Re: [PATCH] perf: make perf.data more self-descriptive (v4)

[Stephane Eranian]

David,

Something I don't quite understand with your issues with endianess.

How does endianess change anything to a int32 (aka int) vs. uint32?

The byte swapping you need to do is regardless of the sign of
the data or am I missing something here?


On Thu, Sep 8, 2011 at 5:27 PM, David Ahern <dsahern@xxxxxxxxx> wrote:
>
>
> On 09/07/2011 01:10 PM, Stephane Eranian wrote:
> >
> > The goal of this patch is to include more information
> > about the host environment into the perf.data so it is
> > more self-descriptive. Overtime, profiles are captured
> > on various machines and it becomes hard to track what
> > was recorded, on what machine and when.
> >
> > This patch provides a way to solve this by extending
> > the perf.data file with basic information about the
> > host machine. To add those extensions, we leverage
> > the feature bits capabilities of the perf.data format.
> > The change is backward compatible with existing perf.data
> > files.
> >
> > We define the following useful new extensions:
> > Â- HEADER_HOSTNAME: the hostname
> > Â- HEADER_OSRELEASE: the kernel release number
> > Â- HEADER_ARCH: the hw architecture
> > Â- HEADER_CPUID: the cpu model description
> > Â- HEADER_NRCPUS: number of online/avail cpus
> > Â- HEADER_CMDLINE: perf command line
> > Â- HEADER_VERSION: perf version
> > Â- HEADER_TOPOLOGY: cpu topology
> > Â- HEADER_EVENT_DESC: full event description (attrs)
> >
> > The small granularity for the entries is to make it easier
> > to extend without breaking backward compatiblity. All entries
> > are provided as ASCII strings, easy to parse, except for the
> > event description.
> >
> > In the second version, we dropped the -I option for the
> > perf record command. The extra information is systematically
> > captured. But it's still displayed optionally by perf report.
> > There are also a couple of cleanups of function prototypes
> > and global variables.
> >
> > In the third version, we:
> > - fixed missing MIPS support for CPUID
> > - used u32 instead of int when writing integers to file
> > - fixed couple of bswap() bugs
> > - refactorize all write/print functions using function pointers
> > - improved write_cmdline() to include actual path to perf binary
> >
> > In the fourth version, we:
> > - fixed a couple of int vs. u32 issues
> > - factorized HEADER_BUILD_ID, HEADER_TRACE_INFO
> > - added -I option to perf script report
> > - reverted change to evsel.c (unrelated to meta-data)
> > - modified write_*() to always write despite errors (adds_feature bit present)
> > - added HEADER_TOTAL_MEM: total memory installed on this system
> > - added HEADER_NUMA_TOPOLOGY: shows NUMA nodes and for each memory, cpu list
> > - renamed HEADER_TOPOLOGY to HEADER_CPU_TOPOLOGY
>
> Still some endianness problems; cross-endian analysis is causing perf to
> abort. I need some time to investigate. I can send you a ppc data file
> if you have the time.
>
> David
>
> >
> >
> > $ perf record -ecycles,instructions noploop 1
> > noploop for 1 seconds
> > [ perf record: Woken up 1 times to write data ]
> > [ perf record: Captured and wrote 0.098 MB perf.data (~4260 samples) ]
> > $ perf report -I --stdio
> > #
> > # captured on: Wed Sep Â7 21:00:10 2011
> > # hostname : quad
> > # os release : 3.1.0-rc4-tip
> > # arch : x86_64
> > # cpuid : Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
> > # nrcpus online : 4
> > # nrcpus avail : 4
> > # event = cycles, type = 0, config = 0x0, config1 = 0x0, config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 536, 537, 538, 539 }
> > # event = instructions, type = 0, config = 0x1, config1 = 0x0, config2 = 0x0, excl_usr = 0, excl_kern = 0, id = { 540, 541, 542,
> > # cmdline : /home/eranian/perfmon/official/tip/build/tools/perf/perf record -ecycles,instructions noploop 1
> > # perf version : 3.1.0-rc4
> > # CPU0 Â sibling cores Â: 0-3
> > # CPU0 Â sibling threads: 0
> > # CPU1 Â sibling cores Â: 0-3
> > # CPU1 Â sibling threads: 1
> > # CPU2 Â sibling cores Â: 0-3
> > # CPU2 Â sibling threads: 2
> > # CPU3 Â sibling cores Â: 0-3
> > # CPU3 Â sibling threads: 3
> > # total memory: 8105344 kB
> > # node0 Â meminfo : 8320608 kB
> > # node0 Â cpu list: 0-3
> > #
> > # Events: 995 Âcycles
> > #
> > # Overhead ÂCommand   ÂShared Object        ÂSymbol
> > # ........ Â....... Â................. Â....................
> > #
> >   99.90% Ânoploop Ânoploop      Â[.] noploop
> > Â Â Â0.10% Ânoploop Â[kernel.kallsyms] Â[k] trace_hardirqs_on
> > Â Â Â0.00% Ânoploop Â[kernel.kallsyms] Â[k] lock_release
> > Â Â Â0.00% Ânoploop Â[kernel.kallsyms] Â[k] intel_pmu_enable_all
> > ...
> >
> >
> > Signed-off-by: Stephane Eranian <eranian@xxxxxxxxxx>
> > ---
> >
> > diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
> > index 04253c0..191e10e 100644
> > --- a/tools/perf/Documentation/perf-report.txt
> > +++ b/tools/perf/Documentation/perf-report.txt
> > @@ -134,6 +134,11 @@ OPTIONS
> > Â Â Â CPUs are specified with -: 0-2. Default is to report samples on all
> > Â Â Â CPUs.
> >
> > +-I::
> > +--show-info::
> > + Â Â Display information about the perf.data file, incl. hostname,
> > + Â Â os release, perf version, machine desc.
> > +
> > ÂSEE ALSO
> > Â--------
> > Âlinkperf:perf-stat[1]
> > diff --git a/tools/perf/Documentation/perf-script.txt b/tools/perf/Documentation/perf-script.txt
> > index db01786..ebf8a23 100644
> > --- a/tools/perf/Documentation/perf-script.txt
> > +++ b/tools/perf/Documentation/perf-script.txt
> > @@ -188,6 +188,13 @@ OPTIONS
> > Â Â Â CPUs are specified with -: 0-2. Default is to report samples on all
> > Â Â Â CPUs.
> >
> > +-I::
> > +--show-info::
> > + Â Â Display information about the perf.data file, incl. hostname,
> > + Â Â os release, perf version, machine desc. It can only be used with the
> > + Â Â perf script report mode and it must be specified after the script name
> > + Â Â like for instance: perf script report syscall-counts -I.
