linux/tools/perf/util/evsel.c

1114 lines
24 KiB
C
Raw Normal View History

/*
* Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
*
* Parts came from builtin-{top,stat,record}.c, see those files for further
* copyright notes.
*
* Released under the GPL v2. (and only v2, not any later version)
*/
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
#include <byteswap.h>
#include <linux/bitops.h>
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
#include "asm/bug.h"
#include "debugfs.h"
#include "event-parse.h"
#include "evsel.h"
#include "evlist.h"
#include "util.h"
#include "cpumap.h"
#include "thread_map.h"
#include "target.h"
#include "../../../include/linux/hw_breakpoint.h"
perf tools: Support for DWARF mode callchain This patch enables perf to use the DWARF unwind code. It extends the perf record '-g' option with following arguments: 'fp' - provides framepointer based user stack backtrace 'dwarf[,size]' - provides DWARF (libunwind) based user stack backtrace. The size specifies the size of the user stack dump. If omitted it is 8192 by default. If libunwind is found during the perf build, then the 'dwarf' argument becomes available for record command. The 'fp' stays as default option in any case. Examples: (perf compiled with libunwind) perf record -g dwarf ls - provides dwarf unwind with 8192 as stack dump size perf record -g dwarf,4096 ls - provides dwarf unwind with 4096 as stack dump size perf record -g -- ls perf record -g fp ls - provides frame pointer unwind Signed-off-by: Jiri Olsa <jolsa@redhat.com> Original-patch-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: "Frank Ch. Eigler" <fche@redhat.com> Cc: Arun Sharma <asharma@fb.com> Cc: Benjamin Redelings <benjamin.redelings@nescent.org> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Robert Richter <robert.richter@amd.com> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/r/1344345647-11536-13-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-07 13:20:47 +00:00
#include "../../include/linux/perf_event.h"
#include "perf_regs.h"
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
static int __perf_evsel__sample_size(u64 sample_type)
{
u64 mask = sample_type & PERF_SAMPLE_MASK;
int size = 0;
int i;
for (i = 0; i < 64; i++) {
if (mask & (1ULL << i))
size++;
}
size *= sizeof(u64);
return size;
}
void hists__init(struct hists *hists)
{
memset(hists, 0, sizeof(*hists));
hists->entries_in_array[0] = hists->entries_in_array[1] = RB_ROOT;
hists->entries_in = &hists->entries_in_array[0];
hists->entries_collapsed = RB_ROOT;
hists->entries = RB_ROOT;
pthread_mutex_init(&hists->lock, NULL);
}
void perf_evsel__init(struct perf_evsel *evsel,
struct perf_event_attr *attr, int idx)
{
evsel->idx = idx;
evsel->attr = *attr;
INIT_LIST_HEAD(&evsel->node);
hists__init(&evsel->hists);
evsel->sample_size = __perf_evsel__sample_size(attr->sample_type);
}
struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx)
{
struct perf_evsel *evsel = zalloc(sizeof(*evsel));
if (evsel != NULL)
perf_evsel__init(evsel, attr, idx);
return evsel;
}
static struct event_format *event_format__new(const char *sys, const char *name)
{
int fd, n;
char *filename;
void *bf = NULL, *nbf;
size_t size = 0, alloc_size = 0;
struct event_format *format = NULL;
if (asprintf(&filename, "%s/%s/%s/format", tracing_events_path, sys, name) < 0)
goto out;
fd = open(filename, O_RDONLY);
if (fd < 0)
goto out_free_filename;
do {
if (size == alloc_size) {
alloc_size += BUFSIZ;
nbf = realloc(bf, alloc_size);
if (nbf == NULL)
goto out_free_bf;
bf = nbf;
}
n = read(fd, bf + size, BUFSIZ);
if (n < 0)
goto out_free_bf;
size += n;
} while (n > 0);
pevent_parse_format(&format, bf, size, sys);
out_free_bf:
free(bf);
close(fd);
out_free_filename:
free(filename);
out:
return format;
}
struct perf_evsel *perf_evsel__newtp(const char *sys, const char *name, int idx)
{
struct perf_evsel *evsel = zalloc(sizeof(*evsel));
if (evsel != NULL) {
struct perf_event_attr attr = {
.type = PERF_TYPE_TRACEPOINT,
};
evsel->tp_format = event_format__new(sys, name);
if (evsel->tp_format == NULL)
goto out_free;
attr.config = evsel->tp_format->id;
perf_evsel__init(evsel, &attr, idx);
evsel->name = evsel->tp_format->name;
}
return evsel;
out_free:
free(evsel);
return NULL;
}
const char *perf_evsel__hw_names[PERF_COUNT_HW_MAX] = {
"cycles",
"instructions",
"cache-references",
"cache-misses",
"branches",
"branch-misses",
"bus-cycles",
"stalled-cycles-frontend",
"stalled-cycles-backend",
"ref-cycles",
};
static const char *__perf_evsel__hw_name(u64 config)
{
if (config < PERF_COUNT_HW_MAX && perf_evsel__hw_names[config])
return perf_evsel__hw_names[config];
return "unknown-hardware";
}
static int perf_evsel__add_modifiers(struct perf_evsel *evsel, char *bf, size_t size)
{
