel or userspace, working just because the map lookup fallback was in place (David S. Miller) - Allow for selecting if 'overwrite' mode should be used in 'perf top' and make the default for it not to be used. This is due to problems with the current implementation where the pausing used ends up making 'perf top' miss PERF_RECORD_{MMAP,FORK,EXEC,etc} events, which with short lifetime threads workloads leads quickly to many "unknown" maps (and thus symbols) to appear in the UI. Workloads with long thread lifetimes and with few metadata events can still use --overwrite to take advantage of the overwrite mode (Arnaldo Carvalho de Melo) - Start 'perf top''s display thread earlier, so that the screen doesn't remain blank for too long at tool start (David S. Miller) - Don't clone maps from parent when synthesizing forks, to avoid the inevitable flurry of overlapping maps as we process the synthesized MMAP2 events that get delivered shortly thereafter. (David S. Miller) - Take pgoff into account when reporting elf to libdwfl, now the unwinding results are the same with elfutils's libdwfl and libunwind (Milian Wolff) - Update lotsa kernel ABI headers (Arnaldo Carvalho de Melo) - 'perf trace' syscall arg beautification improvements to allow for handling args such as mount's 'flags', where maks have to be ignored before considering what is left, that, if only zeroes, is suppressed like other args without such masks (Arnaldo Carvalho de Melo) - Beautify mount's 'source' and 'flags' args (Arnaldo Carvalho de Melo) - Generate mmap's flags bit constants from linux/mman.h and all the arch specific mman.h files, so that no changes in the main 'perf trace' source files is required when new flags get added (Arnaldo Carvalho de Melo) - Consider syscall aliases, so that 'perf trace -e umount' works and we don't have to use 'umount2' (that works as well, just not required) (Arnaldo Carvalho de Melo) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Adrian Hunter authored
In the absence of a fallback, samples must provide a correct cpumode for the 'ip'. Do that now there is no fallback. Signed-off-by:
-
Adrian Hunter authored
In the absence of a fallback, callchains must encode also the callchain context. Do that now there is no fallback. Signed-off-by:
-
David Miller authored
When synthesizing FORK events, we are trying to create thread objects for the already running tasks on the machine. Normally, for a kernel FORK event, we want to clone the parent's maps because that is what the kernel just did. But when synthesizing, this should not be done. If we do, we end up with overlapping maps as we process the sythesized MMAP2 events that get delivered shortly thereafter. Use the FORK event misc flags in an internal way to signal this situation, so we can elide the map clone when appropriate. Signed-off-by:
David S. Miller <davem@davemloft.net> Cc: Don Zickus <dzickus@redhat.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Joe Mario <jmario@redhat.com> Link: http://lkml.kernel.org/r/20181030.222404.2085088822877051075.davem@davemloft.net [ Added comment about flag use in machine__process_fork_event(), use ternary op in thread__clone_map_groups() as suggested by Jiri ] Signed-off-by:
-
David Miller authored
If events are coming in at a rate such that the event processing thread can barely keep up, our initial run of the event ring will almost never terminate and this delays the starting of the display thread. The screen basically stays black until the event thread can get out of it's endless loop. Therefore, start the display thread before we start processing the ring buffer. This also make sure that we always have the user requested real time setting engaged when processing the ring. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
