2013-09-25 15:17:59

by Milosz Tanski

Subject: [PATCH] ceph: hung on ceph fscache invalidate in some cases

In some cases I'm on my ceph client cluster I'm seeing hunk kernel tasks in
the invalidate page code path. This is due to the fact that we don't check if
the page is marked as cache before calling fscache_wait_on_page_write().

This is the log from the hang

INFO: task XXXXXX:12034 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
...
Call Trace:
[<ffffffff81568d09>] schedule+0x29/0x70
[<ffffffffa01d4cbd>] __fscache_wait_on_page_write+0x6d/0xb0 [fscache]
[<ffffffff81083520>] ? add_wait_queue+0x60/0x60
[<ffffffffa029a3e9>] ceph_invalidate_fscache_page+0x29/0x50 [ceph]
[<ffffffffa027df00>] ceph_invalidatepage+0x70/0x190 [ceph]
[<ffffffff8112656f>] ? delete_from_page_cache+0x5f/0x70
[<ffffffff81133cab>] truncate_inode_page+0x8b/0x90
[<ffffffff81133ded>] truncate_inode_pages_range.part.12+0x13d/0x620
[<ffffffff8113431d>] truncate_inode_pages_range+0x4d/0x60
[<ffffffff811343b5>] truncate_inode_pages+0x15/0x20
[<ffffffff8119bbf6>] evict+0x1a6/0x1b0
[<ffffffff8119c3f3>] iput+0x103/0x190
...

Signed-off-by: Milosz Tanski <[email protected]>
---
fs/ceph/cache.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c
index 6bfe65e..360b622 100644
--- a/fs/ceph/cache.c
+++ b/fs/ceph/cache.c
@@ -324,6 +324,9 @@ void ceph_invalidate_fscache_page(struct inode* inode, struct page *page)
{
struct ceph_inode_info *ci = ceph_inode(inode);

+ if (!PageFsCache(page))
+ return;
+
fscache_wait_on_page_write(ci->fscache, page);
fscache_uncache_page(ci->fscache, page);
}
--
1.7.9.5

2013-09-25 17:45:52

by Sage Weil

Subject: Re: [PATCH] ceph: hung on ceph fscache invalidate in some cases

Applied this to the ceph-client.git master branch. I'll send this
upstream shortly after it gets some testing and/or any other fixes pile
up.

Thanks!
sage



On Wed, 25 Sep 2013, Milosz Tanski wrote:

> In some cases I'm on my ceph client cluster I'm seeing hunk kernel tasks in
> the invalidate page code path. This is due to the fact that we don't check if
> the page is marked as cache before calling fscache_wait_on_page_write().
>
> This is the log from the hang
>
> INFO: task XXXXXX:12034 blocked for more than 120 seconds.
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> ...
> Call Trace:
> [<ffffffff81568d09>] schedule+0x29/0x70
> [<ffffffffa01d4cbd>] __fscache_wait_on_page_write+0x6d/0xb0 [fscache]
> [<ffffffff81083520>] ? add_wait_queue+0x60/0x60
> [<ffffffffa029a3e9>] ceph_invalidate_fscache_page+0x29/0x50 [ceph]
> [<ffffffffa027df00>] ceph_invalidatepage+0x70/0x190 [ceph]
> [<ffffffff8112656f>] ? delete_from_page_cache+0x5f/0x70
> [<ffffffff81133cab>] truncate_inode_page+0x8b/0x90
> [<ffffffff81133ded>] truncate_inode_pages_range.part.12+0x13d/0x620
> [<ffffffff8113431d>] truncate_inode_pages_range+0x4d/0x60
> [<ffffffff811343b5>] truncate_inode_pages+0x15/0x20
> [<ffffffff8119bbf6>] evict+0x1a6/0x1b0
> [<ffffffff8119c3f3>] iput+0x103/0x190
> ...
>
> Signed-off-by: Milosz Tanski <[email protected]>
> ---
> fs/ceph/cache.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c
> index 6bfe65e..360b622 100644
> --- a/fs/ceph/cache.c
> +++ b/fs/ceph/cache.c
> @@ -324,6 +324,9 @@ void ceph_invalidate_fscache_page(struct inode* inode, struct page *page)
> {
> struct ceph_inode_info *ci = ceph_inode(inode);
>
> + if (!PageFsCache(page))
> + return;
> +
> fscache_wait_on_page_write(ci->fscache, page);
> fscache_uncache_page(ci->fscache, page);
> }
> --
> 1.7.9.5
>
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>