From c0268e8d1f450e286fc55e77f53a9ede6b72acab Mon Sep 17 00:00:00 2001 From: Joseph Schuchart Date: Thu, 24 Oct 2013 10:10:51 -0300 Subject: perf script python: Fix mem leak due to missing Py_DECREFs on dict entries We are using the Python scripting interface in perf to extract kernel events relevant for performance analysis of HPC codes. We noticed that the "perf script" call allocates a significant amount of memory (in the order of several 100 MiB) during it's run, e.g. 125 MiB for a 25 MiB input file: $> perf record -o perf.data -a -R -g fp \ -e power:cpu_frequency -e sched:sched_switch \ -e sched:sched_migrate_task -e sched:sched_process_exit \ -e sched:sched_process_fork -e sched:sched_process_exec \ -e cycles -m 4096 --freq 4000 $> /usr/bin/time perf script -i perf.data -s dummy_script.py 0.84user 0.13system 0:01.92elapsed 51%CPU (0avgtext+0avgdata 125532maxresident)k 73072inputs+0outputs (57major+33086minor)pagefaults 0swaps Upon further investigation using the valgrind massif tool, we noticed that Python objects that are created in trace-event-python.c via PyString_FromString*() (and their Integer and Long counterparts) are never free'd. The reason for this seem to be missing Py_DECREF calls on the objects that are returned by these functions and stored in the Python dictionaries. The Python dictionaries do not steal references (as opposed to Python tuples and lists) but instead add their own reference. Hence, the reference that is returned by these object creation functions is never released and the memory is leaked. (see [1,2]) The attached patch fixes this by wrapping all relevant calls to PyDict_SetItemString() and decrementing the reference counter immediately after the Python function call. This reduces the allocated memory to a reasonable amount: $> /usr/bin/time perf script -i perf.data -s dummy_script.py 0.73user 0.05system 0:00.79elapsed 99%CPU (0avgtext+0avgdata 49132maxresident)k 0inputs+0outputs (0major+14045minor)pagefaults 0swaps For comparison, with a 120 MiB input file the memory consumption reported by time drops from almost 600 MiB to 146 MiB. The patch has been tested using Linux 3.8.2 with Python 2.7.4 and Linux 3.11.6 with Python 2.7.5. Please let me know if you need any further information. [1] http://docs.python.org/2/c-api/tuple.html#PyTuple_SetItem [2] http://docs.python.org/2/c-api/dict.html#PyDict_SetItemString Signed-off-by: Joseph Schuchart Reviewed-by: Tom Zanussi Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Tom Zanussi Link: http://lkml.kernel.org/r/1381468543-25334-4-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c index cc75a3c..95d91a0 100644 --- a/tools/perf/util/scripting-engines/trace-event-python.c +++ b/tools/perf/util/scripting-engines/trace-event-python.c @@ -56,6 +56,17 @@ static void handler_call_die(const char *handler_name) Py_FatalError("problem in Python trace event handler"); } +/* + * Insert val into into the dictionary and decrement the reference counter. + * This is necessary for dictionaries since PyDict_SetItemString() does not + * steal a reference, as opposed to PyTuple_SetItem(). + */ +static void pydict_set_item_string_decref(PyObject *dict, const char *key, PyObject *val) +{ + PyDict_SetItemString(dict, key, val); + Py_DECREF(val); +} + static void define_value(enum print_arg_type field_type, const char *ev_name, const char *field_name, @@ -279,11 +290,11 @@ static void python_process_tracepoint(union perf_event *perf_event PyTuple_SetItem(t, n++, PyInt_FromLong(pid)); PyTuple_SetItem(t, n++, PyString_FromString(comm)); } else { - PyDict_SetItemString(dict, "common_cpu", PyInt_FromLong(cpu)); - PyDict_SetItemString(dict, "common_s", PyInt_FromLong(s)); - PyDict_SetItemString(dict, "common_ns", PyInt_FromLong(ns)); - PyDict_SetItemString(dict, "common_pid", PyInt_FromLong(pid)); - PyDict_SetItemString(dict, "common_comm", PyString_FromString(comm)); + pydict_set_item_string_decref(dict, "common_cpu", PyInt_FromLong(cpu)); + pydict_set_item_string_decref(dict, "common_s", PyInt_FromLong(s)); + pydict_set_item_string_decref(dict, "common_ns", PyInt_FromLong(ns)); + pydict_set_item_string_decref(dict, "common_pid", PyInt_FromLong(pid)); + pydict_set_item_string_decref(dict, "common_comm", PyString_FromString(comm)); } for (field = event->format.fields; field; field = field->next) { if (field->flags & FIELD_IS_STRING) { @@ -313,7 +324,7 @@ static void python_process_tracepoint(union perf_event *perf_event if (handler) PyTuple_SetItem(t, n++, obj); else - PyDict_SetItemString(dict, field->name, obj); + pydict_set_item_string_decref(dict, field->name, obj); } if (!handler) @@ -370,21 +381,21 @@ static void python_process_general_event(union perf_event *perf_event if (!handler || !PyCallable_Check(handler)) goto exit; - PyDict_SetItemString(dict, "ev_name", PyString_FromString(perf_evsel__name(evsel))); - PyDict_SetItemString(dict, "attr", PyString_FromStringAndSize( + pydict_set_item_string_decref(dict, "ev_name", PyString_FromString(perf_evsel__name(evsel))); + pydict_set_item_string_decref(dict, "attr", PyString_FromStringAndSize( (const char *)&evsel->attr, sizeof(evsel->attr))); - PyDict_SetItemString(dict, "sample", PyString_FromStringAndSize( + pydict_set_item_string_decref(dict, "sample", PyString_FromStringAndSize( (const char *)sample, sizeof(*sample))); - PyDict_SetItemString(dict, "raw_buf", PyString_FromStringAndSize( + pydict_set_item_string_decref(dict, "raw_buf", PyString_FromStringAndSize( (const char *)sample->raw_data, sample->raw_size)); - PyDict_SetItemString(dict, "comm", + pydict_set_item_string_decref(dict, "comm", PyString_FromString(thread->comm)); if (al->map) { - PyDict_SetItemString(dict, "dso", + pydict_set_item_string_decref(dict, "dso", PyString_FromString(al->map->dso->name)); } if (al->sym) { - PyDict_SetItemString(dict, "symbol", + pydict_set_item_string_decref(dict, "symbol", PyString_FromString(al->sym->name)); } -- cgit v0.10.2 From 2fd869f08aec5a8e4cbf01bc3fa345c4e53342d7 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 28 Oct 2013 09:38:12 -0300 Subject: perf tools: Fix up /proc/PID/maps parsing When introducing support for MMAP2 we considered more parts of each map representation in /proc/PID/maps, and when disabling it we forgot to reduce the number of expected parsed/assigned entries in the sscanf call, fix it to expect the right number of desired fields, 5. Reported-by: Markus Trippelsdorf Based-on-a-patch-by: Markus Trippelsdorf Cc: Adrian Hunter Cc: David Ahern Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-vrbo1wik997ahjzl1chm3bdm@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 63df031..49096ea 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -213,7 +213,7 @@ static int perf_event__synthesize_mmap_events(struct perf_tool *tool, &event->mmap.pgoff, execname); - if (n != 8) + if (n != 5) continue; if (prot[2] != 'x') -- cgit v0.10.2 From 9754c4f9b23d5ce6756514acdf134ad61470734a Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Fri, 25 Oct 2013 13:24:53 +0200 Subject: perf hists: Add color overhead for stdio output buffer Following commit tightened up the buffer size for output to strict width of used format columns: 99cf666 perf hists: Fix formatting of long symbol names This works fine until you hit color overhead output which places extra bytes into output buffer. We need to account for color overhead in the output buffer. Adding maximum color byte size to the output buffer size. Signed-off-by: Jiri Olsa Cc: Adrian Hunter Cc: Corey Ashford Cc: David Ahern Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382700293-1803-1-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/ui/stdio/hist.c b/tools/perf/ui/stdio/hist.c index 194e2f4..6c15268 100644 --- a/tools/perf/ui/stdio/hist.c +++ b/tools/perf/ui/stdio/hist.c @@ -315,8 +315,7 @@ static inline void advance_hpp(struct perf_hpp *hpp, int inc) } static int hist_entry__period_snprintf(struct perf_hpp *hpp, - struct hist_entry *he, - bool