To pick the changes from: 9163a0fc ("net: bridge: add support for per-port vlan stats") And silence this build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/if_link.h' differs from latest version at 'include/uapi/linux/if_link.h' Cc: Alexei Starovoitov <ast@kernel.org> Cc: David S. Miller <davem@davemloft.net> Cc: Eric Leblond <eric@regit.org> Cc: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> Link: https://lkml.kernel.org/n/tip-7p53ghippywz7fqkwo3nkzet@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Picking the changes from: 89d35528 ("netlink: Add new socket option to enable strict checking on dumps") To silence this build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/netlink.h' differs from latest version at 'include/uapi/linux/netlink.h' Cc: Alexei Starovoitov <ast@kernel.org> Cc: David Ahern <dsahern@gmail.com> Cc: David S. Miller <davem@davemloft.net> Cc: Eric Leblond <eric@regit.org> Link: https://lkml.kernel.org/n/tip-1xymkfjpmhxfzrs46t8z8mjw@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
For powerpc, s390, x86 and the main uapi linux/kvm.h header, none of them entail changes in tooling. Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-avn7iy8f4tcm2y40sbsdk31m@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To pick up the changes from: 20916d46 ("mm/hugetlb: add mmap() encodings for 32MB and 512MB page sizes") That do not entail changes in in tools, this just shows that we have to consider bits [26:31] of flags to beautify that in tools like 'perf trace' This silences this perf build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/mman.h' differs from latest version at 'include/uapi/linux/mman.h' diff -u tools/include/uapi/linux/mman.h include/uapi/linux/mman.h Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-3rvc39lon93kgt5pl31d8g4x@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Instead of requiring us to go on and edit sources to add new flag. # perf trace -e *mmap sleep 0.1 0.025 ( 0.005 ms): sleep/29876 mmap(len: 163746, prot: READ, flags: PRIVATE, fd: 3) = 0x7faa68ad1000 0.059 ( 0.004 ms): sleep/29876 mmap(len: 8192, prot: READ|WRITE, flags: PRIVATE|ANONYMOUS) = 0x7faa68acf000 0.069 ( 0.006 ms): sleep/29876 mmap(len: 3889792, prot: EXEC|READ, flags: PRIVATE|DENYWRITE, fd: 3) = 0x7faa6851f000 0.086 ( 0.009 ms): sleep/29876 mmap(addr: 0x7faa688cb000, len: 24576, prot: READ|WRITE, flags: PRIVATE|FIXED|DENYWRITE, fd: 3, off: 1753088) = 0x7faa688cb000 0.101 ( 0.005 ms): sleep/29876 mmap(addr: 0x7faa688d1000, len: 14976, prot: READ|WRITE, flags: PRIVATE|FIXED|ANONYMOUS) = 0x7faa688d1000 0.348 ( 0.005 ms): sleep/29876 mmap(len: 111950656, prot: READ, flags: PRIVATE, fd: 3) = 0x7faa61a5b000 # 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: https://lkml.kernel.org/n/tip-ggmoy6vxoygh5yim890ht0kf@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
Now when we run 'make -C tools/perf O=/tmp/build/perf' we end up with: $ cat /tmp/build/perf/trace/beauty/generated/mmap_flags_array.c static const char *mmap_flags[] = { [ilog2(0x40) + 1] = "32BIT", [ilog2(0x01) + 1] = "SHARED", [ilog2(0x02) + 1] = "PRIVATE", [ilog2(0x10) + 1] = "FIXED", [ilog2(0x20) + 1] = "ANONYMOUS", [ilog2(0x100000) + 1] = "FIXED_NOREPLACE", [ilog2(0x0100) + 1] = "GROWSDOWN", [ilog2(0x0800) + 1] = "DENYWRITE", [ilog2(0x1000) + 1] = "EXECUTABLE", [ilog2(0x2000) + 1] = "LOCKED", [ilog2(0x4000) + 1] = "NORESERVE", [ilog2(0x8000) + 1] = "POPULATE", [ilog2(0x10000) + 1] = "NONBLOCK", [ilog2(0x20000) + 1] = "STACK", [ilog2(0x40000) + 1] = "HUGETLB", [ilog2(0x80000) + 1] = "SYNC", }; $ 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: https://lkml.kernel.org/n/tip-t3fn7u3tjsupio6e6vkufx9m@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
It'll use tools/{arch}/*,include copies of mman.h to generate a table to be used by tools, initially by the 'mmap' beautifiers in 'perf trace', but that could also be used to translate from a string constant to the integer value to be used in a eBPF or tracefs tracepoint filter. Tested for all archs using: $ for arch in `ls tools/arch/` ; \ do echo $arch ; tools/perf/trace/beauty/mmap_flags.sh $arch ; \ done | less Example for alpha, an oddball, doesn't include any header, defines all its stuff: $ tools/perf/trace/beauty/mmap_flags.sh alpha static const char *mmap_flags[] = { [ilog2(0x10) + 1] = "ANONYMOUS", [ilog2(0x02000) + 1] = "DENYWRITE", [ilog2(0x04000) + 1] = "EXECUTABLE", [ilog2(0x100) + 1] = "FIXED", [ilog2(0x01000) + 1] = "GROWSDOWN", [ilog2(0x100000) + 1] = "HUGETLB", [ilog2(0x08000) + 1] = "LOCKED", [ilog2(0x40000) + 1] = "NONBLOCK", [ilog2(0x10000) + 1] = "NORESERVE", [ilog2(0x20000) + 1] = "POPULATE", [ilog2(0x02) + 1] = "PRIVATE", [ilog2(0x01) + 