int colon = 0, r = 0;
struct perf_event_attr *attr = &evsel->attr;
bool exclude_guest_default = false;
#define MOD_PRINT(context, mod) do { \
if (!attr->exclude_##context) { \
if (!colon) colon = ++r; \
r += scnprintf(bf + r, size - r, "%c", mod); \
} } while(0)
if (attr->exclude_kernel || attr->exclude_user || attr->exclude_hv) {
MOD_PRINT(kernel, 'k');
MOD_PRINT(user, 'u');
MOD_PRINT(hv, 'h');
exclude_guest_default = true;
}
if (attr->precise_ip) {
if (!colon)
colon = ++r;
r += scnprintf(bf + r, size - r, "%.*s", attr->precise_ip, "ppp");
exclude_guest_default = true;
}
if (attr->exclude_host || attr->exclude_guest == exclude_guest_default) {
MOD_PRINT(host, 'H');
MOD_PRINT(guest, 'G');
}
#undef MOD_PRINT
if (colon)
bf[colon - 1] = ':';
return r;
}
static int perf_evsel__hw_name(struct perf_evsel *evsel, char *bf, size_t size)
{
int r = scnprintf(bf, size, "%s", __perf_evsel__hw_name(evsel->attr.config));
return r + perf_evsel__add_modifiers(evsel, bf + r, size - r);
}
const char *perf_evsel__sw_names[PERF_COUNT_SW_MAX] = {
"cpu-clock",
"task-clock",
"page-faults",
"context-switches",
"cpu-migrations",
"minor-faults",
"major-faults",
"alignment-faults",
"emulation-faults",
};
static const char *__perf_evsel__sw_name(u64 config)
{
if (config < PERF_COUNT_SW_MAX && perf_evsel__sw_names[config])
return perf_evsel__sw_names[config];
return "unknown-software";
}
static int perf_evsel__sw_name(struct perf_evsel *evsel, char *bf, size_t size)
{
int r = scnprintf(bf, size, "%s", __perf_evsel__sw_name(evsel->attr.config));
return r + perf_evsel__add_modifiers(evsel, bf + r, size - r);
}
static int __perf_evsel__bp_name(char *bf, size_t size, u64 addr, u64 type)
{
int r;
r = scnprintf(bf, size, "mem:0x%" PRIx64 ":", addr);
if (type & HW_BREAKPOINT_R)
r += scnprintf(bf + r, size - r, "r");
if (type & HW_BREAKPOINT_W)
r += scnprintf(bf + r, size - r, "w");
if (type & HW_BREAKPOINT_X)
r += scnprintf(bf + r, size - r, "x");
return r;
}
static int perf_evsel__bp_name(struct perf_evsel *evsel, char *bf, size_t size)
{
struct perf_event_attr *attr = &evsel->attr;
int r = __perf_evsel__bp_name(bf, size, attr->bp_addr, attr->bp_type);
return r + perf_evsel__add_modifiers(evsel, bf + r, size - r);
}
const char *perf_evsel__hw_cache[PERF_COUNT_HW_CACHE_MAX]
[PERF_EVSEL__MAX_ALIASES] = {
{ "L1-dcache", "l1-d", "l1d", "L1-data", },
{ "L1-icache", "l1-i", "l1i", "L1-instruction", },
{ "LLC", "L2", },
{ "dTLB", "d-tlb", "Data-TLB", },
{ "iTLB", "i-tlb", "Instruction-TLB", },
{ "branch", "branches", "bpu", "btb", "bpc", },
{ "node", },
};
const char *perf_evsel__hw_cache_op[PERF_COUNT_HW_CACHE_OP_MAX]
[PERF_EVSEL__MAX_ALIASES] = {
{ "load", "loads", "read", },
{ "store", "stores", "write", },
{ "prefetch", "prefetches", "speculative-read", "speculative-load", },
};
const char *perf_evsel__hw_cache_result[PERF_COUNT_HW_CACHE_RESULT_MAX]
[PERF_EVSEL__MAX_ALIASES] = {
{ "refs", "Reference", "ops", "access", },
{ "misses", "miss", },
};
#define C(x) PERF_COUNT_HW_CACHE_##x
#define CACHE_READ (1 << C(OP_READ))
#define CACHE_WRITE (1 << C(OP_WRITE))
#define CACHE_PREFETCH (1 << C(OP_PREFETCH))
#define COP(x) (1 << x)
/*
* cache operartion stat
* L1I : Read and prefetch only
* ITLB and BPU : Read-only
*/
static unsigned long perf_evsel__hw_cache_stat[C(MAX)] = {
[C(L1D)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(L1I)] = (CACHE_READ | CACHE_PREFETCH),
[C(LL)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(DTLB)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(ITLB)] = (CACHE_READ),
[C(BPU)] = (CACHE_READ),
[C(NODE)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
};
bool perf_evsel__is_cache_op_valid(u8 type, u8 op)
{
if (perf_evsel__hw_cache_stat[type] & COP(op))
return true; /* valid */
else
return false; /* invalid */
}
int __perf_evsel__hw_cache_type_op_res_name(u8 type, u8 op, u8 result,
char *bf, size_t size)
{
if (result) {
return scnprintf(bf, size, "%s-%s-%s", perf_evsel__hw_cache[type][0],
perf_evsel__hw_cache_op[op][0],
perf_evsel__hw_cache_result[result][0]);
}
return scnprintf(bf, size, "%s-%s", perf_evsel__hw_cache[type][0],
perf_evsel__hw_cache_op[op][1]);
}
static int __perf_evsel__hw_cache_name(u64 config, char *bf, size_t size)
{
u8 op, result, type = (config >> 0) & 0xff;
const char *err = "unknown-ext-hardware-cache-type";
if (type > PERF_COUNT_HW_CACHE_MAX)
goto out_err;
op = (config >> 8) & 0xff;
err = "unknown-ext-hardware-cache-op";
if (op > PERF_COUNT_HW_CACHE_OP_MAX)
goto out_err;
result = (config >> 16) & 0xff;
err = "unknown-ext-hardware-cache-result";
if (result > PERF_COUNT_HW_CACHE_RESULT_MAX)
goto out_err;
err = "invalid-cache";
if (!perf_evsel__is_cache_op_valid(type, op))
goto out_err;
return __perf_evsel__hw_cache_type_op_res_name(type, op, result, bf, size);
