From 100198322b2eb7fb38750cb0fcae5cd533907410 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 1 Feb 2022 12:53:16 -0300 Subject: perf beauty: Make the prctl arg regexp more strict to cope with PR_SET_VMA This new PR_SET_VMA value isn't in sequence with all the other prctl arguments and instead uses a big, 0x prefixed hex number: 0x53564d41 (S V M A). This makes it harder to generate a string table as it would be rather sparse, so make the regexp more stricter to avoid catching those. A followup patch for 'perf trace' to cope with such oddities will be needed, but then its a matter for the next merge window. The next patch will update the prctl.h file to cope with this perf build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/prctl.h' differs from latest version at 'include/uapi/linux/prctl.h' diff -u tools/include/uapi/linux/prctl.h include/uapi/linux/prctl.h Here is the output of this script: $ tools/perf/trace/beauty/prctl_option.sh static const char *prctl_options[] = { [1] = "SET_PDEATHSIG", [2] = "GET_PDEATHSIG", [3] = "GET_DUMPABLE", [4] = "SET_DUMPABLE", [5] = "GET_UNALIGN", [6] = "SET_UNALIGN", [7] = "GET_KEEPCAPS", [8] = "SET_KEEPCAPS", [9] = "GET_FPEMU", [10] = "SET_FPEMU", [11] = "GET_FPEXC", [12] = "SET_FPEXC", [13] = "GET_TIMING", [14] = "SET_TIMING", [15] = "SET_NAME", [16] = "GET_NAME", [19] = "GET_ENDIAN", [20] = "SET_ENDIAN", [21] = "GET_SECCOMP", [22] = "SET_SECCOMP", [23] = "CAPBSET_READ", [24] = "CAPBSET_DROP", [25] = "GET_TSC", [26] = "SET_TSC", [27] = "GET_SECUREBITS", [28] = "SET_SECUREBITS", [29] = "SET_TIMERSLACK", [30] = "GET_TIMERSLACK", [31] = "TASK_PERF_EVENTS_DISABLE", [32] = "TASK_PERF_EVENTS_ENABLE", [33] = "MCE_KILL", [34] = "MCE_KILL_GET", [35] = "SET_MM", [36] = "SET_CHILD_SUBREAPER", [37] = "GET_CHILD_SUBREAPER", [38] = "SET_NO_NEW_PRIVS", [39] = "GET_NO_NEW_PRIVS", [40] = "GET_TID_ADDRESS", [41] = "SET_THP_DISABLE", [42] = "GET_THP_DISABLE", [43] = "MPX_ENABLE_MANAGEMENT", [44] = "MPX_DISABLE_MANAGEMENT", [45] = "SET_FP_MODE", [46] = "GET_FP_MODE", [47] = "CAP_AMBIENT", [50] = "SVE_SET_VL", [51] = "SVE_GET_VL", [52] = "GET_SPECULATION_CTRL", [53] = "SET_SPECULATION_CTRL", [54] = "PAC_RESET_KEYS", [55] = "SET_TAGGED_ADDR_CTRL", [56] = "GET_TAGGED_ADDR_CTRL", [57] = "SET_IO_FLUSHER", [58] = "GET_IO_FLUSHER", [59] = "SET_SYSCALL_USER_DISPATCH", [60] = "PAC_SET_ENABLED_KEYS", [61] = "PAC_GET_ENABLED_KEYS", [62] = "SCHED_CORE", }; static const char *prctl_set_mm_options[] = { [1] = "START_CODE", [2] = "END_CODE", [3] = "START_DATA", [4] = "END_DATA", [5] = "START_STACK", [6] = "START_BRK", [7] = "BRK", [8] = "ARG_START", [9] = "ARG_END", [10] = "ENV_START", [11] = "ENV_END", [12] = "AUXV", [13] = "EXE_FILE", [14] = "MAP", [15] = "MAP_SIZE", }; $ Cc: Adrian Hunter Cc: Colin Cross Cc: Ian Rogers Cc: Jiri Olsa Cc: Kees Cook Cc: Namhyung Kim Cc: Suren Baghdasaryan Link: https://lore.kernel.org/lkml/YflZqY0rYQ3d1bKt@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/trace/beauty/prctl_option.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf') diff --git a/tools/perf/trace/beauty/prctl_option.sh b/tools/perf/trace/beauty/prctl_option.sh index 3109d7b05e11..3d278785fe57 100755 --- a/tools/perf/trace/beauty/prctl_option.sh +++ b/tools/perf/trace/beauty/prctl_option.sh @@ -4,7 +4,7 @@ [ $# -eq 1 ] && header_dir=$1 || header_dir=tools/include/uapi/linux/ printf "static const char *prctl_options[] = {\n" -regex='^#define[[:space:]]+PR_(\w+)[[:space:]]*([[:xdigit:]]+).*' +regex='^#define[[:space:]]{1}PR_(\w+)[[:space:]]*([[:xdigit:]]+)([[:space:]]*\/.*)?$' egrep $regex ${header_dir}/prctl.h | grep -v PR_SET_PTRACER | \ sed -r "s/$regex/\2 \1/g" | \ sort -n | xargs printf "\t[%s] = \"%s\",\n" -- cgit v1.2.3 From ae65b443f03fca3620cb37c5e019ddca3f89a1ce Mon Sep 17 00:00:00 2001 From: Anshuman Khandual Date: Wed, 2 Feb 2022 16:27:23 +0530 Subject: perf tools: Add missing branch_sample_type to perf_event_attr__fprintf() This updates branch sample type with missing PERF_SAMPLE_BRANCH_TYPE_SAVE. Suggested-by: James Clark Signed-off-by: Anshuman Khandual Acked-by: Jiri Olsa Cc: James Clark Cc: Mark Rutland Cc: Peter Zijlstra Cc: linux-arm-kernel@lists.infradead.org Link: http://lore.kernel.org/lkml/1643799443-15109-1-git-send-email-anshuman.khandual@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/perf_event_attr_fprintf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf') diff --git a/tools/perf/util/perf_event_attr_fprintf.c b/tools/perf/util/perf_event_attr_fprintf.c index 47b7531f51da..98af3fa4ea35 100644 --- a/tools/perf/util/perf_event_attr_fprintf.c +++ b/tools/perf/util/perf_event_attr_fprintf.c @@ -52,7 +52,7 @@ static void __p_branch_sample_type(char *buf, size_t size, u64 value) bit_name(ABORT_TX), bit_name(IN_TX), bit_name(NO_TX), bit_name(COND), bit_name(CALL_STACK), bit_name(IND_JUMP), bit_name(CALL), bit_name(NO_FLAGS), bit_name(NO_CYCLES), - bit_name(HW_INDEX), + bit_name(TYPE_SAVE), bit_name(HW_INDEX), { .name = NULL, } }; #undef bit_name -- cgit v1.2.3 From a663520fcc4bce2814032e3de6c4e2665b9555e5 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 1 Feb 2022 23:08:25 -0800 Subject: perf annotate: Set error stream of objdump process for TUI The stderr should be set to a pipe when using TUI. Otherwise it'd print to stdout and break TUI windows with an error message. Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: Ian Rogers Cc: Jiri Olsa Cc: Peter Zijlstra Link: http://lore.kernel.org/lkml/20220202070828.143303-2-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 01900689dc00..8190a124b99d 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -2036,6 +2036,7 @@ static int symbol__disassemble(struct symbol *sym, struct annotate_args *args) memset(&objdump_process, 0, sizeof(objdump_process)); objdump_process.argv = objdump_argv; objdump_process.out = -1; + objdump_process.err = -1; if (start_command(&objdump_process)) { pr_err("Failure starting to run %s\n", command); err = -1; -- cgit v1.2.3 From d792a7a94c2c3ba045247266bee2e2bced7b495a Mon Sep 17 00:00:00 2001 From: Ameer Hamza Date: Tue, 25 Jan 2022 17:11:41 +0500 Subject: perf session: Check for NULL pointer before dereference Move NULL pointer check before dereferencing the variable. Addresses-Coverity: 1497622 ("Derereference before null check") Reviewed-by: James Clark Signed-off-by: Ameer Hamza Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexey Bayduraev Cc: German Gomez Cc: Ingo Molnar Cc: Jiri Olsa Cc: Leo Yan Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Riccardo Mancini Link: https://lore.kernel.org/r/20220125121141.18347-1-amhamza.mgc@gmail.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'tools/perf') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 2c0d30f08e78..498b05708db5 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1503,11 +1503,12 @@ static int machines__deliver_event(struct machines *machines, ++evlist->stats.nr_unknown_id; return 0; } - dump_sample(evsel, event, sample, perf_env__arch(machine->env)); if (machine == NULL) { ++evlist->stats.nr_unprocessable_samples; + dump_sample(evsel, event, sample, perf_env__arch(NULL)); return 0; } + dump_sample(evsel, event, sample, perf_env__arch(machine->env)); return evlist__deliver_sample(evlist, tool, event, sample, evsel, machine); case PERF_RECORD_MMAP: return tool->mmap(tool, event, sample, machine); -- cgit v1.2.3 From bc9c806e524429a4b98de257179af5e3fc2cb57d Mon Sep 17 00:00:00 2001 From: Leo Yan Date: Fri, 24 Dec 2021 20:40:13 +0800 Subject: perf synthetic-events: Return error if procfs isn't mounted for PID namespaces For perf recording, it retrieves process info by iterating nodes in proc fs. If we run perf in a non-root PID namespace with command: # unshare --fork --pid perf record -e cycles -a -- test_program ... in this case, unshare command creates a child PID namespace and launches perf tool in it, but the issue is the proc fs is not mounted for the non-root PID namespace, this leads to the perf tool gathering process info from its parent PID namespace. We can use below command to observe the process nodes under proc fs: # unshare --pid --fork ls /proc 1 137 1968 2128 3 342 48 62 78 crypto kcore net uptime 10 138 2 2142 30 35 49 63 8 devices keys pagetypeinfo version 11 139 20 2143 304 36 50 64 82 device-tree key-users partitions vmallocinfo 12 14 2011 22 305 37 51 65 83 diskstats kmsg self vmstat 128 140 2038 23 307 39 52 656 84 driver kpagecgroup slabinfo zoneinfo 129 15 2074 24 309 4 53 67 9 execdomains kpagecount softirqs 13 16 2094 241 31 40 54 68 asound fb kpageflags stat 130 164 2096 242 310 41 55 69 buddyinfo filesystems loadavg swaps 131 17 2098 25 317 42 56 70 bus fs locks sys 132 175 21 26 32 43 57 71 cgroups interrupts meminfo sysrq-trigger 133 179 2102 263 329 44 58 75 cmdline iomem misc sysvipc 134 1875 2103 27 330 45 59 76 config.gz ioports modules thread-self 135 19 2117 29 333 46 6 77 consoles irq mounts timer_list 136 1941 2121 298 34 47 60 773 cpuinfo kallsyms mtd tty So it shows many existed tasks, since unshared command has not mounted the proc fs for the new created PID namespace, it still accesses the proc fs of the root PID namespace. This leads to two prominent issues: - Firstly, PID values are mismatched between thread info and samples. The gathered thread info are coming from the proc fs of the root PID namespace, but samples record its PID from the child PID namespace. - The second issue is profiled program 'test_program' returns its forked PID number from the child PID namespace, perf tool wrongly uses this PID number to retrieve the process info via the proc fs of the root PID namespace. To avoid issues, we need to mount proc fs for the child PID namespace with the option '--mount-proc' when use unshare command: # unshare --fork --pid --mount-proc perf record -e cycles -a -- test_program Conversely, when the proc fs of the root PID namespace is used by child namespace, perf tool can detect the multiple PID levels and nsinfo__is_in_root_namespace() returns false, this patch reports error for this case: # unshare --fork --pid perf record -e cycles -a -- test_program Couldn't synthesize bpf events. Perf runs in non-root PID namespace but it tries to gather process info from its parent PID namespace. Please mount the proc file system properly, e.g. add the option '--mount-proc' for unshare command. Reviewed-by: James Clark Signed-off-by: Leo Yan Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andrii Nakryiko Cc: Daniel Borkmann Cc: Ian Rogers Cc: Ingo Molnar Cc: Jiri Olsa Cc: John Fastabend Cc: KP Singh Cc: Mark Rutland Cc: Martin KaFai Lau Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Song Liu Cc: Yonghong Song Link: https://lore.kernel.org/r/20211224124014.2492751-1-leo.yan@linaro.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/synthetic-events.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'tools/perf') diff --git a/tools/perf/util/synthetic-events.c b/tools/perf/util/synthetic-events.c index 70f095624a0b..b654de0841f8 100644 --- a/tools/perf/util/synthetic-events.c +++ b/tools/perf/util/synthetic-events.c @@ -1784,6 +1784,25 @@ int __machine__synthesize_threads(struct machine *machine, struct perf_tool *too perf_event__handler_t process, bool needs_mmap, bool data_mmap, unsigned int nr_threads_synthesize) { + /* + * When perf runs in non-root PID namespace, and the namespace's proc FS + * is not mounted, nsinfo__is_in_root_namespace() returns false. + * In this case, the proc FS is coming for the parent namespace, thus + * perf tool will wrongly gather process info from its parent PID + * namespace. + * + * To avoid the confusion that the perf tool runs in a child PID + * namespace but it synthesizes thread info from its parent PID + * namespace, returns failure with warning. + */ + if (!nsinfo__is_in_root_namespace()) { + pr_err("Perf runs in non-root PID namespace but it tries to "); + pr_err("gather process info from its parent PID namespace.\n"); + pr_err("Please mount the proc file system properly, e.g. "); + pr_err("add the option '--mount-proc' for unshare command.\n"); + return -EPERM; + } + if (target__has_task(target)) return perf_event__synthesize_thread_map(tool, threads, process, machine, needs_mmap, data_mmap); -- cgit v1.2.3 From a2887b9b8d1db7be971e5951e08ffe8563ea412f Mon Sep 17 00:00:00 2001 From: Masanari Iida Date: Sat, 25 Dec 2021 09:55:58 +0900 Subject: perf bpf: Fix a typo in bpf_counter_cgroup.c This patch fixes a spelling typo in error message. Signed-off-by: Masanari Iida Acked-by: Namhyung Kim Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20211225005558.503935-1-standby24x7@gmail.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_counter_cgroup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf') diff --git a/tools/perf/util/bpf_counter_cgroup.c b/tools/perf/util/bpf_counter_cgroup.c index 631e34a0b66f..ac60c08e8e2a 100644 --- a/tools/perf/util/bpf_counter_cgroup.c +++ b/tools/perf/util/bpf_counter_cgroup.c @@ -266,7 +266,7 @@ static int bperf_cgrp__read(struct evsel *evsel) idx = evsel->core.idx; err = bpf_map_lookup_elem(reading_map_fd, &idx, values); if (err) { - pr_err("bpf map lookup falied: idx=%u, event=%s, cgrp=%s\n", + pr_err("bpf map lookup failed: idx=%u, event=%s, cgrp=%s\n", idx, evsel__name(evsel), evsel->cgrp->name); goto out; } -- cgit v1.2.3 From 05b5a9d6285412d97fc61b8ec113d1d4f6b950c2 Mon Sep 17 00:00:00 2001 From: German Gomez Date: Wed, 26 Jan 2022 10:59:26 +0000 Subject: perf tools: Apply correct label to user/kernel symbols in branch mode In branch mode, the branch symbols were being displayed with incorrect cpumode labels. So fix this. For example, before: # perf record -b -a -- sleep 1 # perf report -b Overhead Command Source Shared Object Source Symbol Target Symbol 0.08% swapper [kernel.kallsyms] [k] rcu_idle_enter [k] cpuidle_enter_state ==> 0.08% cmd0 [kernel.kallsyms] [.] psi_group_change [.] psi_group_change 0.08% cmd1 [kernel.kallsyms] [k] psi_group_change [k] psi_group_change After: # perf report -b Overhead Command Source Shared Object Source Symbol Target Symbol 0.08% swapper [kernel.kallsyms] [k] rcu_idle_enter [k] cpuidle_enter_state 0.08% cmd0 [kernel.kallsyms] [k] psi_group_change [k] pei_group_change 0.08% cmd1 [kernel.kallsyms] [k] psi_group_change [k] psi_group_change Reviewed-by: James Clark Signed-off-by: German Gomez Cc: Alexander Shishkin Cc: Jiri Olsa Cc: Mark Rutland Cc: Namhyung Kim Link: https://lore.kernel.org/r/20220126105927.3411216-1-german.gomez@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/machine.c | 2 ++ tools/perf/util/map_symbol.h | 1 + tools/perf/util/sort.c | 4 ++-- 3 files changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/perf') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index f70ba56912d4..394550003693 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2073,6 +2073,7 @@ static void ip__resolve_ams(struct thread *thread, ams->addr = ip; ams->al_addr = al.addr; + ams->al_level = al.level; ams->ms.maps = al.maps; ams->ms.sym = al.sym; ams->ms.map = al.map; @@ -2092,6 +2093,7 @@ static void ip__resolve_data(struct thread *thread, ams->addr = addr; ams->al_addr = al.addr; + ams->al_level = al.level; ams->ms.maps = al.maps; ams->ms.sym = al.sym; ams->ms.map = al.map; diff --git a/tools/perf/util/map_symbol.h b/tools/perf/util/map_symbol.h index 7d22ade082c8..e08817b0c30f 100644 --- a/tools/perf/util/map_symbol.h +++ b/tools/perf/util/map_symbol.h @@ -18,6 +18,7 @@ struct addr_map_symbol { struct map_symbol ms; u64 addr; u64 al_addr; + char al_level; u64 phys_addr; u64 data_page_size; }; diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index cfba8c337783..2da081ef532b 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -915,7 +915,7 @@ static int hist_entry__sym_from_snprintf(struct hist_entry *he, char *bf, struct addr_map_symbol *from = &he->branch_info->from; return _hist_entry__sym_snprintf(&from->ms, from->al_addr, - he->level, bf, size, width); + from->al_level, bf, size, width); } return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A"); @@ -928,7 +928,7 @@ static int hist_entry__sym_to_snprintf(struct hist_entry *he, char *bf, struct addr_map_symbol *to = &he->branch_info->to; return _hist_entry__sym_snprintf(&to->ms, to->al_addr, - he->level, bf, size, width); + to->al_level, bf, size, width); } return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A"); -- cgit v1.2.3 From b2b1aa73ade982c175ac926a1fd34e76ad628b94 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 4 Feb 2022 17:09:41 -0800 Subject: perf stat: Fix display of grouped aliased events An event may have a number of uncore aliases that when added to the evlist are consecutive. If there are multiple uncore events in a group then parse_events__set_leader_for_uncore_aliase will reorder the evlist so that events on the same PMU are adjacent. The collect_all_aliases function assumes that aliases are in blocks so that only the first counter is printed and all others are marked merged. The reordering for groups breaks the assumption and so all counts are printed. This change removes the assumption from collect_all_aliases that the events are in blocks and instead processes the entire evlist. Before: ``` $ perf stat -e '{UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE,UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE},duration_time' -a -A -- sleep 1 Performance counter stats for 'system wide': CPU0 256,866 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 494,413 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 967 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,738 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 285,161 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 429,920 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 955 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,443 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 310,753 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 416,657 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,231 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,573 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 416,067 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 405,966 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,481 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,447 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 312,911 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 408,154 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,086 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,380 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 333,994 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 370,349 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,287 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,335 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 188,107 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 302,423 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 701 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,070 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 307,221 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 383,642 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,036 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,158 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 318,479 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 821,545 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,028 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 2,550 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 227,618 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 372,272 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 903 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,456 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 376,783 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 419,827 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,406 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,453 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 286,583 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 429,956 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 999 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,436 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 313,867 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 370,159 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,114 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,291 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 342,083 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 409,111 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,399 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,684 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 365,828 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 376,037 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,378 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,411 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 382,456 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 621,743 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,232 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,955 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 342,316 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 385,067 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,176 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,268 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 373,588 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 386,163 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,394 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,464 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 381,206 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 546,891 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,266 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,712 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 221,176 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 392,069 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 831 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,456 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 355,401 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 705,595 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,235 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 2,216 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 371,436 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 428,103 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,306 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,442 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 384,352 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 504,200 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,468 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,860 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 228,856 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 287,976 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 832 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,060 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 215,121 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 334,162 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 681 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,026 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 296,179 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 436,083 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,084 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,525 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 262,296 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 416,573 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 986 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,533 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 285,852 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 359,842 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,073 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,326 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 303,379 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 367,222 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,008 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,156 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 273,487 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 425,449 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 932 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,367 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 297,596 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 414,793 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,140 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,601 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 342,365 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 360,422 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,291 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,342 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 327,196 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 580,858 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,122 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 2,014 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 296,564 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 452,817 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,087 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,694 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 375,002 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 389,393 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,478 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 1,540 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 365,213 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 594,685 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 1,401 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 2,222 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 1,000,749,060 ns duration_time 1.000749060 seconds time elapsed ``` After: ``` Performance counter stats for 'system wide': CPU0 20,547,434 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU36 45,202,862 UNC_CHA_TOR_OCCUPANCY.IA_MISS_DRD_REMOTE CPU0 82,001 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU36 159,688 UNC_CHA_TOR_INSERTS.IA_MISS_DRD_REMOTE CPU0 1,000,464,828 ns duration_time 1.000464828 seconds time elapsed ``` Fixes: 3cdc5c2cb924acb4 ("perf parse-events: Handle uncore event aliases in small groups properly") Reviewed-by: Andi Kleen Signed-off-by: Ian Rogers Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Asaf Yaffe Cc: Caleb Biggers Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: John Garry Cc: Kan Liang Cc: Kshipra Bopardikar Cc: Mark Rutland Cc: Maxime Coquelin Cc: Namhyung Kim Cc: Perry Taylor Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Vineet Singh Cc: Zhengjun Xing Link: https://lore.kernel.org/r/20220205010941.1065469-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/stat-display.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) (limited to 'tools/perf') diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c index 5db83e51ceef..9cbe351b141f 100644 --- a/tools/perf/util/stat-display.c +++ b/tools/perf/util/stat-display.c @@ -585,15 +585,16 @@ static void collect_all_aliases(struct perf_stat_config *config, struct evsel *c alias = list_prepare_entry(counter, &(evlist->core.entries), core.node); list_for_each_entry_continue (alias, &evlist->core.entries, core.node) { - if (strcmp(evsel__name(alias), evsel__name(counter)) || - alias->scale != counter->scale || - alias->cgrp != counter->cgrp || - strcmp(alias->unit, counter->unit) || - evsel__is_clock(alias) != evsel__is_clock(counter) || - !strcmp(alias->pmu_name, counter->pmu_name)) - break; - alias->merged_stat = true; - cb(config, alias, data, false); + /* Merge events with the same name, etc. but on different PMUs. */ + if (!strcmp(evsel__name(alias), evsel__name(counter)) && + alias->scale == counter->scale && + alias->cgrp == counter->cgrp && + !strcmp(alias->unit, counter->unit) && + evsel__is_clock(alias) == evsel__is_clock(counter) && + strcmp(alias->pmu_name, counter->pmu_name)) { + alias->merged_stat = true; + cb(config, alias, data, false); + } } } -- cgit v1.2.3 From fceb62124d8fe1f6fb4b64e8f11c095dca8e7ea7 Mon Sep 17 00:00:00 2001 From: Changbin Du Date: Thu, 27 Jan 2022 21:20:10 +0800 Subject: perf ftrace: system_wide collection is not effective by default The ftrace.target.system_wide must be set before invoking evlist__create_maps(), otherwise it has no effect. Fixes: 53be50282269b46c ("perf ftrace: Add 'latency' subcommand") Signed-off-by: Changbin Du Acked-by: Namhyung Kim Cc: Ingo Molnar Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20220127132010.4836-1-changbin.du@gmail.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-ftrace.c | 45 ++++++++++++++++++++++++--------------------- 1 file changed, 24 insertions(+), 21 deletions(-) (limited to 'tools/perf') diff --git a/tools/perf/builtin-ftrace.c b/tools/perf/builtin-ftrace.c index dec24dc0e767..a8785dec5ca6 100644 --- a/tools/perf/builtin-ftrace.c +++ b/tools/perf/builtin-ftrace.c @@ -1115,6 +1115,7 @@ enum perf_ftrace_subcommand { int cmd_ftrace(int argc, const char **argv) { int ret; + int (*cmd_func)(struct perf_ftrace *) = NULL; struct perf_ftrace ftrace = { .tracer = DEFAULT_TRACER, .target = { .uid = UINT_MAX, }, @@ -1221,6 +1222,28 @@ int cmd_ftrace(int argc, const char **argv) goto out_delete_filters; } + switch (subcmd) { + case PERF_FTRACE_TRACE: + if (!argc && target__none(&ftrace.target)) + ftrace.target.system_wide = true; + cmd_func = __cmd_ftrace; + break; + case PERF_FTRACE_LATENCY: + if (list_empty(&ftrace.filters)) { + pr_err("Should provide a function to measure\n"); + parse_options_usage(ftrace_usage, options, "T", 1); + ret = -EINVAL; + goto out_delete_filters; + } + cmd_func = __cmd_latency; + break; + case PERF_FTRACE_NONE: + default: + pr_err("Invalid subcommand\n"); + ret = -EINVAL; + goto out_delete_filters; + } + ret = target__validate(&ftrace.target); if (ret) { char errbuf[512]; @@ -1248,27 +1271,7 @@ int cmd_ftrace(int argc, const char **argv) goto out_delete_evlist; } - switch (subcmd) { - case PERF_FTRACE_TRACE: - if (!argc && target__none(&ftrace.target)) - ftrace.target.system_wide = true; - ret = __cmd_ftrace(&ftrace); - break; - case PERF_FTRACE_LATENCY: - if (list_empty(&ftrace.filters)) { - pr_err("Should provide a function to measure\n"); - parse_options_usage(ftrace_usage, options, "T", 1); - ret = -EINVAL; - goto out_delete_evlist; - } - ret = __cmd_latency(&ftrace); - break; - case PERF_FTRACE_NONE: - default: - pr_err("Invalid subcommand\n"); - ret = -EINVAL; - break; - } + ret = cmd_func(&ftrace); out_delete_evlist: evlist__delete(ftrace.evlist); -- cgit v1.2.3 From de9f498d2b381de1abf654ca3459c4f01227b5cd Mon Sep 17 00:00:00 2001 From: Changbin Du Date: Tue, 8 Feb 2022 22:07:25 +0800 Subject: perf trace: Avoid early exit due SIGCHLD from non-workload processes The function trace__symbols_init() runs "perf-read-vdso32" and that ends up with a SIGCHLD delivered to 'perf'. And this SIGCHLD make perf exit early. 'perf trace' should exit only if the SIGCHLD is from our workload process. So let's use sigaction() instead of signal() to match such condition. Committer notes: Use memset to zero the 'struct sigaction' variable as the '= { 0 }' method isn't accepted in many compiler versions, e.g.: 4 34.02 alpine:3.6 : FAIL clang version 4.0.0 (tags/RELEASE_400/final) builtin-trace.c:4897:35: error: suggest braces around initialization of subobject [-Werror,-Wmissing-braces] struct sigaction sigchld_act = { 0 }; ^ {} builtin-trace.c:4897:37: error: missing field 'sa_mask' initializer [-Werror,-Wmissing-field-initializers] struct sigaction sigchld_act = { 0 }; ^ 2 errors generated. 6 32.60 alpine:3.8 : FAIL gcc version 6.4.0 (Alpine 6.4.0) builtin-trace.c:4897:35: error: suggest braces around initialization of subobject [-Werror,-Wmissing-braces] struct sigaction sigchld_act = { 0 }; ^ {} builtin-trace.c:4897:37: error: missing field 'sa_mask' initializer [-Werror,-Wmissing-field-initializers] struct sigaction sigchld_act = { 0 }; ^ 2 errors generated. 7 34.82 alpine:3.9 : FAIL gcc version 8.3.0 (Alpine 8.3.0) builtin-trace.c:4897:35: error: suggest braces around initialization of subobject [-Werror,-Wmissing-braces] struct sigaction sigchld_act = { 0 }; ^ {} builtin-trace.c:4897:37: error: missing field 'sa_mask' initializer [-Werror,-Wmissing-field-initializers] struct sigaction sigchld_act = { 0 }; ^ 2 errors generated. Signed-off-by: Changbin Du Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20220208140725.3947-1-changbin.du@gmail.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-trace.c | 23 ++++++++++++++++++----- 1 file changed, 18 insertions(+), 5 deletions(-) (limited to 'tools/perf') diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c index 32844d8a0ea5..52b137a184a6 100644 --- a/tools/perf/builtin-trace.c +++ b/tools/perf/builtin-trace.c @@ -1536,13 +1536,20 @@ static size_t trace__fprintf_tstamp(struct trace *trace, u64 tstamp, FILE *fp) return fprintf(fp, " ? "); } +static pid_t workload_pid = -1; static bool done = false; static bool interrupted = false; -static void sig_handler(int sig) +static void sighandler_interrupt(int sig __maybe_unused) { - done = true; - interrupted = sig == SIGINT; + done = interrupted = true; +} + +static void sighandler_chld(int sig __maybe_unused, siginfo_t *info, + void *context __maybe_unused) +{ + if (info->si_pid == workload_pid) + done = true; } static size_t trace__fprintf_comm_tid(struct trace *trace, struct thread *thread, FILE *fp) @@ -3938,7 +3945,6 @@ static int trace__run(struct trace *trace, int argc, const char **argv) bool draining = false; trace->live = true; - signal(SIGCHLD, sig_handler); if (!trace->raw_augmented_syscalls) { if (trace->trace_syscalls && trace__add_syscall_newtp(trace)) @@ -4018,6 +4024,7 @@ static int trace__run(struct trace *trace, int argc, const char **argv) fprintf(trace->output, "Couldn't run the workload!