From 27f3b24de03fc7cec6f2406f8525ad18086c2121 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 16 Dec 2009 17:16:19 -0500 Subject: [PATCH 01/10] perf probe: Fix libdwarf include path for Debian Fix libdwarf include path to fit debian-like systems too. Borislav Petkov reported: > even after installing libdwarf-dev on my debian box here, > make in tools/perf/ still complains that it cannot find libdwarf: > > Makefile:491: No libdwarf.h found or old libdwarf.h found, disables dwarf > support. Please install libdwarf-dev/libdwarf-devel >= 20081231 > > The problem is that the include path on debian is not > /usr/include/libdwarf/ but simply /usr/include because the debian > package libdwarf-dev puts the headers straight into > /usr/include. This patch adds -I/usr/include/libdwarf to BASIC_CFLAGS and fix probe-finder.h to include just libdwarf.h/dwarf.h. This patch also adds a workaround for the undefined _MIPS_SZLONG bug in libdwarf.h. Reported-by: Borislav Petkov Signed-off-by: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Gabor Gombas Cc: systemtap Cc: DLE LKML-Reference: <20091216221618.13816.83296.stgit@dhcp-100-2-132.bos.redhat.com> [ v2: small stylistic fixlets to probe-finder.h ] Signed-off-by: Ingo Molnar --- tools/perf/Makefile | 3 +- tools/perf/util/probe-finder.h | 59 ++++++++++++++++++---------------- 2 files changed, 34 insertions(+), 28 deletions(-) diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 7814dbbd401d..4390d225686d 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -487,10 +487,11 @@ else msg := $(error No libelf.h/libelf found, please install libelf-dev/elfutils-libelf-devel and glibc-dev[el]); endif -ifneq ($(shell sh -c "(echo '\#include '; echo '\#include '; echo 'int main(void) { Dwarf_Debug dbg; Dwarf_Error err; Dwarf_Ranges *rng; dwarf_init(0, DW_DLC_READ, 0, 0, &dbg, &err); dwarf_get_ranges(dbg, 0, &rng, 0, 0, &err); return (long)dbg; }') | $(CC) -x c - $(ALL_CFLAGS) -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -ldwarf -lelf -o /dev/null $(ALL_LDFLAGS) $(EXTLIBS) "$(QUIET_STDERR)" && echo y"), y) +ifneq ($(shell sh -c "(echo '\#ifndef _MIPS_SZLONG'; echo '\#define _MIPS_SZLONG 0'; echo '\#endif'; echo '\#include '; echo '\#include '; echo 'int main(void) { Dwarf_Debug dbg; Dwarf_Error err; Dwarf_Ranges *rng; dwarf_init(0, DW_DLC_READ, 0, 0, &dbg, &err); dwarf_get_ranges(dbg, 0, &rng, 0, 0, &err); return (long)dbg; }') | $(CC) -x c - $(ALL_CFLAGS) -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -I/usr/include/libdwarf -ldwarf -lelf -o /dev/null $(ALL_LDFLAGS) $(EXTLIBS) "$(QUIET_STDERR)" && echo y"), y) msg := $(warning No libdwarf.h found or old libdwarf.h found, disables dwarf support. Please install libdwarf-dev/libdwarf-devel >= 20081231); BASIC_CFLAGS += -DNO_LIBDWARF else + BASIC_CFLAGS += -I/usr/include/libdwarf EXTLIBS += -lelf -ldwarf LIB_OBJS += util/probe-finder.o endif diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h index 5e4050ce2963..a4086aaddb73 100644 --- a/tools/perf/util/probe-finder.h +++ b/tools/perf/util/probe-finder.h @@ -1,9 +1,9 @@ #ifndef _PROBE_FINDER_H #define _PROBE_FINDER_H -#define MAX_PATH_LEN 256 -#define MAX_PROBE_BUFFER 1024 -#define MAX_PROBES 128 +#define MAX_PATH_LEN 256 +#define MAX_PROBE_BUFFER 1024 +#define MAX_PROBES 128 static inline int is_c_varname(const char *name) { @@ -12,48 +12,53 @@ static inline int is_c_varname(const char *name) } struct probe_point { - char *event; /* Event name */ - char *group; /* Event group */ + char *event; /* Event name */ + char *group; /* Event group */ /* Inputs */ - char *file; /* File name */ - int line; /* Line number */ + char *file; /* File name */ + int line; /* Line number */ - char *function; /* Function name */ - int offset; /* Offset bytes */ + char *function; /* Function name */ + int offset; /* Offset bytes */ - int nr_args; /* Number of arguments */ - char **args; /* Arguments */ + int nr_args; /* Number of arguments */ + char **args; /* Arguments */ - int retprobe; /* Return probe */ + int retprobe; /* Return probe */ /* Output */ - int found; /* Number of found probe points */ - char *probes[MAX_PROBES]; /* Output buffers (will be allocated)*/ + int found; /* Number of found probe points */ + char *probes[MAX_PROBES]; /* Output buffers (will be allocated)*/ }; #ifndef NO_LIBDWARF extern int find_probepoint(int fd, struct probe_point *pp); -#include -#include +/* Workaround for undefined _MIPS_SZLONG bug in libdwarf.h: */ +#ifndef _MIPS_SZLONG +# define _MIPS_SZLONG 0 +#endif + +#include +#include struct probe_finder { - struct probe_point *pp; /* Target probe point */ + struct probe_point *pp; /* Target probe point */ /* For function searching */ - Dwarf_Addr addr; /* Address */ - Dwarf_Unsigned fno; /* File number */ - Dwarf_Unsigned lno; /* Line number */ - Dwarf_Off inl_offs; /* Inline offset */ - Dwarf_Die cu_die; /* Current CU */ + Dwarf_Addr addr; /* Address */ + Dwarf_Unsigned fno; /* File number */ + Dwarf_Unsigned lno; /* Line number */ + Dwarf_Off inl_offs; /* Inline offset */ + Dwarf_Die cu_die; /* Current CU */ /* For variable searching */ - Dwarf_Addr cu_base; /* Current CU base address */ - Dwarf_Locdesc fbloc; /* Location of Current Frame Base */ - const char *var; /* Current variable name */ - char *buf; /* Current output buffer */ - int len; /* Length of output buffer */ + Dwarf_Addr cu_base; /* Current CU base address */ + Dwarf_Locdesc fbloc; /* Location of Current Frame Base */ + const char *var; /* Current variable name */ + char *buf; /* Current output buffer */ + int len; /* Length of output buffer */ }; #endif /* NO_LIBDWARF */ From 96c96612e952f63cc0055db9df7d8b5b1ada02be Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 16 Dec 2009 17:24:00 -0500 Subject: [PATCH 02/10] perf probe: Check whether debugfs path is correct Check whether the debugfs path is correct before executing a command, because perf-probe depends on debugfs. Signed-off-by: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Jim Keniston Cc: Ananth N Mavinakayanahalli Cc: Christoph Hellwig Cc: Jason Baron Cc: K.Prasad Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: systemtap Cc: DLE LKML-Reference: <20091216222400.14459.48162.stgit@dhcp-100-2-132.bos.redhat.com> Signed-off-by: Ingo Molnar --- tools/perf/builtin-probe.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c index 7e741f54d798..c1e6774fd3ed 100644 --- a/tools/perf/builtin-probe.c +++ b/tools/perf/builtin-probe.c @@ -38,6 +38,7 @@ #include "util/strlist.h" #include "util/event.h" #include "util/debug.h" +#include "util/debugfs.h" #include "util/symbol.h" #include "util/thread.h" #include "util/session.h" @@ -205,6 +206,9 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) if ((!session.nr_probe && !session.dellist && !session.list_events)) usage_with_options(probe_usage, options); + if (debugfs_valid_mountpoint(debugfs_path) < 0) + die("Failed to find debugfs path."); + if (session.list_events) { if (session.nr_probe != 0 || session.dellist) { pr_warning(" Error: Don't use --list with" From 6f3cf440470650b3841d325acacd0c5ea9504c68 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 16 Dec 2009 17:24:08 -0500 Subject: [PATCH 03/10] kprobe-tracer: Check new event/group name Check new event/group name is same syntax as a C symbol. In other words, checking the name is as like as other tracepoint events. This can prevent user to create an event with useless name (e.g. foo|bar, foo*bar). Signed-off-by: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Jim Keniston Cc: Ananth N Mavinakayanahalli Cc: Christoph Hellwig Cc: Jason Baron Cc: K.Prasad Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: systemtap Cc: DLE LKML-Reference: <20091216222408.14459.68790.stgit@dhcp-100-2-132.bos.redhat.com> [ v2: minor cleanups ] Signed-off-by: Ingo Molnar --- kernel/trace/trace_kprobe.c | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-) diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 7ecab06547a5..375f81a568dc 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -282,6 +282,18 @@ static int kprobe_dispatcher(struct kprobe *kp, struct pt_regs *regs); static int kretprobe_dispatcher(struct kretprobe_instance *ri, struct pt_regs *regs); +/* Check the name is good for event/group */ +static int check_event_name(const char *name) +{ + if (!isalpha(*name) && *name != '_') + return 0; + while (*++name != '\0') { + if (!isalpha(*name) && !isdigit(*name) && *name != '_') + return 0; + } + return 1; +} + /* * Allocate new trace_probe and initialize it (including kprobes). */ @@ -293,10 +305,11 @@ static struct trace_probe *alloc_trace_probe(const char *group, int nargs, int is_return) { struct trace_probe *tp; + int ret = -ENOMEM; tp = kzalloc(SIZEOF_TRACE_PROBE(nargs), GFP_KERNEL); if (!tp) - return ERR_PTR(-ENOMEM); + return ERR_PTR(ret); if (symbol) { tp->symbol = kstrdup(symbol, GFP_KERNEL); @@ -312,14 +325,20 @@ static struct trace_probe *alloc_trace_probe(const char *group, else tp->rp.kp.pre_handler = kprobe_dispatcher; - if (!event) + if (!event || !check_event_name(event)) { + ret = -EINVAL; goto error; + } + tp->call.name = kstrdup(event, GFP_KERNEL); if (!tp->call.name) goto error; - if (!group) + if (!group || !check_event_name(group)) { + ret = -EINVAL; goto error; + } + tp->call.system = kstrdup(group, GFP_KERNEL); if (!tp->call.system) goto error; @@ -330,7 +349,7 @@ error: kfree(tp->call.name); kfree(tp->symbol); kfree(tp); - return ERR_PTR(-ENOMEM); + return ERR_PTR(ret); } static void free_probe_arg(struct probe_arg *arg) @@ -695,10 +714,10 @@ static int create_trace_probe(int argc, char **argv) if (!event) { /* Make a new event name */ if (symbol) - snprintf(buf, MAX_EVENT_NAME_LEN, "%c@%s%+ld", + snprintf(buf, MAX_EVENT_NAME_LEN, "%c_%s_%ld", is_return ? 'r' : 'p', symbol, offset); else - snprintf(buf, MAX_EVENT_NAME_LEN, "%c@0x%p", + snprintf(buf, MAX_EVENT_NAME_LEN, "%c_0x%p", is_return ? 'r' : 'p', addr); event = buf; } From b7702a2136b5f8e0e186e22cae91aaecf98b418c Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Wed, 16 Dec 2009 17:24:15 -0500 Subject: [PATCH 04/10] perf probe: Check new event name Check new event name is same syntax as a C symbol in perf command. In other words, checking the name is as like as other tracepoint events. This can prevent user to create an event with useless name (e.g. foo|bar, foo*bar). Signed-off-by: Masami Hiramatsu Cc: Frederic Weisbecker Cc: Paul Mackerras Cc: Arnaldo Carvalho de Melo Cc: Steven Rostedt Cc: Jim Keniston Cc: Ananth N Mavinakayanahalli Cc: Christoph Hellwig Cc: Jason Baron Cc: K.Prasad Cc: Peter Zijlstra Cc: Srikar Dronamraju Cc: Frederic Weisbecker Cc: systemtap Cc: DLE LKML-Reference: <20091216222415.14459.71383.stgit@dhcp-100-2-132.bos.redhat.com> [ v2: minor cleanups ] Signed-off-by: Ingo Molnar --- tools/perf/util/probe-event.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 2ca62154f79b..29465d440043 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -62,6 +62,18 @@ static int e_snprintf(char *str, size_t size, const char *format, ...) return ret; } +/* Check the name is good for event/group */ +static bool check_event_name(const char *name) +{ + if (!isalpha(*name) && *name != '_') + return false; + while (*++name != '\0') { + if (!isalpha(*name) && !isdigit(*name) && *name != '_') + return false; + } + return true; +} + /* Parse probepoint definition. */ static void parse_perf_probe_probepoint(char *arg, struct probe_point *pp) { @@ -82,6 +94,9 @@ static void parse_perf_probe_probepoint(char *arg, struct probe_point *pp) ptr = strchr(arg, ':'); if (ptr) /* Group name is not supported yet. */ semantic_error("Group name is not supported yet."); + if (!check_event_name(arg)) + semantic_error("%s is bad for event name -it must " + "follow C symbol-naming rule.", arg); pp->event = strdup(arg); arg = tmp; } From 5b74ed4729ad2b2017453add68104a83206caefb Mon Sep 17 00:00:00 2001 From: "Robert P. J. Day" Date: Wed, 16 Dec 2009 10:09:45 -0500 Subject: [PATCH 05/10] perf events: Remove unused perf_counter.h header file Since nothing includes the file and it's also not exported to user space, remove it. Signed-off-by: Robert P. J. Day Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/perf_counter.h | 444 ----------------------------------- 1 file changed, 444 deletions(-) delete mode 100644 include/linux/perf_counter.h diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h deleted file mode 100644 index e3fb25606706..000000000000 --- a/include/linux/perf_counter.h +++ /dev/null @@ -1,444 +0,0 @@ -/* - * NOTE: this file will be removed in a future kernel release, it is - * provided as a courtesy copy of user-space code that relies on the - * old (pre-rename) symbols and constants. - * - * Performance events: - * - * Copyright (C) 2008-2009, Thomas Gleixner - * Copyright (C) 2008-2009, Red Hat, Inc., Ingo Molnar - * Copyright (C) 2008-2009, Red Hat, Inc., Peter Zijlstra - * - * Data type definitions, declarations, prototypes. - * - * Started by: Thomas Gleixner and Ingo Molnar - * - * For licencing details see kernel-base/COPYING - */ -#ifndef _LINUX_PERF_COUNTER_H -#define _LINUX_PERF_COUNTER_H - -#include -#include -#include - -/* - * User-space ABI bits: - */ - -/* - * attr.type - */ -enum perf_type_id { - PERF_TYPE_HARDWARE = 0, - PERF_TYPE_SOFTWARE = 1, - PERF_TYPE_TRACEPOINT = 2, - PERF_TYPE_HW_CACHE = 3, - PERF_TYPE_RAW = 4, - - PERF_TYPE_MAX, /* non-ABI */ -}; - -/* - * Generalized performance counter event types, used by the - * attr.event_id parameter of the sys_perf_counter_open() - * syscall: - */ -enum perf_hw_id { - /* - * Common hardware events, generalized by the kernel: - */ - PERF_COUNT_HW_CPU_CYCLES = 0, - PERF_COUNT_HW_INSTRUCTIONS = 1, - PERF_COUNT_HW_CACHE_REFERENCES = 2, - PERF_COUNT_HW_CACHE_MISSES = 3, - PERF_COUNT_HW_BRANCH_INSTRUCTIONS = 4, - PERF_COUNT_HW_BRANCH_MISSES = 5, - PERF_COUNT_HW_BUS_CYCLES = 6, - - PERF_COUNT_HW_MAX, /* non-ABI */ -}; - -/* - * Generalized hardware cache counters: - * - * { L1-D, L1-I, LLC, ITLB, DTLB, BPU } x - * { read, write, prefetch } x - * { accesses, misses } - */ -enum perf_hw_cache_id { - PERF_COUNT_HW_CACHE_L1D = 0, - PERF_COUNT_HW_CACHE_L1I = 1, - PERF_COUNT_HW_CACHE_LL = 2, - PERF_COUNT_HW_CACHE_DTLB = 3, - PERF_COUNT_HW_CACHE_ITLB = 4, - PERF_COUNT_HW_CACHE_BPU = 5, - - PERF_COUNT_HW_CACHE_MAX, /* non-ABI */ -}; - -enum perf_hw_cache_op_id { - PERF_COUNT_HW_CACHE_OP_READ = 0, - PERF_COUNT_HW_CACHE_OP_WRITE = 1, - PERF_COUNT_HW_CACHE_OP_PREFETCH = 2, - - PERF_COUNT_HW_CACHE_OP_MAX, /* non-ABI */ -}; - -enum perf_hw_cache_op_result_id { - PERF_COUNT_HW_CACHE_RESULT_ACCESS = 0, - PERF_COUNT_HW_CACHE_RESULT_MISS = 1, - - PERF_COUNT_HW_CACHE_RESULT_MAX, /* non-ABI */ -}; - -/* - * Special "software" counters provided by the kernel, even if the hardware - * does not support performance counters. These counters measure various - * physical and sw events of the kernel (and allow the profiling of them as - * well): - */ -enum perf_sw_ids { - PERF_COUNT_SW_CPU_CLOCK = 0, - PERF_COUNT_SW_TASK_CLOCK = 1, - PERF_COUNT_SW_PAGE_FAULTS = 2, - PERF_COUNT_SW_CONTEXT_SWITCHES = 3, - PERF_COUNT_SW_CPU_MIGRATIONS = 4, - PERF_COUNT_SW_PAGE_FAULTS_MIN = 5, - PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, - PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, - PERF_COUNT_SW_EMULATION_FAULTS = 8, - - PERF_COUNT_SW_MAX, /* non-ABI */ -}; - -/* - * Bits that can be set in attr.sample_type to request information - * in the overflow packets. - */ -enum perf_counter_sample_format { - PERF_SAMPLE_IP = 1U << 0, - PERF_SAMPLE_TID = 1U << 1, - PERF_SAMPLE_TIME = 1U << 2, - PERF_SAMPLE_ADDR = 1U << 3, - PERF_SAMPLE_READ = 1U << 4, - PERF_SAMPLE_CALLCHAIN = 1U << 5, - PERF_SAMPLE_ID = 1U << 6, - PERF_SAMPLE_CPU = 1U << 7, - PERF_SAMPLE_PERIOD = 1U << 8, - PERF_SAMPLE_STREAM_ID = 1U << 9, - PERF_SAMPLE_RAW = 1U << 10, - - PERF_SAMPLE_MAX = 1U << 11, /* non-ABI */ -}; - -/* - * The format of the data returned by read() on a perf counter fd, - * as specified by attr.read_format: - * - * struct read_format { - * { u64 value; - * { u64 time_enabled; } && PERF_FORMAT_ENABLED - * { u64 time_running; } && PERF_FORMAT_RUNNING - * { u64 id; } && PERF_FORMAT_ID - * } && !PERF_FORMAT_GROUP - * - * { u64 nr; - * { u64 time_enabled; } && PERF_FORMAT_ENABLED - * { u64 time_running; } && PERF_FORMAT_RUNNING - * { u64 value; - * { u64 id; } && PERF_FORMAT_ID - * } cntr[nr]; - * } && PERF_FORMAT_GROUP - * }; - */ -enum perf_counter_read_format { - PERF_FORMAT_TOTAL_TIME_ENABLED = 1U << 0, - PERF_FORMAT_TOTAL_TIME_RUNNING = 1U << 1, - PERF_FORMAT_ID = 1U << 2, - PERF_FORMAT_GROUP = 1U << 3, - - PERF_FORMAT_MAX = 1U << 4, /* non-ABI */ -}; - -#define PERF_ATTR_SIZE_VER0 64 /* sizeof first published struct */ - -/* - * Hardware event to monitor via a performance monitoring counter: - */ -struct perf_counter_attr { - - /* - * Major type: hardware/software/tracepoint/etc. - */ - __u32 type; - - /* - * Size of the attr structure, for fwd/bwd compat. - */ - __u32 size; - - /* - * Type specific configuration information. - */ - __u64 config; - - union { - __u64 sample_period; - __u64 sample_freq; - }; - - __u64 sample_type; - __u64 read_format; - - __u64 disabled : 1, /* off by default */ - inherit : 1, /* children inherit it */ - pinned : 1, /* must always be on PMU */ - exclusive : 1, /* only group on PMU */ - exclude_user : 1, /* don't count user */ - exclude_kernel : 1, /* ditto kernel */ - exclude_hv : 1, /* ditto hypervisor */ - exclude_idle : 1, /* don't count when idle */ - mmap : 1, /* include mmap data */ - comm : 1, /* include comm data */ - freq : 1, /* use freq, not period */ - inherit_stat : 1, /* per task counts */ - enable_on_exec : 1, /* next exec enables */ - task : 1, /* trace fork/exit */ - watermark : 1, /* wakeup_watermark */ - - __reserved_1 : 49; - - union { - __u32 wakeup_events; /* wakeup every n events */ - __u32 wakeup_watermark; /* bytes before wakeup */ - }; - __u32 __reserved_2; - - __u64 __reserved_3; -}; - -/* - * Ioctls that can be done on a perf counter fd: - */ -#define PERF_COUNTER_IOC_ENABLE _IO ('$', 0) -#define PERF_COUNTER_IOC_DISABLE _IO ('$', 1) -#define PERF_COUNTER_IOC_REFRESH _IO ('$', 2) -#define PERF_COUNTER_IOC_RESET _IO ('$', 3) -#define PERF_COUNTER_IOC_PERIOD _IOW('$', 4, u64) -#define PERF_COUNTER_IOC_SET_OUTPUT _IO ('$', 5) -#define PERF_COUNTER_IOC_SET_FILTER _IOW('$', 6, char *) - -enum perf_counter_ioc_flags { - PERF_IOC_FLAG_GROUP = 1U << 0, -}; - -/* - * Structure of the page that can be mapped via mmap - */ -struct perf_counter_mmap_page { - __u32 version; /* version number of this structure */ - __u32 compat_version; /* lowest version this is compat with */ - - /* - * Bits needed to read the hw counters in user-space. - * - * u32 seq; - * s64 count; - * - * do { - * seq = pc->lock; - * - * barrier() - * if (pc->index) { - * count = pmc_read(pc->index - 1); - * count += pc->offset; - * } else - * goto regular_read; - * - * barrier(); - * } while (pc->lock != seq); - * - * NOTE: for obvious reason this only works on self-monitoring - * processes. - */ - __u32 lock; /* seqlock for synchronization */ - __u32 index; /* hardware counter identifier */ - __s64 offset; /* add to hardware counter value */ - __u64 time_enabled; /* time counter active */ - __u64 time_running; /* time counter on cpu */ - - /* - * Hole for extension of the self monitor capabilities - */ - - __u64 __reserved[123]; /* align to 1k */ - - /* - * Control data for the mmap() data buffer. - * - * User-space reading the @data_head value should issue an rmb(), on - * SMP capable platforms, after reading this value -- see - * perf_counter_wakeup(). - * - * When the mapping is PROT_WRITE the @data_tail value should be - * written by userspace to reflect the last read data. In this case - * the kernel will not over-write unread data. - */ - __u64 data_head; /* head in the data section */ - __u64 data_tail; /* user-space written tail */ -}; - -#define PERF_EVENT_MISC_CPUMODE_MASK (3 << 0) -#define PERF_EVENT_MISC_CPUMODE_UNKNOWN (0 << 0) -#define PERF_EVENT_MISC_KERNEL (1 << 0) -#define PERF_EVENT_MISC_USER (2 << 0) -#define PERF_EVENT_MISC_HYPERVISOR (3 << 0) - -struct perf_event_header { - __u32 type; - __u16 misc; - __u16 size; -}; - -enum perf_event_type { - - /* - * The MMAP events record the PROT_EXEC mappings so that we can - * correlate userspace IPs to code. They have the following structure: - * - * struct { - * struct perf_event_header header; - * - * u32 pid, tid; - * u64 addr; - * u64 len; - * u64 pgoff; - * char filename[]; - * }; - */ - PERF_EVENT_MMAP = 1, - - /* - * struct { - * struct perf_event_header header; - * u64 id; - * u64 