out_err:
return scnprintf(bf, size, "%s", err);
}
static int perf_evsel__hw_cache_name(struct perf_evsel *evsel, char *bf, size_t size)
{
int ret = __perf_evsel__hw_cache_name(evsel->attr.config, bf, size);
return ret + perf_evsel__add_modifiers(evsel, bf + ret, size - ret);
}
static int perf_evsel__raw_name(struct perf_evsel *evsel, char *bf, size_t size)
{
int ret = scnprintf(bf, size, "raw 0x%" PRIx64, evsel->attr.config);
return ret + perf_evsel__add_modifiers(evsel, bf + ret, size - ret);
}
const char *perf_evsel__name(struct perf_evsel *evsel)
{
char bf[128];
if (evsel->name)
return evsel->name;
switch (evsel->attr.type) {
case PERF_TYPE_RAW:
perf_evsel__raw_name(evsel, bf, sizeof(bf));
break;
case PERF_TYPE_HARDWARE:
perf_evsel__hw_name(evsel, bf, sizeof(bf));
break;
case PERF_TYPE_HW_CACHE:
perf_evsel__hw_cache_name(evsel, bf, sizeof(bf));
break;
case PERF_TYPE_SOFTWARE:
perf_evsel__sw_name(evsel, bf, sizeof(bf));
break;
case PERF_TYPE_TRACEPOINT:
scnprintf(bf, sizeof(bf), "%s", "unknown tracepoint");
break;
case PERF_TYPE_BREAKPOINT:
perf_evsel__bp_name(evsel, bf, sizeof(bf));
break;
default:
scnprintf(bf, sizeof(bf), "unknown attr type: %d",
evsel->attr.type);
break;
}
evsel->name = strdup(bf);
return evsel->name ?: "unknown";
}
void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts,
struct perf_evsel *first)
{
struct perf_event_attr *attr = &evsel->attr;
int track = !evsel->idx; /* only the first counter needs these */
attr->disabled = 1;
attr->sample_id_all = opts->sample_id_all_missing ? 0 : 1;
attr->inherit = !opts->no_inherit;
attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED |
PERF_FORMAT_TOTAL_TIME_RUNNING |
PERF_FORMAT_ID;
attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID;
/*
* We default some events to a 1 default interval. But keep
* it a weak assumption overridable by the user.
*/
if (!attr->sample_period || (opts->user_freq != UINT_MAX &&
opts->user_interval != ULLONG_MAX)) {
if (opts->freq) {
attr->sample_type |= PERF_SAMPLE_PERIOD;
attr->freq = 1;
attr->sample_freq = opts->freq;
} else {
attr->sample_period = opts->default_interval;
}
}
if (opts->no_samples)
attr->sample_freq = 0;
if (opts->inherit_stat)
attr->inherit_stat = 1;
if (opts->sample_address) {
attr->sample_type |= PERF_SAMPLE_ADDR;
attr->mmap_data = track;
}
perf tools: Support for DWARF mode callchain This patch enables perf to use the DWARF unwind code. It extends the perf record '-g' option with following arguments: 'fp' - provides framepointer based user stack backtrace 'dwarf[,size]' - provides DWARF (libunwind) based user stack backtrace. The size specifies the size of the user stack dump. If omitted it is 8192 by default. If libunwind is found during the perf build, then the 'dwarf' argument becomes available for record command. The 'fp' stays as default option in any case. Examples: (perf compiled with libunwind) perf record -g dwarf ls - provides dwarf unwind with 8192 as stack dump size perf record -g dwarf,4096 ls - provides dwarf unwind with 4096 as stack dump size perf record -g -- ls perf record -g fp ls - provides frame pointer unwind Signed-off-by: Jiri Olsa <jolsa@redhat.com> Original-patch-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: "Frank Ch. Eigler" <fche@redhat.com> Cc: Arun Sharma <asharma@fb.com> Cc: Benjamin Redelings <benjamin.redelings@nescent.org> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Robert Richter <robert.richter@amd.com> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/r/1344345647-11536-13-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-07 13:20:47 +00:00
if (opts->call_graph) {
attr->sample_type |= PERF_SAMPLE_CALLCHAIN;
perf tools: Support for DWARF mode callchain This patch enables perf to use the DWARF unwind code. It extends the perf record '-g' option with following arguments: 'fp' - provides framepointer based user stack backtrace 'dwarf[,size]' - provides DWARF (libunwind) based user stack backtrace. The size specifies the size of the user stack dump. If omitted it is 8192 by default. If libunwind is found during the perf build, then the 'dwarf' argument becomes available for record command. The 'fp' stays as default option in any case. Examples: (perf compiled with libunwind) perf record -g dwarf ls - provides dwarf unwind with 8192 as stack dump size perf record -g dwarf,4096 ls - provides dwarf unwind with 4096 as stack dump size perf record -g -- ls perf record -g fp ls - provides frame pointer unwind Signed-off-by: Jiri Olsa <jolsa@redhat.com> Original-patch-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: "Frank Ch. Eigler" <fche@redhat.com> Cc: Arun Sharma <asharma@fb.com> Cc: Benjamin Redelings <benjamin.redelings@nescent.org> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Robert Richter <robert.richter@amd.com> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/r/1344345647-11536-13-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-07 13:20:47 +00:00