\n"); goto out_delete_evlist; } + workload_pid = evlist->workload.pid; } err = evlist__open(evlist); @@ -4887,10 +4894,16 @@ int cmd_trace(int argc, const char **argv) const char * const trace_subcommands[] = { "record", NULL }; int err = -1; char bf[BUFSIZ]; + struct sigaction sigchld_act; signal(SIGSEGV, sighandler_dump_stack); signal(SIGFPE, sighandler_dump_stack); - signal(SIGINT, sig_handler); + signal(SIGINT, sighandler_interrupt); + + memset(&sigchld_act, 0, sizeof(sigchld_act)); + sigchld_act.sa_flags = SA_SIGINFO; + sigchld_act.sa_sigaction = sighandler_chld; + sigaction(SIGCHLD, &sigchld_act, NULL); trace.evlist = evlist__new(); trace.sctbl = syscalltbl__new(); -- cgit v1.2.3 From 0b31ea6613ad1e6b9b84d877c1be18e39934e90a Mon Sep 17 00:00:00 2001 From: James Clark Date: Thu, 10 Feb 2022 20:06:19 +0000 Subject: perf cs-etm: No-op refactor of synth opt usage sample_branches and sample_instructions are already saved in the synth_opts struct. Other usages like synth_opts.last_branch don't save a value, so make this more consistent by always going through synth_opts and not saving duplicate values. Reviewed-by: Leo Yan Signed-off-by: James Clark Cc: Alexander Shishkin Cc: Jiri Olsa Cc: John Garry Cc: Mark Rutland Cc: Mathieu Poirier Cc: Mike Leach Cc: Namhyung Kim Cc: Will Deacon Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20220210200620.1227232-1-james.clark@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/cs-etm.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-) (limited to 'tools/perf') diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c index 4f672f7d008c..796a065a500e 100644 --- a/tools/perf/util/cs-etm.c +++ b/tools/perf/util/cs-etm.c @@ -50,8 +50,6 @@ struct cs_etm_auxtrace { u8 timeless_decoding; u8 snapshot_mode; u8 data_queued; - u8 sample_branches; - u8 sample_instructions; int num_cpu; u64 latest_kernel_timestamp; @@ -410,8 +408,8 @@ static void cs_etm__packet_swap(struct cs_etm_auxtrace *etm, { struct cs_etm_packet *tmp; - if (etm->sample_branches || etm->synth_opts.last_branch || - etm->sample_instructions) { + if (etm->synth_opts.branches || etm->synth_opts.last_branch || + etm->synth_opts.instructions) { /* * Swap PACKET with PREV_PACKET: PACKET becomes PREV_PACKET for * the next incoming packet. @@ -1365,7 +1363,6 @@ static int cs_etm__synth_events(struct cs_etm_auxtrace *etm, err = cs_etm__synth_event(session, &attr, id); if (err) return err; - etm->sample_branches = true; etm->branches_sample_type = attr.sample_type; etm->branches_id = id; id += 1; @@ -1389,7 +1386,6 @@ static int cs_etm__synth_events(struct cs_etm_auxtrace *etm, err = cs_etm__synth_event(session, &attr, id); if (err) return err; - etm->sample_instructions = true; etm->instructions_sample_type = attr.sample_type; etm->instructions_id = id; id += 1; @@ -1420,7 +1416,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq, tidq->prev_packet->last_instr_taken_branch) cs_etm__update_last_branch_rb(etmq, tidq); - if (etm->sample_instructions && + if (etm->synth_opts.instructions && tidq->period_instructions >= etm->instructions_sample_period) { /* * Emit instruction sample periodically @@ -1503,7 +1499,7 @@ static int cs_etm__sample(struct cs_etm_queue *etmq, } } - if (etm->sample_branches) { + if (etm->synth_opts.branches) { bool generate_sample = false; /* Generate sample for tracing on packet */ @@ -1582,7 +1578,7 @@ static int cs_etm__flush(struct cs_etm_queue *etmq, } - if (etm->sample_branches && + if (etm->synth_opts.branches && tidq->prev_packet->sample_type == CS_ETM_RANGE) { err = cs_etm__synth_branch_sample(etmq, tidq); if (err) -- cgit v1.2.3 From 9de0736973dd7f0f710d0f5e0a3dfd9fa9ffeb3f Mon Sep 17 00:00:00 2001 From: James Clark Date: Thu, 10 Feb 2022 20:06:20 +0000 Subject: perf cs-etm: Fix corrupt inject files when only last branch option is enabled 'perf inject' with Coresight data generates files that cannot be opened when only the last branch option is specified: perf inject -i perf.data --itrace=l -o inject.data perf script -i inject.data 0x33faa8 [0x8]: failed to process type: 9 [Bad address] This is because cs_etm__synth_instruction_sample() is called even when the sample type for instructions hasn't been setup. Last branch records are attached to instruction samples so it doesn't make sense to generate them when --itrace=i isn't specified anyway. This change disables all calls of cs_etm__synth_instruction_sample() unless --itrace=i is specified, resulting in a file with no samples if only --itrace=l is provided, rather than a bad file. Reviewed-by: Leo Yan Signed-off-by: James Clark Cc: Alexander Shishkin Cc: Jiri Olsa Cc: John Garry Cc: Mark Rutland Cc: Mathieu Poirier Cc: Mike Leach Cc: Namhyung Kim Cc: Will Deacon Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20220210200620.1227232-2-james.clark@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/cs-etm.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf') diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c index 796a065a500e..8b95fb3c4d7b 100644 --- a/tools/perf/util/cs-etm.c +++ b/tools/perf/util/cs-etm.c @@ -1553,6 +1553,7 @@ static int cs_etm__flush(struct cs_etm_queue *etmq, goto swap_packet; if (etmq->etm->synth_opts.last_branch && + etmq->etm->synth_opts.instructions && tidq->prev_packet->sample_type == CS_ETM_RANGE) { u64 addr; @@ -1610,6 +1611,7 @@ static int cs_etm__end_block(struct cs_etm_queue *etmq, * the trace. */ if (etmq->etm->synth_opts.last_branch && + etmq->etm->synth_opts.instructions && tidq->prev_packet->sample_type == CS_ETM_RANGE) { u64 addr; -- cgit v1.2.3 From 047e6032c4f054372311eb20c759cfe0b3ca7ffe Mon Sep 17 00:00:00 2001 From: German Gomez Date: Tue, 25 Jan 2022 10:44:34 +0000 Subject: perf test: Fix arm64 perf_event_attr tests wrt --call-graph initialization The struct perf_event_attr is initialised differently in Arm64 when recording in call-graph fp mode, so update the relevant tests, and add two extra arm64-only tests. Before: $ perf test 17 -v 17: Setup struct perf_event_attr [...] running './tests/attr/test-record-graph-default' expected