color) + struct hist_entry *he) { const char *sep = symbol_conf.field_sep; struct perf_hpp_fmt *fmt; @@ -338,7 +337,7 @@ static int hist_entry__period_snprintf(struct perf_hpp *hpp, } else first = false; - if (color && fmt->color) + if (perf_hpp__use_color() && fmt->color) ret = fmt->color(fmt, hpp, he); else ret = fmt->entry(fmt, hpp, he); @@ -358,12 +357,11 @@ static int hist_entry__fprintf(struct hist_entry *he, size_t size, .buf = bf, .size = size, }; - bool color = !symbol_conf.field_sep; if (size == 0 || size > bfsz) size = hpp.size = bfsz; - ret = hist_entry__period_snprintf(&hpp, he, color); + ret = hist_entry__period_snprintf(&hpp, he); hist_entry__sort_snprintf(he, bf + ret, size - ret, hists); ret = fprintf(fp, "%s\n", bf); @@ -482,6 +480,7 @@ size_t hists__fprintf(struct hists *hists, bool show_header, int max_rows, print_entries: linesz = hists__sort_list_width(hists) + 3 + 1; + linesz += perf_hpp__color_overhead(); line = malloc(linesz); if (line == NULL) { ret = -1; diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 1329b6b..ce8dc61 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -5,6 +5,7 @@ #include #include "callchain.h" #include "header.h" +#include "color.h" extern struct callchain_param callchain_param; @@ -175,6 +176,18 @@ void perf_hpp__init(void); void perf_hpp__column_register(struct perf_hpp_fmt *format); void perf_hpp__column_enable(unsigned col); +static inline size_t perf_hpp__use_color(void) +{ + return !symbol_conf.field_sep; +} + +static inline size_t perf_hpp__color_overhead(void) +{ + return perf_hpp__use_color() ? + (COLOR_MAXLEN + sizeof(PERF_COLOR_RESET)) * PERF_HPP__MAX_INDEX + : 0; +} + struct perf_evlist; struct hist_browser_timer { -- cgit v0.10.2 From 09b0fd45ff63413df94cbd832a765076b201edbb Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Sat, 26 Oct 2013 16:25:33 +0200 Subject: perf record: Split -g and --call-graph Splitting -g and --call-graph for record command, so we could use '-g' with no option. The '-g' option now takes NO argument and enables the configured unwind method, which is currently the frame pointers method. It will be possible to configure unwind method via config file in upcoming patches. All current '-g' arguments is overtaken by --call-graph option. Signed-off-by: Jiri Olsa Tested-by: David Ahern Tested-by: Ingo Molnar Reviewed-by: David Ahern Acked-by: Ingo Molnar Cc: Adrian Hunter Cc: Andi Kleen Cc: Corey Ashford Cc: David Ahern Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382797536-32303-2-git-send-email-jolsa@redhat.com [ reordered -g/--call-graph on --help and expanded the man page according to comments by David Ahern and Namhyung Kim ] Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index e297b74..ca0d3d9 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -90,8 +90,20 @@ OPTIONS Number of mmap data pages. Must be a power of two. -g:: + Enables call-graph (stack chain/backtrace) recording. + --call-graph:: - Do call-graph (stack chain/backtrace) recording. + Setup and enable call-graph (stack chain/backtrace) recording, + implies -g. + + Allows specifying "fp" (frame pointer) or "dwarf" + (DWARF's CFI - Call Frame Information) as the method to collect + the information used to show the call graphs. + + In some systems, where binaries are build with gcc + --fomit-frame-pointer, using the "fp" method will produce bogus + call graphs, using "dwarf", if available (perf tools linked to + the libunwind library) should be used instead. -q:: --quiet:: diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index a41ac415..d046514 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -712,21 +712,12 @@ static int get_stack_size(char *str, unsigned long *_size) } #endif /* LIBUNWIND_SUPPORT */ -int record_parse_callchain_opt(const struct option *opt, - const char *arg, int unset) +int record_parse_callchain(const char *arg, struct perf_record_opts *opts) { - struct perf_record_opts *opts = opt->value; char *tok, *name, *saveptr = NULL; char *buf; int ret = -1; - /* --no-call-graph */ - if (unset) - return 0; - - /* We specified default option if none is provided. */ - BUG_ON(!arg); - /* We need buffer that we know we can write to. */ buf = malloc(strlen(arg) + 1); if (!buf) @@ -764,13 +755,9 @@ int record_parse_callchain_opt(const struct option *opt, ret = get_stack_size(tok, &size); opts->stack_dump_size = size; } - - if (!ret) - pr_debug("callchain: stack dump size %d\n", - opts->stack_dump_size); #endif /* LIBUNWIND_SUPPORT */ } else { - pr_err("callchain: Unknown -g option " + pr_err("callchain: Unknown --call-graph option " "value: %s\n", arg); break; } @@ -778,13 +765,52 @@ int record_parse_callchain_opt(const struct option *opt, } while (0); free(buf); + return ret; +} + +static void callchain_debug(struct perf_record_opts *opts) +{ + pr_debug("callchain: type %d\n", opts->call_graph); + if (opts->call_graph == CALLCHAIN_DWARF) + pr_debug("callchain: stack dump size %d\n", + opts->stack_dump_size); +} + +int record_parse_callchain_opt(const struct option *opt, + const char *arg, + int unset) +{ + struct perf_record_opts *opts = opt->value; + int ret; + + /* --no-call-graph */ + if (unset) { + opts->call_graph = CALLCHAIN_NONE; + pr_debug("callchain: disabled\n"); + return 0; + } + + ret = record_parse_callchain(arg, opts); if (!ret) - pr_debug("callchain: type %d\n", opts->call_graph); + callchain_debug(opts); return ret; } +int record_callchain_opt(const struct option *opt, + const char *arg __maybe_unused, + int unset __maybe_unused) +{ + struct perf_record_opts *opts = opt->value; + + if (opts->call_graph == CALLCHAIN_NONE) + opts->call_graph = CALLCHAIN_FP; + + callchain_debug(opts); + return 0; +} + static const char * const record_usage[] = { "perf record [] []", "perf record [] -- []", @@ -813,12 +839,12 @@ static struct perf_record record = { }, }; -#define CALLCHAIN_HELP "do call-graph (stack chain/backtrace) recording: " +#define CALLCHAIN_HELP "setup and enables call-graph (stack chain/backtrace) recording: " #ifdef LIBUNWIND_SUPPORT -const char record_callchain_help[] = CALLCHAIN_HELP "[fp] dwarf"; +const char record_callchain_help[] = CALLCHAIN_HELP "fp dwarf"; #else -const char record_callchain_help[] = CALLCHAIN_HELP "[fp]"; +const char record_callchain_help[] = CALLCHAIN_HELP "fp"; #endif /* @@ -858,9 +884,12 @@ const struct option record_options[] = { "number of mmap data pages"), OPT_BOOLEAN(0, "group", &record.opts.group, "put the counters into a counter group"), - OPT_CALLBACK_DEFAULT('g', "call-graph", &record.opts, - "mode[,dump_size]", record_callchain_help, - &record_parse_callchain_opt, "fp"), + OPT_CALLBACK_NOOPT('g', NULL, &record.opts, + NULL, "enables call-graph recording" , + &record_callchain_opt), + OPT_CALLBACK(0, "call-graph", &record.opts, + "mode[,dump_size]", record_callchain_help, + &record_parse_callchain_opt), OPT_INCR('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), OPT_BOOLEAN('q', "quiet", &quiet, "don't print any message"), diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 2b585bc..9e99060 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -147,6 +147,9 @@ static inline void callchain_cursor_advance(struct callchain_cursor *cursor) struct option; +int record_parse_callchain(const char *arg, struct perf_record_opts *opts); int record_parse_callchain_opt(const struct option *opt, const char *arg, int unset); +int record_callchain_opt(const struct option *opt, const char *arg, int unset); + extern const char record_callchain_help[]; #endif /* __PERF_CALLCHAIN_H */ -- cgit v0.10.2 From ae779a630977d93fbebfa06216ea47df5b5c62c8 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Sat, 26 Oct 2013 16:25:34 +0200 Subject: perf top: Split -G and --call-graph Splitting -G and --call-graph for record command, so we could use '-G' with no option. The '-G' option now takes NO argument and enables the configured unwind method, which is currently the frame pointers method. It will be possible