if (opts->call_graph == CALLCHAIN_DWARF) {
attr->sample_type |= PERF_SAMPLE_REGS_USER |
PERF_SAMPLE_STACK_USER;
attr->sample_regs_user = PERF_REGS_MASK;
attr->sample_stack_user = opts->stack_dump_size;
attr->exclude_callchain_user = 1;
}
}
if (perf_target__has_cpu(&opts->target))
attr->sample_type |= PERF_SAMPLE_CPU;
if (opts->period)
attr->sample_type |= PERF_SAMPLE_PERIOD;
if (!opts->sample_id_all_missing &&
(opts->sample_time || !opts->no_inherit ||
perf_target__has_cpu(&opts->target)))
attr->sample_type |= PERF_SAMPLE_TIME;
if (opts->raw_samples) {
attr->sample_type |= PERF_SAMPLE_TIME;
attr->sample_type |= PERF_SAMPLE_RAW;
attr->sample_type |= PERF_SAMPLE_CPU;
}
if (opts->no_delay) {
attr->watermark = 0;
attr->wakeup_events = 1;
}
if (opts->branch_stack) {
attr->sample_type |= PERF_SAMPLE_BRANCH_STACK;
attr->branch_sample_type = opts->branch_stack;
}
attr->mmap = track;
attr->comm = track;
if (perf_target__none(&opts->target) &&
(!opts->group || evsel == first)) {
attr->enable_on_exec = 1;
}
}
int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
{
int cpu, thread;
evsel->fd = xyarray__new(ncpus, nthreads, sizeof(int));
if (evsel->fd) {
for (cpu = 0; cpu < ncpus; cpu++) {
for (thread = 0; thread < nthreads; thread++) {
FD(evsel, cpu, thread) = -1;
}
}
}
return evsel->fd != NULL ? 0 : -ENOMEM;
}
int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads)
{
evsel->sample_id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id));
if (evsel->sample_id == NULL)
return -ENOMEM;
evsel->id = zalloc(ncpus * nthreads * sizeof(u64));
if (evsel->id == NULL) {
xyarray__delete(evsel->sample_id);
evsel->sample_id = NULL;
return -ENOMEM;
}
return 0;
}
int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus)
{
evsel->counts = zalloc((sizeof(*evsel->counts) +
(ncpus * sizeof(struct perf_counts_values))));
return evsel->counts != NULL ? 0 : -ENOMEM;
}
void perf_evsel__free_fd(struct perf_evsel *evsel)
{
xyarray__delete(evsel->fd);
evsel->fd = NULL;
}
void perf_evsel__free_id(struct perf_evsel *evsel)
{
xyarray__delete(evsel->sample_id);
evsel->sample_id = NULL;
free(evsel->id);
evsel->id = NULL;
}
void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
{
int cpu, thread;
for (cpu = 0; cpu < ncpus; cpu++)
for (thread = 0; thread < nthreads; ++thread) {
close(FD(evsel, cpu, thread));
FD(evsel, cpu, thread) = -1;
}
}
void perf_evsel__exit(struct perf_evsel *evsel)
{
assert(list_empty(&evsel->node));
xyarray__delete(evsel->fd);
xyarray__delete(evsel->sample_id);
free(evsel->id);
}
void perf_evsel__delete(struct perf_evsel *evsel)
{
perf_evsel__exit(evsel);
close_cgroup(evsel->cgrp);
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
free(evsel->group_name);
if (evsel->tp_format && evsel->name == evsel->tp_format->name) {
evsel->name = NULL;
pevent_free_format(evsel->tp_format);
}
free(evsel->name);
free(evsel);
}
int __perf_evsel__read_on_cpu(struct perf_evsel *evsel,
int cpu, int thread, bool scale)
{
struct perf_counts_values count;
size_t nv = scale ? 3 : 1;
if (FD(evsel, cpu, thread) < 0)
return -EINVAL;
if (evsel->counts == NULL && perf_evsel__alloc_counts(evsel, cpu + 1) < 0)
return -ENOMEM;
if (readn(FD(evsel, cpu, thread), &count, nv * sizeof(u64)) < 0)
return -errno;
if (scale) {
if (count.run == 0)
count.val = 0;
else if (count.run < count.ena)
count.val = (u64)((double)count.val * count.ena / count.run + 0.5);
} else
count.ena = count.run = 0;
evsel->counts->cpu[cpu] = count;
return 0;
}
int __perf_evsel__read(struct perf_evsel *evsel,
int ncpus, int nthreads, bool scale)
{
size_t nv = scale ? 3 : 1;
int cpu, thread;
struct perf_counts_values *aggr = &evsel->counts->aggr, count;
aggr->val = aggr->ena = aggr->run = 0;
for (cpu = 0; cpu < ncpus; cpu++) {
for (thread = 0; thread < nthreads; thread++) {
if (FD(evsel, cpu, thread) < 0)
continue;
if (readn(FD(evsel, cpu, thread),
&count, nv * sizeof(u64)) < 0)
return -errno;
aggr->val += count.val;
if (scale) {
aggr->ena += count.ena;
aggr->run += count.run;
}
}
}
evsel->counts->scaled = 0;
if (scale) {
if (aggr->run == 0) {
evsel->counts->scaled = -1;
aggr->val = 0;
return 0;
}
if (aggr->run < aggr->ena) {
evsel->counts->scaled = 1;
aggr->val = (u64)((double)aggr->val * aggr->ena / aggr->run + 0.5);
}
} else
aggr->ena = aggr->run = 0;
return 0;
}
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
static int get_group_fd(struct perf_evsel *evsel, int cpu, int thread)
{
struct perf_evsel *leader = evsel->leader;
int fd;
if (!leader)
return -1;
/*
* Leader must be already processed/open,
* if not it's a bug.