1] = "SHARED", [ilog2(0x80000) + 1] = "STACK", }; $ Common case, my workstation, defines one entry (MAP_32BIT), then includes mman.h, which gets it to include mman-common.h too: $ tools/perf/trace/beauty/mmap_flags.sh static const char *mmap_flags[] = { [ilog2(0x40) + 1] = "32BIT", [ilog2(0x01) + 1] = "SHARED", [ilog2(0x02) + 1] = "PRIVATE", [ilog2(0x10) + 1] = "FIXED", [ilog2(0x20) + 1] = "ANONYMOUS", [ilog2(0x100000) + 1] = "FIXED_NOREPLACE", [ilog2(0x0100) + 1] = "GROWSDOWN", [ilog2(0x0800) + 1] = "DENYWRITE", [ilog2(0x1000) + 1] = "EXECUTABLE", [ilog2(0x2000) + 1] = "LOCKED", [ilog2(0x4000) + 1] = "NORESERVE", [ilog2(0x8000) + 1] = "POPULATE", [ilog2(0x10000) + 1] = "NONBLOCK", [ilog2(0x20000) + 1] = "STACK", [ilog2(0x40000) + 1] = "HUGETLB", [ilog2(0x80000) + 1] = "SYNC", }; $ uname -m x86_64 $ Sparc, that defines a bunch then includes just mman-common.h: $ tools/perf/trace/beauty/mmap_flags.sh sparc static const char *mmap_flags[] = { [ilog2(0x0800) + 1] = "DENYWRITE", [ilog2(0x1000) + 1] = "EXECUTABLE", [ilog2(0x0200) + 1] = "GROWSDOWN", [ilog2(0x40000) + 1] = "HUGETLB", [ilog2(0x100) + 1] = "LOCKED", [ilog2(0x10000) + 1] = "NONBLOCK", [ilog2(0x40) + 1] = "NORESERVE", [ilog2(0x8000) + 1] = "POPULATE", [ilog2(0x20000) + 1] = "STACK", [ilog2(0x01) + 1] = "SHARED", [ilog2(0x02) + 1] = "PRIVATE", [ilog2(0x10) + 1] = "FIXED", [ilog2(0x20) + 1] = "ANONYMOUS", [ilog2(0x100000) + 1] = "FIXED_NOREPLACE", }; [acme@jouet perf]$ 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: https://lkml.kernel.org/n/tip-xydeh491z8fkgglcmqnl5thj@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
To silence this perf build warning: Warning: Kernel ABI header at 'tools/include/uapi/sound/asound.h' differs from latest version at 'include/uapi/sound/asound.h' diff -u tools/include/uapi/sound/asound.h include/uapi/sound/asound.h Due to this cset: a9840151 ("ALSA: timer: fix wrong comment to refer to 'SNDRV_TIMER_PSFLG_*'") Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Takashi Sakamoto <o-takashi@sakamocchi.jp> Cc: Takashi Iwai <tiwai@suse.de> Link: https://lkml.kernel.org/n/tip-76gsvs0w2g0x723ivqa2xua3@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To get the changes in: 82b355d1 ("y2038: Remove newstat family from default syscall set") Which will make the syscall table used by 'perf trace' for arm64 to be updated from the changes in that patch. This silences these perf build warnings: Warning: Kernel ABI header at 'tools/arch/arm64/include/uapi/asm/unistd.h' differs from latest version at 'arch/arm64/include/uapi/asm/unistd.h' diff -u tools/arch/arm64/include/uapi/asm/unistd.h arch/arm64/include/uapi/asm/unistd.h Warning: Kernel ABI header at 'tools/include/uapi/asm-generic/unistd.h' differs from latest version at 'include/uapi/asm-generic/unistd.h' diff -u tools/include/uapi/asm-generic/unistd.h include/uapi/asm-generic/unistd.h Cc: Kim Phillips <kim.phillips@arm.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lkml.kernel.org/n/tip-3euy7c4yy5mvnp5bm16t9vqg@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To silence this perf build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/fs.h' differs from latest version at 'include/uapi/linux/fs.h' diff -u tools/include/uapi/linux/fs.h include/uapi/linux/fs.h Due to just two comments added by: Fixes: 578bdaab ("crypto: speck - remove Speck") So nothing that entails changes in tools/, that so far uses fs.h to generate the mount and umount syscalls 'flags' argument integer->string tables with: $ tools/perf/trace/beauty/mount_flags.sh static const char *mount_flags[] = { [4096 ? (ilog2(4096) + 1) : 0] = "BIND", <SNIP> [30 + 1] = "ACTIVE", [31 + 1] = "NOUSER", }; $ # trace -e mount,umount mount --bind /proc /mnt 1.228 ( 2.581 ms): mount/1068 mount(dev_name: /mnt, dir_name: 0x55f011c354a0, type: 0x55f011c38170, flags: BIND) = 0 # trace -e mount,umount umount /proc /mnt umount: /proc: target is busy. 1.587 ( 0.010 ms): umount/1070 umount2(name: /proc) = -1 EBUSY Device or resource busy 1.799 (12.660 ms): umount/1070 umount2(name: /mnt) = 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> Cc: Jason A. Donenfeld <Jason@zx2c4.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Link: https://lkml.kernel.org/n/tip-c00bqzclscgah26z2g5zxm73@git.kernel.orgSigned-off-by:
-
David S. Miller authored
When processing using 'perf report -g caller', which is the default, we ended up reverting the callchain entries received from the kernel, but simply reverting throws away the information that tells that from a point onwards the addresses are for userspace, kernel, guest kernel, guest user, hypervisor. The idea is that if we are walking backwards, for each cluster of non-cpumode entries we have to first scan backwards for the next one and use that for the cluster. This seems silly and more expensive than it needs to be but it is enough for a initial fix. The code here is really complicated because it is intimately intertwined with the lbr and branch handling, as well as this callchain order, further fixes will be needed to properly take into account the cpumode in those cases. Another problem with ORDER_CALLER is that the NULL "0" IP that is at the end of most callchains shows up at the top of the histogram because every callchain contains it and with ORDER_CALLER it is the first entry. Signed-off-by:
-
Leo Yan authored
Since commit edeb0c90 ("perf tools: Stop fallbacking to kallsyms for vdso symbols lookup"), the kernel address cannot be properly parsed to kernel symbol with command 'perf script -k vmlinux'. The reason is CoreSight samples is always to set CPU mode as PERF_RECORD_MISC_USER, thus it fails to find corresponding map/dso in below flows: process_sample_event() `-> machine__resolve() `-> thread__find_map(thread, sample->cpumode, sample->ip, al); In this flow it needs to pass argument 'sample->cpumode' to tell what's the CPU mode, before it always passed PERF_RECORD_MISC_USER but without any failure until the commit edeb0c90 ("perf tools: Stop fallbacking to kallsyms for vdso symbols lookup") has been merged. The reason is even with the wrong CPU mode the function thread__find_map() firstly fails to find map but it will rollback to find kernel map for vdso symbols lookup. In the latest code it has removed the fallback code, thus if CPU mode is PERF_RECORD_MISC_USER then it cannot find map anymore with kernel address. This patch is to correct samples CPU mode setting, it creates a new helper function cs_etm__cpu_mode() to tell what's the CPU mode based on the address with the info from machine structure; this patch has a bit extension to check not only kernel and user mode, but also check for host/guest and hypervisor mode. Finally this patch uses the function in instruction and branch samples and also apply in cs_etm__mem_access() for a minor polishing. Signed-off-by:
Leo Yan <leo.yan@linaro.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: David Miller <davem@davemloft.net> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: stable@kernel.org # v4.19 Link: http://lkml.kernel.org/r/1540883908-17018-1-git-send-email-leo.yan@linaro.orgSigned-off-by:
-
Milian Wolff authored
libdwfl parses an ELF file itself and creates mappings for the individual sections. perf on the other hand sees raw mmap events which represent individual sections. When we encounter an address pointing into a mapping with pgoff != 0, we must take that into account and report the file at the non-offset base address. This fixes unwinding with libdwfl in some cases. E.g. for a file like: ``` using namespace std; mutex g_mutex; double worker() { lock_guard<mutex> guard(g_mutex); uniform_real_distribution<double> uniform(-1E5, 1E5); default_random_engine engine; double s = 0; for (int i = 0; i < 1000; ++i) { s += norm(complex<double>(uniform(engine), uniform(engine))); } cout << s << endl; return s; } int main() { vector<std::future<double>> results; for (int i = 0; i < 10000; ++i) { results.push_back(async(launch::async, worker)); } return 0; } ``` Compile it with `g++ -g -O2 -lpthread cpp-locking.cpp -o cpp-locking`, then record it with `perf record --call-graph dwarf -e sched:sched_switch`. When you analyze it with `perf script` and libunwind, you should see: ``` cpp-locking 20038 [005] 54830.236589: sched:sched_switch: prev_comm=cpp-locking prev_pid=20038 prev_prio=120 prev_state=T ==> next_comm=swapper/5 next_pid=0 next_prio=120 ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1670208 schedule+0x28 