user/sven/linux.git/tools/perf, branch v6.1.41

perf tool x86: Fix perf_env memory leak

2023-07-19T14:21:38Z

[ Upstream commit 99d4850062a84564f36923764bb93935ef2ed108 ] Found by leak sanitizer: ``` ==1632594==ERROR: LeakSanitizer: detected memory leaks Direct leak of 21 byte(s) in 1 object(s) allocated from: #0 0x7f2953a7077b in __interceptor_strdup ../../../../src/libsanitizer/asan/asan_interceptors.cpp:439 #1 0x556701d6fbbf in perf_env__read_cpuid util/env.c:369 #2 0x556701d70589 in perf_env__cpuid util/env.c:465 #3 0x55670204bba2 in x86__is_amd_cpu arch/x86/util/env.c:14 #4 0x5567020487a2 in arch__post_evsel_config arch/x86/util/evsel.c:83 #5 0x556701d8f78b in evsel__config util/evsel.c:1366 #6 0x556701ef5872 in evlist__config util/record.c:108 #7 0x556701cd6bcd in test__PERF_RECORD tests/perf-record.c:112 #8 0x556701cacd07 in run_test tests/builtin-test.c:236 #9 0x556701cacfac in test_and_print tests/builtin-test.c:265 #10 0x556701cadddb in __cmd_test tests/builtin-test.c:402 #11 0x556701caf2aa in cmd_test tests/builtin-test.c:559 #12 0x556701d3b557 in run_builtin tools/perf/perf.c:323 #13 0x556701d3bac8 in handle_internal_command tools/perf/perf.c:377 #14 0x556701d3be90 in run_argv tools/perf/perf.c:421 #15 0x556701d3c3f8 in main tools/perf/perf.c:537 #16 0x7f2952a46189 in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58 SUMMARY: AddressSanitizer: 21 byte(s) leaked in 1 allocation(s). ``` Fixes: f7b58cbdb3ff36eb ("perf mem/c2c: Add load store event mappings for AMD") Signed-off-by: Ian Rogers Acked-by: Ravi Bangoria Tested-by: Arnaldo Carvalho de Melo Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Link: https://lore.kernel.org/r/20230613235416.1650755-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf tool x86: Consolidate is_amd check into single function

2023-07-19T14:21:38Z

[ Upstream commit 0cd1ca4650c9cf5f318110f67d39cbebae3693b3 ] There are multiple places where x86 specific code determines AMD vs Intel arch and acts based on that. Consolidate those checks into a single function. Signed-off-by: Ravi Bangoria Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Ali Saidi Cc: Ananth Narayan Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Leo Yan Cc: Madhavan Srinivasan Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Sandipan Das Cc: Santosh Shukla Link: https://lore.kernel.org/r/20230613095506.547-3-ravi.bangoria@amd.com Signed-off-by: Arnaldo Carvalho de Melo Stable-dep-of: 99d4850062a8 ("perf tool x86: Fix perf_env memory leak") Signed-off-by: Sasha Levin

perf dwarf-aux: Fix off-by-one in die_get_varname()

2023-07-19T14:21:38Z

[ Upstream commit 3abfcfd847717d232e36963f31a361747c388fe7 ] The die_get_varname() returns "(unknown_type)" string if it failed to find a type for the variable. But it had a space before the opening parenthesis and it made the closing parenthesis cut off due to the off-by-one in the string length (14). Signed-off-by: Namhyung Kim Fixes: 88fd633cdfa19060 ("perf probe: No need to use formatting strbuf method") Cc: Adrian Hunter Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Masami Hiramatsu Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20230612234102.3909116-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf script: Fix allocation of evsel->priv related to per-event dump files

