linux/tools/perf/util/map.h

218 lines
6.0 KiB
C
Raw Normal View History

#ifndef __PERF_MAP_H
#define __PERF_MAP_H
#include <linux/compiler.h>
#include <linux/list.h>
#include <linux/rbtree.h>
#include <stdio.h>
#include <stdbool.h>
#include "types.h"
enum map_type {
MAP__FUNCTION = 0,
MAP__VARIABLE,
};
#define MAP__NR_TYPES (MAP__VARIABLE + 1)
extern const char *map_type__name[MAP__NR_TYPES];
struct dso;
struct ref_reloc_sym;
struct map_groups;
struct machine;
struct map {
union {
struct rb_node rb_node;
struct list_head node;
};
u64 start;
u64 end;
enum map_type type;
u32 priv;
u64 pgoff;
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
/* ip -> dso rip */
u64 (*map_ip)(struct map *, u64);
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
/* dso rip -> ip */
u64 (*unmap_ip)(struct map *, u64);
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
struct dso *dso;
struct map_groups *groups;
};
struct kmap {
struct ref_reloc_sym *ref_reloc_sym;
struct map_groups *kmaps;
};
struct map_groups {
struct rb_root maps[MAP__NR_TYPES];
struct list_head removed_maps[MAP__NR_TYPES];
struct machine *machine;
};
/* Native host kernel uses -1 as pid index in machine */
#define HOST_KERNEL_ID (-1)
#define DEFAULT_GUEST_KERNEL_ID (0)
struct machine {
struct rb_node rb_node;
pid_t pid;
char *root_dir;
struct list_head user_dsos;
struct list_head kernel_dsos;
struct map_groups kmaps;
struct map *vmlinux_maps[MAP__NR_TYPES];
};
static inline
struct map *machine__kernel_map(struct machine *self, enum map_type type)
{
return self->vmlinux_maps[type];
}
static inline struct kmap *map__kmap(struct map *self)
{
return (struct kmap *)(self + 1);
}
static inline u64 map__map_ip(struct map *map, u64 ip)
{
return ip - map->start + map->pgoff;
}
static inline u64 map__unmap_ip(struct map *map, u64 ip)
{
return ip + map->start - map->pgoff;
}
static inline u64 identity__map_ip(struct map *map __used, u64 ip)
{
return ip;
}
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
perf top: Fix annotate for userspace First, for programs and prelinked libraries, annotate code was fooled by objdump output IPs (src->eip in the code) being wrongly converted to absolute IPs. In such case there were no conversion needed, but in src->eip = strtoull(src->line, NULL, 16); src->eip = map->unmap_ip(map, src->eip); // = eip + map->start - map->pgoff we were reading absolute address from objdump (e.g. 8048604) and then almost doubling it, because eip & map->start are approximately close for small programs. Needless to say, that later, in record_precise_ip() there was no matching with real runtime IPs. And second, like with `perf annotate` the problem with non-prelinked *.so was that we were doing rip -> objdump address conversion wrong. Also, because unlike `perf annotate`, `perf top` code does annotation based on absolute IPs for performance reasons(*), new helper for mapping objdump addresse to IP is introduced. (*) we get samples info in absolute IPs, and since we do lots of hit-testing on absolute IPs at runtime in record_precise_ip(), it's better to convert objdump addresses to IPs once and do no conversion at runtime. I also had to fix how objdump output is parsed (with hardcoded 8/16 characters format, which was inappropriate for ET_DYN dsos with small addresses like '4ac') Also note, that not all objdump output lines has associtated IPs, e.g. look at source lines here: 000004ac <my_strlen>: extern "C" int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { int len = 0; 4b2: c7 45 fc 00 00 00 00 movl $0x0,-0x4(%ebp) 4b9: eb 08 jmp 4c3 <my_strlen+0x17> while (*s) { ++len; 4bb: 83 45 fc 01 addl $0x1,-0x4(%ebp) ++s; 4bf: 83 45 08 01 addl $0x1,0x8(%ebp) So we mark them with eip=0, and ignore such lines in annotate lookup code. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> [ Note: one hunk of this patch was applied by Mike in 57d8188 ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265550376-12665-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-07 21:46:15 +08:00