to configure unwind method via config file in upcoming patches. All current '-G' arguments is overtaken by --call-graph option. NOTE: The documentation for top --call-graph option was wrongly copied from report command. Signed-off-by: Jiri Olsa Tested-by: David Ahern Tested-by: Ingo Molnar Reviewed-by: David Ahern Acked-by: Ingo Molnar Cc: Adrian Hunter Cc: Andi Kleen Cc: Corey Ashford Cc: David Ahern Cc: Ingo Molnar Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1382797536-32303-3-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/Documentation/perf-top.txt b/tools/perf/Documentation/perf-top.txt index 58d6598..6a118e7 100644 --- a/tools/perf/Documentation/perf-top.txt +++ b/tools/perf/Documentation/perf-top.txt @@ -140,20 +140,12 @@ Default is to monitor all CPUS. --asm-raw:: Show raw instruction encoding of assembly instructions. --G [type,min,order]:: +-G:: + Enables call-graph (stack chain/backtrace) recording. + --call-graph:: - Display call chains using type, min percent threshold and order. - type can be either: - - flat: single column, linear exposure of call chains. - - graph: use a graph tree, displaying absolute overhead rates. - - fractal: like graph, but displays relative rates. Each branch of - the tree is considered as a new profiled object. - - order can be either: - - callee: callee based call graph. - - caller: inverted caller based call graph. - - Default: fractal,0.5,callee. + Setup and enable call-graph (stack chain/backtrace) recording, + implies -G. --ignore-callees=:: Ignore callees of the function(s) matching the given regex. diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 2122141..0df298a 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1016,16 +1016,16 @@ out_delete: } static int -parse_callchain_opt(const struct option *opt, const char *arg, int unset) +callchain_opt(const struct option *opt, const char *arg, int unset) { - /* - * --no-call-graph - */ - if (unset) - return 0; - symbol_conf.use_callchain = true; + return record_callchain_opt(opt, arg, unset); +} +static int +parse_callchain_opt(const struct option *opt, const char *arg, int unset) +{ + symbol_conf.use_callchain = true; return record_parse_callchain_opt(opt, arg, unset); } @@ -1106,9 +1106,12 @@ int cmd_top(int argc, const char **argv, const char *prefix __maybe_unused) "sort by key(s): pid, comm, dso, symbol, parent, weight, local_weight"), OPT_BOOLEAN('n', "show-nr-samples", &symbol_conf.show_nr_samples, "Show a column with the number of samples"), - OPT_CALLBACK_DEFAULT('G', "call-graph", &top.record_opts, - "mode[,dump_size]", record_callchain_help, - &parse_callchain_opt, "fp"), + OPT_CALLBACK_NOOPT('G', NULL, &top.record_opts, + NULL, "enables call-graph recording", + &callchain_opt), + OPT_CALLBACK(0, "call-graph", &top.record_opts, + "mode[,dump_size]", record_callchain_help, + &parse_callchain_opt), OPT_CALLBACK(0, "ignore-callees", NULL, "regex", "ignore callees of these functions in call graphs", report_parse_ignore_callees_opt), -- cgit v0.10.2 From 8e50d384cc1d5afd2989cf0f7093756ed7164eb2 Mon Sep 17 00:00:00 2001 From: Zhouyi Zhou Date: Thu, 24 Oct 2013 15:43:33 +0800 Subject: perf tools: Fixup mmap event consumption The tail position of the event buffer should only be modified after actually use that event. If not the event buffer could be invalid before use, and segment fault occurs when invoking perf top -G. Signed-off-by: Zhouyi Zhou Cc: David Ahern Cc: Zhouyi Zhou Link: http://lkml.kernel.org/r/1382600613-32177-1-git-send-email-zhouzhouyi@gmail.com [ Simplified the logic using exit gotos and renamed write_tail method to mmap_consume ] Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c index 935d522..fbc2888 100644 --- a/tools/perf/builtin-kvm.c +++ b/tools/perf/builtin-kvm.c @@ -888,11 +888,18 @@ static s64 perf_kvm__mmap_read_idx(struct perf_kvm_stat *kvm, int idx, while ((event = perf_evlist__mmap_read(kvm->evlist, idx)) != NULL) { err = perf_evlist__parse_sample(kvm->evlist, event, &sample); if (err) { + perf_evlist__mmap_consume(kvm->evlist, idx); pr_err("Failed to parse sample\n"); return -1; } err = perf_session_queue_event(kvm->session, event, &sample, 0); + /* + * FIXME: Here we can't consume the event, as perf_session_queue_event will + * point to it, and it'll get possibly overwritten by the kernel. + */ + perf_evlist__mmap_consume(kvm->evlist, idx); + if (err) { pr_err("Failed to enqueue sample: %d\n", err); return -1; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 0df298a..5a11f13 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -810,7 +810,7 @@ static void perf_top__mmap_read_idx(struct perf_top *top, int idx) ret = perf_evlist__parse_sample(top->evlist, event, &sample); if (ret) { pr_err("Can't parse sample, err = %d\n", ret); - continue; + goto next_event; } evsel = perf_evlist__id2evsel(session->evlist, sample.id); @@ -825,13 +825,13 @@ static void perf_top__mmap_read_idx(struct perf_top *top, int idx) case PERF_RECORD_MISC_USER: ++top->us_samples; if (top->hide_user_symbols) - continue; + goto next_event; machine = &session->machines.host; break; case PERF_RECORD_MISC_KERNEL: ++top->kernel_samples; if (top->hide_kernel_symbols) - continue; + goto next_event; machine = &session->machines.host; break; case PERF_RECORD_MISC_GUEST_KERNEL: @@ -847,7 +847,7 @@ static void perf_top__mmap_read_idx(struct perf_top *top, int idx) */ /* Fall thru */ default: - continue; + goto next_event; } @@ -859,6 +859,8 @@ static void perf_top__mmap_read_idx(struct perf_top *top, int idx) machine__process_event(machine, event); } else ++session->stats.nr_unknown_events; +next_event: + perf_evlist__mmap_consume(top->evlist, idx); } } diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c index 71aa3e3..99c8d9a 100644 --- a/tools/perf/builtin-trace.c +++ b/tools/perf/builtin-trace.c @@ -987,7 +987,7 @@ again: err = perf_evlist__parse_sample(evlist, event, &sample); if (err) { fprintf(trace->output, "Can't parse sample, err = %d, skipping...\n", err); - continue; + goto next_event; } if (trace->base_time == 0) @@ -1001,18 +1001,20 @@ again: evsel = perf_evlist__id2evsel(evlist, sample.id); if (evsel == NULL) { fprintf(trace->output, "Unknown tp ID %" PRIu64 ", skipping...\n", sample.id); - continue; + goto next_event; } if (sample.raw_data == NULL) { fprintf(trace->output, "%s sample with no payload for tid: %d, cpu %d, raw_size=%d, skipping...\n", perf_evsel__name(evsel), sample.tid, sample.cpu, sample.raw_size); - continue; + goto next_event; } handler = evsel->handler.func; handler(trace, evsel, &sample); +next_event: + perf_evlist__mmap_consume(evlist, i); if (done) goto out_unmap_evlist; diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c index 6fb781d..e3fedfa 100644 --- a/tools/perf/tests/code-reading.c +++ b/tools/perf/tests/code-reading.c @@ -290,6 +290,7 @@ static int process_events(struct machine *machine, struct perf_evlist *evlist, for (i = 0; i < evlist->nr_mmaps; i++) { while ((event = perf_evlist__mmap_read(evlist, i)) != NULL) { ret = process_event(machine, evlist, event, state); + perf_evlist__mmap_consume(evlist, i); if (ret < 0) return ret; } diff --git a/tools/perf/tests/keep-tracking.c b/tools/perf/tests/keep-tracking.c index d444ea2..376c356 100644 --- a/tools/perf/tests/keep-tracking.c +++ b/tools/perf/tests/keep-tracking.c @@ -36,6 +36,7 @@ static int find_comm(struct perf_evlist *evlist, const char *comm) (pid_t)event->comm.tid == getpid() && strcmp(event->comm.comm, comm) == 0) found += 1; + perf_evlist__mmap_consume(evlist, i); } } return found; diff --git a/tools/perf/tests/mmap-basic.c b/tools/perf/tests/mmap-basic.c index c4185b9..a7232c2 100644 --- a/tools/perf/tests/mmap-basic.c +++ b/tools/perf/tests/mmap-basic.c @@ -122,6 +122,7 @@ int test__basic_mmap(void) goto out_munmap; } nr_events[evsel->idx]++; + perf_evlist__mmap_consume(evlist, 0); } err = 0; diff --git a/tools/perf/tests/open-syscall-tp-fields.c b/tools/perf/tests/open-syscall-tp-fields.c index