Messages in this thread |  | | | Date | Thu, 04 Mar 2010 15:54:59 -0500 | | From | Masami Hiramatsu <> | | Subject | Re: [PATCH 09/14] perf, x86: use LBR for PEBS IP+1 fixup |
| |
Peter Zijlstra wrote: > On Thu, 2010-03-04 at 11:21 -0500, Masami Hiramatsu wrote: >> Peter Zijlstra wrote: > >>> +#define perf_misc_flags(regs) \ >>> +({ int misc = 0; \ >>> + if (user_mode(regs)) \ >>> + misc |= PERF_RECORD_MISC_USER; \ >>> + else \ >>> + misc |= PERF_RECORD_MISC_KERNEL; \ >>> + if (regs->flags & PERF_EFLAGS_EXACT) \ >>> + misc |= PERF_RECORD_MISC_EXACT; \ >>> + misc; }) >>> + >>> +#define perf_instruction_pointer(regs) ((regs)->ip) >> >> Hmm, why don't you use instruction_pointer() defined in asm/ptrace.h? >> And I couldn't find any user of this macro in this patch... > > perf_instruction_pointer() is used in kernel/perf_event.c, and yeah I > could have used instruction_pointer() but that's yet another wrapper. > > Anyway, Yanmin is poking at doing kvm-guest profiling and will likely > rewrite all of the perf_misc() and perf_instruction_pointer() stuff > soon.
Hmm, still I can't find where it is used (in your patches). Anyway, would you mean that perf_instruction_pointer() will have different implementation by Yanmin?
Thank you,
-- Masami Hiramatsu e-mail: mhiramat@redhat.com
|  |