lost; - * }; - */ - PERF_EVENT_LOST = 2, - - /* - * struct { - * struct perf_event_header header; - * - * u32 pid, tid; - * char comm[]; - * }; - */ - PERF_EVENT_COMM = 3, - - /* - * struct { - * struct perf_event_header header; - * u32 pid, ppid; - * u32 tid, ptid; - * u64 time; - * }; - */ - PERF_EVENT_EXIT = 4, - - /* - * struct { - * struct perf_event_header header; - * u64 time; - * u64 id; - * u64 stream_id; - * }; - */ - PERF_EVENT_THROTTLE = 5, - PERF_EVENT_UNTHROTTLE = 6, - - /* - * struct { - * struct perf_event_header header; - * u32 pid, ppid; - * u32 tid, ptid; - * u64 time; - * }; - */ - PERF_EVENT_FORK = 7, - - /* - * struct { - * struct perf_event_header header; - * u32 pid, tid; - * - * struct read_format values; - * }; - */ - PERF_EVENT_READ = 8, - - /* - * struct { - * struct perf_event_header header; - * - * { u64 ip; } && PERF_SAMPLE_IP - * { u32 pid, tid; } && PERF_SAMPLE_TID - * { u64 time; } && PERF_SAMPLE_TIME - * { u64 addr; } && PERF_SAMPLE_ADDR - * { u64 id; } && PERF_SAMPLE_ID - * { u64 stream_id;} && PERF_SAMPLE_STREAM_ID - * { u32 cpu, res; } && PERF_SAMPLE_CPU - * { u64 period; } && PERF_SAMPLE_PERIOD - * - * { struct read_format values; } && PERF_SAMPLE_READ - * - * { u64 nr, - * u64 ips[nr]; } && PERF_SAMPLE_CALLCHAIN - * - * # - * # The RAW record below is opaque data wrt the ABI - * # - * # That is, the ABI doesn't make any promises wrt to - * # the stability of its content, it may vary depending - * # on event, hardware, kernel version and phase of - * # the moon. - * # - * # In other words, PERF_SAMPLE_RAW contents are not an ABI. - * # - * - * { u32 size; - * char data[size];}&& PERF_SAMPLE_RAW - * }; - */ - PERF_EVENT_SAMPLE = 9, - - PERF_EVENT_MAX, /* non-ABI */ -}; - -enum perf_callchain_context { - PERF_CONTEXT_HV = (__u64)-32, - PERF_CONTEXT_KERNEL = (__u64)-128, - PERF_CONTEXT_USER = (__u64)-512, - - PERF_CONTEXT_GUEST = (__u64)-2048, - PERF_CONTEXT_GUEST_KERNEL = (__u64)-2176, - PERF_CONTEXT_GUEST_USER = (__u64)-2560, - - PERF_CONTEXT_MAX = (__u64)-4095, -}; - -#define PERF_FLAG_FD_NO_GROUP (1U << 0) -#define PERF_FLAG_FD_OUTPUT (1U << 1) - -/* - * In case some app still references the old symbols: - */ - -#define __NR_perf_counter_open __NR_perf_event_open - -#define PR_TASK_PERF_COUNTERS_DISABLE PR_TASK_PERF_EVENTS_DISABLE -#define PR_TASK_PERF_COUNTERS_ENABLE PR_TASK_PERF_EVENTS_ENABLE - -#endif /* _LINUX_PERF_COUNTER_H */ From 61c1917f47f73c968e92d04d15370b1dc3ec4592 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 17 Dec 2009 05:40:33 +0100 Subject: [PATCH 06/10] perf events, x86/stacktrace: Make stack walking optional The current print_context_stack helper that does the stack walking job is good for usual stacktraces as it walks through all the stack and reports even addresses that look unreliable, which is nice when we don't have frame pointers for example. But we have users like perf that only require reliable stacktraces, and those may want a more adapted stack walker, so lets make this function a callback in stacktrace_ops that users can tune for their needs. Signed-off-by: Frederic Weisbecker Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras LKML-Reference: <1261024834-5336-1-git-send-regression-fweisbec@gmail.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/stacktrace.h | 18 ++++++++++++++++++ arch/x86/kernel/cpu/perf_event.c | 1 + arch/x86/kernel/dumpstack.c | 9 +++++---- arch/x86/kernel/dumpstack.h | 6 ------ arch/x86/kernel/dumpstack_32.c | 2 +- arch/x86/kernel/dumpstack_64.c | 4 ++-- arch/x86/kernel/stacktrace.c | 18 ++++++++++-------- arch/x86/oprofile/backtrace.c | 9 +++++---- kernel/trace/trace_sysprof.c | 1 + 9 files changed, 43 insertions(+), 25 deletions(-) diff --git a/arch/x86/include/asm/stacktrace.h b/arch/x86/include/asm/stacktrace.h index cf86a5e73815..6c75151a3cca 100644 --- a/arch/x86/include/asm/stacktrace.h +++ b/arch/x86/include/asm/stacktrace.h @@ -5,6 +5,23 @@ extern int kstack_depth_to_print; int x86_is_stack_id(int id, char *name); +struct thread_info; +struct stacktrace_ops; + +typedef unsigned long (*walk_stack_t)(struct thread_info *tinfo, + unsigned long *stack, + unsigned long bp, + const struct stacktrace_ops *ops, + void *data, + unsigned long *end, + int *graph); + +extern unsigned long +print_context_stack(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end, int *graph); + /* Generic stack tracer with callbacks */ struct stacktrace_ops { @@ -14,6 +31,7 @@ struct stacktrace_ops { void (*address)(void *data, unsigned long address, int reliable); /* On negative return stop dumping */ int (*stack)(void *data, char *name); + walk_stack_t walk_stack; }; void dump_trace(struct task_struct *tsk, struct pt_regs *regs, diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index 45506d5dd8df..d3802ee5a416 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -2336,6 +2336,7 @@ static const struct stacktrace_ops backtrace_ops = { .warning_symbol = backtrace_warning_symbol, .stack = backtrace_stack, .address = backtrace_address, + .walk_stack = print_context_stack, }; #include "../dumpstack.h" diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c index 0a0aa1cec8f1..8aaa119b7cad 100644 --- a/arch/x86/kernel/dumpstack.c +++ b/arch/x86/kernel/dumpstack.c @@ -141,10 +141,11 @@ static void print_trace_address(void *data, unsigned long addr, int reliable) } static const struct stacktrace_ops print_trace_ops = { - .warning = print_trace_warning, - .