Re: [PATCH 2/2] More precise timestamps for nested writes

2015-04-15 Thread Suresh E. Warrier
On 04/15/2015 04:16 AM, Peter Zijlstra wrote: > On Mon, Apr 13, 2015 at 09:38:01PM -0500, Suresh E. Warrier wrote: > >> diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c >> index c9b3005..0a2d862 100644 >> --- a/kernel/trace/ring_buffer.c >> ++

Re: [PATCH 2/2] More precise timestamps for nested writes

2015-04-14 Thread Suresh E. Warrier
On 04/14/2015 12:13 PM, Peter Zijlstra wrote: > On Mon, Apr 13, 2015 at 09:38:01PM -0500, Suresh E. Warrier wrote: >> +static u64 *get_write_timestamp(struct ring_buffer_per_cpu *cpu_buffer, >> +unsigned long *flags) >> +{ >> +if

[PATCH 2/2] More precise timestamps for nested writes

2015-04-13 Thread Suresh E. Warrier
When tracing the behavior of multiple fio jobs running in parallel our performance team observed that some scsi_dispatch_cmd_done events appeared to occur earlier, often several microseconds earlier, than their associated scsi_dispatch_cmd_start event in the trace records. Other interrupt events we

[PATCH 1/2] ring-buffer: Introduce precise nested timestamp sysfs attribute

2015-04-13 Thread Suresh E. Warrier
Introduce a new tunable nested_precise_ts that allows precise time stamps to be recorded for nested trace writes. The tunable is: /sys/kernel/debug/tracing/nested_precise_ts. By default, it is zero and the behavior is disabled. Set the tunable to 1 in order to get accurate timestamps. Signed-off

[PATCH 0/2] ring-buffer: More precise timestamps for nested writes

2015-04-13 Thread Suresh E. Warrier
This patch set provides supports a way to record precise timestamps for nested writes and a tunable to enable the behavior. Nested writes occur when a trace writer is preempted, usually through an interrupt and the interrupt handler also writes its own trace. It has been noticed on large systems w

Re: [PATCH] ring-buffer: More precise time stamps for nested writes

2015-03-26 Thread Suresh E. Warrier
On 03/24/2015 06:19 PM, Steven Rostedt wrote: > On Tue, 24 Mar 2015 18:10:05 -0500 > .. .. > There is no architecture where disabling interrupts is cheap. Actually, > enabling them is the killer. Doing function tracing shows the impact of > this rather well, as it would disable and enable interru

[PATCH] ring-buffer: More precise time stamps for nested writes

2015-03-24 Thread Suresh E. Warrier
When tracing the behavior of multiple fio jobs running in parallel our performance team observed that some scsi_dispatch_cmd_done events appeared to occur earlier, often several microseconds earlier, than their associated scsi_dispatch_cmd_start event in the trace records. Other interrupt events we