Re: [PATCH] aio: propogate post-EIOCBQUEUED errors to completion event

From: Benjamin LaHaise <bcrl@kvack.org>
To: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Chris Mason <chris.mason@oracle.com>,
	"Ananiev, Leonid I" <leonid.i.ananiev@intel.com>,
	Zach Brown <zach.brown@oracle.com>,
	linux-aio@kvack.org, linux-kernel@vger.kernel.org,
	Suparna bhattacharya <suparna@in.ibm.com>,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] aio: propogate post-EIOCBQUEUED errors to completion event
Date: Tue, 20 Feb 2007 11:06:15 -0500	[thread overview]
Message-ID: <20070220160615.GT31205@kvack.org> (raw)
In-Reply-To: <1171987310.6271.23.camel@heimdal.trondhjem.org>

On Tue, Feb 20, 2007 at 11:01:50AM -0500, Trond Myklebust wrote:
> We don't try to resolve "conflicting" writes between ordinary mmap() and
> write(), so why should we be doing it for mmap and O_DIRECT?

Yes we do -- both writes will succeed.  More importantly, if one modifies 
the first 512 bytes of a page and the other modifies the last 512 bytes of 
the page, both writes will show up in the completed result.  This is not the 
case with the -EIO bailout.

> mmap() is designed to violate the ordinary mutex locks for write(), so
> if a conflict arises, whether it be with O_DIRECT or ordinary writes
> then it is a case of "last writer wins".

Except that we're not handling cases like the one mentioned above, which 
is quite broken.

		-ben
-- 
"Time is of no importance, Mr. President, only life is important."
Don't Email: <dont@kvack.org>.