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, + .warning = print_trace_warning, + .warning_symbol = print_trace_warning_symbol, + .stack = print_trace_stack, + .address = print_trace_address, + .walk_stack = print_context_stack, }; void diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h index 81086c227ab7..4fd1420faffa 100644 --- a/arch/x86/kernel/dumpstack.h +++ b/arch/x86/kernel/dumpstack.h @@ -14,12 +14,6 @@ #define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) #endif -extern unsigned long -print_context_stack(struct thread_info *tinfo, - unsigned long *stack, unsigned long bp, - const struct stacktrace_ops *ops, void *data, - unsigned long *end, int *graph); - extern void show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, char *log_lvl); diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index e0ed4c7abb62..ae775ca47b25 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -58,7 +58,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, context = (struct thread_info *) ((unsigned long)stack & (~(THREAD_SIZE - 1))); - bp = print_context_stack(context, stack, bp, ops, data, NULL, &graph); + bp = ops->walk_stack(context, stack, bp, ops, data, NULL, &graph); stack = (unsigned long *)context->previous_esp; if (!stack) diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index b13af53883aa..0ad9597073f5 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -188,8 +188,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, if (ops->stack(data, id) < 0) break; - bp = print_context_stack(tinfo, stack, bp, ops, - data, estack_end, &graph); + bp = ops->walk_stack(tinfo, stack, bp, ops, + data, estack_end, &graph); ops->stack(data, ""); /* * We link to the next stack via the diff --git a/arch/x86/kernel/stacktrace.c b/arch/x86/kernel/stacktrace.c index c3eb207181fe..922eefbb3f6c 100644 --- a/arch/x86/kernel/stacktrace.c +++ b/arch/x86/kernel/stacktrace.c @@ -53,17 +53,19 @@ save_stack_address_nosched(void *data, unsigned long addr, int reliable) } static const struct stacktrace_ops save_stack_ops = { - .warning = save_stack_warning, - .warning_symbol = save_stack_warning_symbol, - .stack = save_stack_stack, - .address = save_stack_address, + .warning = save_stack_warning, + .warning_symbol = save_stack_warning_symbol, + .stack = save_stack_stack, + .address = save_stack_address, + .walk_stack = print_context_stack, }; static const struct stacktrace_ops save_stack_ops_nosched = { - .warning = save_stack_warning, - .warning_symbol = save_stack_warning_symbol, - .stack = save_stack_stack, - .address = save_stack_address_nosched, + .warning = save_stack_warning, + .warning_symbol = save_stack_warning_symbol, + .stack = save_stack_stack, + .address = save_stack_address_nosched, + .walk_stack = print_context_stack, }; /* diff --git a/arch/x86/oprofile/backtrace.c b/arch/x86/oprofile/backtrace.c index 044897be021f..3855096c59b8 100644 --- a/arch/x86/oprofile/backtrace.c +++ b/arch/x86/oprofile/backtrace.c @@ -41,10 +41,11 @@ static void backtrace_address(void *data, unsigned long addr, int reliable) } static struct stacktrace_ops backtrace_ops = { - .warning = backtrace_warning, - .warning_symbol = backtrace_warning_symbol, - .stack = backtrace_stack, - .address = backtrace_address, + .warning = backtrace_warning, + .warning_symbol = backtrace_warning_symbol, + .stack = backtrace_stack, + .address = backtrace_address, + .walk_stack = print_context_stack, }; struct frame_head { diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index f6693969287d..a7974a552ca9 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -93,6 +93,7 @@ static const struct stacktrace_ops backtrace_ops = { .warning_symbol = backtrace_warning_symbol, .stack = backtrace_stack, .address = backtrace_address, + .walk_stack = print_context_stack, }; static int From 06d65bda75341485d32f33da474b0664819ad497 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 17 Dec 2009 05:40:34 +0100 Subject: [PATCH 07/10] perf events, x86/stacktrace: Fix performance/softlockup by providing a special frame pointer-only stack walker It's just wasteful for stacktrace users like perf to walk through every entries on the stack whereas these only accept reliable ones, ie: that the frame pointer validates. Since perf requires pure reliable stacktraces, it needs a stack walker based on frame pointers-only to optimize the stacktrace processing. This might solve some near-lockup scenarios that can be triggered by call-graph tracing timer events. Signed-off-by: Frederic Weisbecker Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras LKML-Reference: <1261024834-5336-2-git-send-regression-fweisbec@gmail.com> [ v2: fix for modular builds and small detail tidyup ] Signed-off-by: Ingo Molnar --- arch/x86/include/asm/stacktrace.h | 6 ++++++ arch/x86/kernel/cpu/perf_event.c | 2 +- arch/x86/kernel/dumpstack.c | 28 ++++++++++++++++++++++++++-- 