From: venkatesh.pallipadi@intel.com
To: len.brown@intel.com, andi@firstfloor.org
Cc: linux-pm@lists.osdl.org
Subject: [patch 1/3] cpuidle: Do not use poll_idle unless user asks for it
Date: Wed, 30 Jul 2008 19:21:42 -0700 [thread overview]
Message-ID: <20080731022215.751308000@intel.com> (raw)
In-Reply-To: 20080731022141.088307000@intel.com
[-- Attachment #1: not_use_poll_idle_unless_asked_for.patch --]
[-- Type: text/plain, Size: 4025 bytes --]
poll_idle was added to CPUIDLE, just as a low latency idle handler, to be
used in cases when user desires CPUs not to enter any idle state at all. It
was supposed to be a run time idle=poll option to the user. But, it was indeed
getting used during normal menu and ladder governor default case, with no
special user setting (Reported by Linus Torvalds).
Change below ensures that poll_idle will not be used unless user explicitly
asks pm_qos infrastructure for zero latency requirement.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
---
drivers/cpuidle/governors/ladder.c | 14 ++++++++++----
drivers/cpuidle/governors/menu.c | 11 +++++++++--
2 files changed, 19 insertions(+), 6 deletions(-)
Index: linux-2.6/drivers/cpuidle/governors/menu.c
===================================================================
--- linux-2.6.orig/drivers/cpuidle/governors/menu.c 2008-07-28 16:52:04.000000000 -0700
+++ linux-2.6/drivers/cpuidle/governors/menu.c 2008-07-30 19:00:48.000000000 -0700
@@ -34,21 +34,28 @@ static DEFINE_PER_CPU(struct menu_device
static int menu_select(struct cpuidle_device *dev)
{
struct menu_device *data = &__get_cpu_var(menu_devices);
+ int latency_req = pm_qos_requirement(PM_QOS_CPU_DMA_LATENCY);
int i;
+ /* Special case when user has set very strict latency requirement */
+ if (unlikely(latency_req == 0)) {
+ data->last_state_idx = 0;
+ return 0;
+ }
+
/* determine the expected residency time */
data->expected_us =
(u32) ktime_to_ns(tick_nohz_get_sleep_length()) / 1000;
/* find the deepest idle state that satisfies our constraints */
- for (i = 1; i < dev->state_count; i++) {
+ for (i = CPUIDLE_DRIVER_STATE_START + 1; i < dev->state_count; i++) {
struct cpuidle_state *s = &dev->states[i];
if (s->target_residency > data->expected_us)
break;
if (s->target_residency > data->predicted_us)
break;
- if (s->exit_latency > pm_qos_requirement(PM_QOS_CPU_DMA_LATENCY))
+ if (s->exit_latency > latency_req)
break;
}
Index: linux-2.6/drivers/cpuidle/governors/ladder.c
===================================================================
--- linux-2.6.orig/drivers/cpuidle/governors/ladder.c 2008-07-28 16:52:04.000000000 -0700
+++ linux-2.6/drivers/cpuidle/governors/ladder.c 2008-07-30 19:01:28.000000000 -0700
@@ -67,10 +67,17 @@ static int ladder_select_state(struct cp
struct ladder_device *ldev = &__get_cpu_var(ladder_devices);
struct ladder_device_state *last_state;
int last_residency, last_idx = ldev->last_state_idx;
+ int latency_req = pm_qos_requirement(PM_QOS_CPU_DMA_LATENCY);
if (unlikely(!ldev))
return 0;
+ /* Special case when user has set very strict latency requirement */
+ if (unlikely(latency_req == 0)) {
+ ladder_do_selection(ldev, last_idx, 0);
+ return 0;
+ }
+
last_state = &ldev->states[last_idx];
if (dev->states[last_idx].flags & CPUIDLE_FLAG_TIME_VALID)
@@ -81,8 +88,7 @@ static int ladder_select_state(struct cp
/* consider promotion */
if (last_idx < dev->state_count - 1 &&
last_residency > last_state->threshold.promotion_time &&
- dev->states[last_idx + 1].exit_latency <=
- pm_qos_requirement(PM_QOS_CPU_DMA_LATENCY)) {
+ dev->states[last_idx + 1].exit_latency <= latency_req) {
last_state->stats.promotion_count++;
last_state->stats.demotion_count = 0;
if (last_state->stats.promotion_count >= last_state->threshold.promotion_count) {
@@ -92,7 +98,7 @@ static int ladder_select_state(struct cp
}
/* consider demotion */
- if (last_idx > 0 &&
+ if (last_idx > CPUIDLE_DRIVER_STATE_START &&
last_residency < last_state->threshold.demotion_time) {
last_state->stats.demotion_count++;
last_state->stats.promotion_count = 0;
@@ -117,7 +123,7 @@ static int ladder_enable_device(struct c
struct ladder_device_state *lstate;
struct cpuidle_state *state;
- ldev->last_state_idx = 0;
+ ldev->last_state_idx = CPUIDLE_DRIVER_STATE_START;
for (i = 0; i < dev->state_count; i++) {
state = &dev->states[i];
--
next prev parent reply other threads:[~2008-07-31 2:21 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-07-31 2:21 [patch 0/3] cpuilde: Misc bug fixes venkatesh.pallipadi
2008-07-31 2:21 ` venkatesh.pallipadi [this message]
2008-07-31 2:21 ` [patch 2/3] cpuidle: Menu governor fix wrong usage of measured_us venkatesh.pallipadi
2008-07-31 2:21 ` [patch 3/3] cpuidle: Make ladder governor honor latency requirements fully venkatesh.pallipadi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080731022215.751308000@intel.com \
--to=venkatesh.pallipadi@intel.com \
--cc=andi@firstfloor.org \
--cc=len.brown@intel.com \
--cc=linux-pm@lists.osdl.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox