mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "ming.ling" <ming.ling@spreadtrum.com>
To: <labbott@redhat.com>, <sumit.semwal@linaro.org>,
	<gregkh@linuxfoundation.org>, <arve@android.com>,
	<riandrews@android.com>
Cc: <devel@driverdev.osuosl.org>, <linux-kernel@vger.kernel.org>,
	<orson.zhai@spreadtrum.com>, <geng.ren@spreadtrum.com>,
	<chunyan.zhang@spreadtrum.com>, <zhizhou.tian@spreadtrum.com>,
	<yuming.han@spreadtrum.com>, <xiajing@spreadst.com>,
	<mark.yang@spreadtrum.com>, <wenan.hu@spreadtrum.com>,
	Ming Ling <ming.ling@spreadtrum.com>
Subject: [PATCH] staging: android: ion: Make ion_free asynchronous more actively.
Date: Tue, 11 Oct 2016 10:31:53 +0800	[thread overview]
Message-ID: <1476153113-19072-1-git-send-email-ming.ling@spreadtrum.com> (raw)

From: Ming Ling <ming.ling@spreadtrum.com>

So far some ion heaps such as carveout_heap, chunk_heap,
system_heap have freed buffers asynchrounously. Freed buffers
are placed on a free list and freed from a low priority background
thread. If allocations from a particular heap fail, the free list
is drained.

But that mechanism does not work well to free those buffers when
cpu is very busy:
	1.That low priority background thread has no chance to run.
	2.And with helps of kinds of memory reclaim mechanisms,
	allocations from a system_heap never fail, so the free list
	can't be drained on that condition.
	3.Ion_heap_shrink_scan can shrink the free list, but it is
	too slow.

Too many freed buffers in heap's free list waiting to be freed will
lead to low memory killer over-killing innocent processes.

For example:
On my Android Mobile phone(arm32, 2G ram)
lowmemorykiller: Killing 'viders.calendar' (2271), adj 15,
to free 45060kB on behalf of 'kswapd0' (60) because
cache 17208kB is below limit 46080kB for oom_score_adj 0.
Free memory is 3424kB above reserved.

Memory footprint is as below:
--------------------------------------------
MemTotal:        1972164 kB
--------------------------------------------
Normal free:43624kB min:3228kB low:5308kB high:6116kB active_anon:
656kB inactive_anon:1308kB active_file:10600kB inactive_file:6628kB
HighMem free:13208kB min:512kB low:4720kB high:6352kB active_anon:
126188kB inactive_anon:74364kB active_file:325720kB inactive_file:
585832k
--------------------------------------------
ion system heap:
--------------------
	total orphaned		9961472
	total			58986496
	deferred free		630575104
--------------------------------------------

There are about 630575104=601MB memory on ion system heap's free
list waiting to be freed. And it kills more processes, until ion low
priority background thread gets a chance to run as below:
lowmemorykiller: Killing 'd.process.media' (1402), adj 11,
lowmemorykiller: Killing 'dsrt.srtmemtest' (2552), adj 9,
lowmemorykiller: Killing 'rd.engineermode' (2167), adj 8,
lowmemorykiller: Killing 'rd.fileexplorer' (2431), adj 7,
lowmemorykiller: Killing 'd.process.acore' (1499), adj 5,
...

Limit ion_heap_deferred_free thread as a low priority thread doesn't
make any sense. Remove that restriction makes ion_free asynchronous
more actively, and more background applications can survive.

Signed-off-by: Ming Ling <ming.ling@spreadtrum.com>
---
 drivers/staging/android/ion/ion_heap.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/staging/android/ion/ion_heap.c b/drivers/staging/android/ion/ion_heap.c
index 4e5c0f1..d68a80c 100644
--- a/drivers/staging/android/ion/ion_heap.c
+++ b/drivers/staging/android/ion/ion_heap.c
@@ -250,8 +250,6 @@ static int ion_heap_deferred_free(void *data)
 
 int ion_heap_init_deferred_free(struct ion_heap *heap)
 {
-	struct sched_param param = { .sched_priority = 0 };
-
 	INIT_LIST_HEAD(&heap->free_list);
 	init_waitqueue_head(&heap->waitqueue);
 	heap->task = kthread_run(ion_heap_deferred_free, heap,
@@ -261,7 +259,6 @@ int ion_heap_init_deferred_free(struct ion_heap *heap)
 		       __func__);
 		return PTR_ERR_OR_ZERO(heap->task);
 	}
-	sched_setscheduler(heap->task, SCHED_IDLE, &param);
 	return 0;
 }
 
-- 
1.9.1

             reply	other threads:[~2016-10-11  2:32 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-11  2:31 ming.ling [this message]
2016-10-13  0:08 ` Laura Abbott

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=1476153113-19072-1-git-send-email-ming.ling@spreadtrum.com \
    --to=ming.ling@spreadtrum.com \
    --cc=arve@android.com \
    --cc=chunyan.zhang@spreadtrum.com \
    --cc=devel@driverdev.osuosl.org \
    --cc=geng.ren@spreadtrum.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=labbott@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mark.yang@spreadtrum.com \
    --cc=orson.zhai@spreadtrum.com \
    --cc=riandrews@android.com \
    --cc=sumit.semwal@linaro.org \
    --cc=wenan.hu@spreadtrum.com \
    --cc=xiajing@spreadst.com \
    --cc=yuming.han@spreadtrum.com \
    --cc=zhizhou.tian@spreadtrum.com \
    /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

all inboxes | Powered by JetHome®