linux/tools/perf/util/symbol.c

3197 lines
71 KiB
C
Raw Normal View History

#include <dirent.h>
#include <errno.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/param.h>
#include <fcntl.h>
#include <unistd.h>
#include <inttypes.h>
perf annotate: Use build-ids to find the right DSO We were still using the pathname found on the MMAP event, that could not be the one we used when recording, so use the build-id cache for that, only falling back to use the pathname in the MMAP event if no build-ids are available. With this we now also are able to do secure, seamless offline annotation. Example: [root@doppio linux-2.6-tip]# perf report -g none -v 2> /dev/null | head -10 8.12% Xorg /usr/lib64/libpixman-1.so.0.14.0 0x0000000000026d02 B [.] pixman_rasterize_edges 4.68% firefox /usr/lib64/xulrunner-1.9.1/libxul.so 0x00000000005dbdba B [.] 0x000000005dbdba 3.70% swapper /lib/modules/2.6.34-rc6/build/vmlinux 0xffffffff81022cea ! [k] read_hpet 2.96% init /lib/modules/2.6.34-rc6/build/vmlinux 0xffffffff81022cea ! [k] read_hpet 2.73% swapper /lib/modules/2.6.34-rc6/build/vmlinux 0xffffffff8100a738 ! [k] mwait_idle_with_hints [root@doppio linux-2.6-tip]# perf annotate -v pixman_rasterize_edges 2>&1 | grep Executing Executing: objdump --start-address=0x000000371ce26670 --stop-address=0x000000371ce2709f -dS /root/.debug/.build-id/bd/6ac5199137aaeb279f864717d8d061477466c1|grep -v /root/.debug/.build-id/bd/6ac5199137aaeb279f864717d8d061477466c1|expand [root@doppio linux-2.6-tip]# perf buildid-list | grep libpixman-1.so.0.14.0 bd6ac5199137aaeb279f864717d8d061477466c1 /usr/lib64/libpixman-1.so.0.14.0 [root@doppio linux-2.6-tip]# Reported-by: Stephane Eranian <eranian@google.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-20 15:15:33 +00:00
#include "build-id.h"
#include "util.h"
#include "debug.h"
#include "symbol.h"
#include "strlist.h"
#include <libelf.h>
#include <gelf.h>
#include <elf.h>
#include <limits.h>
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
#include <sys/utsname.h>
#ifndef KSYM_NAME_LEN
#define KSYM_NAME_LEN 256
#endif
#ifndef NT_GNU_BUILD_ID
#define NT_GNU_BUILD_ID 3
#endif
static void dso_cache__free(struct rb_root *root);
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id);
static int elf_read_build_id(Elf *elf, void *bf, size_t size);
static void dsos__add(struct list_head *head, struct dso *dso);
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type);
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
symbol_filter_t filter);
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
symbol_filter_t filter);
static int vmlinux_path__nr_entries;
static char **vmlinux_path;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
struct symbol_conf symbol_conf = {
.exclude_other = true,
.use_modules = true,
.try_vmlinux_path = true,
.annotate_src = true,
.symfs = "",
};
static enum dso_binary_type binary_type_symtab[] = {
DSO_BINARY_TYPE__KALLSYMS,
DSO_BINARY_TYPE__GUEST_KALLSYMS,
DSO_BINARY_TYPE__JAVA_JIT,
DSO_BINARY_TYPE__DEBUGLINK,
DSO_BINARY_TYPE__BUILD_ID_CACHE,
DSO_BINARY_TYPE__FEDORA_DEBUGINFO,
DSO_BINARY_TYPE__UBUNTU_DEBUGINFO,
DSO_BINARY_TYPE__BUILDID_DEBUGINFO,
DSO_BINARY_TYPE__SYSTEM_PATH_DSO,
DSO_BINARY_TYPE__GUEST_KMODULE,
DSO_BINARY_TYPE__SYSTEM_PATH_KMODULE,
DSO_BINARY_TYPE__NOT_FOUND,
};
#define DSO_BINARY_TYPE__SYMTAB_CNT ARRAY_SIZE(binary_type_symtab)
static enum dso_binary_type binary_type_data[] = {
DSO_BINARY_TYPE__BUILD_ID_CACHE,
DSO_BINARY_TYPE__SYSTEM_PATH_DSO,
DSO_BINARY_TYPE__NOT_FOUND,
};
#define DSO_BINARY_TYPE__DATA_CNT ARRAY_SIZE(binary_type_data)
int dso__name_len(const struct dso *dso)
{
if (!dso)
return strlen("[unknown]");
if (verbose)
return dso->long_name_len;
return dso->short_name_len;
}
bool dso__loaded(const struct dso *dso, enum map_type type)
{
return dso->loaded & (1 << type);
}
bool dso__sorted_by_name(const struct dso *dso, enum map_type type)
{
return dso->sorted_by_name & (1 << type);
}
static void dso__set_sorted_by_name(struct dso *dso, enum map_type type)
{
dso->sorted_by_name |= (1 << type);
}
bool symbol_type__is_a(char symbol_type, enum map_type map_type)
{
symbol_type = toupper(symbol_type);
switch (map_type) {
case MAP__FUNCTION:
return symbol_type == 'T' || symbol_type == 'W';
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
case MAP__VARIABLE:
return symbol_type == 'D';
default:
return false;
}
}
static int prefix_underscores_count(const char *str)
{
const char *tail = str;
while (*tail == '_')
tail++;
return tail - str;
}
#define SYMBOL_A 0
#define SYMBOL_B 1
static int choose_best_symbol(struct symbol *syma, struct symbol *symb)
{
s64 a;
s64 b;
/* Prefer a symbol with non zero length */
a = syma->end - syma->start;
b = symb->end - symb->start;
if ((b == 0) && (a > 0))
return SYMBOL_A;
else if ((a == 0) && (b > 0))
return SYMBOL_B;
/* Prefer a non weak symbol over a weak one */
a = syma->binding == STB_WEAK;
b = symb->binding == STB_WEAK;
if (b && !a)
return SYMBOL_A;
if (a && !b)
return SYMBOL_B;
/* Prefer a global symbol over a non global one */
a = syma->binding == STB_GLOBAL;
b = symb->binding == STB_GLOBAL;
if (a && !b)
return SYMBOL_A;
if (b && !a)
return SYMBOL_B;
/* Prefer a symbol with less underscores */
a = prefix_underscores_count(syma->name);
b = prefix_underscores_count(symb->name);
if (b > a)
return SYMBOL_A;
else if (a > b)
return SYMBOL_B;
/* If all else fails, choose the symbol with the longest name */
if (strlen(syma->name) >= strlen(symb->name))
return SYMBOL_A;
else
return SYMBOL_B;
}
static void symbols__fixup_duplicate(struct rb_root *symbols)
{
struct rb_node *nd;
struct symbol *curr, *next;
nd = rb_first(symbols);
while (nd) {
curr = rb_entry(nd, struct symbol, rb_node);
again:
nd = rb_next(&curr->rb_node);
next = rb_entry(nd, struct symbol, rb_node);
if (!nd)
break;
if (curr->start != next->start)
continue;
if (choose_best_symbol(curr, next) == SYMBOL_A) {
rb_erase(&next->rb_node, symbols);
goto again;
} else {
nd = rb_next(&curr->rb_node);
rb_erase(&curr->rb_node, symbols);
}
}
}
static void symbols__fixup_end(struct rb_root *symbols)
{
struct rb_node *nd, *prevnd = rb_first(symbols);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
struct symbol *curr, *prev;
if (prevnd == NULL)
return;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
curr = rb_entry(prevnd, struct symbol, rb_node);
for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
prev = curr;
curr = rb_entry(nd, struct symbol, rb_node);
if (prev->end == prev->start && prev->end != curr->start)
prev->end = curr->start - 1;
}
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
/* Last entry */
if (curr->end == curr->start)
curr->end = roundup(curr->start, 4096);
}
static void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
{
struct map *prev, *curr;
struct rb_node *nd, *prevnd = rb_first(&mg->maps[type]);
if (prevnd == NULL)
return;
curr = rb_entry(prevnd, struct map, rb_node);
for (nd = rb_next(prevnd); nd; nd = rb_next(nd)) {
prev = curr;
curr = rb_entry(nd, struct map, rb_node);
prev->end = curr->start - 1;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
}
/*
* We still haven't the actual symbols, so guess the
* last map final address.
*/
curr->end = ~0ULL;
}
static void map_groups__fixup_end(struct map_groups *mg)
{
int i;
for (i = 0; i < MAP__NR_TYPES; ++i)
__map_groups__fixup_end(mg, i);
}
static struct symbol *symbol__new(u64 start, u64 len, u8 binding,
const char *name)
{
size_t namelen = strlen(name) + 1;
struct symbol *sym = calloc(1, (symbol_conf.priv_size +
sizeof(*sym) + namelen));
if (sym == NULL)
perf_counter tools: Add 'perf annotate' feature Add new perf sub-command to display annotated source code: $ perf annotate decode_tree_entry ------------------------------------------------ Percent | Source code & Disassembly of /home/mingo/git/git ------------------------------------------------ : : /home/mingo/git/git: file format elf64-x86-64 : : : Disassembly of section .text: : : 00000000004a0da0 <decode_tree_entry>: : *modep = mode; : return str; : } : : static void decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size) : { 3.82 : 4a0da0: 41 54 push %r12 : const char *path; : unsigned int mode, len; : : if (size < 24 || buf[size - 21]) 0.17 : 4a0da2: 48 83 fa 17 cmp $0x17,%rdx : *modep = mode; : return str; : } : : static void decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size) : { 0.00 : 4a0da6: 49 89 fc mov %rdi,%r12 0.00 : 4a0da9: 55 push %rbp 3.37 : 4a0daa: 53 push %rbx : const char *path; : unsigned int mode, len; : : if (size < 24 || buf[size - 21]) 0.08 : 4a0dab: 76 73 jbe 4a0e20 <decode_tree_entry+0x80> 0.00 : 4a0dad: 80 7c 16 eb 00 cmpb $0x0,-0x15(%rsi,%rdx,1) 3.48 : 4a0db2: 75 6c jne 4a0e20 <decode_tree_entry+0x80> : static const char *get_mode(const char *str, unsigned int *modep) : { : unsigned char c; : unsigned int mode = 0; : : if (*str == ' ') 1.94 : 4a0db4: 0f b6 06 movzbl (%rsi),%eax 0.39 : 4a0db7: 3c 20 cmp $0x20,%al 0.00 : 4a0db9: 74 65 je 4a0e20 <decode_tree_entry+0x80> : return NULL; : : while ((c = *str++) != ' ') { 0.06 : 4a0dbb: 89 c2 mov %eax,%edx : if (c < '0' || c > '7') 1.99 : 4a0dbd: 31 ed xor %ebp,%ebp : unsigned int mode = 0; : : if (*str == ' ') : return NULL; : : while ((c = *str++) != ' ') { 1.74 : 4a0dbf: 48 8d 5e 01 lea 0x1(%rsi),%rbx : if (c < '0' || c > '7') 0.00 : 4a0dc3: 8d 42 d0 lea -0x30(%rdx),%eax 0.17 : 4a0dc6: 3c 07 cmp $0x7,%al 0.00 : 4a0dc8: 76 0d jbe 4a0dd7 <decode_tree_entry+0x37> 0.00 : 4a0dca: eb 54 jmp 4a0e20 <decode_tree_entry+0x80> 0.00 : 4a0dcc: 0f 1f 40 00 nopl 0x0(%rax) 16.57 : 4a0dd0: 8d 42 d0 lea -0x30(%rdx),%eax 0.14 : 4a0dd3: 3c 07 cmp $0x7,%al 0.00 : 4a0dd5: 77 49 ja 4a0e20 <decode_tree_entry+0x80> : return NULL; : mode = (mode << 3) + (c - '0'); 3.12 : 4a0dd7: 0f b6 c2 movzbl %dl,%eax : unsigned int mode = 0; : : if (*str == ' ') : return NULL; : : while ((c = *str++) != ' ') { 0.00 : 4a0dda: 0f b6 13 movzbl (%rbx),%edx 16.74 : 4a0ddd: 48 83 c3 01 add $0x1,%rbx : if (c < '0' || c > '7') : return NULL; : mode = (mode << 3) + (c - '0'); The first column is the percentage of samples that arrived on that particular line - relative to the total cost of the function. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-06 13:48:52 +00:00
return NULL;
if (symbol_conf.priv_size)
sym = ((void *)sym) + symbol_conf.priv_size;
sym->start = start;
sym->end = len ? start + len - 1 : start;
sym->binding = binding;
sym->namelen = namelen - 1;
pr_debug4("%s: %s %#" PRIx64 "-%#" PRIx64 "\n",
__func__, name, start, sym->end);
memcpy(sym->name, name, namelen);
return sym;
}
void symbol__delete(struct symbol *sym)
{
free(((void *)sym) - symbol_conf.priv_size);
}
static size_t symbol__fprintf(struct symbol *sym, FILE *fp)
{
return fprintf(fp, " %" PRIx64 "-%" PRIx64 " %c %s\n",
sym->start, sym->end,
sym->binding == STB_GLOBAL ? 'g' :
sym->binding == STB_LOCAL ? 'l' : 'w',
sym->name);
}
size_t symbol__fprintf_symname_offs(const struct symbol *sym,
const struct addr_location *al, FILE *fp)
{
unsigned long offset;
size_t length;
if (sym && sym->name) {
length = fprintf(fp, "%s", sym->name);
if (al) {
offset = al->addr - sym->start;
length += fprintf(fp, "+0x%lx", offset);
}
return length;
} else
return fprintf(fp, "[unknown]");
}
size_t symbol__fprintf_symname(const struct symbol *sym, FILE *fp)
{
return symbol__fprintf_symname_offs(sym, NULL, fp);
}
void dso__set_long_name(struct dso *dso, char *name)
{
if (name == NULL)
return;
dso->long_name = name;
dso->long_name_len = strlen(name);
}
static void dso__set_short_name(struct dso *dso, const char *name)
{
if (name == NULL)
return;
dso->short_name = name;
dso->short_name_len = strlen(name);
}
static void dso__set_basename(struct dso *dso)
{
dso__set_short_name(dso, basename(dso->long_name));
}
struct dso *dso__new(const char *name)
{
struct dso *dso = calloc(1, sizeof(*dso) + strlen(name) + 1);
if (dso != NULL) {
int i;
strcpy(dso->name, name);
dso__set_long_name(dso, dso->name);
dso__set_short_name(dso, dso->name);
for (i = 0; i < MAP__NR_TYPES; ++i)
dso->symbols[i] = dso->symbol_names[i] = RB_ROOT;
dso->cache = RB_ROOT;
dso->symtab_type = DSO_BINARY_TYPE__NOT_FOUND;
dso->data_type = DSO_BINARY_TYPE__NOT_FOUND;
dso->loaded = 0;
dso->sorted_by_name = 0;