*/
BUG_ON(!leader->fd);
fd = FD(leader, cpu, thread);
BUG_ON(fd == -1);
return fd;
}
static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
struct thread_map *threads)
{
int cpu, thread;
unsigned long flags = 0;
int pid = -1, err;
if (evsel->fd == NULL &&
perf_evsel__alloc_fd(evsel, cpus->nr, threads->nr) < 0)
return -ENOMEM;
if (evsel->cgrp) {
flags = PERF_FLAG_PID_CGROUP;
pid = evsel->cgrp->fd;
}
for (cpu = 0; cpu < cpus->nr; cpu++) {
for (thread = 0; thread < threads->nr; thread++) {
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
int group_fd;
if (!evsel->cgrp)
pid = threads->map[thread];
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
group_fd = get_group_fd(evsel, cpu, thread);
FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr,
pid,
cpus->map[cpu],
group_fd, flags);
if (FD(evsel, cpu, thread) < 0) {
err = -errno;
goto out_close;
}
}
}
return 0;
out_close:
do {
while (--thread >= 0) {
close(FD(evsel, cpu, thread));
FD(evsel, cpu, thread) = -1;
}
thread = threads->nr;
} while (--cpu >= 0);
return err;
}
void perf_evsel__close(struct perf_evsel *evsel, int ncpus, int nthreads)
{
if (evsel->fd == NULL)
return;
perf_evsel__close_fd(evsel, ncpus, nthreads);
perf_evsel__free_fd(evsel);
evsel->fd = NULL;
}
static struct {
struct cpu_map map;
int cpus[1];
} empty_cpu_map = {
.map.nr = 1,
.cpus = { -1, },
};
static struct {
struct thread_map map;
int threads[1];
} empty_thread_map = {
.map.nr = 1,
.threads = { -1, },
};
int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
struct thread_map *threads)
{
if (cpus == NULL) {
/* Work around old compiler warnings about strict aliasing */
cpus = &empty_cpu_map.map;
}
if (threads == NULL)
threads = &empty_thread_map.map;
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
return __perf_evsel__open(evsel, cpus, threads);
}
int perf_evsel__open_per_cpu(struct perf_evsel *evsel,
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
struct cpu_map *cpus)
{
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
return __perf_evsel__open(evsel, cpus, &empty_thread_map.map);
}
int perf_evsel__open_per_thread(struct perf_evsel *evsel,
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
struct thread_map *threads)
{
perf tools: Enable grouping logic for parsed events This patch adds a functionality that allows to create event groups based on the way they are specified on the command line. Adding functionality to the '{}' group syntax introduced in earlier patch. The current '--group/-g' option behaviour remains intact. If you specify it for record/stat/top command, all the specified events become members of a single group with the first event as a group leader. With the new '{}' group syntax you can create group like: # perf record -e '{cycles,faults}' ls resulting in single event group containing 'cycles' and 'faults' events, with cycles event as group leader. All groups are created with regards to threads and cpus. Thus recording an event group within a 2 threads on server with 4 CPUs will create 8 separate groups. Examples (first event in brackets is group leader): # 1 group (cpu-clock,task-clock) perf record --group -e cpu-clock,task-clock ls perf record -e '{cpu-clock,task-clock}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls # 1 group (cpu-clock,task-clock,minor-faults,major-faults) perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults) perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \ -e instructions ls # 1 group # (cpu-clock,task-clock,minor-faults,major-faults,instructions) perf record --group -e cpu-clock,task-clock \ -e minor-faults,major-faults -e instructions ls perf record -e '{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls It's possible to use standard event modifier for a group, which spans over all events in the group and updates each event modifier settings, for example: # perf record -r '{faults:k,cache-references}:p' resulting in ':kp' modifier being used for 'faults' and ':p' modifier being used for 'cache-references' event. Reviewed-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ulrich Drepper <drepper@gmail.com> Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-08-08 10:22:36 +00:00
return __perf_evsel__open(evsel, &empty_cpu_map.map, threads);
}
static int perf_event__parse_id_sample(const union perf_event *event, u64 type,
perf evsel: Fix 32 bit values endianity swap for sample_id_all header We swap the sample_id_all header by u64 pointers. Some members of the header happen to be 32 bit values. We need to handle them separatelly. Together with other endianity patches, this change fixies perf report discrepancies on origin and target systems as described in test 1 below, e.g. following perf report diff: ... 0.12% ps [kernel.kallsyms] [k] clear_page - 0.12% awk bash [.] alloc_word_desc + 0.12% awk bash [.] yyparse 0.11% beah-rhts-task libpython2.6.so.1.0 [.] 0x5560e 0.10% perf libc-2.12.so [.] __ctype_toupper_loc - 0.09% rhts-test-runne bash [.] maybe_make_export_env + 0.09% rhts-test-runne bash [.] 0x385a0 0.09% ps [kernel.kallsyms] [k] page_fault ... Note, running following to test perf endianity handling: test 1) - origin system: # perf record -a -- sleep 10 (any perf record will do) # perf report > report.origin # perf archive perf.data - copy the perf.data, report.origin and perf.data.tar.bz2 to a target system and run: # tar xjvf perf.data.tar.bz2 -C ~/.debug # perf report > report.target # diff -u report.origin report.target - the diff should produce no output (besides some white space stuff and possibly different date/TZ output) test 2) - origin system: # perf record -ag -fo /tmp/perf.data -- sleep 1 - mount origin system root to the target system on /mnt/origin - target system: # perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \ --kallsyms /mnt/origin/proc/kallsyms - complete perf.data header is displayed Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338380624-7443-4-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:44 +00:00
struct perf_sample *sample,
bool swapped)
{
const u64 *array = event->sample.array;