3 files changed, 33 insertions(+), 3 deletions(-) diff --git a/arch/x86/include/asm/stacktrace.h b/arch/x86/include/asm/stacktrace.h index 6c75151a3cca..35e89122a42f 100644 --- a/arch/x86/include/asm/stacktrace.h +++ b/arch/x86/include/asm/stacktrace.h @@ -22,6 +22,12 @@ print_context_stack(struct thread_info *tinfo, const struct stacktrace_ops *ops, void *data, unsigned long *end, int *graph); +extern unsigned long +print_context_stack_bp(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end, int *graph); + /* Generic stack tracer with callbacks */ struct stacktrace_ops { diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index d3802ee5a416..c223b7e895d9 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -2336,7 +2336,7 @@ static const struct stacktrace_ops backtrace_ops = { .warning_symbol = backtrace_warning_symbol, .stack = backtrace_stack, .address = backtrace_address, - .walk_stack = print_context_stack, + .walk_stack = print_context_stack_bp, }; #include "../dumpstack.h" diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c index 8aaa119b7cad..c56bc2873030 100644 --- a/arch/x86/kernel/dumpstack.c +++ b/arch/x86/kernel/dumpstack.c @@ -109,6 +109,30 @@ print_context_stack(struct thread_info *tinfo, } return bp; } +EXPORT_SYMBOL_GPL(print_context_stack); + +unsigned long +print_context_stack_bp(struct thread_info *tinfo, + unsigned long *stack, unsigned long bp, + const struct stacktrace_ops *ops, void *data, + unsigned long *end, int *graph) +{ + struct stack_frame *frame = (struct stack_frame *)bp; + unsigned long *ret_addr = &frame->return_address; + + while (valid_stack_ptr(tinfo, ret_addr, sizeof(*ret_addr), end)) { + unsigned long addr = *ret_addr; + + if (__kernel_text_address(addr)) { + ops->address(data, addr, 1); + frame = frame->next_frame; + ret_addr = &frame->return_address; + print_ftrace_graph_addr(addr, data, ops, tinfo, graph); + } + } + return (unsigned long)frame; +} +EXPORT_SYMBOL_GPL(print_context_stack_bp); static void @@ -143,8 +167,8 @@ static void print_trace_address(void *data, unsigned long addr, int reliable) static const struct stacktrace_ops print_trace_ops = { .warning = print_trace_warning, .warning_symbol = print_trace_warning_symbol, - .stack = print_trace_stack, - .address = print_trace_address, + .stack = print_trace_stack, + .address = print_trace_address, .walk_stack = print_context_stack, }; From 5d27c23df09b702868d9a3bff86ec6abd22963ac Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 17 Dec 2009 13:16:32 +0100 Subject: [PATCH 08/10] perf events: Dont report side-band events on each cpu for per-task-per-cpu events Acme noticed that his FORK/MMAP numbers were inflated by about the same factor as his cpu-count. This led to the discovery of a few more sites that need to respect the event->cpu filter. Reported-by: Arnaldo Carvalho de Melo Signed-off-by: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <20091217121830.215333434@chello.nl> Signed-off-by: Ingo Molnar --- kernel/perf_event.c | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-) diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 8ab86988bd24..03cc061398d1 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -1381,6 +1381,9 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx) if (event->state != PERF_EVENT_STATE_ACTIVE) continue; + if (event->cpu != -1 && event->cpu != smp_processor_id()) + continue; + hwc = &event->hw; interrupts = hwc->interrupts; @@ -3265,6 +3268,9 @@ static void perf_event_task_output(struct perf_event *event, static int perf_event_task_match(struct perf_event *event) { + if (event->cpu != -1 && event->cpu != smp_processor_id()) + return 0; + if (event->attr.comm || event->attr.mmap || event->attr.task) return 1; @@ -3290,12 +3296,11 @@ static void perf_event_task_event(struct perf_task_event *task_event) rcu_read_lock(); cpuctx = &get_cpu_var(perf_cpu_context); perf_event_task_ctx(&cpuctx->ctx, task_event); - put_cpu_var(perf_cpu_context); - if (!ctx) ctx = rcu_dereference(task_event->task->perf_event_ctxp); if (ctx) perf_event_task_ctx(ctx, task_event); + put_cpu_var(perf_cpu_context); rcu_read_unlock(); } @@ -3372,6 +3377,9 @@ static void perf_event_comm_output(struct perf_event *event, static int perf_event_comm_match(struct perf_event *event) { + if (event->cpu != -1 && event->cpu != smp_processor_id()) + return 0; + if (event->attr.comm) return 1; @@ -3408,15 +3416,10 @@ static void perf_event_comm_event(struct perf_comm_event *comm_event) rcu_read_lock(); cpuctx = &get_cpu_var(perf_cpu_context); perf_event_comm_ctx(&cpuctx->ctx, comm_event); - put_cpu_var(perf_cpu_context); - - /* - * doesn't really matter which of the child contexts the - * events ends up in. - */ ctx = rcu_dereference(current->perf_event_ctxp); if (ctx) perf_event_comm_ctx(ctx, comm_event); + put_cpu_var(perf_cpu_context); rcu_read_unlock(); } @@ -3491,6 +3494,9 @@ static void perf_event_mmap_output(struct perf_event *event, static int perf_event_mmap_match(struct perf_event *event, struct perf_mmap_event *mmap_event) { + if (event->cpu != -1 && event->cpu != smp_processor_id()) + return 0; + if (event->attr.mmap) return 1; @@ -3564,15 +3570,10 @@ got_name: rcu_read_lock(); cpuctx = &get_cpu_var(perf_cpu_context); perf_event_mmap_ctx(&cpuctx->ctx, mmap_event); - put_cpu_var(perf_cpu_context); - - /* - * doesn't really matter which of the child contexts the - * events ends up in. - */ ctx = rcu_dereference(current->perf_event_ctxp); if (ctx) perf_event_mmap_ctx(ctx, mmap_event); + put_cpu_var(perf_cpu_context); rcu_read_unlock(); kfree(buf); @@ -3863,6 +3864,9 @@ static int perf_swevent_match(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs) { + if (event->cpu != -1 && event->cpu != smp_processor_id()) + return 0; + if (!perf_swevent_is_counting(event)) return 0; From 99e8c5a3b875a34d894a711c9a3669858d6adf45 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 17 Dec 2009 01:33:54 +0100 Subject: [PATCH 09/10] hw-breakpoints: Fix hardware breakpoints -> perf events dependency The kbuild's select command doesn't propagate through the config dependencies. Hence the current rules of hardware breakpoint's config can't ensure perf can never be disabled under us. We have: config X86 selects HAVE_HW_BREAKPOINTS config HAVE_HW_BREAKPOINTS select PERF_EVENTS config PERF_EVENTS [...] x86 will select the breakpoints but that won't propagate to perf events. The user can still disable the latter, but it is necessary for the breakpoints. What we need is: - x86 selects HAVE_HW_BREAKPOINTS and PERF_EVENTS - HAVE_HW_BREAKPOINTS depends on PERF_EVENTS so that we ensure PERF_EVENTS is enabled and frozen for x86. This fixes the following kind of build errors: In file included from arch/x86/kernel/hw_breakpoint.c:31: include/linux/hw_breakpoint.h: In function 'hw_breakpoint_addr': include/linux/hw_breakpoint.h:39: error: 'struct perf_event' has no member named 'attr' v2: Select also ANON_INODES from x86, required for perf Reported-by: Cyrill Gorcunov Reported-by: Michal Marek Reported-by: Andrew Randrianasulu Signed-off-by: Frederic Weisbecker Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras Cc: Randy Dunlap Cc: K.Prasad LKML-Reference: <1261010034-7786-1-git-send-regression-fweisbec@gmail.com> Signed-off-by: Ingo Molnar --- arch/Kconfig | 4 +--- arch/x86/Kconfig | 2 ++ 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/Kconfig b/arch/Kconfig index d82875820a15..9d055b4f0585 100644 --- a/arch/Kconfig +++ b/arch/Kconfig @@ -135,9 +135,7 @@ config HAVE_DEFAULT_NO_SPIN_MUTEXES config HAVE_HW_BREAKPOINT bool - depends on HAVE_PERF_EVENTS - select ANON_INODES - select PERF_EVENTS + depends on PERF_EVENTS config HAVE_USER_RETURN_NOTIFIER bool diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 3b2a5aca4edb..55298e891571 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -50,6 +50,8 @@ config X86 select HAVE_KERNEL_BZIP2 select HAVE_KERNEL_LZMA select HAVE_HW_BREAKPOINT + select PERF_EVENTS + select ANON_INODES select HAVE_ARCH_KMEMCHECK select HAVE_USER_RETURN_NOTIFIER From b5b60fda1e462a849bc37dfbace2888191be82cc Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 18 Dec 2009 13:03:03 -0200 Subject: [PATCH 10/10] perf session: Make events_stats u64 to avoid overflow on 32-bit arches Pekka Enberg reported weird percentages in perf report. It turns out we are overflowing a 32-bit variables in struct events_stats on 32-bit architectures. Before: [acme@ana linux-2.6-tip]$ perf report -i pekka.perf.data 2> /dev/null | head -10 281.96% Xorg b710a561 [.] 0x000000b710a561 140.15% Xorg [kernel] [k] __initramfs_end 51.56% metacity libgobject-2.0.so.0.2000.1 [.] 0x00000000026e46 35.12% evolution libcairo.so.2.10800.6 [.] 0x000000000203bd 33.84% metacity libpthread-2.9.so [.] 0x00000000007a3d After: [acme@ana linux-2.6-tip]$ perf report -i pekka.perf.data 2> /dev/null | head -10 30.04% Xorg b710a561 [.] 0x000000b710a561 14.93% Xorg [kernel] [k] __initramfs_end 5.49% metacity libgobject-2.0.so.0.2000.1 [.] 0x00000000026e46 3.74% evolution libcairo.so.2.10800.6 [.] 0x000000000203bd 3.61% metacity libpthread-2.9.so [.] 0x00000000007a3d Reported-by: Pekka Enberg Tested-by: Pekka Enberg Signed-off-by: Arnaldo Carvalho de Melo Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Peter Zijlstra Cc: Paul Mackerras LKML-Reference: <1261148583-20395-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar --- tools/perf/builtin-report.c | 2 +- tools/perf/util/event.h | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index e50a6b10ee6f..5c2ab5357ec6 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -224,7 +224,7 @@ static int __cmd_report(void) perf_session__collapse_resort(session); perf_session__output_resort(session, session->events_stats.total); - fprintf(stdout, "# Samples: %ld\n#\n", session->events_stats.total); + fprintf(stdout, "# Samples: %Ld\n#\n", session->events_stats.total); perf_session__fprintf_hists(session, NULL, false, stdout); if (sort_order == default_sort_order && parent_pattern == default_parent_pattern) diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 8027309b0422..690a96d0467c 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -95,8 +95,8 @@ typedef union event_union { } event_t; struct events_stats { - unsigned long total; - unsigned long lost; + u64 total; + u64 lost; }; void event__print_totals(void);