perf tools: Fix sparse CPU numbering related bugs
At present, the perf subcommands that do system-wide monitoring
(perf stat, perf record and perf top) don't work properly unless
the online cpus are numbered 0, 1, ..., N-1. These tools ask
for the number of online cpus with sysconf(_SC_NPROCESSORS_ONLN)
and then try to create events for cpus 0, 1, ..., N-1.
This creates problems for systems where the online cpus are
numbered sparsely. For example, a POWER6 system in
single-threaded mode (i.e. only running 1 hardware thread per
core) will have only even-numbered cpus online.
This fixes the problem by reading the /sys/devices/system/cpu/online
file to find out which cpus are online. The code that does that is in
tools/perf/util/cpumap.[ch], and consists of a read_cpu_map()
function that sets up a cpumap[] array and returns the number of
online cpus. If /sys/devices/system/cpu/online can't be read or
can't be parsed successfully, it falls back to using sysconf to
ask how many cpus are online and sets up an identity map in cpumap[].
The perf record, perf stat and perf top code then calls
read_cpu_map() in the system-wide monitoring case (instead of
sysconf) and uses cpumap[] to get the cpu numbers to pass to
perf_event_open.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Anton Blanchard <anton@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
LKML-Reference: <20100310093609.GA3959@brick.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-10 12:36:09 +03:00
# ifndef __PERF_CPUMAP_H
# define __PERF_CPUMAP_H
2012-01-19 20:07:23 +04:00
# include <stdio.h>
2012-09-26 19:41:14 +04:00
# include <stdbool.h>
2015-06-23 01:36:04 +03:00
# include <linux/atomic.h>
2012-01-19 20:07:23 +04:00
2014-04-07 22:55:21 +04:00
# include "perf.h"
# include "util/debug.h"
2011-01-03 22:49:48 +03:00
struct cpu_map {
2015-06-23 01:36:04 +03:00
atomic_t refcnt ;
2011-01-03 22:49:48 +03:00
int nr ;
int map [ ] ;
} ;
struct cpu_map * cpu_map__new ( const char * cpu_list ) ;
struct cpu_map * cpu_map__dummy_new ( void ) ;
2012-09-10 11:53:50 +04:00
struct cpu_map * cpu_map__read ( FILE * file ) ;
2012-01-19 20:07:23 +04:00
size_t cpu_map__fprintf ( struct cpu_map * map , FILE * fp ) ;
2013-02-06 18:46:01 +04:00
int cpu_map__get_socket ( struct cpu_map * map , int idx ) ;
2013-02-14 16:57:29 +04:00
int cpu_map__get_core ( struct cpu_map * map , int idx ) ;
2013-02-06 18:46:01 +04:00
int cpu_map__build_socket_map ( struct cpu_map * cpus , struct cpu_map * * sockp ) ;
2013-02-14 16:57:29 +04:00
int cpu_map__build_core_map ( struct cpu_map * cpus , struct cpu_map * * corep ) ;
2013-02-06 18:46:01 +04:00
2015-06-23 01:36:04 +03:00
struct cpu_map * cpu_map__get ( struct cpu_map * map ) ;
void cpu_map__put ( struct cpu_map * map ) ;
2013-02-06 18:46:01 +04:00
static inline int cpu_map__socket ( struct cpu_map * sock , int s )
{
if ( ! sock | | s > sock - > nr | | s < 0 )
return 0 ;
return sock - > map [ s ] ;
}
2012-01-19 20:07:23 +04:00
2013-02-14 16:57:29 +04:00
static inline int cpu_map__id_to_socket ( int id )
{
return id > > 16 ;
}
static inline int cpu_map__id_to_cpu ( int id )
{
return id & 0xffff ;
}
2012-09-26 19:41:14 +04:00
static inline int cpu_map__nr ( const struct cpu_map * map )
{
return map ? map - > nr : 1 ;
}
2013-05-23 04:42:38 +04:00
static inline bool cpu_map__empty ( const struct cpu_map * map )
2012-09-26 19:41:14 +04:00
{
return map ? map - > map [ 0 ] = = - 1 : true ;
}
2014-04-07 22:55:21 +04:00
int max_cpu_num ;
int max_node_num ;
int * cpunode_map ;
int cpu__setup_cpunode_map ( void ) ;
static inline int cpu__max_node ( void )
{
if ( unlikely ( ! max_node_num ) )
pr_debug ( " cpu_map not initialized \n " ) ;
return max_node_num ;
}
static inline int cpu__max_cpu ( void )
{
if ( unlikely ( ! max_cpu_num ) )
pr_debug ( " cpu_map not initialized \n " ) ;
return max_cpu_num ;
}
static inline int cpu__get_node ( int cpu )
{
if ( unlikely ( cpunode_map = = NULL ) ) {
pr_debug ( " cpu_map not initialized \n " ) ;
return - 1 ;
}
return cpunode_map [ cpu ] ;
}
perf tools: Fix sparse CPU numbering related bugs
At present, the perf subcommands that do system-wide monitoring
(perf stat, perf record and perf top) don't work properly unless
the online cpus are numbered 0, 1, ..., N-1. These tools ask
for the number of online cpus with sysconf(_SC_NPROCESSORS_ONLN)
and then try to create events for cpus 0, 1, ..., N-1.
This creates problems for systems where the online cpus are
numbered sparsely. For example, a POWER6 system in
single-threaded mode (i.e. only running 1 hardware thread per
core) will have only even-numbered cpus online.
This fixes the problem by reading the /sys/devices/system/cpu/online
file to find out which cpus are online. The code that does that is in
tools/perf/util/cpumap.[ch], and consists of a read_cpu_map()
function that sets up a cpumap[] array and returns the number of
online cpus. If /sys/devices/system/cpu/online can't be read or
can't be parsed successfully, it falls back to using sysconf to
ask how many cpus are online and sets up an identity map in cpumap[].
The perf record, perf stat and perf top code then calls
read_cpu_map() in the system-wide monitoring case (instead of
sysconf) and uses cpumap[] to get the cpu numbers to pass to
perf_event_open.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Anton Blanchard <anton@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
LKML-Reference: <20100310093609.GA3959@brick.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2010-03-10 12:36:09 +03:00
# endif /* __PERF_CPUMAP_H */