perf: report should only load text symbols from kallsyms
PATCH 2/2 perf: report should only load text symbols from kallsyms
Quoted 1 line
From
7508bc84d541ac245be633803d393244b6860af6Mon Sep 17 00:00:00 2001
From: Arnaldo Carvalho de Melo <[email protected]>
Date: Tue, 26 May 2009 19:14:47 -0300
Subject: [PATCH] perf: report should only load text symbols from kallsyms
Just like we do for userspace when reading the symtab, reducing the number of entries we insert on the symbols rbtree.
Before:
[acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report > /dev/null
Performance counter stats for 'perf':
218.138382 task clock ticks (msecs)
4 context switches (events)
8 CPU migrations (events)
2136 pagefaults (events)
32746212 CPU cycles (events) (scaled from 67.04%)
11961102 instructions (events) (scaled from 66.19%)
49841 cache references (events) (scaled from 21.96%)
13777 cache misses (events) (scaled from 21.98%)Wall-clock time elapsed: 218.702477 msecs
[acme@emilia ~]$ perf report -i perf_report.perf | head
11.06 perf [.] 0x00000000000057cb /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 9.15 perf [.] 0x00000000000056a0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 8.72 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 8.51 perf [.] 0x0000000000006672 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 3.83 perf [k] 0xffffffff811cfc5a vsnprintf 3.40 perf [.] 0x0000000000005e33 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 3.40 perf [.] 0x0000000000005ec7 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 3.19 perf [k] 0xffffffff811ce1c1 number 2.77 perf [.] 0x0000000000006869 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 2.77 perf [.] 0x000000000000fde3 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: rb_insert_color
[acme@emilia ~]$
After:
acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report > /dev/null
Performance counter stats for 'perf':
190.228511 task clock ticks (msecs)
4 context switches (events)
7 CPU migrations (events)
1625 pagefaults (events)
29578745 CPU cycles (events) (scaled from 66.92%)
10516914 instructions (events) (scaled from 66.47%)
44015 cache references (events) (scaled from 22.04%)
8248 cache misses (events) (scaled from 22.07%)Wall-clock time elapsed: 190.816096 msecs
[acme@emilia ~]$ perf report -i perf_report.perf | head
15.99 perf [.] 0x00000000000057a9 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 10.87 perf [.] 0x000000000000674d /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 8.74 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 5.54 perf [.] 0x0000000000005e42 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 4.48 perf [.] 0x0000000000005ebe /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 4.48 perf [k] 0xffffffff811cfba0 vsnprintf 3.84 perf [.] 0x00000000000056b4 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 3.62 perf [.] 0x00000000000068d0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 3.20 perf [k] 0xffffffff811ce0b3 number 2.56 perf [.] 0x0000000000006d78 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: __cmd_report
[acme@emilia ~]$
--- Documentation/perf_counter/builtin-report.c | 14 +++++++++++--- 1 files changed, 11 insertions(+), 3 deletions(-)
Documentation/perf_counter/builtin-report.c +11 -3
@@ -3,6 +3,7 @@#include <libelf.h> #include <gelf.h> #include <elf.h> +#include <ctype.h> #include "util/list.h" #include "util/rbtree.h"
@@ -408,13 +409,20 @@ static int load_kallsyms(void)int len =hex2long (line, &start); - len +=3 ;/* ' t ' */ -if (len >= line_len) + len++; +if (len +2 >= line_len) +continue ; + +char symbol_type = line[len]; +/* +* We're interested only in code ('T'ext) +*/ +if (toupper (symbol_type) !='T' )continue ;/* * Well fix up the end later, when we have all sorted. */ -struct symbol *sym =symbol__new (start,0xdead , line + len); +struct symbol *sym =symbol__new (start,0xdead , line + len +2 );if (sym ==NULL ) goto out_delete_dso;
Signature · 1 line
1.6.0.6
Quoted 53 lines · Arnaldo Carvalho de Melo
Quoted 1 line
From
7508bc84d541ac245be633803d393244b6860af6Mon Sep 17 00:00:00 2001From: Arnaldo Carvalho de Melo <[email protected]>
Date: Tue, 26 May 2009 19:14:47 -0300
Subject: [PATCH] perf: report should only load text symbols from kallsymsJust like we do for userspace when reading the symtab, reducing the number of entries we insert on the symbols rbtree.