perf evsel: Fix 32 bit values endianity swap for sample_id_all header We swap the sample_id_all header by u64 pointers. Some members of the header happen to be 32 bit values. We need to handle them separatelly. Together with other endianity patches, this change fixies perf report discrepancies on origin and target systems as described in test 1 below, e.g. following perf report diff: ... 0.12% ps [kernel.kallsyms] [k] clear_page - 0.12% awk bash [.] alloc_word_desc + 0.12% awk bash [.] yyparse 0.11% beah-rhts-task libpython2.6.so.1.0 [.] 0x5560e 0.10% perf libc-2.12.so [.] __ctype_toupper_loc - 0.09% rhts-test-runne bash [.] maybe_make_export_env + 0.09% rhts-test-runne bash [.] 0x385a0 0.09% ps [kernel.kallsyms] [k] page_fault ... Note, running following to test perf endianity handling: test 1) - origin system: # perf record -a -- sleep 10 (any perf record will do) # perf report > report.origin # perf archive perf.data - copy the perf.data, report.origin and perf.data.tar.bz2 to a target system and run: # tar xjvf perf.data.tar.bz2 -C ~/.debug # perf report > report.target # diff -u report.origin report.target - the diff should produce no output (besides some white space stuff and possibly different date/TZ output) test 2) - origin system: # perf record -ag -fo /tmp/perf.data -- sleep 1 - mount origin system root to the target system on /mnt/origin - target system: # perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \ --kallsyms /mnt/origin/proc/kallsyms - complete perf.data header is displayed Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338380624-7443-4-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:44 +00:00
union u64_swap u;
array += ((event->header.size -
sizeof(event->header)) / sizeof(u64)) - 1;
if (type & PERF_SAMPLE_CPU) {
perf evsel: Fix 32 bit values endianity swap for sample_id_all header We swap the sample_id_all header by u64 pointers. Some members of the header happen to be 32 bit values. We need to handle them separatelly. Together with other endianity patches, this change fixies perf report discrepancies on origin and target systems as described in test 1 below, e.g. following perf report diff: ... 0.12% ps [kernel.kallsyms] [k] clear_page - 0.12% awk bash [.] alloc_word_desc + 0.12% awk bash [.] yyparse 0.11% beah-rhts-task libpython2.6.so.1.0 [.] 0x5560e 0.10% perf libc-2.12.so [.] __ctype_toupper_loc - 0.09% rhts-test-runne bash [.] maybe_make_export_env + 0.09% rhts-test-runne bash [.] 0x385a0 0.09% ps [kernel.kallsyms] [k] page_fault ... Note, running following to test perf endianity handling: test 1) - origin system: # perf record -a -- sleep 10 (any perf record will do) # perf report > report.origin # perf archive perf.data - copy the perf.data, report.origin and perf.data.tar.bz2 to a target system and run: # tar xjvf perf.data.tar.bz2 -C ~/.debug # perf report > report.target # diff -u report.origin report.target - the diff should produce no output (besides some white space stuff and possibly different date/TZ output) test 2) - origin system: # perf record -ag -fo /tmp/perf.data -- sleep 1 - mount origin system root to the target system on /mnt/origin - target system: # perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \ --kallsyms /mnt/origin/proc/kallsyms - complete perf.data header is displayed Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338380624-7443-4-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:44 +00:00
u.val64 = *array;
if (swapped) {
/* undo swap of u64, then swap on individual u32s */
u.val64 = bswap_64(u.val64);
u.val32[0] = bswap_32(u.val32[0]);
}
sample->cpu = u.val32[0];
array--;
}
if (type & PERF_SAMPLE_STREAM_ID) {
sample->stream_id = *array;
array--;
}
if (type & PERF_SAMPLE_ID) {
sample->id = *array;
array--;
}
if (type & PERF_SAMPLE_TIME) {
sample->time = *array;
array--;
}
if (type & PERF_SAMPLE_TID) {
perf evsel: Fix 32 bit values endianity swap for sample_id_all header We swap the sample_id_all header by u64 pointers. Some members of the header happen to be 32 bit values. We need to handle them separatelly. Together with other endianity patches, this change fixies perf report discrepancies on origin and target systems as described in test 1 below, e.g. following perf report diff: ... 0.12% ps [kernel.kallsyms] [k] clear_page - 0.12% awk bash [.] alloc_word_desc + 0.12% awk bash [.] yyparse 0.11% beah-rhts-task libpython2.6.so.1.0 [.] 0x5560e 0.10% perf libc-2.12.so [.] __ctype_toupper_loc - 0.09% rhts-test-runne bash [.] maybe_make_export_env + 0.09% rhts-test-runne bash [.] 0x385a0 0.09% ps [kernel.kallsyms] [k] page_fault ... Note, running following to test perf endianity handling: test 1) - origin system: # perf record -a -- sleep 10 (any perf record will do) # perf report > report.origin # perf archive perf.data - copy the perf.data, report.origin and perf.data.tar.bz2 to a target system and run: # tar xjvf perf.data.tar.bz2 -C ~/.debug # perf report > report.target # diff -u report.origin report.target - the diff should produce no output (besides some white space stuff and possibly different date/TZ output) test 2) - origin system: # perf record -ag -fo /tmp/perf.data -- sleep 1 - mount origin system root to the target system on /mnt/origin - target system: # perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \ --kallsyms /mnt/origin/proc/kallsyms - complete perf.data header is displayed Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338380624-7443-4-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:44 +00:00
u.val64 = *array;
if (swapped) {
/* undo swap of u64, then swap on individual u32s */
u.val64 = bswap_64(u.val64);
u.val32[0] = bswap_32(u.val32[0]);
u.val32[1] = bswap_32(u.val32[1]);
}
sample->pid = u.val32[0];
sample->tid = u.val32[1];
}
return 0;
}
static bool sample_overlap(const union perf_event *event,
const void *offset, u64 size)
{
const void *base = event;
if (offset + size > base + event->header.size)
return true;
return false;
}
int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event,
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
struct perf_sample *data, bool swapped)
{
u64 type = evsel->attr.sample_type;
u64 regs_user = evsel->attr.sample_regs_user;
const u64 *array;
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
/*
* used for cross-endian analysis. See git commit 65014ab3
* for why this goofiness is needed.
