Re: why are WB_SYNC_NONE COMMITs being done with FLUSH_SYNC set ?

[Christoph Hellwig <hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>]

On Thu, Aug 19, 2010 at 10:58:25AM -0400, Trond Myklebust wrote:
> To me that sounds fine. I've also been trying to wrap my head around the
> differences between 'nonblocking', 'for_background', 'for_reclaim' and
> 'for_kupdate' and how the filesystem is supposed to treat them.

Yeah, it's not clear to me either.  for_background is in fact only used
in nfs, for the priority and the nfs_commit_inode flags, for_kupdate
is only used in nfs, and in a really weird spot in btrfs, and
for_reclaim is used in nfs, and two places in nilfs2 and in shmemfs.

> Aside from the above, I've used 'for_reclaim', 'for_kupdate' and
> 'for_background' in order to adjust the RPC request's queuing priority
> (high in the case of 'for_reclaim' and low for the other two).

Right now writepage calls to the filesystem can come from various
places:

 - the flusher threads
 - VM reclaim (kswapd, memcg, direct reclaim)
 - memory migration
 - filemap_fdatawrite & other calls directly from FS code, also
   including fsync

We have WB_SYNC_ALL set for the second, data integrity pass when doing
a sync from the flusher threads, and when doing data integrity writes
from fs context (most fsync but also a few others).  All these obviously
are high priority.  It's not too easy to set priorities for the others
in my opinion.

--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html