fc5b9fc..524b221 100644 --- a/tools/perf/tests/open-syscall-tp-fields.c +++ b/tools/perf/tests/open-syscall-tp-fields.c @@ -77,8 +77,10 @@ int test__syscall_open_tp_fields(void) ++nr_events; - if (type != PERF_RECORD_SAMPLE) + if (type != PERF_RECORD_SAMPLE) { + perf_evlist__mmap_consume(evlist, i); continue; + } err = perf_evsel__parse_sample(evsel, event, &sample); if (err) { diff --git a/tools/perf/tests/perf-record.c b/tools/perf/tests/perf-record.c index b8a7056..7923b06 100644 --- a/tools/perf/tests/perf-record.c +++ b/tools/perf/tests/perf-record.c @@ -263,6 +263,8 @@ int test__PERF_RECORD(void) type); ++errs; } + + perf_evlist__mmap_consume(evlist, i); } } diff --git a/tools/perf/tests/perf-time-to-tsc.c b/tools/perf/tests/perf-time-to-tsc.c index 0ab61b1..4ca1b93 100644 --- a/tools/perf/tests/perf-time-to-tsc.c +++ b/tools/perf/tests/perf-time-to-tsc.c @@ -122,7 +122,7 @@ int test__perf_time_to_tsc(void) if (event->header.type != PERF_RECORD_COMM || (pid_t)event->comm.pid != getpid() || (pid_t)event->comm.tid != getpid()) - continue; + goto next_event; if (strcmp(event->comm.comm, comm1) == 0) { CHECK__(perf_evsel__parse_sample(evsel, event, @@ -134,6 +134,8 @@ int test__perf_time_to_tsc(void) &sample)); comm2_time = sample.time; } +next_event: + perf_evlist__mmap_consume(evlist, i); } } diff --git a/tools/perf/tests/sw-clock.c b/tools/perf/tests/sw-clock.c index 2e41e2d..6e2b44e 100644 --- a/tools/perf/tests/sw-clock.c +++ b/tools/perf/tests/sw-clock.c @@ -78,7 +78,7 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id) struct perf_sample sample; if (event->header.type != PERF_RECORD_SAMPLE) - continue; + goto next_event; err = perf_evlist__parse_sample(evlist, event, &sample); if (err < 0) { @@ -88,6 +88,8 @@ static int __test__sw_clock_freq(enum perf_sw_ids clock_id) total_periods += sample.period; nr_samples++; +next_event: + perf_evlist__mmap_consume(evlist, 0); } if ((u64) nr_samples == total_periods) { diff --git a/tools/perf/tests/task-exit.c b/tools/perf/tests/task-exit.c index 28fe589..a3e6487 100644 --- a/tools/perf/tests/task-exit.c +++ b/tools/perf/tests/task-exit.c @@ -96,10 +96,10 @@ int test__task_exit(void) retry: while ((event = perf_evlist__mmap_read(evlist, 0)) != NULL) { - if (event->header.type != PERF_RECORD_EXIT) - continue; + if (event->header.type == PERF_RECORD_EXIT) + nr_exit++; - nr_exit++; + perf_evlist__mmap_consume(evlist, 0); } if (!exited || !nr_exit) { diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index f9f77be..e584cd3 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -545,12 +545,19 @@ union perf_event *perf_evlist__mmap_read(struct perf_evlist *evlist, int idx) md->prev = old; - if (!evlist->overwrite) - perf_mmap__write_tail(md, old); - return event; } +void perf_evlist__mmap_consume(struct perf_evlist *evlist, int idx) +{ + if (!evlist->overwrite) { + struct perf_mmap *md = &evlist->mmap[idx]; + unsigned int old = md->prev; + + perf_mmap__write_tail(md, old); + } +} + static void __perf_evlist__munmap(struct perf_evlist *evlist, int idx) { if (evlist->mmap[idx].base != NULL) { diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 880d713..206d093 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -89,6 +89,8 @@ struct perf_sample_id *perf_evlist__id2sid(struct perf_evlist *evlist, u64 id); union perf_event *perf_evlist__mmap_read(struct perf_evlist *self, int idx); +void perf_evlist__mmap_consume(struct perf_evlist *evlist, int idx); + int perf_evlist__open(struct perf_evlist *evlist); void perf_evlist__close(struct perf_evlist *evlist); diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index 71b5412..2ac4bc9 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -822,6 +822,8 @@ static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist, PyObject *pyevent = pyrf_event__new(event); struct pyrf_event *pevent = (struct pyrf_event *)pyevent; + perf_evlist__mmap_consume(evlist, cpu); + if (pyevent == NULL) return PyErr_NoMemory(); -- cgit v0.10.2