/* rip/ip <-> addr suitable for passing to `objdump --start-address=` */
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
u64 map__rip_2objdump(struct map *map, u64 rip);
perf top: Fix annotate for userspace First, for programs and prelinked libraries, annotate code was fooled by objdump output IPs (src->eip in the code) being wrongly converted to absolute IPs. In such case there were no conversion needed, but in src->eip = strtoull(src->line, NULL, 16); src->eip = map->unmap_ip(map, src->eip); // = eip + map->start - map->pgoff we were reading absolute address from objdump (e.g. 8048604) and then almost doubling it, because eip & map->start are approximately close for small programs. Needless to say, that later, in record_precise_ip() there was no matching with real runtime IPs. And second, like with `perf annotate` the problem with non-prelinked *.so was that we were doing rip -> objdump address conversion wrong. Also, because unlike `perf annotate`, `perf top` code does annotation based on absolute IPs for performance reasons(*), new helper for mapping objdump addresse to IP is introduced. (*) we get samples info in absolute IPs, and since we do lots of hit-testing on absolute IPs at runtime in record_precise_ip(), it's better to convert objdump addresses to IPs once and do no conversion at runtime. I also had to fix how objdump output is parsed (with hardcoded 8/16 characters format, which was inappropriate for ET_DYN dsos with small addresses like '4ac') Also note, that not all objdump output lines has associtated IPs, e.g. look at source lines here: 000004ac <my_strlen>: extern "C" int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { int len = 0; 4b2: c7 45 fc 00 00 00 00 movl $0x0,-0x4(%ebp) 4b9: eb 08 jmp 4c3 <my_strlen+0x17> while (*s) { ++len; 4bb: 83 45 fc 01 addl $0x1,-0x4(%ebp) ++s; 4bf: 83 45 08 01 addl $0x1,0x8(%ebp) So we mark them with eip=0, and ignore such lines in annotate lookup code. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> [ Note: one hunk of this patch was applied by Mike in 57d8188 ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265550376-12665-1-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-07 21:46:15 +08:00
u64 map__objdump_2ip(struct map *map, u64 addr);
perf annotate: Fix it for non-prelinked *.so The problem was we were incorrectly calculating objdump addresses for sym->start and sym->end, look: For simple ET_DYN type DSO (*.so) with one function, objdump -dS output is something like this: 000004ac <my_strlen>: int my_strlen(const char *s) 4ac: 55 push %ebp 4ad: 89 e5 mov %esp,%ebp 4af: 83 ec 10 sub $0x10,%esp { i.e. we have relative-to-dso-mapping IPs (=RIP) there. For ET_EXEC type and probably for prelinked libs as well (sorry can't test - I don't use prelink) objdump outputs absolute IPs, e.g. 08048604 <zz_strlen>: extern "C" int zz_strlen(const char *s) 8048604: 55 push %ebp 8048605: 89 e5 mov %esp,%ebp 8048607: 83 ec 10 sub $0x10,%esp { So, if sym->start is always relative to dso mapping(*), we'll have to unmap it for ET_EXEC like cases, and leave as is for ET_DYN cases. (*) and it is - we've explicitely made it relative. Look for adjust_symbols handling in dso__load_sym() Previously we were always unmapping sym->start and for ET_DYN dsos resulting addresses were wrong, and so objdump output was empty. The end result was that perf annotate output for symbols from non-prelinked *.so had always 0.00% percents only, which is wrong. To fix it, let's introduce a helper for converting rip to objdump address, and also let's document what map_ip() and unmap_ip() do -- I had to study sources for several hours to understand it. Signed-off-by: Kirill Smelkov <kirr@landau.phys.spbu.ru> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Mike Galbraith <efault@gmx.de> LKML-Reference: <1265223128-11786-8-git-send-email-acme@infradead.org> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-02-04 02:52:07 +08:00