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb16737cc rwsem_down_read_failed+0xec (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1665e04 call_rwsem_down_read_failed+0x14 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1672a03 down_read+0x13 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb106bd85 __do_page_fault+0x445 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb18015f5 page_fault+0x45 (/lib/modules/4.14.78-1-lts/build/vmlinux) 7f38e4252591 new_heap+0x101 (/usr/lib/libc-2.28.so) 7f38e4252d0b arena_get2.part.4+0x2fb (/usr/lib/libc-2.28.so) 7f38e4255b1c tcache_init.part.6+0xec (/usr/lib/libc-2.28.so) 7f38e42569e5 __GI___libc_malloc+0x115 (inlined) 7f38e4241790 __GI__IO_file_doallocate+0x90 (inlined) 7f38e424fbbf __GI__IO_doallocbuf+0x4f (inlined) 7f38e424ee47 __GI__IO_file_overflow+0x197 (inlined) 7f38e424df36 _IO_new_file_xsputn+0x116 (inlined) 7f38e4242bfb __GI__IO_fwrite+0xdb (inlined) 7f38e463fa6d std::basic_streambuf<char, std::char_traits<char> >::sputn(char const*, long)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> >::_M_put(char const*, long)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> > std::__write<char>(std::ostreambuf_iterator<char, std::char_traits<char> >, char const*, int)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> > std::num_put<char, std::ostreambuf_iterator<char, std::char_traits<char> > >::_M_insert_float<double>(std::ostreambuf_iterator<c> 7f38e464bd70 std::num_put<char, std::ostreambuf_iterator<char, std::char_traits<char> > >::put(std::ostreambuf_iterator<char, std::char_traits<char> >, std::ios_base&, char, double) const+0x90 (inl> 7f38e464bd70 std::ostream& std::ostream::_M_insert<double>(double)+0x90 (/usr/lib/libstdc++.so.6.0.25) 563b9cb502f7 std::ostream::operator<<(double)+0xb7 (inlined) 563b9cb502f7 worker()+0xb7 (/ssd/milian/projects/kdab/rnd/hotspot/build/tests/test-clients/cpp-locking/cpp-locking) 563b9cb506fb double std::__invoke_impl<double, double (*)()>(std::__invoke_other, double (*&&)())+0x2b (inlined) 563b9cb506fb std::__invoke_result<double (*)()>::type std::__invoke<double (*)()>(double (*&&)())+0x2b (inlined) 563b9cb506fb decltype (__invoke((_S_declval<0ul>)())) std::thread::_Invoker<std::tuple<double (*)()> >::_M_invoke<0ul>(std::_Index_tuple<0ul>)+0x2b (inlined) 563b9cb506fb std::thread::_Invoker<std::tuple<double (*)()> >::operator()()+0x2b (inlined) 563b9cb506fb std::__future_base::_Task_setter<std::unique_ptr<std::__future_base::_Result<double>, std::__future_base::_Result_base::_Deleter>, std::thread::_Invoker<std::tuple<double (*)()> >, dou> 563b9cb506fb std::_Function_handler<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> (), std::__future_base::_Task_setter<std::unique_ptr<std::__future_> 563b9cb507e8 std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>::operator()() const+0x28 (inlined) 563b9cb507e8 std::__future_base::_State_baseV2::_M_do_set(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>*, bool*)+0x28 (/ssd/milian/> 7f38e46d24fe __pthread_once_slow+0xbe (/usr/lib/libpthread-2.28.so) 563b9cb51149 __gthread_once+0xe9 (inlined) 563b9cb51149 void std::call_once<void (std::__future_base::_State_baseV2::*)(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>*, bool*)> 563b9cb51149 std::__future_base::_State_baseV2::_M_set_result(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>, bool)+0xe9 (inlined) 563b9cb51149 std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_impl(std::thread::_Invoker<std::tuple<double (*)()> >&&)::{lambda()#1}::op> 563b9cb51149 void std::__invoke_impl<void, std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_impl(std::thread::_Invoker<std::tuple<double> 563b9cb51149 std::__invoke_result<std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_impl(std::thread::_Invoker<std::tuple<double (*)()> >> 563b9cb51149 decltype (__invoke((_S_declval<0ul>)())) std::thread::_Invoker<std::tuple<std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_> 563b9cb51149 std::thread::_Invoker<std::tuple<std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_impl(std::thread::_Invoker<std::tuple<dou> 563b9cb51149 std::thread::_State_impl<std::thread::_Invoker<std::tuple<std::__future_base::_Async_state_impl<std::thread::_Invoker<std::tuple<double (*)()> >, double>::_Async_state_impl(std::thread> 7f38e45f0062 execute_native_thread_routine+0x12 (/usr/lib/libstdc++.so.6.0.25) 7f38e46caa9c start_thread+0xfc (/usr/lib/libpthread-2.28.so) 7f38e42ccb22 __GI___clone+0x42 (inlined) ``` Before this patch, using libdwfl, you would see: ``` cpp-locking 20038 [005] 54830.236589: sched:sched_switch: prev_comm=cpp-locking prev_pid=20038 prev_prio=120 prev_state=T ==> next_comm=swapper/5 next_pid=0 next_prio=120 ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1670208 schedule+0x28 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb16737cc rwsem_down_read_failed+0xec (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1665e04 call_rwsem_down_read_failed+0x14 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1672a03 down_read+0x13 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb106bd85 __do_page_fault+0x445 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb18015f5 page_fault+0x45 (/lib/modules/4.14.78-1-lts/build/vmlinux) 7f38e4252591 new_heap+0x101 (/usr/lib/libc-2.28.so) a041161e77950c5c [unknown] ([unknown]) ``` With this patch applied, we get a bit further in unwinding: ``` cpp-locking 20038 [005] 54830.236589: sched:sched_switch: prev_comm=cpp-locking prev_pid=20038 prev_prio=120 prev_state=T ==> next_comm=swapper/5 next_pid=0 next_prio=120 ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb166fec5 __sched_text_start+0x545 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1670208 schedule+0x28 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb16737cc rwsem_down_read_failed+0xec (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1665e04 call_rwsem_down_read_failed+0x14 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb1672a03 down_read+0x13 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb106bd85 __do_page_fault+0x445 (/lib/modules/4.14.78-1-lts/build/vmlinux) ffffffffb18015f5 page_fault+0x45 (/lib/modules/4.14.78-1-lts/build/vmlinux) 7f38e4252591 new_heap+0x101 (/usr/lib/libc-2.28.so) 7f38e4252d0b arena_get2.part.4+0x2fb (/usr/lib/libc-2.28.so) 7f38e4255b1c tcache_init.part.6+0xec (/usr/lib/libc-2.28.so) 7f38e42569e5 __GI___libc_malloc+0x115 (inlined) 7f38e4241790 __GI__IO_file_doallocate+0x90 (inlined) 7f38e424fbbf __GI__IO_doallocbuf+0x4f (inlined) 7f38e424ee47 __GI__IO_file_overflow+0x197 (inlined) 7f38e424df36 _IO_new_file_xsputn+0x116 (inlined) 7f38e4242bfb __GI__IO_fwrite+0xdb (inlined) 7f38e463fa6d std::basic_streambuf<char, std::char_traits<char> >::sputn(char const*, long)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> >::_M_put(char const*, long)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> > std::__write<char>(std::ostreambuf_iterator<char, std::char_traits<char> >, char const*, int)+0x1cd (inlined) 7f38e463fa6d std::ostreambuf_iterator<char, std::char_traits<char> > std::num_put<char, std::ostreambuf_iterator<char, std::char_traits<char> > >::_M_insert_float<double>(std::ostreambuf_iterator<c> 7f38e464bd70 std::num_put<char, std::ostreambuf_iterator<char, std::char_traits<char> > >::put(std::ostreambuf_iterator<char, std::char_traits<char> >, std::ios_base&, char, double) const+0x90 (inl> 7f38e464bd70 std::ostream& std::ostream::_M_insert<double>(double)+0x90 (/usr/lib/libstdc++.so.6.0.25) 563b9cb502f7 std::ostream::operator<<(double)+0xb7 (inlined) 563b9cb502f7 worker()+0xb7 (/ssd/milian/projects/kdab/rnd/hotspot/build/tests/test-clients/cpp-locking/cpp-locking) 6eab825c1ee3e4ff [unknown] ([unknown]) ``` Note that the backtrace is still stopping too early, when compared to the nice results obtained via libunwind. It's unclear so far what the reason for that is. Committer note: Further comment by Milian on the thread started on the Link: tag below: --- The remaining issue is due to a bug in elfutils: https://sourceware.org/ml/elfutils-devel/2018-q4/msg00089.html With both patches applied, libunwind and elfutils produce the same output for the above scenario. --- Signed-off-by:Milian Wolff <milian.wolff@kdab.com> Acked-by:
-
Arnaldo Carvalho de Melo authored
