Re: [PATCH] Btrfs: always store the mirror we read the eb from

[Josef Bacik <josef@redhat.com>]

On Wed, Apr 18, 2012 at 12:44:07PM +0200, David Sterba wrote:
> On Mon, Apr 16, 2012 at 09:42:26AM -0400, Josef Bacik wrote:
> > --- a/fs/btrfs/disk-io.c
> > +++ b/fs/btrfs/disk-io.c
> > @@ -390,8 +390,7 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
> >  
> >  		if (!failed_mirror) {
> >  			failed = 1;
> > -			printk(KERN_ERR "failed mirror was %d\n", eb->failed_mirror);
> > -			failed_mirror = eb->failed_mirror;
> > +			failed_mirror = eb->read_mirror;
> >  		}
> >  
> >  		mirror_num++;
> 
> this hunk does not apply for me (on top of master or for-linus), there's
> a different context. In your version the
> 
> 	if (!failed_mirror)
> 
> check is moved after
> 
>  392                 num_copies = btrfs_num_copies(&root->fs_info->mapping_tree,
>  393                                               eb->start, eb->len);
>  394                 if (num_copies == 1)
>  395                         break;
>  396
> 
> which was one of the incremental patches sent for testing, but it's
> needed.
> 
> 

Oh duh sorry I saved that patch as a commit in my tree but didn't actually send
it to the list, I will do that.  Thanks,

Josef