xfs
[Top] [All Lists]

Re: [PATCH 1/2] xfs: remove efi from AIL in log recovery error

To: Brian Foster <bfoster@xxxxxxxxxx>
Subject: Re: [PATCH 1/2] xfs: remove efi from AIL in log recovery error
From: Mark Tinguely <tinguely@xxxxxxx>
Date: Fri, 28 Mar 2014 11:21:28 -0500
Cc: xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <20140328160719.GA15537@xxxxxxxxxxxxxxx>
References: <20140325195733.510384972@xxxxxxx> <20140325195819.638326569@xxxxxxx> <20140328152434.GB21961@xxxxxxxxxxxxxxx> <53359812.3090806@xxxxxxx> <20140328160719.GA15537@xxxxxxxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; FreeBSD amd64; rv:9.0) Gecko/20120122 Thunderbird/9.0
On 03/28/14 11:07, Brian Foster wrote:
On Fri, Mar 28, 2014 at 10:41:06AM -0500, Mark Tinguely wrote:
On 03/28/14 10:24, Brian Foster wrote:
On Tue, Mar 25, 2014 at 03:06:34PM -0500, Mark Tinguely wrote:

...

Hi Mark,

If we hit the scenario where we start skipping EFIs after an error, is
the equivalent unpin() call from process_efi() not necessary on the
subsequent EFIs?

Brian

yes, good catch. They will have to be decremented twice. something like:
+               if (!error)
+                       error = xlog_recover_process_efi(log->l_mp, efip);
+               else
+                       xfs_efi_item_unpin(&efip->efi_item, 0);
+               if (error)
...


Ok, looks reasonable to me. An extra sentence or two in the previous
comment to explain what's going on there would be nice as well. ;)

Brian

Probably will flip the if statement logic, but a comment is also a good idea. Thank-you for the feed back.

--Mark.

<Prev in Thread] Current Thread [Next in Thread>