Age | Commit message (Collapse) | Author |
|
Here is formatted version of the patch:
From 23362fa0f7bbfd0497a8ccbb9d06a4520eae8057 Mon Sep 17 00:00:00 2001
From: Igor Zhbanov<i.zhbanov@samsung.com>
Date: Wed, 11 Jul 2012 12:30:27 +0400
Subject: [PATCH] Wrong common field name in GFX Wakeups event
I have found that PowerTOP doesn't displays values in "GFX Wakeups/s" column.
It is because of wrong event field name the handle_trace_point() method
expects to find in a trace event.
|
|
|
|
Various fixes to pass W3C HTML validation. Also unified
whitespaces - preferred tabs over multiple spaces to save
some space.
|
|
|
|
I apologies for numerous grammar mistakes in V1, those are the reason for
re-sending.
| commit 8580d679ede86a1c91f5925b6971a094cea6f832
| Author: Chris E Ferron <chris.e.ferron@linux.intel.com>
| Date: Mon Apr 30 11:38:32 2012 -0700
|
| updated remaining string fields to insure descriptions are valid.
|
extended pevent field offset calculation to sched events.
It turned out to be unnecessary, since offset calculation and 0xffff alignment
required only for DYNAMIC fields -- field type starting with __data_loc.
irq_handler_entry event is defined with __string name, while sched_switch event
is defined with __array char next_comm TASK_COMM_LEN field.
Function get_pevent_field_str() checks field->flags and return `char *'
with proper offset calculated.
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
|
|
|
|
Introduce dbg_printf_pevent_info() for event debugging simplification.
Events are printed to STDERR, one event per line.
(pevent trace_seq_do_printf() uses STDOUT, making it difficult to read through
mixed powertop and trace_seq outputs).
$ powertop 2>events
$ less events
prev_comm=ksoftirqd/0 prev_pid=3 prev_prio=120 prev_state=0x1 next_comm=swapper/0 next_pid=0 next_prio=120
prev_comm=swapper/0 prev_pid=0 prev_prio=120 prev_state=0x0 next_comm=ksoftirqd/0 next_pid=3 next_prio=120
prev_comm=ksoftirqd/0 prev_pid=3 prev_prio=120 prev_state=0x1 next_comm=swapper/0 next_pid=0 next_prio=120
prev_comm=swapper/0 prev_pid=0 prev_prio=120 prev_state=0x0 next_comm=X next_pid=2795 next_prio=119
[..]
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
|
|
|commit 2a86bf81403f54acd213c4f326ab214b372fe061
|Author: Arjan van de Ven <arjan@linux.intel.com>
|Date: Fri Apr 27 16:37:40 2012 -0400
|
| this is a hack, and broken, but it seems the trace event infrastructure throws us a curve ball here
didn't solve the problem for me, as I still obverse some interrupts with empty descriptions.
Patch (code extracted from pevent print_event_fields()) seem to fix the problem.
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
|
|
throws us a curve ball here
|
|
fix up a few gcc warnings
|
|
|
|
|