sample_type=295, got 4391 expected sample_regs_user=0, got 1073741824 FAILED './tests/attr/test-record-graph-default' - match failure test child finished with -1 ---- end ---- After: [...] running './tests/attr/test-record-graph-default-aarch64' test limitation 'aarch64' running './tests/attr/test-record-graph-fp-aarch64' test limitation 'aarch64' running './tests/attr/test-record-graph-default' test limitation '!aarch64' excluded architecture list ['aarch64'] skipped [aarch64] './tests/attr/test-record-graph-default' running './tests/attr/test-record-graph-fp' test limitation '!aarch64' excluded architecture list ['aarch64'] skipped [aarch64] './tests/attr/test-record-graph-fp' [...] Fixes: 7248e308a5758761 ("perf tools: Record ARM64 LR register automatically") Signed-off-by: German Gomez Cc: Alexander Shishkin Cc: Alexandre Truong Cc: Ian Rogers Cc: James Clark Cc: Jiri Olsa Cc: John Fastabend Cc: KP Singh Cc: Mark Rutland Cc: Martin KaFai Lau Cc: Namhyung Kim Cc: Song Liu Cc: Yonghong Song Link: http://lore.kernel.org/lkml/20220125104435.2737-1-german.gomez@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/tests/attr/README | 2 ++ tools/perf/tests/attr/test-record-graph-default | 2 ++ tools/perf/tests/attr/test-record-graph-default-aarch64 | 9 +++++++++ tools/perf/tests/attr/test-record-graph-fp | 2 ++ tools/perf/tests/attr/test-record-graph-fp-aarch64 | 9 +++++++++ 5 files changed, 24 insertions(+) create mode 100644 tools/perf/tests/attr/test-record-graph-default-aarch64 create mode 100644 tools/perf/tests/attr/test-record-graph-fp-aarch64 (limited to 'tools/perf') diff --git a/tools/perf/tests/attr/README b/tools/perf/tests/attr/README index a36f49fb4dbe..1116fc6bf2ac 100644 --- a/tools/perf/tests/attr/README +++ b/tools/perf/tests/attr/README @@ -45,8 +45,10 @@ Following tests are defined (with perf commands): perf record -d kill (test-record-data) perf record -F 100 kill (test-record-freq) perf record -g kill (test-record-graph-default) + perf record -g kill (test-record-graph-default-aarch64) perf record --call-graph dwarf kill (test-record-graph-dwarf) perf record --call-graph fp kill (test-record-graph-fp) + perf record --call-graph fp kill (test-record-graph-fp-aarch64) perf record --group -e cycles,instructions kill (test-record-group) perf record -e '{cycles,instructions}' kill (test-record-group1) perf record -e '{cycles/period=1/,instructions/period=2/}:S' kill (test-record-group2) diff --git a/tools/perf/tests/attr/test-record-graph-default b/tools/perf/tests/attr/test-record-graph-default index 5d8234d50845..f0a18b4ea4f5 100644 --- a/tools/perf/tests/attr/test-record-graph-default +++ b/tools/perf/tests/attr/test-record-graph-default @@ -2,6 +2,8 @@ command = record args = --no-bpf-event -g kill >/dev/null 2>&1 ret = 1 +# arm64 enables registers in the default mode (fp) +arch = !aarch64 [event:base-record] sample_type=295 diff --git a/tools/perf/tests/attr/test-record-graph-default-aarch64 b/tools/perf/tests/attr/test-record-graph-default-aarch64 new file mode 100644 index 000000000000..e98d62efb6f7 --- /dev/null +++ b/tools/perf/tests/attr/test-record-graph-default-aarch64 @@ -0,0 +1,9 @@ +[config] +command = record +args = --no-bpf-event -g kill >/dev/null 2>&1 +ret = 1 +arch = aarch64 + +[event:base-record] +sample_type=4391 +sample_regs_user=1073741824 diff --git a/tools/perf/tests/attr/test-record-graph-fp b/tools/perf/tests/attr/test-record-graph-fp index 5630521c0b0f..a6e60e839205 100644 --- a/tools/perf/tests/attr/test-record-graph-fp +++ b/tools/perf/tests/attr/test-record-graph-fp @@ -2,6 +2,8 @@ command = record args = --no-bpf-event --call-graph fp kill >/dev/null 2>&1 ret = 1 +# arm64 enables registers in fp mode +arch = !aarch64 [event:base-record] sample_type=295 diff --git a/tools/perf/tests/attr/test-record-graph-fp-aarch64 b/tools/perf/tests/attr/test-record-graph-fp-aarch64 new file mode 100644 index 000000000000..cbeea9971285 --- /dev/null +++ b/tools/perf/tests/attr/test-record-graph-fp-aarch64 @@ -0,0 +1,9 @@ +[config] +command = record +args = --no-bpf-event --call-graph fp kill >/dev/null 2>&1 +ret = 1 +arch = aarch64 + +[event:base-record] +sample_type=4391 +sample_regs_user=1073741824 -- cgit v1.2.3 From 31ded1535e3182778a1d0e5c32711f55da3bc512 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 16 Feb 2022 16:01:00 -0300 Subject: perf bpf: Defer freeing string after possible strlen() on it This was detected by the gcc in Fedora Rawhide's gcc: 50 11.01 fedora:rawhide : FAIL gcc version 12.0.1 20220205 (Red Hat 12.0.1-0) (GCC) inlined from 'bpf__config_obj' at util/bpf-loader.c:1242:9: util/bpf-loader.c:1225:34: error: pointer 'map_opt' may be used after 'free' [-Werror=use-after-free] 1225 | *key_scan_pos += strlen(map_opt); | ^~~~~~~~~~~~~~~ util/bpf-loader.c:1223:9: note: call to 'free' here 1223 | free(map_name); | ^~~~~~~~~~~~~~ cc1: all warnings being treated as errors So do the calculations on the pointer before freeing it. Fixes: 04f9bf2bac72480c ("perf bpf-loader: Add missing '*' for key_scan_pos") Cc: Adrian Hunter Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Cc: Wang ShaoBo Link: https://lore.kernel.org/lkml/Yg1VtQxKrPpS3uNA@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf-loader.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'tools/perf') diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c index 7ecfaac7536a..16ec605a9fe4 100644 --- a/tools/perf/util/bpf-loader.c +++ b/tools/perf/util/bpf-loader.c @@ -1220,9 +1220,10 @@ bpf__obj_config_map(struct bpf_object *obj, pr_debug("ERROR: Invalid map config option '%s'\n", map_opt); err = -BPF_LOADER_ERRNO__OBJCONF_MAP_OPT; out: - free(map_name); if (!err) *key_scan_pos += strlen(map_opt); + + free(map_name); return err; } -- cgit v1.2.3