struct symbol;
typedef int (*symbol_filter_t)(struct map *map, struct symbol *sym);
void map__init(struct map *self, enum map_type type,
u64 start, u64 end, u64 pgoff, struct dso *dso);
struct map *map__new(struct list_head *dsos__list, u64 start, u64 len,
u64 pgoff, u32 pid, char *filename,
enum map_type type, char *cwd, int cwdlen);
void map__delete(struct map *self);
struct map *map__clone(struct map *self);
int map__overlap(struct map *l, struct map *r);
size_t map__fprintf(struct map *self, FILE *fp);
int map__load(struct map *self, symbol_filter_t filter);
struct symbol *map__find_symbol(struct map *self,
u64 addr, symbol_filter_t filter);
struct symbol *map__find_symbol_by_name(struct map *self, const char *name,
symbol_filter_t filter);
void map__fixup_start(struct map *self);
void map__fixup_end(struct map *self);
void map__reloc_vmlinux(struct map *self);
size_t __map_groups__fprintf_maps(struct map_groups *self,
enum map_type type, int verbose, FILE *fp);
void maps__insert(struct rb_root *maps, struct map *map);
struct map *maps__find(struct rb_root *maps, u64 addr);
void map_groups__init(struct map_groups *self);
int map_groups__clone(struct map_groups *self,
struct map_groups *parent, enum map_type type);
size_t map_groups__fprintf(struct map_groups *self, int verbose, FILE *fp);
size_t map_groups__fprintf_maps(struct map_groups *self, int verbose, FILE *fp);
typedef void (*machine__process_t)(struct machine *self, void *data);
void machines__process(struct rb_root *self, machine__process_t process, void *data);
struct machine *machines__add(struct rb_root *self, pid_t pid,
const char *root_dir);
struct machine *machines__find_host(struct rb_root *self);
struct machine *machines__find(struct rb_root *self, pid_t pid);
struct machine *machines__findnew(struct rb_root *self, pid_t pid);
char *machine__mmap_name(struct machine *self, char *bf, size_t size);
int machine__init(struct machine *self, const char *root_dir, pid_t pid);
/*
* Default guest kernel is defined by parameter --guestkallsyms
* and --guestmodules
*/
static inline bool machine__is_default_guest(struct machine *self)
{
return self ? self->pid == DEFAULT_GUEST_KERNEL_ID : false;
}
static inline bool machine__is_host(struct machine *self)
{
return self ? self->pid == HOST_KERNEL_ID : false;
}
static inline void map_groups__insert(struct map_groups *self, struct map *map)
{
maps__insert(&self->maps[map->type], map);
map->groups = self;
}
static inline struct map *map_groups__find(struct map_groups *self,
enum map_type type, u64 addr)
{
return maps__find(&self->maps[type], addr);
}
struct symbol *map_groups__find_symbol(struct map_groups *self,
enum map_type type, u64 addr,
struct map **mapp,
symbol_filter_t filter);
struct symbol *map_groups__find_symbol_by_name(struct map_groups *self,
enum map_type type,
const char *name,
struct map **mapp,
symbol_filter_t filter);
static inline
struct symbol *machine__find_kernel_symbol(struct machine *self,
enum map_type type, u64 addr,
struct map **mapp,
symbol_filter_t filter)
{
return map_groups__find_symbol(&self->kmaps, type, addr, mapp, filter);
}
static inline
struct symbol *machine__find_kernel_function(struct machine *self, u64 addr,
struct map **mapp,
symbol_filter_t filter)
{
return machine__find_kernel_symbol(self, MAP__FUNCTION, addr, mapp, filter);
}
static inline
struct symbol *map_groups__find_function_by_name(struct map_groups *self,
const char *name, struct map **mapp,
symbol_filter_t filter)
{
return map_groups__find_symbol_by_name(self, MAP__FUNCTION, name, mapp, filter);
}
int map_groups__fixup_overlappings(struct map_groups *self, struct map *map,
int verbose, FILE *fp);
struct map *map_groups__find_by_name(struct map_groups *self,
enum map_type type, const char *name);
struct map *machine__new_module(struct machine *self, u64 start, const char *filename);
void map_groups__flush(struct map_groups *self);
#endif /* __PERF_MAP_H */