2023-07-19T14:21:37Z

[ Upstream commit 36d3e4138e1b6cc9ab179f3f397b5548f8b1eaae ] When printing output we may want to generate per event files, where the --per-event-dump option should be used, creating perf.data.EVENT.dump files instead of printing to stdout. The callback thar processes event thus expects that evsel->priv->fp should point to either the per-event FILE descriptor or to stdout. The a3af66f51bd0bca7 ("perf script: Fix crash because of missing evsel->priv") changeset fixed a case where evsel->priv wasn't setup, thus set to NULL, causing a segfault when trying to access evsel->priv->fp. But it did it for the non --per-event-dump case by allocating a 'struct perf_evsel_script' just to set its ->fp to stdout. Since evsel->priv is only freed when --per-event-dump is used, we ended up with a memory leak, detected using ASAN. Fix it by using the same method as perf_script__setup_per_event_dump(), and reuse that static 'struct perf_evsel_script'. Also check if evsel_script__new() failed. Fixes: a3af66f51bd0bca7 ("perf script: Fix crash because of missing evsel->priv") Reported-by: Ian Rogers Tested-by: Ian Rogers Cc: Adrian Hunter Cc: Jiri Olsa Cc: Namhyung Kim Cc: Ravi Bangoria Link: https://lore.kernel.org/lkml/ZH+F0wGAWV14zvMP@kernel.org Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf bench: Add missing setlocale() call to allow usage of %'d style formatting

2023-07-19T14:21:36Z

[ Upstream commit 16203e9cd01896b4244100a8e3fb9f6e612ab2b1 ] Without this we were not getting the thousands separator for big numbers. Noticed while developing 'perf bench uprobe', but the use of %' predates that, for instance 'perf bench syscall' uses it. Before: # perf bench uprobe all # Running uprobe/baseline benchmark... # Executed 1000 usleep(1000) calls Total time: 1054082243ns 1054082.243000 nsecs/op # After: # perf bench uprobe all # Running uprobe/baseline benchmark... # Executed 1,000 usleep(1000) calls Total time: 1,053,715,144ns 1,053,715.144000 nsecs/op # Fixes: c2a08203052f8975 ("perf bench: Add basic syscall benchmark") Cc: Adrian Hunter Cc: Andre Fredette Cc: Clark Williams Cc: Dave Tucker Cc: Davidlohr Bueso Cc: Derek Barbosa Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Cc: Tiezhu Yang Link: https://lore.kernel.org/lkml/ZH3lcepZ4tBYr1jv@kernel.org Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf symbols: Symbol lookup with kcore can fail if multiple segments match stext

2023-07-05T17:27:37Z

commit 1c249565426e3a9940102c0ba9f63914f7cda73d upstream. This problem was encountered on an arm64 system with a lot of memory. Without kernel debug symbols installed, and with both kcore and kallsyms available, perf managed to get confused and returned "unknown" for all of the kernel symbols that it tried to look up. On this system, stext fell within the vmalloc segment. The kcore symbol matching code tries to find the first segment that contains stext and uses that to replace the segment generated from just the kallsyms information. In this case, however, there were two: a very large vmalloc segment, and the text segment. This caused perf to get confused because multiple overlapping segments were inserted into the RB tree that holds the discovered segments. However, that alone wasn't sufficient to cause the problem. Even when we could find the segment, the offsets were adjusted in such a way that the newly generated symbols didn't line up with the instruction addresses in the trace. The most obvious solution would be to consult which segment type is text from kcore, but this information is not exposed to users. Instead, select the smallest matching segment that contains stext instead of the first matching segment. This allows us to match the text segment instead of vmalloc, if one is contained within the other. Reviewed-by: Adrian Hunter Signed-off-by: Krister Johansen Cc: Alexander Shishkin Cc: David Reaver Cc: Ian Rogers Cc: Jiri Olsa Cc: Mark Rutland Cc: Michael Petlan Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lore.kernel.org/lkml/20230125183418.GD1963@templeofstupid.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Krister Johansen Signed-off-by: Greg Kroah-Hartman