Before:
[acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report >
/dev/nullPerformance counter stats for 'perf':
218.138382 task clock ticks (msecs) 4 context switches (events) 8 CPU migrations (events) 2136 pagefaults (events) 32746212 CPU cycles (events) (scaled from 67.04%) 11961102 instructions (events) (scaled from 66.19%) 49841 cache references (events) (scaled from 21.96%) 13777 cache misses (events) (scaled from 21.98%)Wall-clock time elapsed: 218.702477 msecs
[acme@emilia ~]$ perf report -i perf_report.perf | head
11.06 perf [.] 0x00000000000057cb /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 9.15 perf [.] 0x00000000000056a0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 8.72 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 8.51 perf [.] 0x0000000000006672 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 3.83 perf [k] 0xffffffff811cfc5a vsnprintf 3.40 perf [.] 0x0000000000005e33 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 3.40 perf [.] 0x0000000000005ec7 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 3.19 perf [k] 0xffffffff811ce1c1 number 2.77 perf [.] 0x0000000000006869 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 2.77 perf [.] 0x000000000000fde3 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: rb_insert_color[acme@emilia ~]$
After:
acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report >
/dev/nullPerformance counter stats for 'perf':
190.228511 task clock ticks (msecs) 4 context switches (events) 7 CPU migrations (events) 1625 pagefaults (events) 29578745 CPU cycles (events) (scaled from 66.92%) 10516914 instructions (events) (scaled from 66.47%) 44015 cache references (events) (scaled from 22.04%) 8248 cache misses (events) (scaled from 22.07%)
Quoted 1 line
Wall-clock time elapsed: 190.816096 msecs
nice!
Quoted 36 lines
[acme@emilia ~]$ perf report -i perf_report.perf | head
15.99 perf [.] 0x00000000000057a9 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 10.87 perf [.] 0x000000000000674d /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 8.74 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 5.54 perf [.] 0x0000000000005e42 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 4.48 perf [.] 0x0000000000005ebe /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 4.48 perf [k] 0xffffffff811cfba0 vsnprintf 3.84 perf [.] 0x00000000000056b4 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 3.62 perf [.] 0x00000000000068d0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 3.20 perf [k] 0xffffffff811ce0b3 number 2.56 perf [.] 0x0000000000006d78 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: __cmd_report[acme@emilia ~]$
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>--- Documentation/perf_counter/builtin-report.c | 14 +++++++++++--- 1 files changed, 11 insertions(+), 3 deletions(-)Documentation/perf_counter/builtin-report.c +1
-2@@ -3,6 +3,7 @@#include <libelf.h> #include <gelf.h> #include <elf.h> +#include <ctype.h> #include "util/list.h" #include "util/rbtree.h" @@ -408,13 +409,20 @@ static int load_kallsyms(void)int len =hex2long (line, &start); - len +=3 ;/* ' t ' */ -if (len >= line_len)
Documentation/perf_counter/builtin-report.c static int load_kallsyms(void) · line 421
Quoted 10 lines
Documentation/perf_counter/builtin-report.c +9
-0+ len++; +if (len +2 >= line_len) +continue ; + +char symbol_type = line[len]; +/* +* We're interested only in code ('T'ext) +*/ +if (toupper (symbol_type) !='T' )continue ;
We would also include ' W ' symbols, right?
Ingo
[tip:perfcounters/core] perf_counter tools: Introduce stricter C code checking
In reply to Arnaldo Carvalho de MeloCommit-ID: 16f762a2ac5ecf8a11f6f0332e46cc3459220da5 Gitweb: http://git.kernel.org/tip/16f762a2ac5ecf8a11f6f0332e46cc3459220da5 Author: Ingo Molnar <[email protected]> AuthorDate: Wed, 27 May 2009 09:10:38 +0200 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 27 May 2009 08:10:35 +0200
perf_counter tools: Introduce stricter C code checking
Tighten up our C code requirements:
- disallow warnings
- disallow declarations-mixed-with-statements
- require proper prototypes
- require C99 (with gcc extensions)
Fix up a ton of problems these measures unearth:
- unused functions
- needlessly global functions
- missing prototypes
- code mixed with declarations
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Corey Ashford <[email protected]>
Cc: Marcelo Tosatti <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: John Kacur <[email protected]>
Cc: Steven Rostedt <[email protected]>
LKML-Reference: <[email protected]>
--- Documentation/perf_counter/Makefile | 2 +- Documentation/perf_counter/builtin-help.c | 2 +- Documentation/perf_counter/builtin-record.c | 38 ++++++----- Documentation/perf_counter/builtin-report.c | 103 +++++++++++++-------------- Documentation/perf_counter/builtin-stat.c | 3 +- Documentation/perf_counter/builtin-top.c | 2 +- Documentation/perf_counter/util/abspath.c | 2 +- Documentation/perf_counter/util/cache.h | 2 + Documentation/perf_counter/util/util.h | 2 + 9 files changed, 82 insertions(+), 74 deletions(-)
Documentation/perf_counter/Makefile +1 -1
@@ -159,7 +159,7 @@ uname_V := $(shell sh -c 'uname -v 2>/dev/null || echo not') # CFLAGS and LDFLAGS are for the users to override from the command line. -CFLAGS = -ggdb3 -Wall +CFLAGS = -ggdb3 -Wall -Werror -Wstrict-prototypes -Wmissing-declarations -Wmissing-prototypes -std=gnu99 -Wdeclaration-after-statement LDFLAGS = -lpthread -lrt -lelf ALL_CFLAGS = $(CFLAGS) ALL_LDFLAGS = $(LDFLAGS)
Documentation/perf_counter/builtin-help.c +1 -1
@@ -399,7 +399,7 @@ static void get_html_page_path(struct strbuf *page_path, const char *page)* HTML. */ #ifndef open_html -void open_html (const char *path) +static void open_html (const char *path) {execl_perf_cmd ("web--browse" ,"-c" ,"help.browser" , path,NULL ); }
Documentation/perf_counter/builtin-record.c +21 -17
@@ -1,6 +1,7 @@#include "perf.h" +#include "builtin.h" #include "util/util.h" #include "util/parse-options.h" #include "util/parse-events.h"
@@ -144,26 +145,32 @@ static int nr_poll;static int nr_cpu;struct mmap_event { -struct perf_event_header header; -__u32 pid, tid; -__u64 start; -__u64 len; -__u64 pgoff; -char filename[PATH_MAX ]; +struct perf_event_header header; +__u32 pid; +__u32 tid; +__u64 start; +__u64 len; +__u64 pgoff; +char filename[PATH_MAX ]; }; +struct comm_event { -struct perf_event_header header; -__u32 pid,tid; -char comm[16 ]; +struct perf_event_header header; +__u32 pid; +__u32 tid; +char comm[16 ]; };static pid_t pid_synthesize_comm_event (pid_t pid) { +struct comm_event comm_ev;char filename[PATH_MAX ]; +pid_t spid, ppid;char bf[BUFSIZ ]; -struct comm_event comm_ev; +int fd, nr, ret; +char comm[18 ];size_t size; -int fd; +char state;snprintf (filename,sizeof (filename),"/proc/%d/stat" , pid);
@@ -178,12 +185,8 @@ static pid_t pid_synthesize_comm_event(pid_t pid) }close (fd); -pid_t spid, ppid; -char state; -char comm[18 ]; -memset (&comm_ev,0 ,sizeof (comm_ev)); -int nr =sscanf (bf,"%d %s %c %d %d " , + nr =sscanf (bf,"%d %s %c %d %d " , &spid, comm, &state, &ppid, &comm_ev.pid);if (nr !=5 ) { fprintf(stderr, "couldn't get COMM and pgid, malformed %s\n",
@@ -198,7 +201,8 @@ static pid_t pid_synthesize_comm_event(pid_t pid)memcpy (comm_ev.comm, comm +1 , size); size =ALIGN (size,sizeof (uint64_t )); comm_ev.header.size =sizeof (comm_ev) - (sizeof (comm_ev.comm) - size); -int ret =write (output, &comm_ev, comm_ev.header.size); + + ret =write (output, &comm_ev, comm_ev.header.size);if (ret <0 ) {perror ("failed to write" );exit (-1 );
Documentation/perf_counter/builtin-report.c +51 -52
272 lines not on this page · raw message
Documentation/perf_counter/builtin-stat.c +2 -1
@@ -30,6 +30,7 @@*/ #include "perf.h" +#include "builtin.h" #include "util/util.h" #include "util/parse-options.h" #include "util/parse-events.h"
@@ -108,7 +109,7 @@ static void create_perfstat_counter(int counter) } } -int do_perfstat (int argc,const char **argv) +static int do_perfstat (int argc,const char **argv) {unsigned long long t0, t1;int counter;
Documentation/perf_counter/builtin-top.c +1 -1
@@ -42,8 +42,8 @@* Released under the GPL v2. (and only v2, not any later version) */ -#include "perf.h" +#include "builtin.h" #include "util/util.h" #include "util/util.h" #include "util/parse-options.h"
Documentation/perf_counter/util/abspath.c +1 -1
@@ -5,7 +5,7 @@* symlink to a directory, we do not want to say it is a directory when * dealing with tracked content in the working tree. */ -int is_directory (const char *path) +static int is_directory (const char *path) {struct stat st;return (!stat (path, &st) &&S_ISDIR (st.st_mode));
Documentation/perf_counter/util/cache.h +2 -0
@@ -104,6 +104,8 @@ char *strip_path_suffix(const char *path, const char *suffix);extern char *mkpath (const char *fmt, ...) __attribute__((format (printf,1 ,2 )));extern char *perf_path (const char *fmt, ...) __attribute__((format (printf,1 ,2 ))); +/* perf_mkstemp() - create tmp file honoring TMPDIR variable */ +extern int perf_mkstemp (char *path,size_t len,const char *template);extern char *mksnpath (char *buf,size_t n,const char *fmt, ...) __attribute__((format (printf,3 ,4 )));
Documentation/perf_counter/util/util.h +2 -0
@@ -309,6 +309,8 @@ extern ssize_t xread(int fd, void *buf, size_t len);extern ssize_t xwrite (int fd,const void *buf,size_t len);extern int xdup (int fd);extern FILE *xfdopen (int fd,const char *mode); +extern int xmkstemp (char *template); +static inline size_t xsize_t (off_t len) {return (size_t )len;
[tip:perfcounters/core] perf report: Only load text symbols from kallsyms
In reply to Arnaldo Carvalho de MeloCommit-ID: 03f6316d32738ec5eda2e6f628c12d1c01e61a87 Gitweb: http://git.kernel.org/tip/03f6316d32738ec5eda2e6f628c12d1c01e61a87 Author: Arnaldo Carvalho de Melo <[email protected]> AuthorDate: Tue, 26 May 2009 19:21:55 -0300 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 27 May 2009 09:10:36 +0200
perf report: Only load text symbols from kallsyms
Just like we do for userspace when reading the symtab, reducing the number of entries we insert on the symbols rbtree.
Before:
[acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report > /dev/null
Performance counter stats for 'perf':
218.138382 task clock ticks (msecs)
4 context switches (events)
8 CPU migrations (events)
2136 pagefaults (events)
32746212 CPU cycles (events) (scaled from 67.04%)
11961102 instructions (events) (scaled from 66.19%)
49841 cache references (events) (scaled from 21.96%)
13777 cache misses (events) (scaled from 21.98%)Wall-clock time elapsed: 218.702477 msecs
[acme@emilia ~]$ perf report -i perf_report.perf | head
11.06 perf [.] 0x00000000000057cb /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 9.15 perf [.] 0x00000000000056a0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 8.72 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 8.51 perf [.] 0x0000000000006672 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 3.83 perf [k] 0xffffffff811cfc5a vsnprintf 3.40 perf [.] 0x0000000000005e33 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 3.40 perf [.] 0x0000000000005ec7 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 3.19 perf [k] 0xffffffff811ce1c1 number 2.77 perf [.] 0x0000000000006869 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 2.77 perf [.] 0x000000000000fde3 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: rb_insert_color
[acme@emilia ~]$
After:
acme@emilia ~]$ rm -f perf_report.perf ; perf record -o perf_report.perf perf stat perf report > /dev/null
Performance counter stats for 'perf':
190.228511 task clock ticks (msecs)
4 context switches (events)
7 CPU migrations (events)
1625 pagefaults (events)
29578745 CPU cycles (events) (scaled from 66.92%)
10516914 instructions (events) (scaled from 66.47%)
44015 cache references (events) (scaled from 22.04%)
8248 cache misses (events) (scaled from 22.07%)Wall-clock time elapsed: 190.816096 msecs
[acme@emilia ~]$ perf report -i perf_report.perf | head
15.99 perf [.] 0x00000000000057a9 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__find_symbol 10.87 perf [.] 0x000000000000674d /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: thread__symbol_incnew 8.74 perf [k] 0xffffffff8101b1d2 intel_pmu_enable_all 5.54 perf [.] 0x0000000000005e42 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex 4.48 perf [.] 0x0000000000005ebe /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: hex2long 4.48 perf [k] 0xffffffff811cfba0 vsnprintf 3.84 perf [.] 0x00000000000056b4 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: dso__insert_symbol 3.62 perf [.] 0x00000000000068d0 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: threads__findnew 3.20 perf [k] 0xffffffff811ce0b3 number 2.56 perf [.] 0x0000000000006d78 /home/acme/git/linux-2.6-tip/Documentation/perf_counter/perf: __cmd_report
[acme@emilia ~]$
[ Impact: optimization ]
LKML-Reference: <[email protected]>
--- Documentation/perf_counter/builtin-report.c | 14 +++++++++++--- 1 files changed, 11 insertions(+), 3 deletions(-)
Documentation/perf_counter/builtin-report.c +11 -3
@@ -3,6 +3,7 @@#include <libelf.h> #include <gelf.h> #include <elf.h> +#include <ctype.h> #include "util/list.h" #include "util/rbtree.h"
@@ -408,13 +409,20 @@ static int load_kallsyms(void)int len =hex2long (line, &start); - len +=3 ;/* ' t ' */ -if (len >= line_len) + len++; +if (len +2 >= line_len) +continue ; + +char symbol_type = line[len]; +/* +* We're interested only in code ('T'ext) +*/ +if (toupper (symbol_type) !='T' )continue ;/* * Well fix up the end later, when we have all sorted. */ -struct symbol *sym =symbol__new (start,0xdead , line + len); +struct symbol *sym =symbol__new (start,0xdead , line + len +2 );if (sym ==NULL ) goto out_delete_dso;
[tip:perfcounters/core] perf report: Only load text symbols from kallsyms, fix
In reply to Arnaldo Carvalho de MeloCommit-ID: af83632f98aefd1ae4d8ca3c7c285ccf6a7d3956 Gitweb: http://git.kernel.org/tip/af83632f98aefd1ae4d8ca3c7c285ccf6a7d3956 Author: Ingo Molnar <[email protected]> AuthorDate: Wed, 27 May 2009 08:38:48 +0200 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 27 May 2009 09:10:37 +0200
perf report: Only load text symbols from kallsyms, fix
- allow 'W' symbols too
- Convert initializations to C99 style
- whitespace cleanups
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Corey Ashford <[email protected]>
Cc: Marcelo Tosatti <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: John Kacur <[email protected]>
Cc: Steven Rostedt <[email protected]>
LKML-Reference: <[email protected]>
--- Documentation/perf_counter/builtin-report.c | 30 ++++++++++++++++---------- 1 files changed, 18 insertions(+), 12 deletions(-)
Documentation/perf_counter/builtin-report.c +18 -12
@@ -384,21 +384,26 @@ int hex2long(char *ptr, unsigned long *long_val)static int load_kallsyms (void ) { +struct rb_node *nd, *prevnd; +char *line =NULL ; +FILE *file; +size_t n; + kernel_dso =dso__new ("[kernel]" );if (kernel_dso ==NULL )return -1 ; -FILE *file =fopen ("/proc/kallsyms" ,"r" ); - + file =fopen ("/proc/kallsyms" ,"r" );if (file ==NULL ) goto out_delete_dso; -char *line =NULL ; -size_t n; -while (!feof (file)) {unsigned long start; -int line_len =getline (&line, &n, file); +struct symbol *sym; +int line_len, len; +char symbol_type; + + line_len =getline (&line, &n, file);if (line_len <0 )break ;
@@ -407,22 +412,22 @@ static int load_kallsyms(void) line[--line_len] ='\0' ;/* \n */ -int len =hex2long (line, &start); - + len =hex2long (line, &start); + len++;if (len +2 >= line_len)continue ; -char symbol_type = line[len]; + symbol_type =toupper (line[len]);/* * We're interested only in code ('T'ext) */ -if (toupper (symbol_type) !='T' ) +if (symbol_type !='T' && symbol_type !='W' )continue ;/* * Well fix up the end later, when we have all sorted. */ -struct symbol *sym =symbol__new (start,0xdead , line + len +2 ); + sym =symbol__new (start,0xdead , line + len +2 );if (sym ==NULL ) goto out_delete_dso;
@@ -434,7 +439,7 @@ static int load_kallsyms(void)* Now that we have all sorted out, just set the ->end of all * symbols */ -struct rb_node *nd, *prevnd =rb_first (&kernel_dso->syms); + prevnd =rb_first (&kernel_dso->syms);if (prevnd ==NULL ) goto out_delete_line;
@@ -450,6 +455,7 @@ static int load_kallsyms(void)dsos__add (kernel_dso);free (line);fclose (file); +return 0 ; out_delete_line:
Re: [tip:perfcounters/core] perf_counter tools: Introduce stricter C code checking
In reply to tip-bot for Ingo MolnarQuoted 5 lines · tip-bot for Ingo Molnar
perf_countertools: Introduce stricter C code checkingTighten up our C code requirements:
- disallow warnings
This causes failures when I compile it as a 64-bit executable on powerpc:
CC builtin-record.obuiltin-record.c: In function 'pid_synthesize_mmap_events':
builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 3 has type '__u64 *' builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 4 has type '__u64 *' builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 9 has type '__u64 *'
This is because u64 is an unsigned long in userspace for a 64-bit build, not unsigned long long. I'm not sure how best to solve this problem.
If I compile it as a 32-bit executable, it doesn't generate warnings, but when I try to run "perf top" (this is on a 64-bit kernel, of course, since 32-bit powerpc kernels don't currently support perf_counters), I get:
# perf top left: 0000000000000000 ip: 00000000000891a4 right: 00000000ffffffff KernelTop refresh period: 2 seconds perf: builtin-top.c:453: record_ip: Assertion `left <= ip && ip <= right' failed. Aborted
Paul.
Re: [tip:perfcounters/core] perf_counter tools: Introduce stricter C code checking
Quoted 17 lines · Paul Mackerras
Quoted 5 lines · tip-bot for Ingo Molnar
perf_countertools: Introduce stricter C code checkingTighten up our C code requirements:
- disallow warnings
This causes failures when I compile it as a 64-bit executable on powerpc:
CC builtin-record.o
builtin-record.c: In function 'pid_synthesize_mmap_events':
builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 3 has type '__u64*' builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 4 has type '__u64*' builtin-record.c:241: warning: format '%llx' expects type 'long long unsigned int *', but argument 9 has type '__u64*'
Quoted 3 lines
This is because u64 is an unsigned long in userspace for a 64-bit build, not unsigned long long. I'm not sure how best to solve this problem.
We could perhaps use __u64 consistently? (can we?)
Quoted 5 lines
If I compile it as a 32-bit executable, it doesn't generate warnings, but when I try to run "perf top" (this is on a 64-bit kernel, of course, since 32-bit powerpc kernels don't currently support
perf_counters), I get:
Quoted 7 lines
# perf top left: 0000000000000000 ip: 00000000000891a4 right: 00000000ffffffff KernelTop refresh period: 2 seconds perf: builtin-top.c:453: record_ip: Assertion `left <= ip && ip <= right' failed. Aborted
mind trying a 'git bisect run' session - which commit broke things for you? Or is this related to the type problems?
Ingo