On 20.10.2016 15:38, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> The bdi flusher should be throttled only depends on
> own bdi and is decoupled with others.
>
> separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
> PGDAT_FILE_WRITEBACK avoid scanning
On 20.10.2016 15:38, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> The bdi flusher should be throttled only depends on
> own bdi and is decoupled with others.
>
> separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
> PGDAT_FILE_WRITEBACK avoid scanning anon lru and it is ok
> then
On Thu 20-10-16 20:38:05, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> The bdi flusher should be throttled only depends on
> own bdi and is decoupled with others.
>
> separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
> PGDAT_FILE_WRITEBACK avoid
On Thu 20-10-16 20:38:05, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> The bdi flusher should be throttled only depends on
> own bdi and is decoupled with others.
>
> separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
> PGDAT_FILE_WRITEBACK avoid scanning anon lru and it is ok
>
From: z00281421
The bdi flusher should be throttled only depends on
own bdi and is decoupled with others.
separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
PGDAT_FILE_WRITEBACK avoid scanning anon lru and it is ok
then throttled on file WRITEBACK.
i think
From: z00281421
The bdi flusher should be throttled only depends on
own bdi and is decoupled with others.
separate PGDAT_WRITEBACK into PGDAT_ANON_WRITEBACK and
PGDAT_FILE_WRITEBACK avoid scanning anon lru and it is ok
then throttled on file WRITEBACK.
i think above may be not right.
On Tue 18-10-16 19:08:05, zhouxianrong wrote:
> Call trace:
> [] __switch_to+0x80/0x98
> [] __schedule+0x314/0x854
> [] schedule+0x48/0xa4
> [] schedule_timeout+0x158/0x2c8
> [] io_schedule_timeout+0xbc/0x14c
> [] wait_iff_congested+0x1d4/0x1ec
> [] shrink_inactive_list+0x530/0x760
> []
On Tue 18-10-16 19:08:05, zhouxianrong wrote:
> Call trace:
> [] __switch_to+0x80/0x98
> [] __schedule+0x314/0x854
> [] schedule+0x48/0xa4
> [] schedule_timeout+0x158/0x2c8
> [] io_schedule_timeout+0xbc/0x14c
> [] wait_iff_congested+0x1d4/0x1ec
> [] shrink_inactive_list+0x530/0x760
> []
Call trace:
[] __switch_to+0x80/0x98
[] __schedule+0x314/0x854
[] schedule+0x48/0xa4
[] schedule_timeout+0x158/0x2c8
[] io_schedule_timeout+0xbc/0x14c
[] wait_iff_congested+0x1d4/0x1ec
[] shrink_inactive_list+0x530/0x760
[] shrink_lruvec+0x534/0x76c
[] shrink_zone+0x88/0x1b8
[]
Call trace:
[] __switch_to+0x80/0x98
[] __schedule+0x314/0x854
[] schedule+0x48/0xa4
[] schedule_timeout+0x158/0x2c8
[] io_schedule_timeout+0xbc/0x14c
[] wait_iff_congested+0x1d4/0x1ec
[] shrink_inactive_list+0x530/0x760
[] shrink_lruvec+0x534/0x76c
[] shrink_zone+0x88/0x1b8
[]
On Tue, Oct 18, 2016 at 03:12:45PM +0800, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> bdi flusher may enter page alloc slow path due to writepage and kmalloc.
> in that case the flusher as a direct reclaimer should not be throttled here
> because it can
On Tue, Oct 18, 2016 at 03:12:45PM +0800, zhouxianr...@huawei.com wrote:
> From: z00281421
>
> bdi flusher may enter page alloc slow path due to writepage and kmalloc.
> in that case the flusher as a direct reclaimer should not be throttled here
> because it can not to reclaim clean file pages
> @@ -1908,7 +1908,7 @@ void wb_workfn(struct work_struct *work)
> long pages_written;
>
> set_worker_desc("flush-%s", dev_name(wb->bdi->dev));
> - current->flags |= PF_SWAPWRITE;
If flags carries PF_LESS_THROTTLE before modified, then you
have to restore it.
> +
> @@ -1908,7 +1908,7 @@ void wb_workfn(struct work_struct *work)
> long pages_written;
>
> set_worker_desc("flush-%s", dev_name(wb->bdi->dev));
> - current->flags |= PF_SWAPWRITE;
If flags carries PF_LESS_THROTTLE before modified, then you
have to restore it.
> +
From: z00281421
bdi flusher may enter page alloc slow path due to writepage and kmalloc.
in that case the flusher as a direct reclaimer should not be throttled here
because it can not to reclaim clean file pages or anaonymous pages
for next moment; furthermore
From: z00281421
bdi flusher may enter page alloc slow path due to writepage and kmalloc.
in that case the flusher as a direct reclaimer should not be throttled here
because it can not to reclaim clean file pages or anaonymous pages
for next moment; furthermore writeback rate of dirty pages
16 matches
Mail list logo