| To: | Brian Foster <bfoster@xxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH RFC 1/2] xfs: return committed status from xfs_trans_roll() |
| From: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
| Date: | Thu, 30 Jul 2015 10:21:01 -0700 |
| Cc: | Dave Chinner <david@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx |
| Delivered-to: | xfs@xxxxxxxxxxx |
| In-reply-to: | <20150728134006.GD38784@xxxxxxxxxxxxxxx> |
| References: | <1437682410-51778-3-git-send-email-bfoster@xxxxxxxxxx> <1437682410-51778-2-git-send-email-bfoster@xxxxxxxxxx> <20150728004009.GB24249@dastard> <20150728134006.GD38784@xxxxxxxxxxxxxxx> |
| User-agent: | Mutt/1.5.23 (2014-03-12) |
On Tue, Jul 28, 2015 at 09:40:06AM -0400, Brian Foster wrote: > > Hmmm - something I just noticed: if we only have one EFD per EFI, > > why do we do we have that layer of extent counting before dropping > > real references? > > > > I wondered this myself, but hadn't made it deep enough to see if we used > the reference count elsewhere. Because the elders (no pun on Alex, sorry :)) didn't realize we a) don't even ever look at at the logged extents in EFD b) have a 1:1 relationship between EFIs and EFDs I tried to sort some of that out about a year ago, but I didn't manage to get far. have just one. Note that not having to log these |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH 4/6] xfsprogs: avoid use of si_tid in struct xlog_split_item, Christoph Hellwig |
|---|---|
| Next by Date: | Re: [PATCH 03/24] xfs: add refcount btree stats infrastructure, Darrick J. Wong |
| Previous by Thread: | Re: [PATCH RFC 1/2] xfs: return committed status from xfs_trans_roll(), Brian Foster |
| Next by Thread: | [PATCH RFC 2/2] xfs: fix efi/efd error handling to avoid fs shutdown hangs, Brian Foster |
| Indexes: | [Date] [Thread] [Top] [All Lists] |