*/
union u64_swap u;
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
memset(data, 0, sizeof(*data));
data->cpu = data->pid = data->tid = -1;
data->stream_id = data->id = data->time = -1ULL;
perf evsel: Fix an issue where perf report fails to show the proper percentage This patch fixes an issue where perf report shows nan% for certain perf.data files. The below is from a report for a do_fork probe: -nan% sshd [kernel.kallsyms] [k] do_fork -nan% packagekitd [kernel.kallsyms] [k] do_fork -nan% dbus-daemon [kernel.kallsyms] [k] do_fork -nan% bash [kernel.kallsyms] [k] do_fork A git bisect shows commit f3bda2c as the cause. However, looking back through the git history, I saw commit 640c03c which seems to have removed the required initialization for perf_sample->period. The problem only started showing after commit f3bda2c. The below patch re-introduces the initialization and it fixes the problem for me. With the below patch, for the same perf.data: 73.08% bash [kernel.kallsyms] [k] do_fork 8.97% 11-dhclient [kernel.kallsyms] [k] do_fork 6.41% sshd [kernel.kallsyms] [k] do_fork 3.85% 20-chrony [kernel.kallsyms] [k] do_fork 2.56% sendmail [kernel.kallsyms] [k] do_fork This patch applies over current linux-tip commit 9949284. Problem introduced in: $ git describe 640c03c v2.6.37-rc3-83-g640c03c Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Robert Richter <robert.richter@amd.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: stable@kernel.org Link: http://lkml.kernel.org/r/20120203170113.5190.25558.stgit@localhost6.localdomain6 Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-02-03 17:01:13 +00:00
data->period = 1;
if (event->header.type != PERF_RECORD_SAMPLE) {
if (!evsel->attr.sample_id_all)
return 0;
perf evsel: Fix 32 bit values endianity swap for sample_id_all header We swap the sample_id_all header by u64 pointers. Some members of the header happen to be 32 bit values. We need to handle them separatelly. Together with other endianity patches, this change fixies perf report discrepancies on origin and target systems as described in test 1 below, e.g. following perf report diff: ... 0.12% ps [kernel.kallsyms] [k] clear_page - 0.12% awk bash [.] alloc_word_desc + 0.12% awk bash [.] yyparse 0.11% beah-rhts-task libpython2.6.so.1.0 [.] 0x5560e 0.10% perf libc-2.12.so [.] __ctype_toupper_loc - 0.09% rhts-test-runne bash [.] maybe_make_export_env + 0.09% rhts-test-runne bash [.] 0x385a0 0.09% ps [kernel.kallsyms] [k] page_fault ... Note, running following to test perf endianity handling: test 1) - origin system: # perf record -a -- sleep 10 (any perf record will do) # perf report > report.origin # perf archive perf.data - copy the perf.data, report.origin and perf.data.tar.bz2 to a target system and run: # tar xjvf perf.data.tar.bz2 -C ~/.debug # perf report > report.target # diff -u report.origin report.target - the diff should produce no output (besides some white space stuff and possibly different date/TZ output) test 2) - origin system: # perf record -ag -fo /tmp/perf.data -- sleep 1 - mount origin system root to the target system on /mnt/origin - target system: # perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \ --kallsyms /mnt/origin/proc/kallsyms - complete perf.data header is displayed Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1338380624-7443-4-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:44 +00:00
return perf_event__parse_id_sample(event, type, data, swapped);
}
array = event->sample.array;
if (evsel->sample_size + sizeof(event->header) > event->header.size)
return -EFAULT;
if (type & PERF_SAMPLE_IP) {
data->ip = event->ip.ip;
array++;
}
if (type & PERF_SAMPLE_TID) {
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
u.val64 = *array;
if (swapped) {
/* undo swap of u64, then swap on individual u32s */
u.val64 = bswap_64(u.val64);
u.val32[0] = bswap_32(u.val32[0]);
u.val32[1] = bswap_32(u.val32[1]);
}
data->pid = u.val32[0];
data->tid = u.val32[1];
array++;
}
if (type & PERF_SAMPLE_TIME) {
data->time = *array;
array++;
}
data->addr = 0;
if (type & PERF_SAMPLE_ADDR) {
data->addr = *array;
array++;
}
data->id = -1ULL;
if (type & PERF_SAMPLE_ID) {
data->id = *array;
array++;
}
if (type & PERF_SAMPLE_STREAM_ID) {
data->stream_id = *array;
array++;
}
if (type & PERF_SAMPLE_CPU) {
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
u.val64 = *array;
if (swapped) {
/* undo swap of u64, then swap on individual u32s */
u.val64 = bswap_64(u.val64);
u.val32[0] = bswap_32(u.val32[0]);
}
data->cpu = u.val32[0];
array++;
}
if (type & PERF_SAMPLE_PERIOD) {
data->period = *array;
array++;
}
if (type & PERF_SAMPLE_READ) {
fprintf(stderr, "PERF_SAMPLE_READ is unsupported for now\n");
return -1;
}
if (type & PERF_SAMPLE_CALLCHAIN) {
if (sample_overlap(event, array, sizeof(data->callchain->nr)))
return -EFAULT;
data->callchain = (struct ip_callchain *)array;
if (sample_overlap(event, array, data->callchain->nr))
