- 10 May, 2024 7 commits
-
-
Ian Rogers authored
If a usage string is built in parse_options_subcommand, also free it. Fixes: 901421a5 ("perf tools: Remove subcmd dependencies on strbuf") Signed-off-by:
Ian Rogers <irogers@google.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Josh Poimboeuf <jpoimboe@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240509052015.1914670-1-irogers@google.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Ian Rogers authored
Leak sanitizer complains about the strdup-ed arguments not being freed and given cmd_record doesn't modify the given strings, remove the strdups. Original discussion in this patch: https://lore.kernel.org/lkml/20240430184156.1824083-1-irogers@google.com/Suggested-by:
Namhyung Kim <namhyung@kernel.org> Signed-off-by:
-
Madadi Vineeth Reddy authored
perf sched: Rename 'switches' column header to 'count' and add usage description, options for latency Rename 'Switches' to 'Count' and document metrics shown for perf sched latency output. Also add options possible with perf sched latency. Initially, after seeing the output of 'perf sched latency', the term 'Switches' seemed like it's the number of context switches-in for a particular task, but upon going through the code, it was observed that it's actually keeping track of number of times a delay was calculated so that it is used in calculation of the average delay. Actually, the switches here is a subset of number of context switches-in because there are some cases where the count is not incremented in switch-in handler 'add_sched_in_event'. For example when a task is switched-in while it's state is not ready to run(!= THREAD_WAIT_CPU). commit d9340c1d ("perf sched: Display time in milliseconds, reorganize output") changed it from the original count to switches. So, renamed switches to count to make things a bit more clearer and added the metrics description of latency in the document. Reviewed-by:
Aditya Gupta <adityag@linux.ibm.com> Signed-off-by:
Madadi Vineeth Reddy <vineethr@linux.ibm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240328090005.8321-1-vineethr@linux.ibm.comSigned-off-by:
-
Namhyung Kim authored
On large systems, cgroups can be created and deleted often. That means there's a race between perf tools and cgroups when it gets the cgroup name and opens the cgroup. I got a report that 'perf stat' with many cgroups failed quite often due to the missing cgroups on such a large machine. I think we can ignore such cgroups when expanding events and use id 0 if it fails to read the cgroup id. IIUC 0 is not a vaild cgroup id so it won't update event counts for the failed cgroups. Signed-off-by:
-
Dominique Martinet authored
Tracepoints can start with digits, although we don't have many of these: $ rg -g '*.h' '\bTRACE_EVENT\([0-9]' net/mac802154/trace.h 53:TRACE_EVENT(802154_drv_return_int, ... net/ieee802154/trace.h 66:TRACE_EVENT(802154_rdev_add_virtual_intf, ... include/trace/events/9p.h 124:TRACE_EVENT(9p_client_req, ... Just allow names to start with digits too so e.g. "perf trace -e '9p:*'" works Reviewed-by:
Dominique Martinet <asmadeus@codewreck.org> Tested-by:
-
Dominique Martinet authored
The next commit will allow tracepoints starting with digits, but most systems do not have any available by default so tests should skip the actual "check if it exists in /sys/kernel/debug/tracing" step. In order to do that, add a new boolean flag specifying if we should actually "format" the probe or not. Originally-by:
Jiri Olsa <jolsa@kernel.org> Reviewed-by:
-
Dominique Martinet authored
The next patch will add another flag to parse_state that we will want to pass to evsel__newtp_idx(), so pass the whole parse_state all the way down instead of giving only the index Originally-by:
-
- 09 May, 2024 10 commits
-
-
James Clark authored
The linked commit updated dso__load_vmlinux() to call dso__set_long_name() before loading the symbols. Loading the symbols may not succeed but dso__set_long_name() takes ownership of the string. The two callers of this function free the string themselves on failure cases, resulting in the following error: $ perf record -- ls $ perf report free(): double free detected in tcache 2 Fix it by always taking ownership of the string, even on failure. This means the string is either freed at the very first early exit condition, or later when the dso is deleted or the long name is replaced. Now no special return value is needed to signify that the caller needs to free the string. Fixes: e59fea47 ("perf symbols: Fix DSO kernel load and symbol process to correctly map DSO to its long_name, type and adjust_symbols") Reviewed-by:
James Clark <james.clark@arm.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20240507141210.195939-5-james.clark@arm.comSigned-off-by:
-
James Clark authored
When loading kcore, the main vmlinux map is updated in the same loop that merges the remaining maps. If a map that overlaps is merged in before kcore, the list can become unsortable when the main map addresses are updated. This will later trigger the check_invariants() assert: $ perf record $ perf report util/maps.c:96: check_invariants: Assertion `map__end(prev) <= map__start(map) || map__start(prev) == map__start(map)' failed. Aborted Fix it by moving the main map update prior to the loop so that maps__merge_in() can split it if necessary. Fixes: 659ad349 ("perf maps: Switch from rbtree to lazily sorted array for addresses") Signed-off-by: -
James Clark authored
maps__merge_in() hard codes the steps to free the maps_by_name list. It seems to not map__put() each element before freeing, and it sets maps_by_name_sorted to true after freeing, which may be harmless but is inconsistent with maps__init() and other functions. maps__maps_by_name_addr() is also quite hard to read because we already have maps__maps_by_name() and maps__maps_by_address(), but the function is only used in that place so delete it. Reviewed-by:
-
James Clark authored
Make the order of operations remove, update, add. Updating addresses before the map is removed causes the ordering check to fail when the map is removed. This can be reproduced when running Perf on an Arm system with a static kernel and Perf uses kcore rather than other sources: $ perf record -- ls $ perf report util/maps.c:96: check_invariants: Assertion `map__end(prev) <= map__start(map) || map__start(prev) == map__start(map)' failed Fixes: 659ad349 ("perf maps: Switch from rbtree to lazily sorted array for addresses") Signed-off-by: -
Ian Rogers authored
In is_valid_tracepoint, rather than scanning "/sys/kernel/tracing/events/*/*" skipping any path where "/sys/kernel/tracing/events/*/*/id" doesn't exist, and then testing if "*:*" matches the tracepoint name, just use the given tracepoint name replace the ':' with '/' and see if the id file exists. This turns a nested directory search into a single file available test. Rather than return 1 for valid and 0 for invalid, return true and false. Signed-off-by:
-
James Clark authored
check_allowed_ops() is used from both HAVE_DWARF_GETLOCATIONS_SUPPORT and HAVE_DWARF_CFI_SUPPORT sections, so move it into the right place so that it's available when either are defined. This shows up when doing a static cross compile for arm64: $ make ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- LDFLAGS="-static" \ EXTRA_PERFLIBS="-lexpat" util/dwarf-aux.c:1723:6: error: implicit declaration of function 'check_allowed_ops' Fixes: 55442cc2 ("perf dwarf-aux: Check allowed DWARF Ops") Reviewed-by:Masami Hiramatsu <mhiramat@kernel.org> Acked-by:
-
Ian Rogers authored
Freeing the thread on failure won't work with reference count checking, use thread__delete(). Don't allocate the comm_str, use a stack allocation instead. Fixes: f6005caf ("perf thread: Add reference count checking") Signed-off-by:
-
Ian Rogers authored
In some cases evsel->name is lazily initialized in evsel__name(). If not initialized passing NULL to strstr() leads to a SEGV. Fixes: ccb17cae ("perf report: Set PERF_SAMPLE_DATA_SRC bit for Arm SPE event") Signed-off-by:
-
Ian Rogers authored
Searching for the entry in the array needs to avoid the intermediate pointer with reference count checking. Refactor the array removal to binary search for the entry. Change the array to hold an entry with a reference count (so the intermediate pointer can work) and remove from the array when the reference count on a comm_str falls to 1. Fixes: 13ca6287 ("perf comm: Add reference count checking to 'struct comm_str'") Signed-off-by:
-
Ian Rogers authored
If the title is NULL then it can lead to a SEGV. Fixes: 769e6a1e ("perf ui browser: Don't save pointer to stack memory") Signed-off-by:
-
- 08 May, 2024 1 commit
-
-
Namhyung Kim authored
It's confusing both pointers and arrays are printed as *. Let's print array types with [] so that we can identify them easily. Although it's interchangable, sometimes it can cause confusion with size like in the below example. Note that it is not the same with C syntax where it goes to the variable names, but we want to have it in the type names (like in Go language). Before: mov [20] 0x68(reg5) -> reg0 type='struct page**' size=0x80 (die:0x4e61d32) After: mov [20] 0x68(reg5) -> reg0 type='struct page*[]' size=0x80 (die:0x4e61d32) Signed-off-by:
-
- 07 May, 2024 16 commits
-
-
Ian Rogers authored
'struct mem_info' is reference counted while 'struct branch_info' and he_cache (struct hist_entry **) are not. Break apart the priv field in 'struct hist_entry_iter' so that we can know which values are owned by the iter and do the appropriate free or put. Move hide_unresolved to marginally shrink the size of the now grown struct. Signed-off-by:
-
Ian Rogers authored
Add reference count checking and switch 'struct mem_info' usage to use accessor functions. Signed-off-by:
-
Ian Rogers authored
Move mem-info to its own header rather than having it split between mem-events and symbol. Signed-off-by:
-
Ian Rogers authored
Reference count checking of an rbtree is troublesome as each pointer should have a reference, switch to using a sorted array. Remove an indirection by embedding the reference count with the string. Use pthread_once to safely initialize the comm_strs and reader writer mutex. Signed-off-by:
-
Ian Rogers authored
It is assigned a value of 1 and never incremented. Remove and replace puts with delete. Signed-off-by:
-
Ian Rogers authored
block_info__get() has no callers so the refcount is only ever one. As such remove the reference counting logic and turn puts to deletes. Signed-off-by:
-
Ian Rogers authored
Freeing hash map doesn't free the entries added to the hashmap, add the missing free(). Fixes: d3e7cad6 ("perf annotate: Add a hashmap for symbol histogram") Signed-off-by:
-
Ian Rogers authored
ui_browser__show() is capturing the input title that is stack allocated memory in hist_browser__run(). Avoid a use after return by strdup-ing the string. Committer notes: Further explanation from Ian Rogers: My command line using tui is: $ sudo bash -c 'rm /tmp/asan.log*; export ASAN_OPTIONS="log_path=/tmp/asan.log"; /tmp/perf/perf mem record -a sleep 1; /tmp/perf/perf mem report' I then go to the perf annotate view and quit. This triggers the asan error (from the log file): ``` ==1254591==ERROR: AddressSanitizer: stack-use-after-return on address 0x7f2813331920 at pc 0x7f28180 65991 bp 0x7fff0a21c750 sp 0x7fff0a21bf10 READ of size 80 at 0x7f2813331920 thread T0 #0 0x7f2818065990 in __interceptor_strlen ../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:461 #1 0x7f2817698251 in SLsmg_write_wrapped_string (/lib/x86_64-linux-gnu/libslang.so.2+0x98251) #2 0x7f28176984b9 in SLsmg_write_nstring (/lib/x86_64-linux-gnu/libslang.so.2+0x984b9) #3 0x55c94045b365 in ui_browser__write_nstring ui/browser.c:60 #4 0x55c94045c558 in __ui_browser__show_title ui/browser.c:266 #5 0x55c94045c776 in ui_browser__show ui/browser.c:288 #6 0x55c94045c06d in ui_browser__handle_resize ui/browser.c:206 #7 0x55c94047979b in do_annotate ui/browsers/hists.c:2458 #8 0x55c94047fb17 in evsel__hists_browse ui/browsers/hists.c:3412 #9 0x55c940480a0c in perf_evsel_menu__run ui/browsers/hists.c:3527 #10 0x55c940481108 in __evlist__tui_browse_hists ui/browsers/hists.c:3613 #11 0x55c9404813f7 in evlist__tui_browse_hists ui/browsers/hists.c:3661 #12 0x55c93ffa253f in report__browse_hists tools/perf/builtin-report.c:671 #13 0x55c93ffa58ca in __cmd_report tools/perf/builtin-report.c:1141 #14 0x55c93ffaf159 in cmd_report tools/perf/builtin-report.c:1805 #15 0x55c94000c05c in report_events tools/perf/builtin-mem.c:374 #16 0x55c94000d96d in cmd_mem tools/perf/builtin-mem.c:516 #17 0x55c9400e44ee in run_builtin tools/perf/perf.c:350 #18 0x55c9400e4a5a in handle_internal_command tools/perf/perf.c:403 #19 0x55c9400e4e22 in run_argv tools/perf/perf.c:447 #20 0x55c9400e53ad in main tools/perf/perf.c:561 #21 0x7f28170456c9 in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58 #22 0x7f2817045784 in __libc_start_main_impl ../csu/libc-start.c:360 #23 0x55c93ff544c0 in _start (/tmp/perf/perf+0x19a4c0) (BuildId: 84899b0e8c7d3a3eaa67b2eb35e3d8b2f8cd4c93) Address 0x7f2813331920 is located in stack of thread T0 at offset 32 in frame #0 0x55c94046e85e in hist_browser__run ui/browsers/hists.c:746 This frame has 1 object(s): [32, 192) 'title' (line 747) <== Memory access at offset 32 is inside this variable HINT: this may be a false positive if your program uses some custom stack unwind mechanism, swapcontext or vfork ``` hist_browser__run isn't on the stack so the asan error looks legit. There's no clean init/exit on struct ui_browser so I may be trading a use-after-return for a memory leak, but that seems look a good trade anyway. Fixes: 05e8b080 ("perf ui browser: Stop using 'self'") Signed-off-by: -
He Zhe authored
'perf bench internals inject-build-id' suffers from the following error when only one DSO is collected. # perf bench internals inject-build-id -v Collected 1 DSOs traps: internals-injec[2305] trap divide error ip:557566ba6394 sp:7ffd4de97fe0 error:0 in perf[557566b2a000+23d000] Build-id injection benchmark Iteration #1 Floating point exception This patch removes the unnecessary minus one from the divisor which also corrects the randomization range. Signed-off-by:He Zhe <zhe.he@windriver.com> Fixes: 0bf02a0d ("perf bench: Add build-id injection benchmark") Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lore.kernel.org/r/20240507065026.2652929-1-zhe.he@windriver.comSigned-off-by:
-
Arnaldo Carvalho de Melo authored
When freeing a->b it is good practice to set a->b to NULL using zfree(&a->b) so that when we have a bug where a reference to a freed 'a' pointer is kept somewhere, we can more quickly cause a segfault if some code tries to use a->b. Convert one such case in the 'perf probe' codebase. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lore.kernel.org/lkml/ZjpBnkL2wO3QJa5W@x1Signed-off-by:
-
James Clark authored
Currently it's only possible to initialize with the default number of queues and then use auxtrace_queues__add_event() to grow the array. But that's problematic if you don't have a real event to pass into that function yet. The queues hold a void *priv member to store custom state, and for Coresight we want to create decoders upfront before receiving data, so add a new function that allows pre-allocating queues. One reason to do this is because we might need to store metadata (HW_ID events) that effects other queues, but never actually receive auxtrace data on that queue. Reviewed-by:
Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by:
Ganapatrao Kulkarni <gankulkarni@os.amperecomputing.com> Acked-by:
Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexandre Torgue <alexandre.torgue@foss.st.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: John Garry <john.g.garry@oracle.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Maxime Coquelin <mcoquelin.stm32@gmail.com> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Steve Clevenger <scclevenger@os.amperecomputing.com> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Will Deacon <will@kernel.org> Link: https://lore.kernel.org/r/20240429152207.479221-3-james.clark@arm.comSigned-off-by:
-
James Clark authored
The likely fix for this is to update perf so print a helpful message. Signed-off-by:
-
Athira Rajeev authored
Fix a comment in function which explains how multi_regs field gets set for an instruction. In the example, "mov %rsi, 8(%rbx,%rcx,4)", the comment mistakenly referred to "dst_multi_regs = 0". Correct it to use "src_multi_regs = 0" Signed-off-by:
Athira Rajeev <atrajeev@linux.vnet.ibm.com> Acked-by:
-
Arnaldo Carvalho de Melo authored
When freeing a->b it is good practice to set a->b to NULL using zfree(&a->b) so that when we have a bug where a reference to a freed 'a' pointer is kept somewhere, we can more quickly cause a segfault if some code tries to use a->b. Convert one such case in the 'perf kwork' codebase. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Yang Jihong <yangjihong1@huawei.com> Link: https://lore.kernel.org/lkml/Zjmc5EiN6zmWZj4r@x1Signed-off-by:
-
Arnaldo Carvalho de Melo authored
When freeing a->b it is good practice to set a->b to NULL using zfree(&a->b) so that when we have a bug where a reference to a freed 'a' pointer is kept somewhere, we can more quickly cause a segfault if some code tries to use a->b. Convert one such case in the callchain code. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lore.kernel.org/lkml/ZjmcGobQ8E52EyjJ@x1Signed-off-by:
-
Arnaldo Carvalho de Melo authored
When freeing a->b it is good practice to set a->b to NULL using zfree(&a->b) so that when we have a bug where a reference to a freed 'a' pointer is kept somewhere, we can more quickly cause a segfault if some code tries to use a->b. This is mostly done but some new cases were introduced recently, convert them to zfree(). Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Namhyung Kim <namhyung@kernel.org> Link: https://lore.kernel.org/lkml/ZjmbHHrjIm5YRIBv@x1Signed-off-by:
-
- 06 May, 2024 6 commits
-
-
Ian Rogers authored
The dso pointer in 'struct dso_data' is necessary for reference count checking to account for the dso_data forming a global list of open dso's with references to the dso. The dso pointer also allows for the indirection that reference count checking needs. Outside of reference count checking the indirection isn't needed and container_of() is more efficient and saves space. The reference count won't be increased by placing items onto the global list, matching how things were before the reference count checking change, but we assert the dso is in dsos holding it live (and that the set of open dsos is a subset of all dsos for the machine). Update the DSO data tests so that they use a dsos struct to make the invariant true. Signed-off-by:
-
Ian Rogers authored
dso__load_sym_internal() passed curr_mapp as an out argument to dso__process_kernel_symbol(). The out argument was never used so remove it to simplify the reference counting logic. Simplify reference counting issues with curr_dso by ensuring the value it points to has a +1 reference count, and then putting as necessary. This avoids some reference counting games when the dso is created making the code more obviously correct with some possible introduced overhead due to the reference counting get/puts. This, however, silences reference count checking and we can always optimize from a seemingly correct point. Signed-off-by:
-
Ian Rogers authored
The dso__put() after the map creation causes a use after put in dso__set_loaded(). To ensure there is a +1 reference count on both sides of the if-else, do a dso__get() on the found map's dso. Signed-off-by:
-
Ian Rogers authored
A dso__put() is needed for the dsos__find() when the map is created and a buildid is sought. Fixes: f649ed80 ("perf dsos: Tidy reference counting and locking") Signed-off-by:
-
Ian Rogers authored
Add reference count checking to struct dso, this can help with implementing correct reference counting discipline. To avoid RC_CHK_ACCESS everywhere, add accessor functions for the variables in struct dso. The majority of the change is mechanical in nature and not easy to split up. Committer testing: 'perf test' up to this patch shows no regressions. But: util/symbol.c: In function ‘dso__load_bfd_symbols’: util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’ 1683 | dso__set_adjust_symbols(dso); | ^~~~~~~~~~~~~~~~~~~~~~~ In file included from util/symbol.c:21: util/dso.h:268:20: note: declared here 268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val) | ^~~~~~~~~~~~~~~~~~~~~~~ make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1 MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/ make[6]: *** Waiting for unfinished jobs.... This was updated: - symbols__fixup_end(&dso->symbols, false); - symbols__fixup_duplicate(&dso->symbols); - dso->adjust_symbols = 1; + symbols__fixup_end(dso__symbols(dso), false); + symbols__fixup_duplicate(dso__symbols(dso)); + dso__set_adjust_symbols(dso); But not build tested with BUILD_NONDISTRO and libbfd devel files installed (binutils-devel on fedora). Add the missing argument: symbols__fixup_end(dso__symbols(dso), false); symbols__fixup_duplicate(dso__symbols(dso)); - dso__set_adjust_symbols(dso); + dso__set_adjust_symbols(dso, true); Signed-off-by: -
Ian Rogers authored
Switch to using the bsearch library function rather than having a hand written binary search. Const-ify some static functions to avoid compiler warnings. Signed-off-by:
-
