lkml.org 
[lkml]   [2012]   [Oct]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 15/34] perf, core: Add a concept of a weightened sample
From
Date
On Thu, 2012-10-18 at 16:19 -0700, Andi Kleen wrote:
> @@ -601,6 +602,7 @@ static inline void perf_sample_data_init(struct perf_sample_data *data,
> data->regs_user.abi = PERF_SAMPLE_REGS_ABI_NONE;
> data->regs_user.regs = NULL;
> data->stack_user_size = 0;
> + data->weight = 0;
> }
>
> extern void perf_output_sample(struct perf_output_handle *handle,

> @@ -562,6 +565,7 @@ enum perf_event_type {
> * { u64 stream_id;} && PERF_SAMPLE_STREAM_ID
> * { u32 cpu, res; } && PERF_SAMPLE_CPU
> * { u64 period; } && PERF_SAMPLE_PERIOD
> + * { u64 weight; } && PERF_SAMPLE_WEIGHT
> *
> * { struct read_format values; } && PERF_SAMPLE_READ
> *

So the only issues I have are that his makes every sample more expensive
by having to 0 out that weight data and the sample placement.

I don't think avoiding that weight init is really worth the pain that'll
cause, so we'll just leave it there.

As to the placement, I suppose it makes sense, although Stephane once
complained about these fields not being in PERF_SAMPLE numeric order.
Since we're not that anyway, he'll have to deal with it.. Stephane, any
strong arguments against this placement?

Also, Stephane, you said you had something similar in you LL patches, do
you mean to re-use this or should we re-base this on top of your
patches.. ?


\
 
 \ /
  Last update: 2012-10-23 16:21    [W:0.234 / U:4.476 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site