return -EFAULT;
array += 1 + data->callchain->nr;
}
if (type & PERF_SAMPLE_RAW) {
const u64 *pdata;
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
u.val64 = *array;
if (WARN_ONCE(swapped,
"Endianness of raw data not corrected!\n")) {
/* undo swap of u64, then swap on individual u32s */
u.val64 = bswap_64(u.val64);
u.val32[0] = bswap_32(u.val32[0]);
u.val32[1] = bswap_32(u.val32[1]);
}
if (sample_overlap(event, array, sizeof(u32)))
return -EFAULT;
perf tool: Fix endianness handling of u32 data in samples Currently, analyzing PPC data files on x86 the cpu field is always 0 and the tid and pid are backwards. For example, analyzing a PPC file on PPC the pid/tid fields show: rsyslogd 1210/1212 and analyzing the same PPC file using an x86 perf binary shows: rsyslogd 1212/1210 The problem is that the swap_op method for samples is perf_event__all64_swap which assumes all elements in the sample_data struct are u64s. cpu, tid and pid are u32s and need to be handled individually. Given that the swap is done before the sample is parsed, the simplest solution is to undo the 64-bit swap of those elements when the sample is parsed and do the proper swap. The RAW data field is generic and perf cannot have programmatic knowledge of how to treat that data. Instead a warning is given to the user. Thanks to Anton Blanchard for providing a data file for a mult-CPU PPC system so I could verify the fix for the CPU fields. v3 -> v4: - fixed use of WARN_ONCE v2 -> v3: - used WARN_ONCE for message regarding raw data - removed struct wrapper around union - fixed whitespace issues v1 -> v2: - added a union for undoing the byte-swap on u64 and redoing swap on u32's to address compiler errors (see git commit 65014ab3) Cc: Anton Blanchard <anton@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/1315321946-16993-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern <dsahern@gmail.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-09-06 15:12:26 +00:00
data->raw_size = u.val32[0];
pdata = (void *) array + sizeof(u32);
if (sample_overlap(event, pdata, data->raw_size))
return -EFAULT;
data->raw_data = (void *) pdata;
array = (void *)array + data->raw_size + sizeof(u32);
}
if (type & PERF_SAMPLE_BRANCH_STACK) {
u64 sz;
data->branch_stack = (struct branch_stack *)array;
array++; /* nr */
sz = data->branch_stack->nr * sizeof(struct branch_entry);
sz /= sizeof(u64);
array += sz;
}
if (type & PERF_SAMPLE_REGS_USER) {
/* First u64 tells us if we have any regs in sample. */
u64 avail = *array++;
if (avail) {
data->user_regs.regs = (u64 *)array;
array += hweight_long(regs_user);
}
}
if (type & PERF_SAMPLE_STACK_USER) {
u64 size = *array++;
data->user_stack.offset = ((char *)(array - 1)
- (char *) event);
if (!size) {
data->user_stack.size = 0;
} else {
data->user_stack.data = (char *)array;
array += size / sizeof(*array);
data->user_stack.size = *array;
}
}
return 0;
}
int perf_event__synthesize_sample(union perf_event *event, u64 type,
const struct perf_sample *sample,
bool swapped)
{
u64 *array;
/*
* used for cross-endian analysis. See git commit 65014ab3
* for why this goofiness is needed.
*/
union u64_swap u;
array = event->sample.array;
if (type & PERF_SAMPLE_IP) {
event->ip.ip = sample->ip;
array++;
}
if (type & PERF_SAMPLE_TID) {
u.val32[0] = sample->pid;
u.val32[1] = sample->tid;
if (swapped) {
/*
* Inverse of what is done in perf_evsel__parse_sample
*/
u.val32[0] = bswap_32(u.val32[0]);
u.val32[1] = bswap_32(u.val32[1]);
u.val64 = bswap_64(u.val64);
}
*array = u.val64;
array++;
}
if (type & PERF_SAMPLE_TIME) {
*array = sample->time;
array++;
}
if (type & PERF_SAMPLE_ADDR) {
*array = sample->addr;
array++;
}
if (type & PERF_SAMPLE_ID) {
*array = sample->id;
array++;
}
if (type & PERF_SAMPLE_STREAM_ID) {
*array = sample->stream_id;
array++;
}
if (type & PERF_SAMPLE_CPU) {
u.val32[0] = sample->cpu;
if (swapped) {
/*
* Inverse of what is done in perf_evsel__parse_sample
*/
u.val32[0] = bswap_32(u.val32[0]);
u.val64 = bswap_64(u.val64);
}
*array = u.val64;
array++;
}
if (type & PERF_SAMPLE_PERIOD) {
*array = sample->period;
array++;
}
return 0;
}
struct format_field *perf_evsel__field(struct perf_evsel *evsel, const char *name)
{
return pevent_find_field(evsel->tp_format, name);
}
char *perf_evsel__strval(struct perf_evsel *evsel, struct perf_sample *sample,
const char *name)
{
struct format_field *field = perf_evsel__field(evsel, name);
int offset;
if (!field)
return NULL;
offset = field->offset;
if (field->flags & FIELD_IS_DYNAMIC) {
offset = *(int *)(sample->raw_data + field->offset);
offset &= 0xffff;
}
return sample->raw_data + offset;
}
u64 perf_evsel__intval(struct perf_evsel *evsel, struct perf_sample *sample,
const char *name)
{
struct format_field *field = perf_evsel__field(evsel, name);
u64 val;
if (!field)
return 0;
val = pevent_read_number(evsel->tp_format->pevent,
sample->raw_data + field->offset, field->size);
return val;
}