> > +
> > ÂSEE ALSO
> > Â--------
> > Âlinkperf:perf-record[1], linkperf:perf-script-perl[1],
> > diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
> > index 6b0519f..524ebbf 100644
> > --- a/tools/perf/builtin-record.c
> > +++ b/tools/perf/builtin-record.c
> > @@ -513,6 +513,18 @@ static int __cmd_record(int argc, const char **argv)
> > Â Â Â if (have_tracepoints(&evsel_list->entries))
> > Â Â Â Â Â Â Â perf_header__set_feat(&session->header, HEADER_TRACE_INFO);
> >
> > + Â Â perf_header__set_feat(&session->header, HEADER_HOSTNAME);
> > + Â Â perf_header__set_feat(&session->header, HEADER_OSRELEASE);
> > + Â Â perf_header__set_feat(&session->header, HEADER_ARCH);
> > + Â Â perf_header__set_feat(&session->header, HEADER_CPUID);
> > + Â Â perf_header__set_feat(&session->header, HEADER_NRCPUS);
> > + Â Â perf_header__set_feat(&session->header, HEADER_EVENT_DESC);
> > + Â Â perf_header__set_feat(&session->header, HEADER_CMDLINE);
> > + Â Â perf_header__set_feat(&session->header, HEADER_VERSION);
> > + Â Â perf_header__set_feat(&session->header, HEADER_CPU_TOPOLOGY);
> > + Â Â perf_header__set_feat(&session->header, HEADER_TOTAL_MEM);
> > + Â Â perf_header__set_feat(&session->header, HEADER_NUMA_TOPOLOGY);
> > +
> > Â Â Â /* 512 kiB: default amount of unprivileged mlocked memory */
> > Â Â Â if (mmap_pages == UINT_MAX)
> > Â Â Â Â Â Â Â mmap_pages = (512 * 1024) / page_size;
> > @@ -782,6 +794,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
> > Â Â Â int err = -ENOMEM;
> > Â Â Â struct perf_evsel *pos;
> >
> > + Â Â perf_header__set_cmdline(argc, argv);
> > +
> > Â Â Â evsel_list = perf_evlist__new(NULL, NULL);
> > Â Â Â if (evsel_list == NULL)
> > Â Â Â Â Â Â Â return -ENOMEM;
> > diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
> > index d7ff277..7721030 100644
> > --- a/tools/perf/builtin-report.c
> > +++ b/tools/perf/builtin-report.c
> > @@ -40,6 +40,7 @@ static char     const *input_name = "perf.data";
> > Âstatic bool     Âforce, use_tui, use_stdio;
> > Âstatic bool     Âhide_unresolved;
> > Âstatic bool     Âdont_use_callchains;
> > +static bool     Âshow_info;
> >
> > Âstatic bool     Âshow_threads;
> > Âstatic struct perf_read_values    show_threads_values;
> > @@ -276,6 +277,9 @@ static int __cmd_report(void)
> > Â Â Â Â Â Â Â Â Â Â Â goto out_delete;
> > Â Â Â }
> >
> > + Â Â if (show_info)
> > + Â Â Â Â Â Â perf_session__fprintf_info(session, stdout);
> > +
> > Â Â Â if (show_threads)
> > Â Â Â Â Â Â Â perf_read_values_init(&show_threads_values);
> >
> > @@ -487,6 +491,8 @@ static const struct option options[] = {
> > Â Â Â OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
> > Â Â Â Â Â Â Â Â Â "Look for files with symbols relative to this directory"),
> > Â Â Â OPT_STRING('c', "cpu", &cpu_list, "cpu", "list of cpus to profile"),
> > + Â Â OPT_BOOLEAN('I', "show-info", &show_info,
> > + Â Â Â Â Â Â Â Â Â Â "display information about perf.data file"),
> > Â Â Â OPT_END()
> > Â};
> >
> > diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
> > index 09024ec..dac86ab 100644
> > --- a/tools/perf/builtin-script.c
> > +++ b/tools/perf/builtin-script.c
> > @@ -22,6 +22,7 @@ static u64 Â Â Â Â Â Â Â Â Âlast_timestamp;
> > Âstatic u64 Â Â Â Â Â Â Â Â Â nr_unordered;
> > Âextern const struct option  record_options[];
> > Âstatic bool         Âno_callchain;
> > +static bool         Âshow_info = false;
> > Âstatic const char      Â*cpu_list;
> > Âstatic DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
> >
> > @@ -1083,7 +1084,8 @@ static const struct option options[] = {
> > Â Â Â Â Â Â Â Â Â Â"comma separated output fields prepend with 'type:'. Valid types: hw,sw,trace,raw. Fields: comm,tid,pid,time,cpu,event,trace,ip,sym,dso,addr",
> > Â Â Â Â Â Â Â Â Â Âparse_output_fields),
> > Â Â Â OPT_STRING('c', "cpu", &cpu_list, "cpu", "list of cpus to profile"),
> > -
> > + Â Â OPT_BOOLEAN('I', "show-info", &show_info,
> > + Â Â Â Â Â Â Â Â Â Â "display host information from perf.data file"),
> > Â Â Â OPT_END()
> > Â};
> >
> > @@ -1268,6 +1270,9 @@ int cmd_script(int argc, const char **argv, const char *prefix __used)
> > Â Â Â Â Â Â Â Â Â Â Â return -1;
> > Â Â Â }
> >
> > + Â Â if (show_info)
> > + Â Â Â Â Â Â perf_session__fprintf_info(session, stdout);
> > +
> > Â Â Â if (!no_callchain)
> > Â Â Â Â Â Â Â symbol_conf.use_callchain = true;
> > Â Â Â else
> > diff --git a/tools/perf/builtin.h b/tools/perf/builtin.h
> > index 4702e24..b382bd5 100644
> > --- a/tools/perf/builtin.h
> > +++ b/tools/perf/builtin.h
> > @@ -4,7 +4,6 @@
> > Â#include "util/util.h"
> > Â#include "util/strbuf.h"
> >
> > -extern const char perf_version_string[];
> > Âextern const char perf_usage_string[];
> > Âextern const char perf_more_info_string[];
> >
> > diff --git a/tools/perf/perf.h b/tools/perf/perf.h
> > index a5fc660..08b0b5e 100644
> > --- a/tools/perf/perf.h
> > +++ b/tools/perf/perf.h
> > @@ -9,18 +9,21 @@ void get_term_dimensions(struct winsize *ws);
> > Â#include "../../arch/x86/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile("lock; addl $0,0(%%esp)" ::: "memory")
> > Â#define cpu_relax() Âasm volatile("rep; nop" ::: "memory");
> > +#define CPUINFO_PROC "model name"
> > Â#endif
> >
> > Â#if defined(__x86_64__)
> > Â#include "../../arch/x86/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile("lfence" ::: "memory")
> > Â#define cpu_relax() Âasm volatile("rep; nop" ::: "memory");
> > +#define CPUINFO_PROC "model name"
> > Â#endif
> >
> > Â#ifdef __powerpc__
> > Â#include "../../arch/powerpc/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile ("sync" ::: "memory")
> > Â#define cpu_relax() Âasm volatile ("" ::: "memory");
> > +#define CPUINFO_PROC "cpu"
> > Â#endif
> >
> > Â#ifdef __s390__
> > @@ -37,30 +40,35 @@ void get_term_dimensions(struct winsize *ws);
> > Â# define rmb() Â Â Â Â Â Â Â asm volatile("" ::: "memory")
> > Â#endif
> > Â#define cpu_relax() Âasm volatile("" ::: "memory")
> > +#define CPUINFO_PROC "cpu type"
> > Â#endif
> >
> > Â#ifdef __hppa__
> > Â#include "../../arch/parisc/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile("" ::: "memory")
> > Â#define cpu_relax() Âasm volatile("" ::: "memory");
> > +#define CPUINFO_PROC "cpu"
> > Â#endif
> >
> > Â#ifdef __sparc__
> > Â#include "../../arch/sparc/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile("":::"memory")
> > Â#define cpu_relax() Âasm volatile("":::"memory")
> > +#define CPUINFO_PROC "cpu"
> > Â#endif
> >
> > Â#ifdef __alpha__
> > Â#include "../../arch/alpha/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile("mb" ::: "memory")
> > Â#define cpu_relax() Âasm volatile("" ::: "memory")
> > +#define CPUINFO_PROC "cpu model"
> > Â#endif
> >
> > Â#ifdef __ia64__
> > Â#include "../../arch/ia64/include/asm/unistd.h"
> > Â#define rmb() Â Â Â Â Â Â Â Âasm volatile ("mf" ::: "memory")
> > Â#define cpu_relax() Âasm volatile ("hint @pause" ::: "memory")
> > +#define CPUINFO_PROC "model name"
> > Â#endif
> >
> > Â#ifdef __arm__
> > @@ -71,6 +79,7 @@ void get_term_dimensions(struct winsize *ws);
> > Â */
> > Â#define rmb() Â Â Â Â Â Â Â Â((void(*)(void))0xffff0fa0)()
> > Â#define cpu_relax() Âasm volatile("":::"memory")
> > +#define CPUINFO_PROC "Processor"
> > Â#endif
> >
> > Â#ifdef __mips__
> > @@ -83,6 +92,7 @@ void get_term_dimensions(struct winsize *ws);
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â : /* no input */ Â Â Â Â Â Â Â Â Â Â Â Â\
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â : "memory")
> > Â#define cpu_relax() Âasm volatile("" ::: "memory")
> > +#define CPUINFO_PROC "cpu model"
> > Â#endif
> >
> > Â#include <time.h>
> > @@ -171,5 +181,6 @@ struct ip_callchain {
> > Â};
> >
> > Âextern bool perf_host, perf_guest;
> > +extern const char perf_version_string[];
> >
> > Â#endif
> > diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> > index b6c1ad1..73fa59e 100644
> > --- a/tools/perf/util/header.c
> > +++ b/tools/perf/util/header.c
> > @@ -7,6 +7,7 @@
> > Â#include <stdlib.h>
> > Â#include <linux/list.h>
> > Â#include <linux/kernel.h>
> > +#include <sys/utsname.h>
> >
> > Â#include "evlist.h"
> > Â#include "evsel.h"
> > @@ -17,12 +18,19 @@
> > Â#include "session.h"
> > Â#include "symbol.h"
> > Â#include "debug.h"
> > +#include "cpumap.h"
> >
> > Âstatic bool no_buildid_cache = false;
> >
> > Âstatic int event_count;
> > Âstatic struct perf_trace_event_type *events;
> >
> > +static u32 header_argc;
> > +static const char **header_argv;
> > +
> > +static int dsos__write_buildid_table(struct perf_header *header, int fd);
> > +static int perf_session__cache_build_ids(struct perf_session *session);
> > +
> > Âint perf_header__push_event(u64 id, const char *name)
> > Â{
> > Â Â Â if (strlen(name) > MAX_EVENT_NAME)
> > @@ -110,6 +118,843 @@ static int write_padded(int fd, const void *bf, size_t count,
> > Â Â Â return err;
> > Â}
> >
> > +static int do_write_string(int fd, const char *str)
> > +{
> > + Â Â u32 len, olen;
> > + Â Â int ret;
> > +
> > + Â Â olen = strlen(str) + 1;
> > + Â Â len = ALIGN(olen, NAME_ALIGN);
> > +
> > + Â Â /* write len, incl. \0 */
> > + Â Â ret = do_write(fd, &len, sizeof(len));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â return write_padded(fd, str, olen, len);
> > +}
> > +
> > +static char *do_read_string(int fd, struct perf_header *ph)
> > +{
> > + Â Â ssize_t sz, ret;
> > + Â Â u32 len;
> > + Â Â char *buf;
> > +
> > + Â Â sz = read(fd, &len, sizeof(len));
> > + Â Â if (sz < (ssize_t)sizeof(len))
> > + Â Â Â Â Â Â return NULL;
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â len = bswap_32(len);
> > +
> > + Â Â buf = malloc(len);
> > + Â Â if (!buf)
> > + Â Â Â Â Â Â return NULL;
> > +
> > + Â Â ret = read(fd, buf, len);
> > + Â Â if (ret == (ssize_t)len) {
> > + Â Â Â Â Â Â /*
> > + Â Â Â Â Â Â Â* strings are padded by zeroes
> > + Â Â Â Â Â Â Â* thus the actual strlen of buf
> > + Â Â Â Â Â Â Â* may be less than len
> > + Â Â Â Â Â Â Â*/
> > + Â Â Â Â Â Â return buf;
> > + Â Â }
> > +
> > + Â Â free(buf);
> > + Â Â return NULL;
> > +}
> > +
> > +int
> > +perf_header__set_cmdline(int argc, const char **argv)
> > +{
> > + Â Â int i;
> > +
> > + Â Â header_argc = (u32)argc;
> > +
> > + Â Â /* do not include NULL termination */
> > + Â Â header_argv = calloc(argc, sizeof(char *));
> > + Â Â if (!header_argv)
> > + Â Â Â Â Â Â return -ENOMEM;
> > +
> > + Â Â /*
> > + Â Â Â* must copy argv contents because it gets moved
> > + Â Â Â* around during option parsing
> > + Â Â Â*/
> > + Â Â for (i = 0; i < argc ; i++)
> > + Â Â Â Â Â Â header_argv[i] = argv[i];
> > +
> > + Â Â return 0;
> > +}
> > +
> > +static int write_trace_info(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist)
> > +{
> > + Â Â return read_tracing_data(fd, &evlist->entries);
> > +}
> > +
> > +
> > +static int write_build_id(int fd, struct perf_header *h,
> > + Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â struct perf_session *session;
> > + Â Â int err;
> > +
> > + Â Â session = container_of(h, struct perf_session, header);
> > +
> > + Â Â err = dsos__write_buildid_table(h, fd);
> > + Â Â if (err < 0) {
> > + Â Â Â Â Â Â pr_debug("failed to write buildid table\n");
> > + Â Â Â Â Â Â return err;
> > + Â Â }
> > + Â Â if (!no_buildid_cache)
> > + Â Â Â Â Â Â perf_session__cache_build_ids(session);
> > +
> > + Â Â return 0;
> > +}
> > +
> > +static int write_hostname(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â struct utsname uts;
> > + Â Â int ret;
> > +
> > + Â Â ret = uname(&uts);
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â return do_write_string(fd, uts.nodename);
> > +}
> > +
> > +static int write_osrelease(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist __used)
> > +{
> > + Â Â struct utsname uts;
> > + Â Â int ret;
> > +
> > + Â Â ret = uname(&uts);
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â return do_write_string(fd, uts.release);
> > +}
> > +
> > +static int write_arch(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â struct utsname uts;
> > + Â Â int ret;
> > +
> > + Â Â ret = uname(&uts);
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â return do_write_string(fd, uts.machine);
> > +}
> > +
> > +static int write_version(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist __used)
> > +{
> > + Â Â return do_write_string(fd, perf_version_string);
> > +}
> > +
> > +static int write_cpuid(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist __used)
> > +{
> > +#ifndef CPUINFO_PROC
> > +#define CPUINFO_PROC NULL
> > +#endif
> > + Â Â FILE *file;
> > + Â Â char buf[256];
> > + Â Â char *s, *p;
> > + Â Â const char *search = CPUINFO_PROC;
> > +
> > + Â Â file = fopen("/proc/cpuinfo", "r");
> > + Â Â if (!file)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â if (search) {
> > + Â Â Â Â Â Â while (fgets(buf, sizeof(buf), file)) {
> > + Â Â Â Â Â Â Â Â Â Â if (strstr(buf, search))
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â goto found;
> > + Â Â Â Â Â Â }
> > + Â Â }
> > + Â Â strcpy(buf, "unknown type");
> > +found:
> > + Â Â fclose(file);
> > +
> > + Â Â s = buf;
> > +
> > + Â Â p = strchr(buf, ':');
> > + Â Â if (p && *(p+1) == ' ' && *(p+2))
> > + Â Â Â Â Â Â s = p + 2;
> > + Â Â p = strchr(s, '\n');
> > + Â Â if (p)
> > + Â Â Â Â Â Â *p = '\0';
> > +
> > + Â Â /* squash extra space characters (branding string) */
> > + Â Â p = s;
> > + Â Â while (*p) {
> > + Â Â Â Â Â Â if (isspace(*p)) {
> > + Â Â Â Â Â Â Â Â Â Â char *r = p + 1;
> > + Â Â Â Â Â Â Â Â Â Â char *q = r;
> > + Â Â Â Â Â Â Â Â Â Â *p = ' ';
> > + Â Â Â Â Â Â Â Â Â Â while (*q && isspace(*q))
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â q++;
> > + Â Â Â Â Â Â Â Â Â Â if (q != (p+1))
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â while ((*r++ = *q++));
> > + Â Â Â Â Â Â }
> > + Â Â Â Â Â Â p++;
> > + Â Â }
> > + Â Â return do_write_string(fd, s);
> > +}
> > +
> > +static int write_nrcpus(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â long nr;
> > + Â Â u32 nrc, nra;
> > + Â Â int ret;
> > +
> > + Â Â nr = sysconf(_SC_NPROCESSORS_CONF);
> > + Â Â if (nr < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â nrc = (u32)(nr & UINT_MAX);
> > +
> > + Â Â ret = sysconf(_SC_NPROCESSORS_ONLN);
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â nra = (u32)(nr & UINT_MAX);
> > +
> > + Â Â ret = do_write(fd, &nrc, sizeof(nrc));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â return do_write(fd, &nra, sizeof(nra));
> > +}
> > +
> > +static int write_event_desc(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist)
> > +{
> > + Â Â struct perf_evsel *attr;
> > + Â Â u32 nre = 0, nri, sz;
> > + Â Â int ret;
> > +
> > + Â Â list_for_each_entry(attr, &evlist->entries, node)
> > + Â Â Â Â Â Â nre++;
> > +
> > + Â Â /*
> > + Â Â Â* write number of events
> > + Â Â Â*/
> > + Â Â ret = do_write(fd, &nre, sizeof(nre));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â /*
> > + Â Â Â* size of perf_event_attr struct
> > + Â Â Â*/
> > + Â Â sz = (u32)sizeof(attr->attr);
> > + Â Â ret = do_write(fd, &sz, sizeof(sz));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â list_for_each_entry(attr, &evlist->entries, node) {
> > +
> > + Â Â Â Â Â Â ret = do_write(fd, &attr->attr, sz);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â Â Â Â Â /*
> > + Â Â Â Â Â Â Â* write number of unique id per event
> > + Â Â Â Â Â Â Â* there is one id per instance of an event
> > + Â Â Â Â Â Â Â*
> > + Â Â Â Â Â Â Â* copy into an nri to be independent of the
> > + Â Â Â Â Â Â Â* type of ids,
> > + Â Â Â Â Â Â Â*/
> > + Â Â Â Â Â Â nri = attr->ids;
> > + Â Â Â Â Â Â ret = do_write(fd, &nri, sizeof(nri));
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > +
> > + Â Â Â Â Â Â /*
> > + Â Â Â Â Â Â Â* write event string as passed on cmdline
> > + Â Â Â Â Â Â Â*/
> > + Â Â Â Â Â Â ret = do_write_string(fd, attr->name);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â Â Â Â Â /*
> > + Â Â Â Â Â Â Â* write unique ids for this event
> > + Â Â Â Â Â Â Â*/
> > + Â Â Â Â Â Â ret = do_write(fd, attr->id, attr->ids * sizeof(u64));
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â }
> > + Â Â return 0;
> > +}
> > +
> > +static int write_cmdline(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist __used)
> > +{
> > + Â Â char buf[MAXPATHLEN];
> > + Â Â char proc[32];
> > + Â Â u32 i, n;
> > + Â Â int ret;
> > +
> > + Â Â /*
> > + Â Â Â* actual atual path to perf binary
> > + Â Â Â*/
> > + Â Â sprintf(proc, "/proc/%d/exe", getpid());
> > + Â Â ret = readlink(proc, buf, sizeof(buf));
> > + Â Â if (ret <= 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â /* readlink() does not add null termination */
> > + Â Â buf[ret] = '\0';
> > +
> > + Â Â /* account for binary path */
> > + Â Â n = header_argc + 1;
> > +
> > + Â Â ret = do_write(fd, &n, sizeof(n));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â ret = do_write_string(fd, buf);
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return ret;
> > +
> > + Â Â for (i = 0 ; i < header_argc; i++) {
> > + Â Â Â Â Â Â ret = do_write_string(fd, header_argv[i]);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â }
> > + Â Â return 0;
> > +}
> > +
> > +static const char *topo_fmt[] = {
> > + Â Â "/sys/devices/system/cpu/cpu%d/topology/core_siblings_list",
> > + Â Â "/sys/devices/system/cpu/cpu%d/topology/thread_siblings_list",
> > + Â Â NULL
> > +};
> > +
> > +static int write_topo_cpu(int fd, int cpu)
> > +{
> > + Â Â FILE *fp;
> > + Â Â char filename[MAXPATHLEN];
> > + Â Â char buf[MAXPATHLEN], *p, *c;
> > + Â Â const char **q = topo_fmt;
> > + Â Â int ret;
> > +
> > + Â Â while (*q) {
> > +
> > + Â Â Â Â Â Â sprintf(filename, *q, cpu);
> > +
> > + Â Â Â Â Â Â fp = fopen(filename, "r");
> > + Â Â Â Â Â Â if (!fp)
> > + Â Â Â Â Â Â Â Â Â Â return -1;
> > +
> > + Â Â Â Â Â Â c = fgets(buf, sizeof(buf), fp);
> > + Â Â Â Â Â Â if (!c) {
> > + Â Â Â Â Â Â Â Â Â Â fclose(fp);
> > + Â Â Â Â Â Â Â Â Â Â return -1;
> > + Â Â Â Â Â Â }
> > +
> > + Â Â Â Â Â Â p = strchr(buf, '\n');
> > + Â Â Â Â Â Â if (p)
> > + Â Â Â Â Â Â Â Â Â Â *p = '\0';
> > +
> > + Â Â Â Â Â Â fclose(fp);
> > +
> > + Â Â Â Â Â Â ret = do_write_string(fd, c);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â Â Â Â Â q++;
> > + Â Â }
> > + Â Â return 0;
> > +}
> > +
> > +static int write_cpu_topology(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â u32 nr, i;
> > + Â Â long ncpus;
> > + Â Â int ret;
> > +
> > + Â Â ncpus = sysconf(_SC_NPROCESSORS_CONF);
> > + Â Â if (ncpus < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â /* hopefully, this will keep compilers happy about the cast */
> > + Â Â nr = (u32)(ncpus & UINT_MAX);
> > +
> > + Â Â ret = do_write(fd, &nr, sizeof(nr));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â for (i = 0; i < nr; i++) {
> > + Â Â Â Â Â Â ret = do_write(fd, &i, sizeof(i));
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > +
> > + Â Â Â Â Â Â ret = write_topo_cpu(fd, i);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â return ret;
> > + Â Â }
> > + Â Â return 0;
> > +}
> > +
> > +static int write_total_mem(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â char buf[MAXPATHLEN], *c;
> > + Â Â FILE *fp;
> > + Â Â int ret = -1, n;
> > + Â Â uint64_t mem;
> > +
> > + Â Â fp = fopen("/proc/meminfo", "r");
> > + Â Â if (!fp)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â while ((c = fgets(buf, sizeof(buf), fp))) {
> > + Â Â Â Â Â Â c = strstr(buf, "MemTotal:");
> > + Â Â Â Â Â Â if (c)
> > + Â Â Â Â Â Â Â Â Â Â break;
> > + Â Â }
> > + Â Â fclose(fp);
> > +
> > + Â Â n = sscanf(buf, "%*s %"PRIu64, &mem);
> > + Â Â if (n == 1)
> > + Â Â Â Â Â Â ret = do_write(fd, &mem, sizeof(mem));
> > +
> > + Â Â return ret;
> > +}
> > +
> > +static int write_topo_node(int fd, int node)
> > +{
> > + Â Â char filename[MAXPATHLEN];
> > + Â Â char buf[MAXPATHLEN], *p, *c;
> > + Â Â FILE *fp;
> > + Â Â uint64_t mem;
> > + Â Â int ret;
> > +
> > + Â Â sprintf(filename, "/sys/devices/system/node/node%d/meminfo", node);
> > + Â Â fp = fopen(filename, "r");
> > + Â Â if (!fp)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â while ((c = fgets(buf, sizeof(buf), fp))) {
> > + Â Â Â Â Â Â c = strstr(buf, "MemTotal:");
> > + Â Â Â Â Â Â if (c)
> > + Â Â Â Â Â Â Â Â Â Â break;
> > + Â Â }
> > + Â Â fclose(fp);
> > + Â Â if (!c)
> > + Â Â Â Â Â Â return -1;
> > + Â Â ret = sscanf(buf, "%*s %*d %*s %"PRIu64, &mem);
> > + Â Â if (ret != 1)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â ret = do_write(fd, &mem, sizeof(mem));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â sprintf(filename, "/sys/devices/system/node/node%d/cpulist", node);
> > + Â Â fp = fopen(filename, "r");
> > + Â Â if (!fp)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â c = fgets(buf, sizeof(buf), fp);
> > + Â Â fclose(fp);
> > + Â Â if (!c)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â p = strchr(buf, '\n');
> > + Â Â if (p)
> > + Â Â Â Â Â Â *p = '\0';
> > +
> > + Â Â return do_write_string(fd, c);
> > +}
> > +
> > +static int write_numa_topology(int fd, struct perf_header *h __used,
> > + Â Â Â Â Â Â Â Â Â Â Â struct perf_evlist *evlist __used)
> > +{
> > + Â Â char buf[MAXPATHLEN];
> > + Â Â FILE *fp;
> > + Â Â struct cpu_map *node_map;
> > + Â Â char *c;
> > + Â Â u32 nr, i, j;
> > + Â Â int ret;
> > +
> > + Â Â fp = fopen("/sys/devices/system/node/online", "r");
> > + Â Â if (!fp)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â c = fgets(buf, sizeof(buf), fp);
> > + Â Â fclose(fp);
> > + Â Â if (!c)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â c = strchr(buf, '\n');
> > + Â Â if (c)
> > + Â Â Â Â Â Â *c = '\0';
> > +
> > + Â Â node_map = cpu_map__new(buf);
> > + Â Â if (!node_map)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â nr = (u32)node_map->nr;
> > + Â Â ret = do_write(fd, &nr, sizeof(nr));
> > + Â Â if (ret < 0)
> > + Â Â Â Â Â Â return -1;
> > +
> > + Â Â for (i = 0; i < nr; i++) {
> > + Â Â Â Â Â Â j = (u32)node_map->map[i];
> > + Â Â Â Â Â Â ret = do_write(fd, &j, sizeof(j));
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â ret = write_topo_node(fd, i);
> > + Â Â Â Â Â Â if (ret < 0)
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > + Â Â }
> > + Â Â ret = 0;
> > +error:
> > + Â Â free(node_map);
> > + Â Â return ret;
> > +}
> > +
> > +static void print_hostname(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â char *str = do_read_string(fd, ph);
> > + Â Â fprintf(fp, "# hostname : %s\n", str);
> > + Â Â free(str);
> > +}
> > +
> > +static void print_osrelease(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â char *str = do_read_string(fd, ph);
> > + Â Â fprintf(fp, "# os release : %s\n", str);
> > + Â Â free(str);
> > +}
> > +
> > +static void print_arch(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â char *str = do_read_string(fd, ph);
> > + Â Â fprintf(fp, "# arch : %s\n", str);
> > + Â Â free(str);
> > +}
> > +
> > +static void print_cpuid(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â char *str = do_read_string(fd, ph);
> > + Â Â fprintf(fp, "# cpuid : %s\n", str);
> > + Â Â free(str);
> > +}
> > +
> > +static void print_nrcpus(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â ssize_t ret;
> > + Â Â u32 nr;
> > +
> > + Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â if (ret != (ssize_t)sizeof(nr))
> > + Â Â Â Â Â Â nr = -1; /* interpreted as error */
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â fprintf(fp, "# nrcpus online : %u\n", nr);
> > +
> > + Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â if (ret != (ssize_t)sizeof(nr))
> > + Â Â Â Â Â Â nr = -1; /* interpreted as error */
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â fprintf(fp, "# nrcpus avail : %u\n", nr);
> > +}
> > +
> > +static void print_version(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â char *str = do_read_string(fd, ph);
> > + Â Â fprintf(fp, "# perf version : %s\n", str);
> > + Â Â free(str);
> > +}
> > +
> > +static void print_cmdline(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â ssize_t ret;
> > + Â Â char *str;
> > + Â Â u32 nr, i;
> > +
> > + Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â if (ret != (ssize_t)sizeof(nr)) {
> > + Â Â Â Â Â Â fprintf(fp, "# cmdline : unknown\n");
> > + Â Â Â Â Â Â return;
> > + Â Â }
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â fprintf(fp, "# cmdline : ");
> > +
> > + Â Â for (i = 0; i < nr; i++) {
> > + Â Â Â Â Â Â str = do_read_string(fd, ph);
> > + Â Â Â Â Â Â fprintf(fp, "%s ", str);
> > + Â Â Â Â Â Â free(str);
> > + Â Â }
> > + Â Â fputc('\n', fp);
> > +}
> > +
> > +static void print_cpu_topology(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â ssize_t ret;
> > + Â Â u32 nr, c, i;
> > + Â Â char *str;
> > +
> > + Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â if (ret != (ssize_t)sizeof(nr)) {
> > + Â Â Â Â Â Â fprintf(fp, "# cpu topology Â: not available\n");
> > + Â Â Â Â Â Â return;
> > + Â Â }
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â for (i = 0; i < nr; i++) {
> > +
> > + Â Â Â Â Â Â ret = read(fd, &c, sizeof(c));
> > + Â Â Â Â Â Â if (ret != (ssize_t)sizeof(c))
> > + Â Â Â Â Â Â Â Â Â Â c = -1; /* interpreted as error */
> > +
> > + Â Â Â Â Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â c = bswap_32(c);
> > +
> > + Â Â Â Â Â Â str = do_read_string(fd, ph);
> > + Â Â Â Â Â Â fprintf(fp, "# CPU%u sibling cores Â: %s\n", c, str);
> > + Â Â Â Â Â Â free(str);
> > +
> > + Â Â Â Â Â Â str = do_read_string(fd, ph);
> > + Â Â Â Â Â Â fprintf(fp, "# CPU%u sibling threads: %s\n", c, str);
> > + Â Â Â Â Â Â free(str);
> > + Â Â }
> > +}
> > +
> > +static void print_event_desc(struct perf_header *ph, int fd, FILE *fp)
> > +{
> > + Â Â struct perf_event_attr attr;
> > + Â Â uint64_t id;
> > + Â Â void *buf = NULL;
> > + Â Â char *str;
> > + Â Â u32 nre, sz, nr, i, j, msz;
> > + Â Â int ret;
> > +
> > + Â Â /* number of events */
> > + Â Â ret = read(fd, &nre, sizeof(nre));
> > + Â Â if (ret != (ssize_t)sizeof(nre))
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â nre = bswap_32(nre);
> > +
> > + Â Â ret = read(fd, &sz, sizeof(sz));
> > + Â Â if (ret != (ssize_t)sizeof(sz))
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â sz = bswap_32(sz);
> > +
> > + Â Â /*
> > + Â Â Â* ensure it is at least to our ABI rev
> > + Â Â Â*/
> > + Â Â if (sz < (u32)sizeof(attr))
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â memset(&attr, 0, sizeof(attr));
> > +
> > + Â Â /* read entire region to sync up to next field */
> > + Â Â buf = malloc(sz);
> > + Â Â if (!buf)
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â msz = sizeof(attr);
> > + Â Â if (sz < msz)
> > + Â Â Â Â Â Â msz = sz;
> > +
> > + Â Â for (i = 0 ; i < nre; i++) {
> > +
> > + Â Â Â Â Â Â ret = read(fd, buf, sz);
> > + Â Â Â Â Â Â if (ret != (ssize_t)sz)
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â perf_event__attr_swap(buf);
> > +
> > + Â Â Â Â Â Â memcpy(&attr, buf, msz);
> > +
> > + Â Â Â Â Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â Â Â Â Â if (ret != (ssize_t)sizeof(nr))
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â Â Â Â Â str = do_read_string(fd, ph);
> > + Â Â Â Â Â Â fprintf(fp, "# event = %s, ", str);
> > + Â Â Â Â Â Â free(str);
> > +
> > + Â Â Â Â Â Â fprintf(fp, "type = %d, config = 0x%"PRIx64
> > + Â Â Â Â Â Â Â Â Â Â Â Â ", config1 = 0x%"PRIx64", config2 = 0x%"PRIx64,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â attr.type,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â (u64)attr.config,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â (u64)attr.config1,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â (u64)attr.config2);
> > +
> > + Â Â Â Â Â Â fprintf(fp, ", excl_usr = %d, excl_kern = %d",
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â attr.exclude_user,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â attr.exclude_kernel);
> > +
> > + Â Â Â Â Â Â if (nr)
> > + Â Â Â Â Â Â Â Â Â Â fprintf(fp, ", id = {");
> > +
> > + Â Â Â Â Â Â for (j = 0 ; j < nr; j++) {
> > + Â Â Â Â Â Â Â Â Â Â ret = read(fd, &id, sizeof(id));
> > + Â Â Â Â Â Â Â Â Â Â if (ret != (ssize_t)sizeof(id))
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â Â Â Â Â if (ph->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â id = bswap_64(id);
> > +
> > + Â Â Â Â Â Â Â Â Â Â if (j)
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â fputc(',', fp);
> > +
> > + Â Â Â Â Â Â Â Â Â Â fprintf(fp, " %"PRIu64, id);
> > + Â Â Â Â Â Â }
> > + Â Â Â Â Â Â if (nr && j == nr)
> > + Â Â Â Â Â Â Â Â Â Â fprintf(fp, " }");
> > + Â Â Â Â Â Â fputc('\n', fp);
> > + Â Â }
> > + Â Â free(buf);
> > + Â Â return;
> > +error:
> > + Â Â fprintf(fp, "# event desc: not available or unable to read\n");
> > +}
> > +
> > +static void print_total_mem(struct perf_header *h __used, int fd, FILE *fp)
> > +{
> > + Â Â uint64_t mem;
> > + Â Â ssize_t ret;
> > +
> > + Â Â ret = read(fd, &mem, sizeof(mem));
> > + Â Â if (ret != sizeof(mem))
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â if (h->needs_swap)
> > + Â Â Â Â Â Â mem = bswap_64(mem);
> > +
> > + Â Â fprintf(fp, "# total memory: %"PRIu64" kB\n", mem);
> > + Â Â return;
> > +error:
> > + Â Â fprintf(fp, "# total memory: unknown\n");
> > +}
> > +
> > +static void print_numa_topology(struct perf_header *h __used, int fd, FILE *fp)
> > +{
> > + Â Â ssize_t ret;
> > + Â Â u32 nr, c, i;
> > + Â Â char *str;
> > + Â Â uint64_t mem;
> > +
> > + Â Â /* nr nodes */
> > + Â Â ret = read(fd, &nr, sizeof(nr));
> > + Â Â if (ret != (ssize_t)sizeof(nr))
> > + Â Â Â Â Â Â goto error;
> > +
> > + Â Â if (h->needs_swap)
> > + Â Â Â Â Â Â nr = bswap_32(nr);
> > +
> > + Â Â for (i = 0; i < nr; i++) {
> > +
> > + Â Â Â Â Â Â /* node number */
> > + Â Â Â Â Â Â ret = read(fd, &c, sizeof(c));
> > + Â Â Â Â Â Â if (ret != (ssize_t)sizeof(c))
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â if (h->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â c = bswap_32(c);
> > +
> > + Â Â Â Â Â Â ret = read(fd, &mem, sizeof(mem));
> > + Â Â Â Â Â Â if (ret != sizeof(mem))
> > + Â Â Â Â Â Â Â Â Â Â goto error;
> > +
> > + Â Â Â Â Â Â if (h->needs_swap)
> > + Â Â Â Â Â Â Â Â Â Â mem = bswap_64(mem);
> > +
> > + Â Â Â Â Â Â fprintf(fp, "# node%u meminfo : %"PRIu64" kB\n", c, mem);
> > +
> > + Â Â Â Â Â Â str = do_read_string(fd, h);
> > + Â Â Â Â Â Â fprintf(fp, "# node%u cpu list: %s\n", c, str);
> > + Â Â Â Â Â Â free(str);
> > + Â Â }
> > + Â Â return;
> > +error:
> > + Â Â fprintf(fp, "# numa topology : not available\n");
> > +}
> > +
> > +struct feature_ops {
> > + Â Â int (*write)(int fd, struct perf_header *h, struct perf_evlist *evlist);
> > + Â Â void (*print)(struct perf_header *h, int fd, FILE *fp);
> > +};
> > +
> > +#define FEAT_OP(n, w, p) [n] = { .write = w, .print = p }
> > +
> > +static struct feature_ops feat_ops[HEADER_LAST_FEATURE] = {
> > + Â Â FEAT_OP(HEADER_TRACE_INFO, write_trace_info, NULL),
> > + Â Â FEAT_OP(HEADER_BUILD_ID, write_build_id, NULL),
> > + Â Â FEAT_OP(HEADER_HOSTNAME, write_hostname, print_hostname),
> > + Â Â FEAT_OP(HEADER_OSRELEASE, write_osrelease, print_osrelease),
> > + Â Â FEAT_OP(HEADER_ARCH, write_arch, print_arch),
> > + Â Â FEAT_OP(HEADER_CPUID, write_cpuid, print_cpuid),
> > + Â Â FEAT_OP(HEADER_NRCPUS, write_nrcpus, print_nrcpus),
> > + Â Â FEAT_OP(HEADER_EVENT_DESC, write_event_desc, print_event_desc),
> > + Â Â FEAT_OP(HEADER_CMDLINE, write_cmdline, print_cmdline),
> > + Â Â FEAT_OP(HEADER_VERSION, write_version, print_version),
> > + Â Â FEAT_OP(HEADER_CPU_TOPOLOGY, write_cpu_topology, print_cpu_topology),
> > + Â Â FEAT_OP(HEADER_TOTAL_MEM, write_total_mem, print_total_mem),
> > + Â Â FEAT_OP(HEADER_NUMA_TOPOLOGY, write_numa_topology, print_numa_topology),
> > +};
> > +
> > +static int perf_header_fprintf_info(struct perf_file_section *section,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â struct perf_header *ph,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int feat, int fd, void *data)
> > +{
> > + Â Â FILE *fp = data;
> > +
> > + Â Â if (lseek(fd, section->offset, SEEK_SET) == (off_t)-1) {
> > + Â Â Â Â Â Â pr_debug("Failed to lseek to %" PRIu64 " offset for feature "
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â "%d, continuing...\n", section->offset, feat);
> > + Â Â Â Â Â Â return 0;
> > + Â Â }
> > + Â Â if (feat < HEADER_TRACE_INFO || feat >= HEADER_LAST_FEATURE) {
> > + Â Â Â Â Â Â pr_warning("unknown feature %d\n", feat);
> > + Â Â Â Â Â Â return -1;
> > + Â Â }
> > +
> > + Â Â if (feat_ops[feat].print)
> > + Â Â Â Â Â Â feat_ops[feat].print(ph, fd, fp);
> > +
> > + Â Â return 0;
> > +}
> > +
> > +int perf_header__fprintf_info(struct perf_session *session, FILE *fp __used)
> > +{
> > + Â Â int fd = session->fd;
> > + Â Â struct perf_header *header = &session->header;
> > + Â Â perf_header__process_sections(header, fd, fp, perf_header_fprintf_info);
> > + Â Â return 0;
> > +}
> > +
> > Â#define dsos__for_each_with_build_id(pos, head) Â Â Â\
> > Â Â Â list_for_each_entry(pos, head, node) Â Â\
> > Â Â Â Â Â Â Â if (!pos->has_build_id) Â Â Â Â \
> > @@ -356,15 +1201,41 @@ static bool perf_session__read_build_ids(struct perf_session *session, bool with
> > Â Â Â return ret;
> > Â}
> >
> > +static int do_write_feat(int fd, struct perf_header *h, int type,
> > + Â Â Â Â Â Â Â Â Â Â Âstruct perf_file_section **p,
> > + Â Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist)
> > +{
> > + Â Â int err;
> > + Â Â int ret = 0;
> > +
> > + Â Â if (perf_header__has_feat(h, type)) {
> > +
> > + Â Â Â Â Â Â (*p)->offset = lseek(fd, 0, SEEK_CUR);
> > +
> > + Â Â Â Â Â Â err = feat_ops[type].write(fd, h, evlist);
> > + Â Â Â Â Â Â if (err < 0) {
> > + Â Â Â Â Â Â Â Â Â Â pr_debug("failed to write feature %d\n", type);
> > +
> > + Â Â Â Â Â Â Â Â Â Â /* undo anything written */
> > + Â Â Â Â Â Â Â Â Â Â lseek(fd, (*p)->offset, SEEK_SET);
> > +
> > + Â Â Â Â Â Â Â Â Â Â return -1;
> > + Â Â Â Â Â Â }
> > + Â Â Â Â Â Â (*p)->size = lseek(fd, 0, SEEK_CUR) - (*p)->offset;
> > + Â Â Â Â Â Â (*p)++;
> > + Â Â }
> > + Â Â return ret;
> > +}
> > +
> > Âstatic int perf_header__adds_write(struct perf_header *header,
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âstruct perf_evlist *evlist, int fd)
> > Â{
> > Â Â Â int nr_sections;
> > Â Â Â struct perf_session *session;
> > - Â Â struct perf_file_section *feat_sec;
> > + Â Â struct perf_file_section *feat_sec, *p;
> > Â Â Â int sec_size;
> > Â Â Â u64 sec_start;
> > - Â Â int idx = 0, err;
> > + Â Â int err;
> >
> > Â Â Â session = container_of(header, struct perf_session, header);
> >
> > @@ -376,7 +1247,7 @@ static int perf_header__adds_write(struct perf_header *header,
> > Â Â Â if (!nr_sections)
> > Â Â Â Â Â Â Â return 0;
> >
> > - Â Â feat_sec = calloc(sizeof(*feat_sec), nr_sections);
> > + Â Â feat_sec = p = calloc(sizeof(*feat_sec), nr_sections);
> > Â Â Â if (feat_sec == NULL)
> > Â Â Â Â Â Â Â return -ENOMEM;
> >
> > @@ -385,36 +1256,65 @@ static int perf_header__adds_write(struct perf_header *header,
> > Â Â Â sec_start = header->data_offset + header->data_size;
> > Â Â Â lseek(fd, sec_start + sec_size, SEEK_SET);
> >
> > - Â Â if (perf_header__has_feat(header, HEADER_TRACE_INFO)) {
> > - Â Â Â Â Â Â struct perf_file_section *trace_sec;
> > -
> > - Â Â Â Â Â Â trace_sec = &feat_sec[idx++];
> > + Â Â err = do_write_feat(fd, header, HEADER_TRACE_INFO, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â goto out_free;
> >
> > - Â Â Â Â Â Â /* Write trace info */
> > - Â Â Â Â Â Â trace_sec->offset = lseek(fd, 0, SEEK_CUR);
> > - Â Â Â Â Â Â read_tracing_data(fd, &evlist->entries);
> > - Â Â Â Â Â Â trace_sec->size = lseek(fd, 0, SEEK_CUR) - trace_sec->offset;
> > + Â Â err = do_write_feat(fd, header, HEADER_BUILD_ID, &p, evlist);
> > + Â Â if (err) {
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_BUILD_ID);
> > + Â Â Â Â Â Â goto out_free;
> > Â Â Â }
> >
> > - Â Â if (perf_header__has_feat(header, HEADER_BUILD_ID)) {
> > - Â Â Â Â Â Â struct perf_file_section *buildid_sec;
> > + Â Â err = do_write_feat(fd, header, HEADER_HOSTNAME, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_HOSTNAME);
> >
> > - Â Â Â Â Â Â buildid_sec = &feat_sec[idx++];
> > + Â Â err = do_write_feat(fd, header, HEADER_OSRELEASE, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_OSRELEASE);
> >
> > - Â Â Â Â Â Â /* Write build-ids */
> > - Â Â Â Â Â Â buildid_sec->offset = lseek(fd, 0, SEEK_CUR);
> > - Â Â Â Â Â Â err = dsos__write_buildid_table(header, fd);
> > - Â Â Â Â Â Â if (err < 0) {
> > - Â Â Â Â Â Â Â Â Â Â pr_debug("failed to write buildid table\n");
> > - Â Â Â Â Â Â Â Â Â Â goto out_free;
> > - Â Â Â Â Â Â }
> > - Â Â Â Â Â Â buildid_sec->size = lseek(fd, 0, SEEK_CUR) -
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â buildid_sec->offset;
> > - Â Â Â Â Â Â if (!no_buildid_cache)
> > - Â Â Â Â Â Â Â Â Â Â perf_session__cache_build_ids(session);
> > - Â Â }
> > + Â Â err = do_write_feat(fd, header, HEADER_ARCH, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_ARCH);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_CPUID, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_CPUID);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_NRCPUS, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_NRCPUS);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_EVENT_DESC, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_EVENT_DESC);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_CMDLINE, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_CMDLINE);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_VERSION, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_VERSION);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_CPU_TOPOLOGY, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_CPU_TOPOLOGY);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_TOTAL_MEM, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_TOTAL_MEM);
> > +
> > + Â Â err = do_write_feat(fd, header, HEADER_NUMA_TOPOLOGY, &p, evlist);
> > + Â Â if (err)
> > + Â Â Â Â Â Â perf_header__clear_feat(header, HEADER_NUMA_TOPOLOGY);
> >
> > Â Â Â lseek(fd, sec_start, SEEK_SET);
> > + Â Â /*
> > + Â Â Â* may write more than needed due to dropped feature, but
> > + Â Â Â* this is okay, reader will skip the mising entries
> > + Â Â Â*/
> > Â Â Â err = do_write(fd, feat_sec, sec_size);
> > Â Â Â if (err < 0)
> > Â Â Â Â Â Â Â pr_debug("failed to write feature section\n");
> > @@ -554,9 +1454,10 @@ static int perf_header__getbuffer64(struct perf_header *header,
> > Â}
> >
> > Âint perf_header__process_sections(struct perf_header *header, int fd,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â void *data,
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int (*process)(struct perf_file_section *section,
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âstruct perf_header *ph,
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âint feat, int fd))
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â struct perf_header *ph,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int feat, int fd, void *data))
> > Â{
> > Â Â Â struct perf_file_section *feat_sec;
> > Â Â Â int nr_sections;
> > @@ -584,7 +1485,7 @@ int perf_header__process_sections(struct perf_header *header, int fd,
> > Â Â Â Â Â Â Â if (perf_header__has_feat(header, feat)) {
> > Â Â Â Â Â Â Â Â Â Â Â struct perf_file_section *sec = &feat_sec[idx++];
> >
> > - Â Â Â Â Â Â Â Â Â Â err = process(sec, header, feat, fd);
> > + Â Â Â Â Â Â Â Â Â Â err = process(sec, header, feat, fd, data);
> > Â Â Â Â Â Â Â Â Â Â Â if (err < 0)
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â break;
> > Â Â Â Â Â Â Â }
> > @@ -796,7 +1697,7 @@ out:
> >
> > Âstatic int perf_file_section__process(struct perf_file_section *section,
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â struct perf_header *ph,
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int feat, int fd)
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int feat, int fd, void *data __used)
> > Â{
> > Â Â Â if (lseek(fd, section->offset, SEEK_SET) == (off_t)-1) {
> > Â Â Â Â Â Â Â pr_debug("Failed to lseek to %" PRIu64 " offset for feature "
> > @@ -935,7 +1836,8 @@ int perf_session__read_header(struct perf_session *session, int fd)
> > Â Â Â Â Â Â Â event_count = Âf_header.event_types.size / sizeof(struct perf_trace_event_type);
> > Â Â Â }
> >
> > - Â Â perf_header__process_sections(header, fd, perf_file_section__process);
> > + Â Â perf_header__process_sections(header, fd, NULL,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â perf_file_section__process);
> >
> > Â Â Â lseek(fd, header->data_offset, SEEK_SET);
> >
> > diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h
> > index 1886256..455480b 100644
> > --- a/tools/perf/util/header.h
> > +++ b/tools/perf/util/header.h
> > @@ -12,6 +12,19 @@
> > Âenum {
> > Â Â Â HEADER_TRACE_INFO = 1,
> > Â Â Â HEADER_BUILD_ID,
> > +
> > + Â Â HEADER_HOSTNAME,
> > + Â Â HEADER_OSRELEASE,
> > + Â Â HEADER_ARCH,
> > + Â Â HEADER_CPUID,
> > + Â Â HEADER_NRCPUS,
> > + Â Â HEADER_EVENT_DESC,
> > + Â Â HEADER_CMDLINE,
> > + Â Â HEADER_VERSION,
> > + Â Â HEADER_CPU_TOPOLOGY,
> > + Â Â HEADER_TOTAL_MEM,
> > + Â Â HEADER_NUMA_TOPOLOGY,
> > +
> > Â Â Â HEADER_LAST_FEATURE,
> > Â};
> >
> > @@ -68,10 +81,15 @@ void perf_header__set_feat(struct perf_header *header, int feat);
> > Âvoid perf_header__clear_feat(struct perf_header *header, int feat);
> > Âbool perf_header__has_feat(const struct perf_header *header, int feat);
> >
> > +int perf_header__set_cmdline(int argc, const char **argv);
> > +
> > Âint perf_header__process_sections(struct perf_header *header, int fd,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â void *data,
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int (*process)(struct perf_file_section *section,
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âstruct perf_header *ph,
> > - Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Âint feat, int fd));
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â struct perf_header *ph,
> > + Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â int feat, int fd, void *data));
> > +
> > +int perf_header__fprintf_info(struct perf_session *session, FILE *fp);
> >
> > Âint build_id_cache__add_s(const char *sbuild_id, const char *debugdir,
> > Â Â Â Â Â Â Â Â Â Â Â Â const char *name, bool is_kallsyms);
> > diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> > index 72458d9..54613a2 100644
> > --- a/tools/perf/util/session.c
> > +++ b/tools/perf/util/session.c
> > @@ -1326,3 +1326,21 @@ int perf_session__cpu_bitmap(struct perf_session *session,
> >
> > Â Â Â return 0;
> > Â}
> > +
> > +void perf_session__fprintf_info(struct perf_session *session, FILE *fp)
> > +{
> > + Â Â struct stat st;
> > + Â Â int ret;
> > +
> > + Â Â if (session == NULL || fp == NULL)
> > + Â Â Â Â Â Â return;
> > +
> > + Â Â ret = fstat(session->fd, &st);
> > + Â Â if (ret == -1)
> > + Â Â Â Â Â Â return;
> > +
> > + Â Â fprintf(fp, "#\n# captured on: %s", ctime(&st.st_ctime));
> > + Â Â ret = perf_header__fprintf_info(session, fp);
> > + Â Â if (ret == 0)
> > + Â Â Â Â Â Â fprintf(fp, "#\n");
> > +}
> > diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
> > index 170601e..33a3a46 100644
> > --- a/tools/perf/util/session.h
> > +++ b/tools/perf/util/session.h
> > @@ -176,4 +176,5 @@ void perf_session__print_ip(union perf_event *event,
> > Âint perf_session__cpu_bitmap(struct perf_session *session,
> > Â Â Â Â Â Â Â Â Â Â Â Â Â Âconst char *cpu_list, unsigned long *cpu_bitmap);
> >
> > +void perf_session__fprintf_info(struct perf_session *session, FILE *fp);
> > Â#endif /* __PERF_SESSION_H */
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/