[BUG] ext2/3/4: dio reads stale data when we do some append dio writes

[Zheng Liu]

On Tue, Nov 19, 2013 at 11:09:29PM +1100, Dave Chinner wrote:
> On Tue, Nov 19, 2013 at 07:51:22PM +0800, Zheng Liu wrote:
> > On Tue, Nov 19, 2013 at 03:18:26AM -0800, Christoph Hellwig wrote:
> > > On Tue, Nov 19, 2013 at 07:19:47PM +0800, Zheng Liu wrote:
> > > > Yes, I know that XFS has a shared/exclusive lock.  I guess that is why
> > > > it can pass the test.  But another question is why xfs fails when we do
> > > > some append dio writes with doing buffered read.
> > > 
> > > Can you provide a test case for that issue?
> > 
> > Simple.  Reader just need to open this file without O_DIRECT flag.  I
> > paste the full code snippet below.  Please take care of this line:
> > 	readfd = open(argv[1], /*O_DIRECT|*/O_RDONLY, S_IRWXU);
> > 
> > The result of this program on my own sand box looks like below:
> >         encounter an error: offset 0
> ....
> > 		if (ret >= 0) {
> > 			for (j = 0; j < ret; j++) {
> > 				if (rbuf[i] != 'a') {
> > 					fprintf(stderr, "encounter an error: offset %ld\n",
> > 						i);
> > 					goto err;
> 
> Should be checking rbuf[j], perhaps?

Oops, it's my fault.  Yes. it should check rbuf[j].

Thanks,
                                                - Zheng