dso->has_build_id = 0;
dso->kernel = DSO_TYPE_USER;
perf symbols: Handle different endians properly during symbol load Currently we dont care about the file object's endianness. It's possible we read buildid file object from different architecture than we are currentlly running on. So we need to care about properly reading such object's data - handle different endianness properly. Adding: needs_swap DSO field dso__swap_init function to initialize DSO's needs_swap DSO__SWAP to read the data with proper swaps 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 1) - 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> 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/1338380624-7443-2-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:42 +00:00
dso->needs_swap = DSO_SWAP__UNSET;
INIT_LIST_HEAD(&dso->node);
}
return dso;
}
static void symbols__delete(struct rb_root *symbols)
{
struct symbol *pos;
struct rb_node *next = rb_first(symbols);
while (next) {
pos = rb_entry(next, struct symbol, rb_node);
next = rb_next(&pos->rb_node);
rb_erase(&pos->rb_node, symbols);
symbol__delete(pos);
}
}
void dso__delete(struct dso *dso)
{
int i;
for (i = 0; i < MAP__NR_TYPES; ++i)
symbols__delete(&dso->symbols[i]);
if (dso->sname_alloc)
free((char *)dso->short_name);
if (dso->lname_alloc)
free(dso->long_name);
dso_cache__free(&dso->cache);
free(dso);
}
void dso__set_build_id(struct dso *dso, void *build_id)
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
{
memcpy(dso->build_id, build_id, sizeof(dso->build_id));
dso->has_build_id = 1;
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
}
static void symbols__insert(struct rb_root *symbols, struct symbol *sym)
{
struct rb_node **p = &symbols->rb_node;
struct rb_node *parent = NULL;
const u64 ip = sym->start;
struct symbol *s;
while (*p != NULL) {
parent = *p;
s = rb_entry(parent, struct symbol, rb_node);
if (ip < s->start)
p = &(*p)->rb_left;
else
p = &(*p)->rb_right;
}
rb_link_node(&sym->rb_node, parent, p);
rb_insert_color(&sym->rb_node, symbols);
}
static struct symbol *symbols__find(struct rb_root *symbols, u64 ip)
{
struct rb_node *n;
if (symbols == NULL)
return NULL;
n = symbols->rb_node;
while (n) {
struct symbol *s = rb_entry(n, struct symbol, rb_node);
if (ip < s->start)
n = n->rb_left;
else if (ip > s->end)
n = n->rb_right;
else
return s;
}
return NULL;
}
struct symbol_name_rb_node {
struct rb_node rb_node;
struct symbol sym;
};
static void symbols__insert_by_name(struct rb_root *symbols, struct symbol *sym)
{
struct rb_node **p = &symbols->rb_node;
struct rb_node *parent = NULL;
struct symbol_name_rb_node *symn, *s;
symn = container_of(sym, struct symbol_name_rb_node, sym);
while (*p != NULL) {
parent = *p;
s = rb_entry(parent, struct symbol_name_rb_node, rb_node);
if (strcmp(sym->name, s->sym.name) < 0)
p = &(*p)->rb_left;
else
p = &(*p)->rb_right;
}
rb_link_node(&symn->rb_node, parent, p);
rb_insert_color(&symn->rb_node, symbols);
}
static void symbols__sort_by_name(struct rb_root *symbols,
struct rb_root *source)
{
struct rb_node *nd;
for (nd = rb_first(source); nd; nd = rb_next(nd)) {
struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
symbols__insert_by_name(symbols, pos);
}
}
static struct symbol *symbols__find_by_name(struct rb_root *symbols,
const char *name)
{
struct rb_node *n;
if (symbols == NULL)
return NULL;
n = symbols->rb_node;
while (n) {
struct symbol_name_rb_node *s;
int cmp;
s = rb_entry(n, struct symbol_name_rb_node, rb_node);
cmp = strcmp(name, s->sym.name);
if (cmp < 0)
n = n->rb_left;
else if (cmp > 0)
n = n->rb_right;
else
return &s->sym;
}
return NULL;
}
struct symbol *dso__find_symbol(struct dso *dso,
enum map_type type, u64 addr)
{
return symbols__find(&dso->symbols[type], addr);
}
struct symbol *dso__find_symbol_by_name(struct dso *dso, enum map_type type,
const char *name)
{
return symbols__find_by_name(&dso->symbol_names[type], name);
}
void dso__sort_by_name(struct dso *dso, enum map_type type)
{
dso__set_sorted_by_name(dso, type);
return symbols__sort_by_name(&dso->symbol_names[type],
&dso->symbols[type]);
}
int build_id__sprintf(const u8 *build_id, int len, char *bf)
{
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
char *bid = bf;
const u8 *raw = build_id;
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
int i;
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
for (i = 0; i < len; ++i) {
sprintf(bid, "%02x", *raw);
++raw;
bid += 2;
}
return raw - build_id;
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
}
size_t dso__fprintf_buildid(struct dso *dso, FILE *fp)
perf symbols: Use the buildids if present With this change 'perf record' will intercept PERF_RECORD_MMAP calls, creating a linked list of DSOs, then when the session finishes, it will traverse this list and read the buildids, stashing them at the end of the file and will set up a new feature bit in the header bitmask. 'perf report' will then notice this feature and populate the 'dsos' list and set the build ids. When reading the symtabs it will refuse to load from a file that doesn't have the same build id. This improves the reliability of the profiler output, as symbols and profiling data is more guaranteed to match. Example: [root@doppio ~]# perf report | head /home/acme/bin/perf with build id b1ea544ac3746e7538972548a09aadecc5753868 not found, continuing without symbols # Samples: 2621434559 # # Overhead Command Shared Object Symbol # ........ ............... ............................. ...... # 7.91% init [kernel] [k] read_hpet 7.64% init [kernel] [k] mwait_idle_with_hints 7.60% swapper [kernel] [k] read_hpet 7.60% swapper [kernel] [k] mwait_idle_with_hints 3.65% init [kernel] [k] 0xffffffffa02339d9 [root@doppio ~]# In this case the 'perf' binary was an older one, vanished, so its symbols probably wouldn't match or would cause subtly different (and misleading) output. Next patches will support the kernel as well, reading the build id notes for it and the modules from /sys. Another patch should also introduce a new plumbing command: 'perf list-buildids' that will then be used in porcelain that is distro specific to fetch -debuginfo packages where such buildids are present. This will in turn allow for one to run 'perf record' in one machine and 'perf report' in another. Future work on having the buildid sent directly from the kernel in the PERF_RECORD_MMAP event is needed to close races, as the DSO can be changed during a 'perf record' session, but this patch at least helps with non-corner cases and current/older kernels. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Frank Ch. Eigler <fche@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Jim Keniston <jkenisto@us.ibm.com> Cc: K. Prasad <prasad@linux.vnet.ibm.com> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Roland McGrath <roland@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Steven Rostedt <rostedt@goodmis.org> LKML-Reference: <1257367843-26224-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-11-04 20:50:43 +00:00
{
char sbuild_id[BUILD_ID_SIZE * 2 + 1];
build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id);
return fprintf(fp, "%s", sbuild_id);
}
size_t dso__fprintf_symbols_by_name(struct dso *dso,
enum map_type type, FILE *fp)
{
size_t ret = 0;
struct rb_node *nd;
struct symbol_name_rb_node *pos;
for (nd = rb_first(&dso->symbol_names[type]); nd; nd = rb_next(nd)) {
pos = rb_entry(nd, struct symbol_name_rb_node, rb_node);
fprintf(fp, "%s\n", pos->sym.name);
}
return ret;
}
size_t dso__fprintf(struct dso *dso, enum map_type type, FILE *fp)
{
struct rb_node *nd;
size_t ret = fprintf(fp, "dso: %s (", dso->short_name);
if (dso->short_name != dso->long_name)
ret += fprintf(fp, "%s, ", dso->long_name);
ret += fprintf(fp, "%s, %sloaded, ", map_type__name[type],
dso->loaded ? "" : "NOT ");
ret += dso__fprintf_buildid(dso, fp);
ret += fprintf(fp, ")\n");
for (nd = rb_first(&dso->symbols[type]); nd; nd = rb_next(nd)) {
struct symbol *pos = rb_entry(nd, struct symbol, rb_node);
ret += symbol__fprintf(pos, fp);
}
return ret;
}
int kallsyms__parse(const char *filename, void *arg,
int (*process_symbol)(void *arg, const char *name,
char type, u64 start, u64 end))
{
char *line = NULL;
size_t n;
int err = -1;
FILE *file = fopen(filename, "r");
if (file == NULL)
goto out_failure;
err = 0;
while (!feof(file)) {
u64 start;
int line_len, len;
char symbol_type;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
char *symbol_name;
line_len = getline(&line, &n, file);
if (line_len < 0 || !line)
break;
line[--line_len] = '\0'; /* \n */
len = hex2u64(line, &start);
len++;
if (len + 2 >= line_len)
continue;
symbol_type = line[len];
len += 2;
symbol_name = line + len;
len = line_len - len;
if (len >= KSYM_NAME_LEN) {
err = -1;
break;
}
/*
* module symbols are not sorted so we add all
* symbols with zero length and rely on
* symbols__fixup_end() to fix it up.
*/
err = process_symbol(arg, symbol_name,
symbol_type, start, start);
if (err)
break;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
}
free(line);
fclose(file);
return err;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
out_failure:
return -1;
}
struct process_kallsyms_args {
struct map *map;
struct dso *dso;
};
static u8 kallsyms2elf_type(char type)
{
if (type == 'W')
return STB_WEAK;
return isupper(type) ? STB_GLOBAL : STB_LOCAL;
}
static int map__process_kallsym_symbol(void *arg, const char *name,
char type, u64 start, u64 end)
{
struct symbol *sym;
struct process_kallsyms_args *a = arg;
struct rb_root *root = &a->dso->symbols[a->map->type];
if (!symbol_type__is_a(type, a->map->type))
return 0;
sym = symbol__new(start, end - start + 1,
kallsyms2elf_type(type), name);
if (sym == NULL)
return -ENOMEM;
/*
* We will pass the symbols to the filter later, in
* map__split_kallsyms, when we have split the maps per module
*/
symbols__insert(root, sym);
return 0;
}
/*
* Loads the function entries in /proc/kallsyms into kernel_map->dso,
* so that we can in the next step set the symbol ->end address and then
* call kernel_maps__split_kallsyms.
*/
static int dso__load_all_kallsyms(struct dso *dso, const char *filename,
struct map *map)
{
struct process_kallsyms_args args = { .map = map, .dso = dso, };
return kallsyms__parse(filename, &args, map__process_kallsym_symbol);
}
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
/*
* Split the symbols into maps, making sure there are no overlaps, i.e. the
* kernel range is broken in several maps, named [kernel].N, as we don't have
* the original ELF section names vmlinux have.
*/
static int dso__split_kallsyms(struct dso *dso, struct map *map,
symbol_filter_t filter)
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
{
struct map_groups *kmaps = map__kmap(map)->kmaps;
struct machine *machine = kmaps->machine;
struct map *curr_map = map;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
struct symbol *pos;
int count = 0, moved = 0;
struct rb_root *root = &dso->symbols[map->type];
struct rb_node *next = rb_first(root);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
int kernel_range = 0;
while (next) {
char *module;
pos = rb_entry(next, struct symbol, rb_node);
next = rb_next(&pos->rb_node);
module = strchr(pos->name, '\t');
if (module) {
if (!symbol_conf.use_modules)
goto discard_symbol;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
*module++ = '\0';
if (strcmp(curr_map->dso->short_name, module)) {
if (curr_map != map &&
dso->kernel == DSO_TYPE_GUEST_KERNEL &&
machine__is_default_guest(machine)) {
/*
* We assume all symbols of a module are
* continuous in * kallsyms, so curr_map
* points to a module and all its
* symbols are in its kmap. Mark it as
* loaded.
*/
dso__set_loaded(curr_map->dso,
curr_map->type);
}
curr_map = map_groups__find_by_name(kmaps,
map->type, module);
if (curr_map == NULL) {
pr_debug("%s/proc/{kallsyms,modules} "
"inconsistency while looking "
"for \"%s\" module!\n",
machine->root_dir, module);
curr_map = map;
goto discard_symbol;
}
if (curr_map->dso->loaded &&
!machine__is_default_guest(machine))
goto discard_symbol;
}
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
/*
* So that we look just like we get from .ko files,
* i.e. not prelinked, relative to map->start.