perf ftrace latency: Remove unnecessary "--" from --use-nsec option

2023-06-09T08:34:02Z

[ Upstream commit 8d73259ef23f449329294dc187932f7470268126 ] The option name should not have the dashes. Current version shows four dashes for the option. $ perf ftrace latency -h Usage: perf ftrace [] [] or: perf ftrace [] -- [] [] or: perf ftrace {trace|latency} [] [] or: perf ftrace {trace|latency} [] -- [] [] -b, --use-bpf Use BPF to measure function latency -n, ----use-nsec Use nano-second histogram -T, --trace-funcs Show latency of given function Fixes: 84005bb6148618cc ("perf ftrace latency: Add -n/--use-nsec option") Signed-off-by: Namhyung Kim Tested-by: Arnaldo Carvalho de Melo Cc: Adrian Hunter Cc: Changbin Du Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20230525212038.3535851-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf stat: Separate bperf from bpf_profiler

2023-05-17T09:53:41Z

[ Upstream commit ecc68ee216c6c5b2f84915e1441adf436f1b019b ] It seems that perf stat -b doesn't produce any results: $ perf stat -e cycles -b 4 -I 10000 -vvv Control descriptor is not initialized cycles: 0 0 0 time counts unit events 10.007641640 cycles Looks like this happens because fentry/fexit progs are getting loaded, but the corresponding perf event is not enabled and not added into the events bpf map. I think there is some mixing up between two type of bpf support, one for bperf and one for bpf_profiler. Both are identified via evsel__is_bpf, based on which perf events are enabled, but for the latter (bpf_profiler) a perf event is required. Using evsel__is_bperf to check only bperf produces expected results: $ perf stat -e cycles -b 4 -I 10000 -vvv Control descriptor is not initialized ------------------------------------------------------------ perf_event_attr: size 136 sample_type IDENTIFIER read_format TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING disabled 1 exclude_guest 1 ------------------------------------------------------------ sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 3 ------------------------------------------------------------ [...perf_event_attr for other CPUs...] ------------------------------------------------------------ cycles: 309426 169009 169009 time counts unit events 10.010091271 309426 cycles The final numbers correspond (at least in the level of magnitude) to the same metric obtained via bpftool. Fixes: 112cb56164bc2108 ("perf stat: Introduce config stat.bpf-counter-events") Reviewed-by: Song Liu Signed-off-by: Dmitrii Dolgov <9erthalion6@gmail.com> Tested-by: Song Liu Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Namhyung Kim Cc: Song Liu Link: https://lore.kernel.org/r/20230412182316.11628-1-9erthalion6@gmail.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf tracepoint: Fix memory leak in is_valid_tracepoint()

2023-05-17T09:53:41Z

[ Upstream commit 9b86c49710eec7b4fbb78a0232b2dd0972a2b576 ] When is_valid_tracepoint() returns 1, need to call put_events_file() to free `dir_path`. Fixes: 25a7d914274de386 ("perf parse-events: Use get/put_events_file()") Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20230421025953.173826-1-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf symbols: Fix return incorrect build_id size in elf_read_build_id()

2023-05-17T09:53:40Z

[ Upstream commit 1511e4696acb715a4fe48be89e1e691daec91c0e ] In elf_read_build_id(), if gnu build_id is found, should return the size of the actually copied data. If descsz is greater thanBuild_ID_SIZE, write_buildid data access may occur. Fixes: be96ea8ffa788dcc ("perf symbols: Fix issue with binaries using 16-bytes buildids (v2)") Reported-by: Will Ochowicz Signed-off-by: Yang Jihong Tested-by: Will Ochowicz Acked-by: Adrian Hunter Cc: Alexander Shishkin Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Leo Yan Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Link: https://lore.kernel.org/lkml/CWLP265MB49702F7BA3D6D8F13E4B1A719C649@CWLP265MB4970.GBRP265.PROD.OUTLOOK.COM/T/ Link: https://lore.kernel.org/r/20230427012841.231729-1-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin