On Wed, Oct 03, 2012 at 03:30:07PM +0200, Jiri Olsa wrote: > On Wed, Oct 03, 2012 at 03:22:17PM +0200, Peter Zijlstra wrote: > > On Wed, 2012-10-03 at 15:13 +0200, Jiri Olsa wrote: > > > @@ -1190,8 +1191,8 @@ static inline void perf_sample_data_init(struct > > > perf_sample_data *data, > > > data->raw = NULL; > > > data->br_stack = NULL; > > > data->period = period; > > > - data->regs_user.abi = PERF_SAMPLE_REGS_ABI_NONE; > > > - data->regs_user.regs = NULL; > > > + /* Sets abi to PERF_SAMPLE_REGS_ABI_NONE. */ > > > + memset(&data->regs_user, 0, sizeof(data->regs_user)); > > > data->stack_user_size = 0; > > > } > > > > Hmm, this will slow down all events, regardless of whether they use any > > of that stuff or not. Since the one user actually does something like: > > > > data->regs_user = *pt_regs; > > > > except it does a memcpy() for some obscure reason, it really doesn't > > matter what is in there when uninitialized, right? > > right, the init can stay as it was > > jirka
I made the change and I split it into 2 patches, should be more readable 1/2 perf x86_64: Fix rsp register for system call fast path 2/2 perf: Simplify the sample's user regs/stack retrieval thanks, jirka Cc: Frederic Weisbecker <[email protected]> Cc: Ingo Molnar <[email protected]> Cc: Paul Mackerras <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Arnaldo Carvalho de Melo <[email protected]> Cc: Oleg Nesterov <[email protected]> Cc: Jiri Olsa <[email protected]> --- arch/x86/kernel/cpu/perf_event.c | 37 +++++++++++++++++++ include/linux/perf_event.h | 1 + kernel/events/core.c | 74 ++++++++++++++++++++------------------ 3 files changed, 77 insertions(+), 35 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