*/
pos->start = curr_map->map_ip(curr_map, pos->start);
pos->end = curr_map->map_ip(curr_map, pos->end);
} else if (curr_map != map) {
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
char dso_name[PATH_MAX];
struct dso *ndso;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (count == 0) {
curr_map = map;
goto filter_symbol;
}
if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
snprintf(dso_name, sizeof(dso_name),
"[guest.kernel].%d",
kernel_range++);
else
snprintf(dso_name, sizeof(dso_name),
"[kernel].%d",
kernel_range++);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
ndso = dso__new(dso_name);
if (ndso == NULL)
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
return -1;
ndso->kernel = dso->kernel;
curr_map = map__new2(pos->start, ndso, map->type);
if (curr_map == NULL) {
dso__delete(ndso);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
return -1;
}
curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
map_groups__insert(kmaps, curr_map);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
++kernel_range;
}
filter_symbol:
if (filter && filter(curr_map, pos)) {
discard_symbol: rb_erase(&pos->rb_node, root);
symbol__delete(pos);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
} else {
if (curr_map != map) {
rb_erase(&pos->rb_node, root);
symbols__insert(&curr_map->dso->symbols[curr_map->type], pos);
++moved;
} else
++count;
}
}
if (curr_map != map &&
dso->kernel == DSO_TYPE_GUEST_KERNEL &&
machine__is_default_guest(kmaps->machine)) {
dso__set_loaded(curr_map->dso, curr_map->type);
}
return count + moved;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
}
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
static bool symbol__restricted_filename(const char *filename,
const char *restricted_filename)
{
bool restricted = false;
if (symbol_conf.kptr_restrict) {
char *r = realpath(filename, NULL);
if (r != NULL) {
restricted = strcmp(r, restricted_filename) == 0;
free(r);
return restricted;
}
}
return restricted;
}
int dso__load_kallsyms(struct dso *dso, const char *filename,
struct map *map, symbol_filter_t filter)
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
{
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
if (symbol__restricted_filename(filename, "/proc/kallsyms"))
return -1;
if (dso__load_all_kallsyms(dso, filename, map) < 0)
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
return -1;
symbols__fixup_duplicate(&dso->symbols[map->type]);
symbols__fixup_end(&dso->symbols[map->type]);
if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
dso->symtab_type = DSO_BINARY_TYPE__GUEST_KALLSYMS;
else
dso->symtab_type = DSO_BINARY_TYPE__KALLSYMS;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
return dso__split_kallsyms(dso, map, filter);
}
static int dso__load_perf_map(struct dso *dso, struct map *map,
symbol_filter_t filter)
{
char *line = NULL;
size_t n;
FILE *file;
int nr_syms = 0;
file = fopen(dso->long_name, "r");
if (file == NULL)
goto out_failure;
while (!feof(file)) {
u64 start, size;
struct symbol *sym;
int line_len, len;
line_len = getline(&line, &n, file);
if (line_len < 0)
break;
if (!line)
goto out_failure;
line[--line_len] = '\0'; /* \n */
len = hex2u64(line, &start);
len++;
if (len + 2 >= line_len)
continue;
len += hex2u64(line + len, &size);
len++;
if (len + 2 >= line_len)
continue;
sym = symbol__new(start, size, STB_GLOBAL, line + len);
if (sym == NULL)
goto out_delete_line;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (filter && filter(map, sym))
symbol__delete(sym);
else {
symbols__insert(&dso->symbols[map->type], sym);
nr_syms++;
}
}
free(line);
fclose(file);
return nr_syms;
out_delete_line:
free(line);
out_failure:
return -1;
}
/**
* elf_symtab__for_each_symbol - iterate thru all the symbols
*
* @syms: struct elf_symtab instance to iterate
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
* @idx: uint32_t idx
* @sym: GElf_Sym iterator
*/
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
#define elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) \
for (idx = 0, gelf_getsym(syms, idx, &sym);\
idx < nr_syms; \
idx++, gelf_getsym(syms, idx, &sym))
static inline uint8_t elf_sym__type(const GElf_Sym *sym)
{
return GELF_ST_TYPE(sym->st_info);
}
static inline int elf_sym__is_function(const GElf_Sym *sym)
{
return elf_sym__type(sym) == STT_FUNC &&
sym->st_name != 0 &&
sym->st_shndx != SHN_UNDEF;
}
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
static inline bool elf_sym__is_object(const GElf_Sym *sym)
{
return elf_sym__type(sym) == STT_OBJECT &&
sym->st_name != 0 &&
sym->st_shndx != SHN_UNDEF;
}
static inline int elf_sym__is_label(const GElf_Sym *sym)
{
return elf_sym__type(sym) == STT_NOTYPE &&
sym->st_name != 0 &&
sym->st_shndx != SHN_UNDEF &&
sym->st_shndx != SHN_ABS;
}
static inline const char *elf_sec__name(const GElf_Shdr *shdr,
const Elf_Data *secstrs)
{
return secstrs->d_buf + shdr->sh_name;
}
static inline int elf_sec__is_text(const GElf_Shdr *shdr,
const Elf_Data *secstrs)
{
return strstr(elf_sec__name(shdr, secstrs), "text") != NULL;
}
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
static inline bool elf_sec__is_data(const GElf_Shdr *shdr,
const Elf_Data *secstrs)
{
return strstr(elf_sec__name(shdr, secstrs), "data") != NULL;
}
static inline const char *elf_sym__name(const GElf_Sym *sym,
const Elf_Data *symstrs)
{
return symstrs->d_buf + sym->st_name;
}
static Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep,
GElf_Shdr *shp, const char *name,
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
size_t *idx)
{
Elf_Scn *sec = NULL;
size_t cnt = 1;
while ((sec = elf_nextscn(elf, sec)) != NULL) {
char *str;
gelf_getshdr(sec, shp);
str = elf_strptr(elf, ep->e_shstrndx, shp->sh_name);
if (!strcmp(name, str)) {
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
if (idx)
*idx = cnt;
break;
}
++cnt;
}
return sec;
}
#define elf_section__for_each_rel(reldata, pos, pos_mem, idx, nr_entries) \
for (idx = 0, pos = gelf_getrel(reldata, 0, &pos_mem); \
idx < nr_entries; \
++idx, pos = gelf_getrel(reldata, idx, &pos_mem))
#define elf_section__for_each_rela(reldata, pos, pos_mem, idx, nr_entries) \
for (idx = 0, pos = gelf_getrela(reldata, 0, &pos_mem); \
idx < nr_entries; \
++idx, pos = gelf_getrela(reldata, idx, &pos_mem))
/*
* We need to check if we have a .dynsym, so that we can handle the
* .plt, synthesizing its symbols, that aren't on the symtabs (be it
* .dynsym or .symtab).
* And always look at the original dso, not at debuginfo packages, that
* have the PLT data stripped out (shdr_rel_plt.sh_type == SHT_NOBITS).
*/
static int
dso__synthesize_plt_symbols(struct dso *dso, char *name, struct map *map,
symbol_filter_t filter)
{
uint32_t nr_rel_entries, idx;
GElf_Sym sym;
u64 plt_offset;
GElf_Shdr shdr_plt;
struct symbol *f;
GElf_Shdr shdr_rel_plt, shdr_dynsym;
Elf_Data *reldata, *syms, *symstrs;
Elf_Scn *scn_plt_rel, *scn_symstrs, *scn_dynsym;
size_t dynsym_idx;
GElf_Ehdr ehdr;
char sympltname[1024];
Elf *elf;
int nr = 0, symidx, fd, err = 0;
fd = open(name, O_RDONLY);
if (fd < 0)
goto out;
elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
if (elf == NULL)
goto out_close;
if (gelf_getehdr(elf, &ehdr) == NULL)
goto out_elf_end;
scn_dynsym = elf_section_by_name(elf, &ehdr, &shdr_dynsym,
".dynsym", &dynsym_idx);
if (scn_dynsym == NULL)
goto out_elf_end;
scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
".rela.plt", NULL);
if (scn_plt_rel == NULL) {
scn_plt_rel = elf_section_by_name(elf, &ehdr, &shdr_rel_plt,
".rel.plt", NULL);
if (scn_plt_rel == NULL)
goto out_elf_end;
}
err = -1;
if (shdr_rel_plt.sh_link != dynsym_idx)
goto out_elf_end;
if (elf_section_by_name(elf, &ehdr, &shdr_plt, ".plt", NULL) == NULL)
goto out_elf_end;
/*
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
* Fetch the relocation section to find the idxes to the GOT
* and the symbols in the .dynsym they refer to.
*/
reldata = elf_getdata(scn_plt_rel, NULL);
if (reldata == NULL)
goto out_elf_end;
syms = elf_getdata(scn_dynsym, NULL);
if (syms == NULL)
goto out_elf_end;
scn_symstrs = elf_getscn(elf, shdr_dynsym.sh_link);
if (scn_symstrs == NULL)
goto out_elf_end;
symstrs = elf_getdata(scn_symstrs, NULL);
if (symstrs == NULL)
goto out_elf_end;
nr_rel_entries = shdr_rel_plt.sh_size / shdr_rel_plt.sh_entsize;
plt_offset = shdr_plt.sh_offset;
if (shdr_rel_plt.sh_type == SHT_RELA) {
GElf_Rela pos_mem, *pos;
elf_section__for_each_rela(reldata, pos, pos_mem, idx,
nr_rel_entries) {
symidx = GELF_R_SYM(pos->r_info);
plt_offset += shdr_plt.sh_entsize;
gelf_getsym(syms, symidx, &sym);
snprintf(sympltname, sizeof(sympltname),
"%s@plt", elf_sym__name(&sym, symstrs));
f = symbol__new(plt_offset, shdr_plt.sh_entsize,
STB_GLOBAL, sympltname);
if (!f)
goto out_elf_end;
if (filter && filter(map, f))
symbol__delete(f);
else {
symbols__insert(&dso->symbols[map->type], f);
++nr;
}
}
} else if (shdr_rel_plt.sh_type == SHT_REL) {
GElf_Rel pos_mem, *pos;
elf_section__for_each_rel(reldata, pos, pos_mem, idx,
nr_rel_entries) {
symidx = GELF_R_SYM(pos->r_info);
plt_offset += shdr_plt.sh_entsize;
gelf_getsym(syms, symidx, &sym);
snprintf(sympltname, sizeof(sympltname),
"%s@plt", elf_sym__name(&sym, symstrs));
f = symbol__new(plt_offset, shdr_plt.sh_entsize,
STB_GLOBAL, sympltname);
if (!f)
goto out_elf_end;
if (filter && filter(map, f))
symbol__delete(f);
else {
symbols__insert(&dso->symbols[map->type], f);
++nr;
}
}
}
err = 0;
out_elf_end:
elf_end(elf);
out_close:
close(fd);
if (err == 0)
return nr;
out:
pr_debug("%s: problems reading %s PLT info.\n",
__func__, dso->long_name);
return 0;
}
static bool elf_sym__is_a(GElf_Sym *sym, enum map_type type)
{
switch (type) {
case MAP__FUNCTION:
return elf_sym__is_function(sym);
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
case MAP__VARIABLE:
return elf_sym__is_object(sym);
default:
return false;
}
}
static bool elf_sec__is_a(GElf_Shdr *shdr, Elf_Data *secstrs,
enum map_type type)
{
switch (type) {
case MAP__FUNCTION:
return elf_sec__is_text(shdr, secstrs);
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
case MAP__VARIABLE:
return elf_sec__is_data(shdr, secstrs);
default:
return false;
}
}
static size_t elf_addr_to_index(Elf *elf, GElf_Addr addr)
{
Elf_Scn *sec = NULL;
GElf_Shdr shdr;
size_t cnt = 1;
while ((sec = elf_nextscn(elf, sec)) != NULL) {
gelf_getshdr(sec, &shdr);
if ((addr >= shdr.sh_addr) &&
(addr < (shdr.sh_addr + shdr.sh_size)))
return cnt;
++cnt;
}
return -1;
}
perf symbols: Handle different endians properly during symbol load Currently we dont care about the file object's endianness. It's possible we read buildid file object from different architecture than we are currentlly running on. So we need to care about properly reading such object's data - handle different endianness properly. Adding: needs_swap DSO field dso__swap_init function to initialize DSO's needs_swap DSO__SWAP to read the data with proper swaps 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 1) - 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> 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/1338380624-7443-2-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:42 +00:00
static int dso__swap_init(struct dso *dso, unsigned char eidata)
{
static unsigned int const endian = 1;
dso->needs_swap = DSO_SWAP__NO;
switch (eidata) {
case ELFDATA2LSB:
/* We are big endian, DSO is little endian. */
if (*(unsigned char const *)&endian != 1)
dso->needs_swap = DSO_SWAP__YES;
break;
case ELFDATA2MSB:
/* We are little endian, DSO is big endian. */
if (*(unsigned char const *)&endian != 0)
dso->needs_swap = DSO_SWAP__YES;
break;
default:
pr_err("unrecognized DSO data encoding %d\n", eidata);
return -EINVAL;
}
return 0;
}
static int dso__load_sym(struct dso *dso, struct map *map, const char *name,
int fd, symbol_filter_t filter, int kmodule,
int want_symtab)
{
struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
struct map *curr_map = map;
struct dso *curr_dso = dso;
Elf_Data *symstrs, *secstrs;
uint32_t nr_syms;
int err = -1;
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
uint32_t idx;
GElf_Ehdr ehdr;
GElf_Shdr shdr, opdshdr;
Elf_Data *syms, *opddata = NULL;
GElf_Sym sym;
Elf_Scn *sec, *sec_strndx, *opdsec;
Elf *elf;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
int nr = 0;
size_t opdidx = 0;
elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
if (elf == NULL) {
pr_debug("%s: cannot read %s ELF file.\n", __func__, name);
goto out_close;
}
if (gelf_getehdr(elf, &ehdr) == NULL) {
pr_debug("%s: cannot get elf header.\n", __func__);
goto out_elf_end;
}
perf symbols: Handle different endians properly during symbol load Currently we dont care about the file object's endianness. It's possible we read buildid file object from different architecture than we are currentlly running on. So we need to care about properly reading such object's data - handle different endianness properly. Adding: needs_swap DSO field dso__swap_init function to initialize DSO's needs_swap DSO__SWAP to read the data with proper swaps 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 1) - 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> 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/1338380624-7443-2-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:42 +00:00
if (dso__swap_init(dso, ehdr.e_ident[EI_DATA]))
goto out_elf_end;
/* Always reject images with a mismatched build-id: */
