Messages in this thread | | | Date | Wed, 13 Jan 2010 11:09:26 +0100 | From | Ingo Molnar <> | Subject | Re: [PATCH 0/5] lockdep: Add information of file and line to lockdep_map |
| |
* Peter Zijlstra <peterz@infradead.org> wrote:
> On Thu, 2010-01-07 at 19:39 +0900, Hitoshi Mitake wrote: > > There are a lot of lock instances with same names (e.g. port_lock). > > This patch series add __FILE__ and __LINE__ to lockdep_map, > > and these will be used for trace lock events. > > > > Example use from perf lock map: > > > > | 0xffffea0004c992b8: __pte_lockptr(page) (src: include/linux/mm.h, line: 952) > > | 0xffffea0004b112b8: __pte_lockptr(page) (src: include/linux/mm.h, line: 952) > > | 0xffffea0004a3f2b8: __pte_lockptr(page) (src: include/linux/mm.h, line: 952) > > | 0xffffea0004cd5228: __pte_lockptr(page) (src: include/linux/mm.h, line: 952) > > | 0xffff8800b91e2b28: &sb->s_type->i_lock_key (src: fs/inode.c, line: 166) > > | 0xffff8800bb9d7ae0: key (src: kernel/wait.c, line: 16) > > | 0xffff8800aa07dae0: &dentry->d_lock (src: fs/dcache.c, line: 944) > > | 0xffff8800b07fbae0: &dentry->d_lock (src: fs/dcache.c, line: 944) > > | 0xffff8800b07f3ae0: &dentry->d_lock (src: fs/dcache.c, line: 944) > > | 0xffff8800bf15fae0: &sighand->siglock (src: kernel/fork.c, line: 1490) > > | 0xffff8800b90f7ae0: &dentry->d_lock (src: fs/dcache.c, line: 944) > > | ... > > > > (This output of perf lock map is produced by my local version, > > I'll send this later.) > > > > And sadly, as Peter Zijlstra predicted, this produces certain overhead. > > > > Before appling this series: > > | % sudo ./perf lock rec perf bench sched messaging > > | # Running sched/messaging benchmark... > > | # 20 sender and receiver processes per group > > | # 10 groups == 400 processes run > > | > > | Total time: 3.834 [sec] > > After: > > sudo ./perf lock rec perf bench sched messaging > > | # Running sched/messaging benchmark... > > | # 20 sender and receiver processes per group > > | # 10 groups == 400 processes run > > | > > | Total time: 5.415 [sec] > > | [ perf record: Woken up 0 times to write data ] > > | [ perf record: Captured and wrote 53.512 MB perf.data (~2337993 samples) ] > > > > But raw exec of perf bench sched messaging is this: > > | % perf bench sched messaging > > | # Running sched/messaging benchmark... > > | # 20 sender and receiver processes per group > > | # 10 groups == 400 processes run > > | > > | Total time: 0.498 [sec] > > > > Tracing lock events already produces amount of overhead. > > I think the overhead produced by this series is not a fatal problem, > > radically optimization is required... > > Right, these patches look OK, for the tracing overhead, you could possibly > hash the file:line into a u64 and reduce the tracepoint size, that should > improve the situation I tihnk, because I seem to remember the only thing > that really matters for speed is the size of things.
ok, great. I looked into merging these bits into perf/lock and perf/lock into tip:master - but the recent upstream raw-spinlock changes interact with the new patches.
I also merged latest perf into perf/lock and there's some new build failures:
builtin-lock.c:14:27: error: util/data_map.h: No such file or directory cc1: warnings being treated as errors builtin-lock.c: In function 'process_sample_event': builtin-lock.c:279: error: implicit declaration of function 'threads__findnew' builtin-lock.c:279: error: nested extern declaration of 'threads__findnew' builtin-lock.c:279: error: assignment makes pointer from integer without a cast builtin-lock.c: At top level: builtin-lock.c:357: error: variable 'file_handler' has initializer but incomplete type builtin-lock.c:358: error: unknown field 'process_sample_event' specified in initializer builtin-lock.c:358: error: excess elements in struct initializer builtin-lock.c:358: error: (near initialization for 'file_handler') builtin-lock.c:359: error: unknown field 'sample_type_check' specified in initializer builtin-lock.c:359: error: excess elements in struct initializer builtin-lock.c:359: error: (near initialization for 'file_handler') builtin-lock.c: In function 'read_events': builtin-lock.c:364: error: implicit declaration of function 'register_idle_thread' builtin-lock.c:364: error: nested extern declaration of 'register_idle_thread' builtin-lock.c:365: error: implicit declaration of function 'register_perf_file_handler' builtin-lock.c:365: error: nested extern declaration of 'register_perf_file_handler' builtin-lock.c:367: error: implicit declaration of function 'mmap_dispatch_perf_file' builtin-lock.c:367: error: nested extern declaration of 'mmap_dispatch_perf_file' builtin-lock.c:368: error: 'event__cwdlen' undeclared (first use in this function) builtin-lock.c:368: error: (Each undeclared identifier is reported only once builtin-lock.c:368: error: for each function it appears in.) builtin-lock.c:368: error: 'event__cwd' undeclared (first use in this function) builtin-lock.c: In function 'cmd_lock': builtin-lock.c:429: error: too many arguments to function 'symbol__init' make: *** [builtin-lock.o] Error 1 make: *** Waiting for unfinished jobs....
once those are resolved and we have the merged in patches we can graduate this topic into tip:master.
Thanks,
Ingo
| |