Enabling --overwrite mode allows us to to use just the most recent records, which helps in high core count machines such as Knights Landing/Mill, but right now is being disabled by default as the pausing used in this technique is leading to loss of metadata events such as PERF_RECORD_MMAP which makes 'perf top' unable to resolve samples, leading to lots of unknown samples appearing on the UI. Enabling this may be useful if you are in such machines and profiling a workload that doesn't creates short lived threads and/or doesn't uses many executable mmap operations. Work is being planed to solve this situation, till then, this will remain disabled by default. Reported-by:
Kan Liang <kan.liang@intel.com> Link: https://lkml.kernel.org/r/4f84468f-37d9-cf1b-12c1-514ef74b6a48@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> Fixes: ebebbf08 ("perf top: Switch default mode to overwrite mode") Link: https://lkml.kernel.org/n/tip-ehvf77vi1si9409r7p4wx788@git.kernel.orgSigned-off-by:
-
- 30 Oct, 2018 10 commits
-
-
Arnaldo Carvalho de Melo authored
In ebebbf08 ("perf top: Switch default mode to overwrite mode") we forgot to leave a way to disable that new default, add a --overwrite option that can be disabled using --no-overwrite, since the code already in such a way that we can readily disable this mode. This is useful when investigating bugs with this mode like the recent report from David Miller where lots of unknown symbols appear due to disabling the events while processing them which disables all record types, not just PERF_RECORD_SAMPLE, which makes it impossible to resolve maps when we lose PERF_RECORD_MMAP records. This can be easily seen while building a kernel, when there are lots of short lived processes. Reported-by:
-
Arnaldo Carvalho de Melo authored
The pathname beautifiers so far support just one augmented pathname per syscall, so do it just for mount's first arg, later this will get fixed. With: # perf probe -l probe:vfs_getname (on getname_flags:73@acme/git/linux/fs/namei.c with pathname) # Later this will get added to augmented_syscalls.c (eBPF): In one xterm: # perf trace -e mount,umount 2687.331 ( 3.544 ms): mount/8892 mount(dev_name: /mnt, dir_name: 0x561f9ac184a0, type: 0x561f9ac1b170, flags: BIND) = 0 3912.126 ( 8.807 ms): umount/8895 umount2(name: /mnt) = 0 ^C# In the other: $ sudo mount --bind /proc /mnt $ sudo umount /mnt Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-qsvhrm2es635cl4zicqjeth2@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
By using the SCA_FILENAME beautifier, that works when either the probe:vfs_getname probe is in place or with the eBPF program tools/perf/examples/bpf/augmented_syscalls.c: # perf probe -l probe:vfs_getname (on getname_flags:73@acme/git/linux/fs/namei.c with pathname) # perf trace -e umount 9630.332 ( 9.521 ms): umount/8082 umount2(name: /mnt) = 0 # The augmented syscalls one will be done in the next patch. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-hegbzlpd2nrn584l5jxn7sy2@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
When trying to trace the 'umount' syscall on x86_64 I noticed that it was failing: # trace -e umount umount /mnt event syntax error: 'umount' \___ parser error Run 'perf list' for a list of valid events Usage: perf trace [<options>] [<command>] or: perf trace [<options>] -- <command> [<options>] or: perf trace record [<options>] [<command>] or: perf trace record [<options>] -- <command> [<options>] -e, --event <event> event/syscall selector. use 'perf list' to list available events # This is because in the x86-64 we have it just as 'umount2': $ grep umount arch/x86/entry/syscalls/syscall_64.tbl 166 common umount2 __x64_sys_umount $ So if the syscall name fails, try fallbacking to looking at the aliases we have in the syscall_fmts table to then re-lookup, now: # trace -e umount umount -f /mnt umount: /mnt: not mounted. 1.759 ( 0.004 ms): umount/18365 umount2(name: 0x55fbfcbc4480, flags: 1) = -1 EINVAL Invalid argument # Time to beautify the flags arg :-) Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-ukweodgzbmjd25lfkgryeft1@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
# trace -e mount mount -o ro -t debugfs nodev /mnt 0.000 ( 1.040 ms): mount/27235 mount(dev_name: 0x5601cc8c64e0, dir_name: 0x5601cc8c6500, type: 0x5601cc8c6480, flags: RDONLY) = 0 # trace -e mount mount -o remount,relatime -t debugfs nodev /mnt 0.000 ( 2.946 ms): mount/27262 mount(dev_name: 0x55f4a73d64e0, dir_name: 0x55f4a73d6500, type: 0x55f4a73d6480, flags: REMOUNT|RELATIME) = 0 # trace -e mount mount -o remount,strictatime -t debugfs nodev /mnt 0.000 ( 2.934 ms): mount/27265 mount(dev_name: 0x5617f71d94e0, dir_name: 0x5617f71d9500, type: 0x5617f71d9480, flags: REMOUNT|STRICTATIME) = 0 # trace -e mount mount -o remount,suid,silent -t debugfs nodev /mnt 0.000 ( 0.049 ms): mount/27273 mount(dev_name: 0x55ad65df24e0, dir_name: 0x55ad65df2500, type: 0x55ad65df2480, flags: REMOUNT|SILENT) = 0 # trace -e mount mount -o remount,rw,sync,lazytime -t debugfs nodev /mnt 0.000 ( 2.684 ms): mount/27281 mount(dev_name: 0x561216055530, dir_name: 0x561216055550, type: 0x561216055510, flags: SYNCHRONOUS|REMOUNT|LAZYTIME) = 0 # trace -e mount mount -o remount,dirsync -t debugfs nodev /mnt 0.000 ( 3.512 ms): mount/27314 mount(dev_name: 0x55c4e7188480, dir_name: 0x55c4e7188530, type: 0x55c4e71884a0, flags: REMOUNT|DIRSYNC, data: 0x55c4e71884e0) = 0 # Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-i5ncao73c0bd02qprgrq6wb9@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
Take mount's 'flags' arg, to cope with this semantic, as defined in do_mount in fs/namespace.c: /* * Pre-0.97 versions of mount() didn't have a flags word. When the * flags word was introduced its top half was required to have the * magic value 0xC0ED, and this remained so until 2.4.0-test9. * Therefore, if this magic number is present, it carries no * information and must be discarded. */ We need to mask this arg, and then see if it is zero, when we simply don't print the arg name and value. The next patch will use this for mount's 'flag' arg. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-btue14k5jemayuykfrwsnh85@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Generalizing pkey_alloc__scnprintf_access_rights(), so that we can use it with other flags-like arguments, such as mount's mountflags argument. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-o3ymi3104m8moaz9865g09w9@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
The intention is to have this as a library, since it is not perf specific at all. I did the switch for the files where I'm the only contributor, with the exception of a few lines changed by Jiri Olsa. Acked-by:
-
Arnaldo Carvalho de Melo authored
It'll use tools/include copy of linux/fs.h to generate a table to be used by tools, initially by the 'mount' and 'umount' beautifiers in 'perf trace', but that could also be used to translate from a string constant to the integer value to be used in a eBPF or tracefs tracepoint filter. When used without any args it produces: $ tools/perf/trace/beauty/mount_flags.sh static const char *mount_flags[] = { [1 ? (ilog2(1) + 1) : 0] = "RDONLY", [2 ? (ilog2(2) + 1) : 0] = "NOSUID", [4 ? (ilog2(4) + 1) : 0] = "NODEV", [8 ? (ilog2(8) + 1) : 0] = "NOEXEC", [16 ? (ilog2(16) + 1) : 0] = "SYNCHRONOUS", [32 ? (ilog2(32) + 1) : 0] = "REMOUNT", [64 ? (ilog2(64) + 1) : 0] = "MANDLOCK", [128 ? (ilog2(128) + 1) : 0] = "DIRSYNC", [1024 ? (ilog2(1024) + 1) : 0] = "NOATIME", [2048 ? (ilog2(2048) + 1) : 0] = "NODIRATIME", [4096 ? (ilog2(4096) + 1) : 0] = "BIND", [8192 ? (ilog2(8192) + 1) : 0] = "MOVE", [16384 ? (ilog2(16384) + 1) : 0] = "REC", [32768 ? (ilog2(32768) + 1) : 0] = "SILENT", [16 + 1] = "POSIXACL", [17 + 1] = "UNBINDABLE", [18 + 1] = "PRIVATE", [19 + 1] = "SLAVE", [20 + 1] = "SHARED", [21 + 1] = "RELATIME", [22 + 1] = "KERNMOUNT", [23 + 1] = "I_VERSION", [24 + 1] = "STRICTATIME", [25 + 1] = "LAZYTIME", [26 + 1] = "SUBMOUNT", [27 + 1] = "NOREMOTELOCK", [28 + 1] = "NOSEC", [29 + 1] = "BORN", [30 + 1] = "ACTIVE", [31 + 1] = "NOUSER", }; $ Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-mgutbbkmip9gfnmd28ikg7xt@git.kernel.orgSigned-off-by: -
Arnaldo Carvalho de Melo authored
We'll use it to create tables for the 'flags' argument to the 'mount' and 'umount' syscalls. Add it to check_headers.sh so that when a new protocol gets added we get a notification during the build process. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Benjamin Peterson <benjamin@python.org> 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: https://lkml.kernel.org/n/tip-yacf9jvkwfwg2g95r2us3xb3@git.kernel.orgSigned-off-by:
-