if (dso->has_build_id) {
u8 build_id[BUILD_ID_SIZE];
if (elf_read_build_id(elf, build_id, BUILD_ID_SIZE) < 0)
goto out_elf_end;
if (!dso__build_id_equal(dso, build_id))
goto out_elf_end;
}
sec = elf_section_by_name(elf, &ehdr, &shdr, ".symtab", NULL);
if (sec == NULL) {
if (want_symtab)
goto out_elf_end;
sec = elf_section_by_name(elf, &ehdr, &shdr, ".dynsym", NULL);
if (sec == NULL)
goto out_elf_end;
}
opdsec = elf_section_by_name(elf, &ehdr, &opdshdr, ".opd", &opdidx);
if (opdshdr.sh_type != SHT_PROGBITS)
opdsec = NULL;
if (opdsec)
opddata = elf_rawdata(opdsec, NULL);
syms = elf_getdata(sec, NULL);
if (syms == NULL)
goto out_elf_end;
sec = elf_getscn(elf, shdr.sh_link);
if (sec == NULL)
goto out_elf_end;
symstrs = elf_getdata(sec, NULL);
if (symstrs == NULL)
goto out_elf_end;
sec_strndx = elf_getscn(elf, ehdr.e_shstrndx);
if (sec_strndx == NULL)
goto out_elf_end;
secstrs = elf_getdata(sec_strndx, NULL);
if (secstrs == NULL)
goto out_elf_end;
nr_syms = shdr.sh_size / shdr.sh_entsize;
memset(&sym, 0, sizeof(sym));
if (dso->kernel == DSO_TYPE_USER) {
dso->adjust_symbols = (ehdr.e_type == ET_EXEC ||
perf_counter tools: Adjust symbols in ET_EXEC files too Ingo Molnar wrote: > i just bisected a 'perf report' bug that would cause us to not > resolve all user-space symbols in a 'git gc' run to: > > f5812a7a336fb952d819e4427b9a2dce02368e82 is first bad commit > commit f5812a7a336fb952d819e4427b9a2dce02368e82 > Author: Arnaldo Carvalho de Melo <acme@redhat.com> > Date: Tue Jun 30 11:43:17 2009 -0300 > > perf_counter tools: Adjust only prelinked symbol's addresses Rename ->prelinked to ->adjust_symbols and making what was done only for prelinked libraries also to ET_EXEC binaries, such as /usr/bin/git: [acme@doppio pahole]$ readelf -h /usr/bin/git | grep Type Type: EXEC (Executable file) [acme@doppio pahole]$ And after installing the 'git-debuginfo' package, I get correct results: [acme@doppio linux-2.6-tip]$ perf report --sort comm,dso,symbol -d /usr/bin/git | head -20 # # (1139614 samples) # # Overhead Command Shared Object Symbol # ........ ................ ......................... ...... # 34.98% git /usr/bin/git [.] send_sideband 33.39% git /usr/bin/git [.] enter_repo 6.81% git /usr/bin/git [.] diff_opt_parse 4.95% git /usr/bin/git [.] is_repository_shallow 3.24% git /usr/bin/git [.] odb_mkstemp 1.39% git /usr/bin/git [.] output 1.34% git /usr/bin/git [.] xmmap 1.25% git /usr/bin/git [.] receive_pack_config 1.16% git /usr/bin/git [.] git_pathdup 0.90% git /usr/bin/git [.] read_object_with_reference 0.86% git /usr/bin/git [.] show_patch_diff 0.85% git /usr/bin/git 0x00000000095e2e 0.69% git /usr/bin/git [.] display [acme@doppio linux-2.6-tip]$ I'll check what are the last cases where we can't resolve symbols, like this 0x00000000095e2e later. And I guess this will fix the problems Mike were seeing too: [acme@doppio linux-2.6-tip]$ readelf -h ../build/perf/vmlinux | grep Type Type: EXEC (Executable file) [acme@doppio linux-2.6-tip]$ Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-03 00:24:14 +00:00
elf_section_by_name(elf, &ehdr, &shdr,
".gnu.prelink_undo",
NULL) != NULL);
} else {
dso->adjust_symbols = 0;
}
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
elf_symtab__for_each_symbol(syms, nr_syms, idx, sym) {
struct symbol *f;
perf tools: Handle relocatable kernels DSOs don't have this problem because the kernel emits a PERF_MMAP for each new executable mapping it performs on monitored threads. To fix the kernel case we simulate the same behaviour, by having 'perf record' to synthesize a PERF_MMAP for the kernel, encoded like this: [root@doppio ~]# perf record -a -f sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.344 MB perf.data (~15038 samples) ] [root@doppio ~]# perf report -D | head -10 0xd0 [0x40]: event: 1 . . ... raw event: size 64 bytes . 0000: 01 00 00 00 00 00 40 00 00 00 00 00 00 00 00 00 ......@........ . 0010: 00 00 00 81 ff ff ff ff 00 00 00 00 00 00 00 00 ............... . 0020: 00 00 00 00 00 00 00 00 5b 6b 65 72 6e 65 6c 2e ........ [kernel . 0030: 6b 61 6c 6c 73 79 6d 73 2e 5f 74 65 78 74 5d 00 kallsyms._text] . 0xd0 [0x40]: PERF_RECORD_MMAP 0/0: [0xffffffff81000000((nil)) @ (nil)]: [kernel.kallsyms._text] I.e. we identify such event as having: .pid = 0 .filename = [kernel.kallsyms.REFNAME] .start = REFNAME addr in /proc/kallsyms at 'perf record' time and use now a hardcoded value of '.text' for REFNAME. Then, later, in 'perf report', if there are any kernel hits and thus we need to resolve kernel symbols, we search for REFNAME and if its address changed, relocation happened and we thus must change the kernel mapping routines to one that uses .pgoff as the relocation to apply. This way we use the same mechanism used for the other DSOs and don't have to do a two pass in all the kernel symbols. Reported-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1262717431-1246-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-05 18:50:31 +00:00
const char *elf_name = elf_sym__name(&sym, symstrs);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
char *demangled = NULL;
int is_label = elf_sym__is_label(&sym);
const char *section_name;
if (kmap && kmap->ref_reloc_sym && kmap->ref_reloc_sym->name &&
strcmp(elf_name, kmap->ref_reloc_sym->name) == 0)
kmap->ref_reloc_sym->unrelocated_addr = sym.st_value;
perf tools: Handle relocatable kernels DSOs don't have this problem because the kernel emits a PERF_MMAP for each new executable mapping it performs on monitored threads. To fix the kernel case we simulate the same behaviour, by having 'perf record' to synthesize a PERF_MMAP for the kernel, encoded like this: [root@doppio ~]# perf record -a -f sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.344 MB perf.data (~15038 samples) ] [root@doppio ~]# perf report -D | head -10 0xd0 [0x40]: event: 1 . . ... raw event: size 64 bytes . 0000: 01 00 00 00 00 00 40 00 00 00 00 00 00 00 00 00 ......@........ . 0010: 00 00 00 81 ff ff ff ff 00 00 00 00 00 00 00 00 ............... . 0020: 00 00 00 00 00 00 00 00 5b 6b 65 72 6e 65 6c 2e ........ [kernel . 0030: 6b 61 6c 6c 73 79 6d 73 2e 5f 74 65 78 74 5d 00 kallsyms._text] . 0xd0 [0x40]: PERF_RECORD_MMAP 0/0: [0xffffffff81000000((nil)) @ (nil)]: [kernel.kallsyms._text] I.e. we identify such event as having: .pid = 0 .filename = [kernel.kallsyms.REFNAME] .start = REFNAME addr in /proc/kallsyms at 'perf record' time and use now a hardcoded value of '.text' for REFNAME. Then, later, in 'perf report', if there are any kernel hits and thus we need to resolve kernel symbols, we search for REFNAME and if its address changed, relocation happened and we thus must change the kernel mapping routines to one that uses .pgoff as the relocation to apply. This way we use the same mechanism used for the other DSOs and don't have to do a two pass in all the kernel symbols. Reported-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1262717431-1246-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-01-05 18:50:31 +00:00
if (!is_label && !elf_sym__is_a(&sym, map->type))
continue;
/* Reject ARM ELF "mapping symbols": these aren't unique and
* don't identify functions, so will confuse the profile
* output: */
if (ehdr.e_machine == EM_ARM) {
if (!strcmp(elf_name, "$a") ||
!strcmp(elf_name, "$d") ||
!strcmp(elf_name, "$t"))
continue;
}
if (opdsec && sym.st_shndx == opdidx) {
u32 offset = sym.st_value - opdshdr.sh_addr;
u64 *opd = opddata->d_buf + offset;
perf symbols: Handle different endians properly during symbol load Currently we dont care about the file object's endianness. It's possible we read buildid file object from different architecture than we are currentlly running on. So we need to care about properly reading such object's data - handle different endianness properly. Adding: needs_swap DSO field dso__swap_init function to initialize DSO's needs_swap DSO__SWAP to read the data with proper swaps 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 1) - 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> 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/1338380624-7443-2-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-05-30 12:23:42 +00:00
sym.st_value = DSO__SWAP(dso, u64, *opd);
sym.st_shndx = elf_addr_to_index(elf, sym.st_value);
}
sec = elf_getscn(elf, sym.st_shndx);
if (!sec)
goto out_elf_end;
gelf_getshdr(sec, &shdr);
if (is_label && !elf_sec__is_a(&shdr, secstrs, map->type))
continue;
section_name = elf_sec__name(&shdr, secstrs);
perf_counter tools: Add 'perf annotate' feature Add new perf sub-command to display annotated source code: $ perf annotate decode_tree_entry ------------------------------------------------ Percent | Source code & Disassembly of /home/mingo/git/git ------------------------------------------------ : : /home/mingo/git/git: file format elf64-x86-64 : : : Disassembly of section .text: : : 00000000004a0da0 <decode_tree_entry>: : *modep = mode; : return str; : } : : static void decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size) : { 3.82 : 4a0da0: 41 54 push %r12 : const char *path; : unsigned int mode, len; : : if (size < 24 || buf[size - 21]) 0.17 : 4a0da2: 48 83 fa 17 cmp $0x17,%rdx : *modep = mode; : return str; : } : : static void decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size) : { 0.00 : 4a0da6: 49 89 fc mov %rdi,%r12 0.00 : 4a0da9: 55 push %rbp 3.37 : 4a0daa: 53 push %rbx : const char *path; : unsigned int mode, len; : : if (size < 24 || buf[size - 21]) 0.08 : 4a0dab: 76 73 jbe 4a0e20 <decode_tree_entry+0x80> 0.00 : 4a0dad: 80 7c 16 eb 00 cmpb $0x0,-0x15(%rsi,%rdx,1) 3.48 : 4a0db2: 75 6c jne 4a0e20 <decode_tree_entry+0x80> : static const char *get_mode(const char *str, unsigned int *modep) : { : unsigned char c; : unsigned int mode = 0; : : if (*str == ' ') 1.94 : 4a0db4: 0f b6 06 movzbl (%rsi),%eax 0.39 : 4a0db7: 3c 20 cmp $0x20,%al 0.00 : 4a0db9: 74 65 je 4a0e20 <decode_tree_entry+0x80> : return NULL; : : while ((c = *str++) != ' ') { 0.06 : 4a0dbb: 89 c2 mov %eax,%edx : if (c < '0' || c > '7') 1.99 : 4a0dbd: 31 ed xor %ebp,%ebp : unsigned int mode = 0; : : if (*str == ' ') : return NULL; : : while ((c = *str++) != ' ') { 1.74 : 4a0dbf: 48 8d 5e 01 lea 0x1(%rsi),%rbx : if (c < '0' || c > '7') 0.00 : 4a0dc3: 8d 42 d0 lea -0x30(%rdx),%eax 0.17 : 4a0dc6: 3c 07 cmp $0x7,%al 0.00 : 4a0dc8: 76 0d jbe 4a0dd7 <decode_tree_entry+0x37> 0.00 : 4a0dca: eb 54 jmp 4a0e20 <decode_tree_entry+0x80> 0.00 : 4a0dcc: 0f 1f 40 00 nopl 0x0(%rax) 16.57 : 4a0dd0: 8d 42 d0 lea -0x30(%rdx),%eax 0.14 : 4a0dd3: 3c 07 cmp $0x7,%al 0.00 : 4a0dd5: 77 49 ja 4a0e20 <decode_tree_entry+0x80> : return NULL; : mode = (mode << 3) + (c - '0'); 3.12 : 4a0dd7: 0f b6 c2 movzbl %dl,%eax : unsigned int mode = 0; : : if (*str == ' ') : return NULL; : : while ((c = *str++) != ' ') { 0.00 : 4a0dda: 0f b6 13 movzbl (%rbx),%edx 16.74 : 4a0ddd: 48 83 c3 01 add $0x1,%rbx : if (c < '0' || c > '7') : return NULL; : mode = (mode << 3) + (c - '0'); The first column is the percentage of samples that arrived on that particular line - relative to the total cost of the function. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-06 13:48:52 +00:00
/* On ARM, symbols for thumb functions have 1 added to
* the symbol address as a flag - remove it */
if ((ehdr.e_machine == EM_ARM) &&
(map->type == MAP__FUNCTION) &&
(sym.st_value & 1))
--sym.st_value;
if (dso->kernel != DSO_TYPE_USER || kmodule) {
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
char dso_name[PATH_MAX];
if (strcmp(section_name,
(curr_dso->short_name +
dso->short_name_len)) == 0)
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
goto new_symbol;
if (strcmp(section_name, ".text") == 0) {
curr_map = map;
curr_dso = dso;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
goto new_symbol;
}
snprintf(dso_name, sizeof(dso_name),
"%s%s", dso->short_name, section_name);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
curr_map = map_groups__find_by_name(kmap->kmaps, map->type, dso_name);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (curr_map == NULL) {
u64 start = sym.st_value;
if (kmodule)
start += map->start + shdr.sh_offset;
curr_dso = dso__new(dso_name);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (curr_dso == NULL)
goto out_elf_end;
curr_dso->kernel = dso->kernel;
curr_dso->long_name = dso->long_name;
curr_dso->long_name_len = dso->long_name_len;
curr_map = map__new2(start, curr_dso,
map->type);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (curr_map == NULL) {
dso__delete(curr_dso);
goto out_elf_end;
}
curr_map->map_ip = identity__map_ip;
curr_map->unmap_ip = identity__map_ip;
curr_dso->symtab_type = dso->symtab_type;
map_groups__insert(kmap->kmaps, curr_map);
dsos__add(&dso->node, curr_dso);
dso__set_loaded(curr_dso, map->type);
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
} else
curr_dso = curr_map->dso;
goto new_symbol;
}
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (curr_dso->adjust_symbols) {
pr_debug4("%s: adjusting symbol: st_value: %#" PRIx64 " "
"sh_addr: %#" PRIx64 " sh_offset: %#" PRIx64 "\n", __func__,
(u64)sym.st_value, (u64)shdr.sh_addr,
(u64)shdr.sh_offset);
sym.st_value -= shdr.sh_addr - shdr.sh_offset;
}
/*
* We need to figure out if the object was created from C++ sources
* DWARF DW_compile_unit has this, but we don't always have access
* to it...
*/
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
demangled = bfd_demangle(NULL, elf_name, DMGL_PARAMS | DMGL_ANSI);
if (demangled != NULL)
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 10:26:57 +00:00
elf_name = demangled;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
new_symbol:
f = symbol__new(sym.st_value, sym.st_size,
GELF_ST_BIND(sym.st_info), elf_name);
free(demangled);
if (!f)
goto out_elf_end;
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
if (filter && filter(curr_map, f))
symbol__delete(f);
else {
symbols__insert(&curr_dso->symbols[curr_map->type], f);
nr++;
}
}
perf tools: Improve kernel/modules symbol lookup This removes the ovelapping of vmlinux addresses with modules, using the ELF section name when using --vmlinux and creating a unique DSO name when using /proc/kallsyms ([kernel].N). This is done by creating multiple 'struct map' instances for address ranges backed by DSOs that have just the symbols for that range and a name that is derived from the ELF section name.o Now it is possible to ask for just the symbols in some particular kernel section: $ perf report -m --vmlinux ../build/tip-recvmmsg/vmlinux \ --dsos [kernel].vsyscall_fn | head -15 52.73% Xorg [.] vread_hpet 18.61% firefox [.] vread_hpet 14.50% npviewer.bin [.] vread_hpet 6.83% compiz [.] vread_hpet 5.73% glxgears [.] vread_hpet 0.63% java [.] vread_hpet 0.30% gnome-terminal [.] vread_hpet 0.23% perf [.] vread_hpet 0.18% xchat [.] vread_hpet $ Now we don't have to first lookup the list of modules and then, if it fails, vmlinux symbols, its just a simple lookup for the map then the symbols, just like for threads. Reports generated using /proc/kallsyms and --vmlinux should provide the same results, modulo the DSO name for sections other than ".text". But they don't right now because things like: ffffffff81011c20-ffffffff81012068 system_call ffffffff81011c30-ffffffff81011c9b system_call_after_swapgs ffffffff81011c9c-ffffffff81011cb6 system_call_fastpath ffffffff81011cb7-ffffffff81011cbb ret_from_sys_call I.e. overlapping symbols, again some ASM special case that we have to fixup. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1254934136-8503-1-git-send-email-acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-07 16:48:56 +00:00
/*
* For misannotated, zeroed, ASM function sizes.
*/
if (nr > 0) {
symbols__fixup_duplicate(&dso->symbols[map->type]);
symbols__fixup_end(&dso->symbols[map->type]);
if (kmap) {
/*
* We need to fixup this here too because we create new
* maps here, for things like vsyscall sections.
*/
__map_groups__fixup_end(kmap->kmaps, map->type);
}
}
err = nr;
out_elf_end:
elf_end(elf);
out_close:
return err;
}
void symbol__elf_init(void)
{
elf_version(EV_CURRENT);
}
static bool dso__build_id_equal(const struct dso *dso, u8 *build_id)
{
return memcmp(dso->build_id, build_id, sizeof(dso->build_id)) == 0;
}
bool __dsos__read_build_ids(struct list_head *head, bool with_hits)
{
bool have_build_id = false;
struct dso *pos;
list_for_each_entry(pos, head, node) {
if (with_hits && !pos->hit)
continue;
if (pos->has_build_id) {
have_build_id = true;
continue;
}
if (filename__read_build_id(pos->long_name, pos->build_id,
sizeof(pos->build_id)) > 0) {
have_build_id = true;
pos->has_build_id = true;
}
}
return have_build_id;
}
/*
* Align offset to 4 bytes as needed for note name and descriptor data.
*/
#define NOTE_ALIGN(n) (((n) + 3) & -4U)
static int elf_read_build_id(Elf *elf, void *bf, size_t size)
2009-08-05 22:02:49 +00:00
{
int err = -1;
2009-08-05 22:02:49 +00:00
GElf_Ehdr ehdr;
GElf_Shdr shdr;
Elf_Data *data;
2009-08-05 22:02:49 +00:00
Elf_Scn *sec;
Elf_Kind ek;
void *ptr;
2009-08-05 22:02:49 +00:00
if (size < BUILD_ID_SIZE)
goto out;
ek = elf_kind(elf);
if (ek != ELF_K_ELF)
goto out;
2009-08-05 22:02:49 +00:00
if (gelf_getehdr(elf, &ehdr) == NULL) {
pr_err("%s: cannot get elf header.\n", __func__);
goto out;
2009-08-05 22:02:49 +00:00
}
/*
* Check following sections for notes:
* '.note.gnu.build-id'
* '.notes'
* '.note' (VDSO specific)
*/
do {
sec = elf_section_by_name(elf, &ehdr, &shdr,
".note.gnu.build-id", NULL);
if (sec)
break;
sec = elf_section_by_name(elf, &ehdr, &shdr,
".notes", NULL);
if (sec)
break;
sec = elf_section_by_name(elf, &ehdr, &shdr,
".note", NULL);
if (sec)
break;
return err;
} while (0);
2009-08-05 22:02:49 +00:00
data = elf_getdata(sec, NULL);
if (data == NULL)
goto out;
ptr = data->d_buf;
while (ptr < (data->d_buf + data->d_size)) {
GElf_Nhdr *nhdr = ptr;
size_t namesz = NOTE_ALIGN(nhdr->n_namesz),
descsz = NOTE_ALIGN(nhdr->n_descsz);
const char *name;
ptr += sizeof(*nhdr);
name = ptr;
ptr += namesz;
if (nhdr->n_type == NT_GNU_BUILD_ID &&
nhdr->n_namesz == sizeof("GNU")) {
if (memcmp(name, "GNU", sizeof("GNU")) == 0) {
size_t sz = min(size, descsz);
memcpy(bf, ptr, sz);
memset(bf + sz, 0, size - sz);
err = descsz;
break;
}
}
ptr += descsz;
}
out:
return err;
}
int filename__read_build_id(const char *filename, void *bf, size_t size)
{
int fd, err = -1;
Elf *elf;
if (size < BUILD_ID_SIZE)
goto out;
fd = open(filename, O_RDONLY);
if (fd < 0)
goto out;
elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
if (elf == NULL) {
pr_debug2("%s: cannot read %s ELF file.\n", __func__, filename);
goto out_close;
}
err = elf_read_build_id(elf, bf, size);
elf_end(elf);
out_close:
close(fd);
out:
return err;
}
int sysfs__read_build_id(const char *filename, void *build_id, size_t size)
{
int fd, err = -1;
if (size < BUILD_ID_SIZE)
goto out;
fd = open(filename, O_RDONLY);
if (fd < 0)
goto out;
while (1) {
char bf[BUFSIZ];
GElf_Nhdr nhdr;
size_t namesz, descsz;
if (read(fd, &nhdr, sizeof(nhdr)) != sizeof(nhdr))
break;
namesz = NOTE_ALIGN(nhdr.n_namesz);
descsz = NOTE_ALIGN(nhdr.n_descsz);
if (nhdr.n_type == NT_GNU_BUILD_ID &&
nhdr.n_namesz == sizeof("GNU")) {
if (read(fd, bf, namesz) != (ssize_t)namesz)
break;
if (memcmp(bf, "GNU", sizeof("GNU")) == 0) {
size_t sz = min(descsz, size);
if (read(fd, build_id, sz) == (ssize_t)sz) {
memset(build_id + sz, 0, size - sz);
err = 0;
break;
}
} else if (read(fd, bf, descsz) != (ssize_t)descsz)
break;
} else {
int n = namesz + descsz;
if (read(fd, bf, n) != n)
break;
}
}
close(fd);
out:
return err;
}
static int filename__read_debuglink(const char *filename,
char *debuglink, size_t size)
{
int fd, err = -1;
Elf *elf;
GElf_Ehdr ehdr;
GElf_Shdr shdr;
Elf_Data *data;
Elf_Scn *sec;
Elf_Kind ek;
fd = open(filename, O_RDONLY);
if (fd < 0)
goto out;
elf = elf_begin(fd, PERF_ELF_C_READ_MMAP, NULL);
if (elf == NULL) {
pr_debug2("%s: cannot read %s ELF file.\n", __func__, filename);
goto out_close;
}
ek = elf_kind(elf);
if (ek != ELF_K_ELF)
goto out_close;
if (gelf_getehdr(elf, &ehdr) == NULL) {
pr_err("%s: cannot get elf header.\n", __func__);
goto out_close;
}
sec = elf_section_by_name(elf, &ehdr, &shdr,
".gnu_debuglink", NULL);
if (sec == NULL)
goto out_close;
data = elf_getdata(sec, NULL);
if (data == NULL)
goto out_close;
/* the start of this section is a zero-terminated string */
strncpy(debuglink, data->d_buf, size);
elf_end(elf);
out_close:
close(fd);
out:
return err;
}
char dso__symtab_origin(const struct dso *dso)
{
static const char origin[] = {
[DSO_BINARY_TYPE__KALLSYMS] = 'k',
[DSO_BINARY_TYPE__JAVA_JIT] = 'j',
[DSO_BINARY_TYPE__DEBUGLINK] = 'l',
[DSO_BINARY_TYPE__BUILD_ID_CACHE] = 'B',
[DSO_BINARY_TYPE__FEDORA_DEBUGINFO] = 'f',
[DSO_BINARY_TYPE__UBUNTU_DEBUGINFO] = 'u',
[DSO_BINARY_TYPE__BUILDID_DEBUGINFO] = 'b',
[DSO_BINARY_TYPE__SYSTEM_PATH_DSO] = 'd',
[DSO_BINARY_TYPE__SYSTEM_PATH_KMODULE] = 'K',
[DSO_BINARY_TYPE__GUEST_KALLSYMS] = 'g',
[DSO_BINARY_TYPE__GUEST_KMODULE] = 'G',
};
if (dso == NULL || dso->symtab_type == DSO_BINARY_TYPE__NOT_FOUND)
return '!';
return origin[dso->symtab_type];
}
int dso__binary_type_file(struct dso *dso, enum dso_binary_type type,
char *root_dir, char *file, size_t size)
{
char build_id_hex[BUILD_ID_SIZE * 2 + 1];
int ret = 0;
switch (type) {
case DSO_BINARY_TYPE__DEBUGLINK: {
char *debuglink;
strncpy(file, dso->long_name, size);
debuglink = file + dso->long_name_len;
while (debuglink != file && *debuglink != '/')
debuglink--;
if (*debuglink == '/')
debuglink++;
filename__read_debuglink(dso->long_name, debuglink,
size - (debuglink - file));
}
break;
case DSO_BINARY_TYPE__BUILD_ID_CACHE:
/* skip the locally configured cache if a symfs is given */
if (symbol_conf.symfs[0] ||
(dso__build_id_filename(dso, file, size) == NULL))
ret = -1;
break;
case DSO_BINARY_TYPE__FEDORA_DEBUGINFO:
snprintf(file, size, "%s/usr/lib/debug%s.debug",
symbol_conf.symfs, dso->long_name);
break;
case DSO_BINARY_TYPE__UBUNTU_DEBUGINFO:
snprintf(file, size, "%s/usr/lib/debug%s",
symbol_conf.symfs, dso->long_name);
break;
case DSO_BINARY_TYPE__BUILDID_DEBUGINFO:
if (!dso->has_build_id) {
ret = -1;
break;
}
build_id__sprintf(dso->build_id,
sizeof(dso->build_id),
build_id_hex);
snprintf(file, size,
"%s/usr/lib/debug/.build-id/%.2s/%s.debug",
symbol_conf.symfs, build_id_hex, build_id_hex + 2);
break;
case DSO_BINARY_TYPE__SYSTEM_PATH_DSO:
snprintf(file, size, "%s%s",
symbol_conf.symfs, dso->long_name);
break;
case DSO_BINARY_TYPE__GUEST_KMODULE:
snprintf(file, size, "%s%s%s", symbol_conf.symfs,
root_dir, dso->long_name);
break;
case DSO_BINARY_TYPE__SYSTEM_PATH_KMODULE:
snprintf(file, size, "%s%s", symbol_conf.symfs,
dso->long_name);
break;
default:
case DSO_BINARY_TYPE__KALLSYMS:
case DSO_BINARY_TYPE__GUEST_KALLSYMS:
case DSO_BINARY_TYPE__JAVA_JIT:
case DSO_BINARY_TYPE__NOT_FOUND:
ret = -1;
break;
}
return ret;
}
int dso__load(struct dso *dso, struct map *map, symbol_filter_t filter)
{
char *name;
int ret = -1;
int fd;
u_int i;
struct machine *machine;
char *root_dir = (char *) "";
int want_symtab;
dso__set_loaded(dso, map->type);
if (dso->kernel == DSO_TYPE_KERNEL)
return dso__load_kernel_sym(dso, map, filter);
else if (dso->kernel == DSO_TYPE_GUEST_KERNEL)
return dso__load_guest_kernel_sym(dso, map, filter);
if (map->groups && map->groups->machine)
machine = map->groups->machine;
else
machine = NULL;
name = malloc(PATH_MAX);
if (!name)
return -1;
dso->adjust_symbols = 0;
if (strncmp(dso->name, "/tmp/perf-", 10) == 0) {
struct stat st;
if (lstat(dso->name, &st) < 0)
return -1;
if (st.st_uid && (st.st_uid != geteuid())) {
pr_warning("File %s not owned by current user or root, "
"ignoring it.\n", dso->name);
return -1;
}
ret = dso__load_perf_map(dso, map, filter);
dso->symtab_type = ret > 0 ? DSO_BINARY_TYPE__JAVA_JIT :
DSO_BINARY_TYPE__NOT_FOUND;
return ret;
}
if (machine)
root_dir = machine->root_dir;
/* Iterate over candidate debug images.
* On the first pass, only load images if they have a full symtab.
* Failing that, do a second pass where we accept .dynsym also
*/
want_symtab = 1;
restart:
for (i = 0; i < DSO_BINARY_TYPE__SYMTAB_CNT; i++) {
dso->symtab_type = binary_type_symtab[i];
if (dso__binary_type_file(dso, dso->symtab_type,
root_dir, name, PATH_MAX))
continue;
/* Name is now the name of the next image to try */
fd = open(name, O_RDONLY);
if (fd < 0)
continue;
ret = dso__load_sym(dso, map, name, fd, filter, 0,
want_symtab);
close(fd);
/*
* Some people seem to have debuginfo files _WITHOUT_ debug
* info!?!?
*/
if (!ret)
continue;
if (ret > 0) {
int nr_plt;
nr_plt = dso__synthesize_plt_symbols(dso, name, map, filter);
if (nr_plt > 0)
ret += nr_plt;
break;
}
}
/*
* If we wanted a full symtab but no image had one,
* relax our requirements and repeat the search.
*/
if (ret <= 0 && want_symtab) {
want_symtab = 0;
goto restart;
}
free(name);
if (ret < 0 && strstr(dso->name, " (deleted)") != NULL)
perf tools: Fix dso__new handle() to handle deleted DSOs It is better than showing the map addr, this way at least we know that we can't get the symtabs because the DSO was deleted (system update) while an app still used such DSO. Yeah, don't do that, but if you do, you'll figure it out quicker this way. [acme@doppio linux-2.6-tip]$ perf report | head -15 # Samples: 3796 # # Overhead Command Shared Object Symbol # ........ ....... ................................................................... ...... # 23.55% pidgin /lib64/libglib-2.0.so.0.2000.4.#prelink#.Pd98lu (deleted) [.] 0x00000000038844 21.55% pidgin /lib64/libpthread-2.10.1.so.#prelink#.AFwK8Q (deleted) [.] 0x0000000000a42d 10.85% pidgin [kernel] [.] vread_hpet 7.85% pidgin /lib64/libgobject-2.0.so.0.2000.4.#prelink#.o1vpU7 (deleted) [.] 0x00000000014de8 3.35% pidgin /lib64/libc-2.10.1.so (deleted) [.] 0x0000000007a875 3.19% pidgin /lib64/libdbus-1.so.3.4.0.#prelink#.6mwgZP (deleted) [.] 0x0000000001d254 3.06% pidgin /usr/lib64/libgtk-x11-2.0.so.0.1600.5.#prelink#.511hAl (deleted) [.] 0x000000002334e7 2.90% pidgin /usr/lib64/libgdk-x11-2.0.so.0.1600.5.#prelink#.5qlMo1 (deleted) [.] 0x00000000037b2d 1.84% pidgin [kernel] [k] do_sys_poll 1.45% pidgin /usr/lib64/libX11.so.6.2.0.#prelink#.iR59Rx (deleted) [.] 0x0000000004c751 [acme@doppio linux-2.6-tip]$ Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Luis Claudio R. Gonçalves <lclaudio@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Frédéric Weisbecker <fweisbec@gmail.com> LKML-Reference: <20090811200436.GA3478@ghostprotocols.net> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-11 20:04:36 +00:00
return 0;
return ret;
}
struct map *map_groups__find_by_name(struct map_groups *mg,
enum map_type type, const char *name)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
{
struct rb_node *nd;
for (nd = rb_first(&mg->maps[type]); nd; nd = rb_next(nd)) {
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
struct map *map = rb_entry(nd, struct map, rb_node);
if (map->dso && strcmp(map->dso->short_name, name) == 0)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
return map;
}
return NULL;
}
static int dso__kernel_module_get_build_id(struct dso *dso,
const char *root_dir)
{
char filename[PATH_MAX];
/*
* kernel module short names are of the form "[module]" and
* we need just "module" here.
*/
const char *name = dso->short_name + 1;
snprintf(filename, sizeof(filename),
"%s/sys/module/%.*s/notes/.note.gnu.build-id",
root_dir, (int)strlen(name) - 1, name);
if (sysfs__read_build_id(filename, dso->build_id,
sizeof(dso->build_id)) == 0)
dso->has_build_id = true;
return 0;
}
static int map_groups__set_modules_path_dir(struct map_groups *mg,
const char *dir_name)
{
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
struct dirent *dent;
DIR *dir = opendir(dir_name);
int ret = 0;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (!dir) {
pr_debug("%s: cannot open %s dir\n", __func__, dir_name);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
return -1;
}
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
while ((dent = readdir(dir)) != NULL) {
char path[PATH_MAX];
struct stat st;
/*sshfs might return bad dent->d_type, so we have to stat*/
snprintf(path, sizeof(path), "%s/%s", dir_name, dent->d_name);
if (stat(path, &st))
continue;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (S_ISDIR(st.st_mode)) {
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (!strcmp(dent->d_name, ".") ||
!strcmp(dent->d_name, ".."))
continue;
ret = map_groups__set_modules_path_dir(mg, path);
if (ret < 0)
goto out;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
} else {
char *dot = strrchr(dent->d_name, '.'),
dso_name[PATH_MAX];
struct map *map;
char *long_name;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (dot == NULL || strcmp(dot, ".ko"))
continue;
snprintf(dso_name, sizeof(dso_name), "[%.*s]",
(int)(dot - dent->d_name), dent->d_name);
strxfrchar(dso_name, '-', '_');
map = map_groups__find_by_name(mg, MAP__FUNCTION,
dso_name);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (map == NULL)
continue;
long_name = strdup(path);
if (long_name == NULL) {
ret = -1;
goto out;
}
dso__set_long_name(map->dso, long_name);
map->dso->lname_alloc = 1;
dso__kernel_module_get_build_id(map->dso, "");
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
}
}
out:
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
closedir(dir);
return ret;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
}
static char *get_kernel_version(const char *root_dir)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
{
char version[PATH_MAX];
FILE *file;
char *name, *tmp;
const char *prefix = "Linux version ";
sprintf(version, "%s/proc/version", root_dir);
file = fopen(version, "r");
if (!file)
return NULL;
version[0] = '\0';
tmp = fgets(version, sizeof(version), file);
fclose(file);
name = strstr(version, prefix);
if (!name)
return NULL;
name += strlen(prefix);
tmp = strchr(name, ' ');
if (tmp)
*tmp = '\0';
return strdup(name);
}
static int machine__set_modules_path(struct machine *machine)
{
char *version;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
char modules_path[PATH_MAX];
version = get_kernel_version(machine->root_dir);
if (!version)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
return -1;
snprintf(modules_path, sizeof(modules_path), "%s/lib/modules/%s/kernel",
machine->root_dir, version);
free(version);
return map_groups__set_modules_path_dir(&machine->kmaps, modules_path);
}
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
/*
* Constructor variant for modules (where we know from /proc/modules where
* they are loaded) and for vmlinux, where only after we load all the
* symbols we'll know where it starts and ends.
*/
static struct map *map__new2(u64 start, struct dso *dso, enum map_type type)
{
struct map *map = calloc(1, (sizeof(*map) +
(dso->kernel ? sizeof(struct kmap) : 0)));
if (map != NULL) {
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
/*
* ->end will be filled after we load all the symbols
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
*/
map__init(map, type, start, 0, 0, dso);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
}
return map;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
}
struct map *machine__new_module(struct machine *machine, u64 start,
const char *filename)
{
struct map *map;
struct dso *dso = __dsos__findnew(&machine->kernel_dsos, filename);
if (dso == NULL)
return NULL;
map = map__new2(start, dso, MAP__FUNCTION);
if (map == NULL)
return NULL;
if (machine__is_host(machine))
dso->symtab_type = DSO_BINARY_TYPE__SYSTEM_PATH_KMODULE;
else
dso->symtab_type = DSO_BINARY_TYPE__GUEST_KMODULE;
map_groups__insert(&machine->kmaps, map);
return map;
}
static int machine__create_modules(struct machine *machine)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
{
char *line = NULL;
size_t n;
FILE *file;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
struct map *map;
const char *modules;
char path[PATH_MAX];
if (machine__is_default_guest(machine))
modules = symbol_conf.default_guest_modules;
else {
sprintf(path, "%s/proc/modules", machine->root_dir);
modules = path;
}
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
if (symbol__restricted_filename(path, "/proc/modules"))
return -1;
file = fopen(modules, "r");
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (file == NULL)
return -1;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
while (!feof(file)) {
char name[PATH_MAX];
u64 start;
char *sep;
int line_len;
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
line_len = getline(&line, &n, file);
if (line_len < 0)
break;
if (!line)
goto out_failure;
line[--line_len] = '\0'; /* \n */
sep = strrchr(line, 'x');
if (sep == NULL)
continue;
hex2u64(sep + 1, &start);
sep = strchr(line, ' ');
if (sep == NULL)
continue;
*sep = '\0';
snprintf(name, sizeof(name), "[%s]", line);
map = machine__new_module(machine, start, name);
if (map == NULL)
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
goto out_delete_line;
dso__kernel_module_get_build_id(map->dso, machine->root_dir);
}
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
free(line);
fclose(file);
return machine__set_modules_path(machine);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
out_delete_line:
free(line);
out_failure:
return -1;
}
int dso__load_vmlinux(struct dso *dso, struct map *map,
const char *vmlinux, symbol_filter_t filter)
{
int err = -1, fd;
char symfs_vmlinux[PATH_MAX];
snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s%s",
symbol_conf.symfs, vmlinux);
fd = open(symfs_vmlinux, O_RDONLY);
if (fd < 0)
return -1;
dso__set_long_name(dso, (char *)vmlinux);
dso__set_loaded(dso, map->type);
err = dso__load_sym(dso, map, symfs_vmlinux, fd, filter, 0, 0);
close(fd);
if (err > 0)
pr_debug("Using %s for symbols\n", symfs_vmlinux);
return err;
}
int dso__load_vmlinux_path(struct dso *dso, struct map *map,
symbol_filter_t filter)
{
int i, err = 0;
char *filename;
pr_debug("Looking at the vmlinux_path (%d entries long)\n",
vmlinux_path__nr_entries + 1);
filename = dso__build_id_filename(dso, NULL, 0);
if (filename != NULL) {
err = dso__load_vmlinux(dso, map, filename, filter);
if (err > 0) {
dso__set_long_name(dso, filename);
goto out;
}
free(filename);
}
for (i = 0; i < vmlinux_path__nr_entries; ++i) {
err = dso__load_vmlinux(dso, map, vmlinux_path[i], filter);
if (err > 0) {
dso__set_long_name(dso, strdup(vmlinux_path[i]));
break;
}
}
out:
return err;
}
static int dso__load_kernel_sym(struct dso *dso, struct map *map,
symbol_filter_t filter)
{
int err;
const char *kallsyms_filename = NULL;
char *kallsyms_allocated_filename = NULL;
/*
* Step 1: if the user specified a kallsyms or vmlinux filename, use
* it and only it, reporting errors to the user if it cannot be used.
*
* For instance, try to analyse an ARM perf.data file _without_ a
* build-id, or if the user specifies the wrong path to the right
* vmlinux file, obviously we can't fallback to another vmlinux (a
* x86_86 one, on the machine where analysis is being performed, say),
* or worse, /proc/kallsyms.
*
* If the specified file _has_ a build-id and there is a build-id
* section in the perf.data file, we will still do the expected
* validation in dso__load_vmlinux and will bail out if they don't
* match.
*/
if (symbol_conf.kallsyms_name != NULL) {
kallsyms_filename = symbol_conf.kallsyms_name;
goto do_kallsyms;
}
if (symbol_conf.vmlinux_name != NULL) {
err = dso__load_vmlinux(dso, map,
symbol_conf.vmlinux_name, filter);
if (err > 0) {
dso__set_long_name(dso,
strdup(symbol_conf.vmlinux_name));
goto out_fixup;
}
return err;
}
if (vmlinux_path != NULL) {
err = dso__load_vmlinux_path(dso, map, filter);
if (err > 0)
goto out_fixup;
}
/* do not try local files if a symfs was given */
if (symbol_conf.symfs[0] != 0)
return -1;
/*
* Say the kernel DSO was created when processing the build-id header table,
* we have a build-id, so check if it is the same as the running kernel,
* using it if it is.
*/
if (dso->has_build_id) {
u8 kallsyms_build_id[BUILD_ID_SIZE];
char sbuild_id[BUILD_ID_SIZE * 2 + 1];
if (sysfs__read_build_id("/sys/kernel/notes", kallsyms_build_id,
sizeof(kallsyms_build_id)) == 0) {
if (dso__build_id_equal(dso, kallsyms_build_id)) {
kallsyms_filename = "/proc/kallsyms";
goto do_kallsyms;
}
}
/*
* Now look if we have it on the build-id cache in
* $HOME/.debug/[kernel.kallsyms].
*/
build_id__sprintf(dso->build_id, sizeof(dso->build_id),
sbuild_id);
if (asprintf(&kallsyms_allocated_filename,
"%s/.debug/[kernel.kallsyms]/%s",
getenv("HOME"), sbuild_id) == -1) {
pr_err("Not enough memory for kallsyms file lookup\n");
return -1;
}
kallsyms_filename = kallsyms_allocated_filename;
if (access(kallsyms_filename, F_OK)) {
pr_err("No kallsyms or vmlinux with build-id %s "
"was found\n", sbuild_id);
free(kallsyms_allocated_filename);
return -1;
}
} else {
/*
* Last resort, if we don't have a build-id and couldn't find
* any vmlinux file, try the running kernel kallsyms table.
*/
kallsyms_filename = "/proc/kallsyms";
}
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
do_kallsyms:
err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
if (err > 0)
pr_debug("Using %s for symbols\n", kallsyms_filename);
free(kallsyms_allocated_filename);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
if (err > 0) {
out_fixup:
if (kallsyms_filename != NULL)
dso__set_long_name(dso, strdup("[kernel.kallsyms]"));
map__fixup_start(map);
map__fixup_end(map);
perf tools: Rewrite and improve support for kernel modules Representing modules as struct map entries, backed by a DSO, etc, using /proc/modules to find where the module is loaded. DSOs now can have a short and long name, so that in verbose mode we can show exactly which .ko or vmlinux image was used. As kernel modules now are a DSO separate from the kernel, we can ask for just the hits for a particular set of kernel modules, just like we can do with shared libraries: [root@doppio linux-2.6-tip]# perf report -n --vmlinux /home/acme/git/build/tip-recvmmsg/vmlinux --modules --dsos \[drm\] | head -15 84.58% 13266 Xorg [k] drm_clflush_pages 4.02% 630 Xorg [k] trace_kmalloc.clone.0 3.95% 619 Xorg [k] drm_ioctl 2.07% 324 Xorg [k] drm_addbufs 1.68% 263 Xorg [k] drm_gem_close_ioctl 0.77% 120 Xorg [k] drm_setmaster_ioctl 0.70% 110 Xorg [k] drm_lastclose 0.68% 106 Xorg [k] drm_open 0.54% 85 Xorg [k] drm_mm_search_free [root@doppio linux-2.6-tip]# Specifying --dsos /lib/modules/2.6.31-tip/kernel/drivers/gpu/drm/drm.ko would have the same effect. Allowing specifying just 'drm.ko' is left for another patch. Processing kallsyms so that per kernel module struct map are instantiated was also left for another patch. That will allow removing the module name from each of its symbols. struct symbol was reduced by removing the ->module backpointer and moving it (well now the map) to struct symbol_entry in perf top, that is its only user right now. The total linecount went down by ~500 lines. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Avi Kivity <avi@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-02 06:29:58 +00:00
}
return err;
}
static int dso__load_guest_kernel_sym(struct dso *dso, struct map *map,
symbol_filter_t filter)
{
int err;
const char *kallsyms_filename = NULL;
struct machine *machine;
char path[PATH_MAX];
if (!map->groups) {
pr_debug("Guest kernel map hasn't the point to groups\n");
return -1;
}
machine = map->groups->machine;
if (machine__is_default_guest(machine)) {
/*
* if the user specified a vmlinux filename, use it and only
* it, reporting errors to the user if it cannot be used.
* Or use file guest_kallsyms inputted by user on commandline
*/
if (symbol_conf.default_guest_vmlinux_name != NULL) {
err = dso__load_vmlinux(dso, map,
symbol_conf.default_guest_vmlinux_name, filter);
goto out_try_fixup;
}
kallsyms_filename = symbol_conf.default_guest_kallsyms;
if (!kallsyms_filename)
return -1;
} else {
sprintf(path, "%s/proc/kallsyms", machine->root_dir);
kallsyms_filename = path;
}
err = dso__load_kallsyms(dso, kallsyms_filename, map, filter);
if (err > 0)
pr_debug("Using %s for symbols\n", kallsyms_filename);
out_try_fixup:
if (err > 0) {
if (kallsyms_filename != NULL) {
machine__mmap_name(machine, path, sizeof(path));
dso__set_long_name(dso, strdup(path));
}
map__fixup_start(map);
map__fixup_end(map);
}
return err;
}
static void dsos__add(struct list_head *head, struct dso *dso)
{
list_add_tail(&dso->node, head);
}
static struct dso *dsos__find(struct list_head *head, const char *name)
{
struct dso *pos;
list_for_each_entry(pos, head, node)
if (strcmp(pos->long_name, name) == 0)
return pos;
return NULL;
}
struct dso *__dsos__findnew(struct list_head *head, const char *name)
{
struct dso *dso = dsos__find(head, name);
if (!dso) {
dso = dso__new(name);
if (dso != NULL) {
dsos__add(head, dso);
dso__set_basename(dso);
}
}
return dso;
}
size_t __dsos__fprintf(struct list_head *head, FILE *fp)
{
struct dso *pos;
size_t ret = 0;
list_for_each_entry(pos, head, node) {
int i;
for (i = 0; i < MAP__NR_TYPES; ++i)
ret += dso__fprintf(pos, i, fp);
}
return ret;
}
size_t machines__fprintf_dsos(struct rb_root *machines, FILE *fp)
{
struct rb_node *nd;
size_t ret = 0;
for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
struct machine *pos = rb_entry(nd, struct machine, rb_node);
ret += __dsos__fprintf(&pos->kernel_dsos, fp);
ret += __dsos__fprintf(&pos->user_dsos, fp);
}
return ret;
}
static size_t __dsos__fprintf_buildid(struct list_head *head, FILE *fp,
bool with_hits)
{
struct dso *pos;
size_t ret = 0;
list_for_each_entry(pos, head, node) {
if (with_hits && !pos->hit)
continue;
ret += dso__fprintf_buildid(pos, fp);
ret += fprintf(fp, " %s\n", pos->long_name);
}
return ret;
}
size_t machine__fprintf_dsos_buildid(struct machine *machine, FILE *fp,
bool with_hits)
{
return __dsos__fprintf_buildid(&machine->kernel_dsos, fp, with_hits) +
__dsos__fprintf_buildid(&machine->user_dsos, fp, with_hits);
}
size_t machines__fprintf_dsos_buildid(struct rb_root *machines,
FILE *fp, bool with_hits)
{
struct rb_node *nd;
size_t ret = 0;
for (nd = rb_first(machines); nd; nd = rb_next(nd)) {
struct machine *pos = rb_entry(nd, struct machine, rb_node);
ret += machine__fprintf_dsos_buildid(pos, fp, with_hits);
}
return ret;
}
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
static struct dso*
dso__kernel_findnew(struct machine *machine, const char *name,
const char *short_name, int dso_type)
{
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
/*
* The kernel dso could be created by build_id processing.
*/
struct dso *dso = __dsos__findnew(&machine->kernel_dsos, name);
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
/*
* We need to run this in all cases, since during the build_id
* processing we had no idea this was the kernel dso.
*/
if (dso != NULL) {
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
dso__set_short_name(dso, short_name);
dso->kernel = dso_type;
}
return dso;
}
void dso__read_running_kernel_build_id(struct dso *dso, struct machine *machine)
{
char path[PATH_MAX];
if (machine__is_default_guest(machine))
return;
sprintf(path, "%s/sys/kernel/notes", machine->root_dir);
if (sysfs__read_build_id(path, dso->build_id,
sizeof(dso->build_id)) == 0)
dso->has_build_id = true;
}
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
static struct dso *machine__get_kernel(struct machine *machine)
{
const char *vmlinux_name = NULL;
struct dso *kernel;
if (machine__is_host(machine)) {
vmlinux_name = symbol_conf.vmlinux_name;
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
if (!vmlinux_name)
vmlinux_name = "[kernel.kallsyms]";
kernel = dso__kernel_findnew(machine, vmlinux_name,
"[kernel]",
DSO_TYPE_KERNEL);
} else {
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
char bf[PATH_MAX];
if (machine__is_default_guest(machine))
vmlinux_name = symbol_conf.default_guest_vmlinux_name;
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
if (!vmlinux_name)
vmlinux_name = machine__mmap_name(machine, bf,
sizeof(bf));
kernel = dso__kernel_findnew(machine, vmlinux_name,
"[guest.kernel]",
DSO_TYPE_GUEST_KERNEL);
}
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
if (kernel != NULL && (!kernel->has_build_id))
dso__read_running_kernel_build_id(kernel, machine);
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
return kernel;
}
struct process_args {
u64 start;
};
static int symbol__in_kernel(void *arg, const char *name,
char type __used, u64 start, u64 end __used)
{
struct process_args *args = arg;
if (strchr(name, '['))
return 0;
args->start = start;
return 1;
}
/* Figure out the start address of kernel map from /proc/kallsyms */
static u64 machine__get_kernel_start_addr(struct machine *machine)
{
const char *filename;
char path[PATH_MAX];
struct process_args args;
if (machine__is_host(machine)) {
filename = "/proc/kallsyms";
} else {
if (machine__is_default_guest(machine))
filename = (char *)symbol_conf.default_guest_kallsyms;
else {
sprintf(path, "%s/proc/kallsyms", machine->root_dir);
filename = path;
}
}
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
if (symbol__restricted_filename(filename, "/proc/kallsyms"))
return 0;
if (kallsyms__parse(filename, &args, symbol__in_kernel) <= 0)
return 0;
return args.start;
}
int __machine__create_kernel_maps(struct machine *machine, struct dso *kernel)
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
{
enum map_type type;
u64 start = machine__get_kernel_start_addr(machine);
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
for (type = 0; type < MAP__NR_TYPES; ++type) {
struct kmap *kmap;
machine->vmlinux_maps[type] = map__new2(start, kernel, type);
if (machine->vmlinux_maps[type] == NULL)
return -1;
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
machine->vmlinux_maps[type]->map_ip =
machine->vmlinux_maps[type]->unmap_ip =
identity__map_ip;
kmap = map__kmap(machine->vmlinux_maps[type]);
kmap->kmaps = &machine->kmaps;
map_groups__insert(&machine->kmaps,
machine->vmlinux_maps[type]);
perf symbols: Add support for 'variable' symtabs Example: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } Added just after symbol__init() call in 'perf top', then: { u64 addr = strtoull(sym_filter, NULL, 16); struct map *map = map_groups__find(kmaps, MAP__VARIABLE, addr); if (map == NULL) pr_err("couldn't find map!\n"); else { struct symbol *sym = map__find_symbol(map, addr, NULL); if (sym == NULL) pr_err("couldn't find addr!\n"); else pr_info("addr %#Lx is in %s global var\n", addr, sym->name); } exit(0); } [root@doppio linux-2.6-tip]# grep ' [dD] ' /proc/kallsyms | grep ' sched' ffffffff817827d8 d sched_nr_latency ffffffff81782ce0 d sched_domains_mutex ffffffff8178c070 d schedstr.22423 ffffffff817909a0 d sched_register_mutex ffffffff81823490 d sched_feat_names ffffffff81823558 d scheduler_running ffffffff818235b8 d sched_clock_running ffffffff818235bc D sched_clock_stable ffffffff81824f00 d sched_switch_trace [root@doppio linux-2.6-tip]# perf top -s 0xffffffff817827d9 addr 0xffffffff817827d9 is in sched_nr_latency global var [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# [root@doppio linux-2.6-tip]# perf top -s ffffffff81782ce0 --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff81782ce0 is in sched_domains_mutex global var [root@doppio linux-2.6-tip]# perf top -s ffffffff818235bc --vmlinux OFF The file OFF cannot be used, trying to use /proc/kallsyms...addr 0xffffffff818235bc is in sched_clock_stable global var [root@doppio linux-2.6-tip]# So it works with both /proc/kallsyms and with ELF symtabs, either the one on the vmlinux explicitely passed via --vmlinux or in one in the vmlinux_path that matches the buildid for the running kernel or the one found in the buildid header section in a perf.data file. Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frédéric Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> LKML-Reference: <1260550239-5372-4-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-11 16:50:39 +00:00
}
return 0;
}
void machine__destroy_kernel_maps(struct machine *machine)
{
enum map_type type;
for (type = 0; type < MAP__NR_TYPES; ++type) {
struct kmap *kmap;
if (machine->vmlinux_maps[type] == NULL)
continue;
kmap = map__kmap(machine->vmlinux_maps[type]);
map_groups__remove(&machine->kmaps,
machine->vmlinux_maps[type]);
if (kmap->ref_reloc_sym) {
/*
* ref_reloc_sym is shared among all maps, so free just
* on one of them.
*/
if (type == MAP__FUNCTION) {
free((char *)kmap->ref_reloc_sym->name);
kmap->ref_reloc_sym->name = NULL;
free(kmap->ref_reloc_sym);
}
kmap->ref_reloc_sym = NULL;
}
map__delete(machine->vmlinux_maps[type]);
machine->vmlinux_maps[type] = NULL;
}
}
int machine__create_kernel_maps(struct machine *machine)
{
perf report: Use properly build_id kernel binaries If we bring the recorded perf data together with kernel binary from another machine using: on server A: perf archive on server B: tar xjvf perf.data.tar.bz2 -C ~/.debug the build_id kernel dso is not properly recognized during the "perf report" command on server B. The reason is, that build_id dsos are added during the session initialization, while the kernel maps are created during the sample event processing. The machine__create_kernel_maps functions ends up creating new dso object for kernel, but it does not check if we already have one added by build_id processing. Also the build_id reading ABI quirk added in commit: - commit b25114817a73bbd2b84ce9dba02ee1ef8989a947 perf build-id: Add quirk to deal with perf.data file format breakage populates the "struct build_id_event::pid" with 0, which is later interpreted as DEFAULT_GUEST_KERNEL_ID. This is not always correct, so it's better to guess the pid value based on the "struct build_id_event::header::misc" value. - Tested with data generated on x86 kernel version v2.6.34 and reported back on x86_64 current kernel. - Not tested for guest kernel case. Note the problem stays for PERF_RECORD_MMAP events recorded by perf that does not use proper pid (HOST_KERNEL_ID/DEFAULT_GUEST_KERNEL_ID). They are misinterpreted within the current perf code. Probably there's not much we can do about that. Cc: Avi Kivity <avi@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20110601194346.GB1934@jolsa.brq.redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-06-01 19:43:46 +00:00
struct dso *kernel = machine__get_kernel(machine);
if (kernel == NULL ||
__machine__create_kernel_maps(machine, kernel) < 0)
return -1;
if (symbol_conf.use_modules && machine__create_modules(machine) < 0) {
if (machine__is_host(machine))
pr_debug("Problems creating module maps, "
"continuing anyway...\n");
else
pr_debug("Problems creating module maps for guest %d, "
"continuing anyway...\n", machine->pid);
}
/*
* Now that we have all the maps created, just set the ->end of them:
*/
map_groups__fixup_end(&machine->kmaps);
return 0;
}
static void vmlinux_path__exit(void)
{
while (--vmlinux_path__nr_entries >= 0) {
free(vmlinux_path[vmlinux_path__nr_entries]);
vmlinux_path[vmlinux_path__nr_entries] = NULL;
}
free(vmlinux_path);
vmlinux_path = NULL;
}
static int vmlinux_path__init(void)
{
struct utsname uts;
char bf[PATH_MAX];
vmlinux_path = malloc(sizeof(char *) * 5);
if (vmlinux_path == NULL)
return -1;
vmlinux_path[vmlinux_path__nr_entries] = strdup("vmlinux");
if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
goto out_fail;
++vmlinux_path__nr_entries;
vmlinux_path[vmlinux_path__nr_entries] = strdup("/boot/vmlinux");
if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
goto out_fail;
++vmlinux_path__nr_entries;
/* only try running kernel version if no symfs was given */
if (symbol_conf.symfs[0] != 0)
return 0;
if (uname(&uts) < 0)
return -1;
snprintf(bf, sizeof(bf), "/boot/vmlinux-%s", uts.release);
vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
goto out_fail;
++vmlinux_path__nr_entries;
snprintf(bf, sizeof(bf), "/lib/modules/%s/build/vmlinux", uts.release);
vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
goto out_fail;
++vmlinux_path__nr_entries;
snprintf(bf, sizeof(bf), "/usr/lib/debug/lib/modules/%s/vmlinux",
uts.release);
vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
goto out_fail;
++vmlinux_path__nr_entries;
return 0;
out_fail:
vmlinux_path__exit();
return -1;
}
size_t machine__fprintf_vmlinux_path(struct machine *machine, FILE *fp)
{
int i;
size_t printed = 0;
struct dso *kdso = machine->vmlinux_maps[MAP__FUNCTION]->dso;
if (kdso->has_build_id) {
char filename[PATH_MAX];
if (dso__build_id_filename(kdso, filename, sizeof(filename)))
printed += fprintf(fp, "[0] %s\n", filename);
}
for (i = 0; i < vmlinux_path__nr_entries; ++i)
printed += fprintf(fp, "[%d] %s\n",
i + kdso->has_build_id, vmlinux_path[i]);
return printed;
}
static int setup_list(struct strlist **list, const char *list_str,
const char *list_name)
{
if (list_str == NULL)
return 0;
*list = strlist__new(true, list_str);
if (!*list) {
pr_err("problems parsing %s list\n", list_name);
return -1;
}
return 0;
}
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
static bool symbol__read_kptr_restrict(void)
{
bool value = false;
if (geteuid() != 0) {
FILE *fp = fopen("/proc/sys/kernel/kptr_restrict", "r");
if (fp != NULL) {
char line[8];
if (fgets(line, sizeof(line), fp) != NULL)
value = atoi(line) != 0;
fclose(fp);
}
}
return value;
}
int symbol__init(void)
{
const char *symfs;
if (symbol_conf.initialized)
return 0;
symbol_conf.priv_size = ALIGN(symbol_conf.priv_size, sizeof(u64));
symbol__elf_init();
if (symbol_conf.sort_by_name)
symbol_conf.priv_size += (sizeof(struct symbol_name_rb_node) -
sizeof(struct symbol));
if (symbol_conf.try_vmlinux_path && vmlinux_path__init() < 0)
return -1;
if (symbol_conf.field_sep && *symbol_conf.field_sep == '.') {
pr_err("'.' is the only non valid --field-separator argument\n");
return -1;
}
if (setup_list(&symbol_conf.dso_list,
symbol_conf.dso_list_str, "dso") < 0)
return -1;
if (setup_list(&symbol_conf.comm_list,
symbol_conf.comm_list_str, "comm") < 0)
goto out_free_dso_list;
if (setup_list(&symbol_conf.sym_list,
symbol_conf.sym_list_str, "symbol") < 0)
goto out_free_comm_list;
/*
* A path to symbols of "/" is identical to ""
* reset here for simplicity.
*/
symfs = realpath(symbol_conf.symfs, NULL);
if (symfs == NULL)
symfs = symbol_conf.symfs;
if (strcmp(symfs, "/") == 0)
symbol_conf.symfs = "";
if (symfs != symbol_conf.symfs)
free((void *)symfs);
perf symbols: Handle /proc/sys/kernel/kptr_restrict Perf uses /proc/modules to figure out where kernel modules are loaded. With the advent of kptr_restrict, non root users get zeroes for all module start addresses. So check if kptr_restrict is non zero and don't generate the syntethic PERF_RECORD_MMAP events for them. Warn the user about it in perf record and in perf report. In perf report the reference relocation symbol being zero means that kptr_restrict was set, thus /proc/kallsyms has only zeroed addresses, so don't use it to fixup symbol addresses when using a valid kallsyms (in the buildid cache) or vmlinux (in the vmlinux path) build-id located automatically or specified by the user. Provide an explanation about it in 'perf report' if kernel samples were taken, checking if a suitable vmlinux or kallsyms was found/specified. Restricted /proc/kallsyms don't go to the buildid cache anymore. Example: [acme@emilia ~]$ perf record -F 100000 sleep 1 WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted, check /proc/sys/kernel/kptr_restrict. Samples in kernel functions may not be resolved if a suitable vmlinux file is not found in the buildid cache or in the vmlinux path. Samples in kernel modules won't be resolved at all. If some relocation was applied (e.g. kexec) symbols may be misresolved even with a suitable vmlinux or kallsyms file. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.005 MB perf.data (~231 samples) ] [acme@emilia ~]$ [acme@emilia ~]$ perf report --stdio Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. If some relocation was applied (e.g. kexec) symbols may be misresolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ..................... # 20.24% sleep [kernel.kallsyms] [k] page_fault 20.04% sleep [kernel.kallsyms] [k] filemap_fault 19.78% sleep [kernel.kallsyms] [k] __lru_cache_add 19.69% sleep ld-2.12.so [.] memcpy 14.71% sleep [kernel.kallsyms] [k] dput 4.70% sleep [kernel.kallsyms] [k] flush_signal_handlers 0.73% sleep [kernel.kallsyms] [k] perf_event_comm 0.11% sleep [kernel.kallsyms] [k] native_write_msr_safe # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ This is because it found a suitable vmlinux (build-id checked) in /lib/modules/2.6.39-rc7+/build/vmlinux (use -v in perf report to see the long file name). If we remove that file from the vmlinux path: [root@emilia ~]# mv /lib/modules/2.6.39-rc7+/build/vmlinux \ /lib/modules/2.6.39-rc7+/build/vmlinux.OFF [acme@emilia ~]$ perf report --stdio [kernel.kallsyms] with build id 57298cdbe0131f6871667ec0eaab4804dcf6f562 not found, continuing without symbols Kernel address maps (/proc/{kallsyms,modules}) were restricted, check /proc/sys/kernel/kptr_restrict before running 'perf record'. As no suitable kallsyms nor vmlinux was found, kernel samples can't be resolved. Samples in kernel modules can't be resolved as well. # Events: 13 cycles # # Overhead Command Shared Object Symbol # ........ ....... ................. ...... # 80.31% sleep [kernel.kallsyms] [k] 0xffffffff8103425a 19.69% sleep ld-2.12.so [.] memcpy # # (For a higher level overview, try: perf report --sort comm,dso) # [acme@emilia ~]$ Reported-by: Stephane Eranian <eranian@google.com> Suggested-by: David Miller <davem@davemloft.net> Cc: Dave Jones <davej@redhat.com> Cc: David Miller <davem@davemloft.net> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Kees Cook <kees.cook@canonical.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Pekka Enberg <penberg@cs.helsinki.fi> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Tom Zanussi <tzanussi@gmail.com> Link: http://lkml.kernel.org/n/tip-mt512joaxxbhhp1odop04yit@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-05-26 12:53:51 +00:00
symbol_conf.kptr_restrict = symbol__read_kptr_restrict();
symbol_conf.initialized = true;
return 0;
out_free_comm_list:
strlist__delete(symbol_conf.comm_list);
out_free_dso_list:
strlist__delete(symbol_conf.dso_list);
return -1;
}
void symbol__exit(void)
{
if (!symbol_conf.initialized)
return;
strlist__delete(symbol_conf.sym_list);
strlist__delete(symbol_conf.dso_list);
strlist__delete(symbol_conf.comm_list);
vmlinux_path__exit();
symbol_conf.sym_list = symbol_conf.dso_list = symbol_conf.comm_list = NULL;
symbol_conf.initialized = false;
}
int machines__create_kernel_maps(struct rb_root *machines, pid_t pid)
{
struct machine *machine = machines__findnew(machines, pid);
if (machine == NULL)
return -1;
return machine__create_kernel_maps(machine);
}
static int hex(char ch)
{
if ((ch >= '0') && (ch <= '9'))
return ch - '0';
if ((ch >= 'a') && (ch <= 'f'))
return ch - 'a' + 10;
if ((ch >= 'A') && (ch <= 'F'))
return ch - 'A' + 10;
return -1;
}
/*
* While we find nice hex chars, build a long_val.
* Return number of chars processed.
*/
int hex2u64(const char *ptr, u64 *long_val)
{
const char *p = ptr;
*long_val = 0;
while (*p) {
const int hex_val = hex(*p);
if (hex_val < 0)
break;
*long_val = (*long_val << 4) | hex_val;
p++;
}
return p - ptr;
}
char *strxfrchar(char *s, char from, char to)
{
char *p = s;
while ((p = strchr(p, from)) != NULL)
*p++ = to;
return s;
}
int machines__create_guest_kernel_maps(struct rb_root *machines)
{
int ret = 0;
struct dirent **namelist = NULL;
int i, items = 0;
char path[PATH_MAX];
pid_t pid;
char *endp;
if (symbol_conf.default_guest_vmlinux_name ||
symbol_conf.default_guest_modules ||
symbol_conf.default_guest_kallsyms) {
machines__create_kernel_maps(machines, DEFAULT_GUEST_KERNEL_ID);
}
if (symbol_conf.guestmount) {
items = scandir(symbol_conf.guestmount, &namelist, NULL, NULL);
if (items <= 0)
return -ENOENT;
for (i = 0; i < items; i++) {
if (!isdigit(namelist[i]->d_name[0])) {
/* Filter out . and .. */
continue;
}
pid = (pid_t)strtol(namelist[i]->d_name, &endp, 10);
if ((*endp != '\0') ||
(endp == namelist[i]->d_name) ||
(errno == ERANGE)) {
pr_debug("invalid directory (%s). Skipping.\n",
namelist[i]->d_name);
continue;
}
sprintf(path, "%s/%s/proc/kallsyms",
symbol_conf.guestmount,
namelist[i]->d_name);
ret = access(path, R_OK);
if (ret) {
pr_debug("Can't access file %s\n", path);
goto failure;
}
machines__create_kernel_maps(machines, pid);
}
failure:
free(namelist);
}
return ret;
}
void machines__destroy_guest_kernel_maps(struct rb_root *machines)
{
struct rb_node *next = rb_first(machines);
while (next) {
struct machine *pos = rb_entry(next, struct machine, rb_node);
next = rb_next(&pos->rb_node);
rb_erase(&pos->rb_node, machines);
machine__delete(pos);
}
}
int machine__load_kallsyms(struct machine *machine, const char *filename,
enum map_type type, symbol_filter_t filter)
{
struct map *map = machine->vmlinux_maps[type];
int ret = dso__load_kallsyms(map->dso, filename, map, filter);
if (ret > 0) {
dso__set_loaded(map->dso, type);
/*
* Since /proc/kallsyms will have multiple sessions for the
* kernel, with modules between them, fixup the end of all
* sections.
*/
__map_groups__fixup_end(&machine->kmaps, type);
}
return ret;
}
int machine__load_vmlinux_path(struct machine *machine, enum map_type type,
symbol_filter_t filter)
{
struct map *map = machine->vmlinux_maps[type];
int ret = dso__load_vmlinux_path(map->dso, map, filter);
if (ret > 0) {
dso__set_loaded(map->dso, type);
map__reloc_vmlinux(map);
}
return ret;
}
struct map *dso__new_map(const char *name)
{
struct map *map = NULL;
struct dso *dso = dso__new(name);
if (dso)
map = map__new2(0, dso, MAP__FUNCTION);
return map;
}
static int open_dso(struct dso *dso, struct machine *machine)
{
char *root_dir = (char *) "";
char *name;
int fd;
name = malloc(PATH_MAX);
if (!name)
return -ENOMEM;
if (machine)
root_dir = machine->root_dir;
if (dso__binary_type_file(dso, dso->data_type,
root_dir, name, PATH_MAX)) {
free(name);
return -EINVAL;
}
fd = open(name, O_RDONLY);
free(name);
return fd;
}
int dso__data_fd(struct dso *dso, struct machine *machine)
{
int i = 0;
if (dso->data_type != DSO_BINARY_TYPE__NOT_FOUND)
return open_dso(dso, machine);
do {
int fd;
dso->data_type = binary_type_data[i++];
fd = open_dso(dso, machine);
if (fd >= 0)
return fd;
} while (dso->data_type != DSO_BINARY_TYPE__NOT_FOUND);
return -EINVAL;
}
static void
dso_cache__free(struct rb_root *root)
{
struct rb_node *next = rb_first(root);
while (next) {
struct dso_cache *cache;
cache = rb_entry(next, struct dso_cache, rb_node);
next = rb_next(&cache->rb_node);
rb_erase(&cache->rb_node, root);
free(cache);
}
}
static struct dso_cache*
dso_cache__find(struct rb_root *root, u64 offset)
{
struct rb_node **p = &root->rb_node;
struct rb_node *parent = NULL;
struct dso_cache *cache;
while (*p != NULL) {
u64 end;
parent = *p;
cache = rb_entry(parent, struct dso_cache, rb_node);
end = cache->offset + DSO__DATA_CACHE_SIZE;
if (offset < cache->offset)
p = &(*p)->rb_left;
else if (offset >= end)
p = &(*p)->rb_right;
else
return cache;
}
return NULL;
}
static void
dso_cache__insert(struct rb_root *root, struct dso_cache *new)
{
struct rb_node **p = &root->rb_node;
struct rb_node *parent = NULL;
struct dso_cache *cache;
u64 offset = new->offset;
while (*p != NULL) {
u64 end;
parent = *p;
cache = rb_entry(parent, struct dso_cache, rb_node);
end = cache->offset + DSO__DATA_CACHE_SIZE;
if (offset < cache->offset)
p = &(*p)->rb_left;
else if (offset >= end)
p = &(*p)->rb_right;
}
rb_link_node(&new->rb_node, parent, p);
rb_insert_color(&new->rb_node, root);
}
static ssize_t
dso_cache__memcpy(struct dso_cache *cache, u64 offset,
u8 *data, u64 size)
{
u64 cache_offset = offset - cache->offset;
u64 cache_size = min(cache->size - cache_offset, size);
memcpy(data, cache->data + cache_offset, cache_size);
return cache_size;
}
static ssize_t
dso_cache__read(struct dso *dso, struct machine *machine,
u64 offset, u8 *data, ssize_t size)
{
struct dso_cache *cache;
ssize_t ret;
int fd;
fd = dso__data_fd(dso, machine);
if (fd < 0)
return -1;
do {
u64 cache_offset;
ret = -ENOMEM;
cache = zalloc(sizeof(*cache) + DSO__DATA_CACHE_SIZE);
if (!cache)
break;
cache_offset = offset & DSO__DATA_CACHE_MASK;
ret = -EINVAL;
if (-1 == lseek(fd, cache_offset, SEEK_SET))
break;
ret = read(fd, cache->data, DSO__DATA_CACHE_SIZE);
if (ret <= 0)
break;
cache->offset = cache_offset;
cache->size = ret;
dso_cache__insert(&dso->cache, cache);
ret = dso_cache__memcpy(cache, offset, data, size);
} while (0);
if (ret <= 0)
free(cache);
close(fd);
return ret;
}
static ssize_t dso_cache_read(struct dso *dso, struct machine *machine,
u64 offset, u8 *data, ssize_t size)
{
struct dso_cache *cache;
cache = dso_cache__find(&dso->cache, offset);
if (cache)
return dso_cache__memcpy(cache, offset, data, size);
else
return dso_cache__read(dso, machine, offset, data, size);
}
ssize_t dso__data_read_offset(struct dso *dso, struct machine *machine,
u64 offset, u8 *data, ssize_t size)
{
ssize_t r = 0;
u8 *p = data;
do {
ssize_t ret;
ret = dso_cache_read(dso, machine, offset, p, size);
if (ret < 0)
return ret;
/* Reached EOF, return what we have. */
if (!ret)
break;
BUG_ON(ret > size);
r += ret;
p += ret;
offset += ret;
size -= ret;
} while (size);
return r;
}
ssize_t dso__data_read_addr(struct dso *dso, struct map *map,
struct machine *machine, u64 addr,
u8 *data, ssize_t size)
{
u64 offset = map->map_ip(map, addr);
return dso__data_read_offset(dso, machine, offset, data, size);
}