| To: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH 2/2] remove ancient sys_getdents code paths |
| From: | Kyle McMartin <kmcmarti@xxxxxxxxxx> |
| Date: | Tue, 25 Mar 2014 12:08:08 -0400 |
| Cc: | Brian Foster <bfoster@xxxxxxxxxx>, xfs@xxxxxxxxxxx |
| Delivered-to: | xfs@xxxxxxxxxxx |
| In-reply-to: | <20140325153422.GA31551@xxxxxxxxxxxxx> |
| References: | <20140324181920.GG23291@xxxxxxxxxxxxxxxxxxxxxxx> <20140325141553.GB18691@xxxxxxxxxxxxxxx> <20140325153422.GA31551@xxxxxxxxxxxxx> |
| User-agent: | Mutt/1.5.21 (2010-09-15) |
On Tue, Mar 25, 2014 at 08:34:22AM -0700, Christoph Hellwig wrote: > On Tue, Mar 25, 2014 at 10:15:55AM -0400, Brian Foster wrote: > > I wonder if we could clean up the d_name offset checks to check it once > > and use a flag. Or use a signed type for size_diff and trigger off that. > > Just a thought, fwiw. > > I'd just leave the code as-is and just remove the ifdef cruft. If > anyone is motivated enough to do real work on this area the callers > should simply be switched to readdir(3) and friends instead of > bikeshedding the current cruft. I just don't want to put even more load > on Kyle how just wants to get xfsdump running on arm64.. > Heh, actually, I started doing exactly that last night, after looking at all the getdents_wrap callers... That's probably a bit more... error-sensitive patch though, so I'd prefer if at least 1/2 and/or 2/2 went in separately for debuggability. --Kyle |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: [PATCH] xfs: fix buffer use after free on IO error, Eric Sandeen |
|---|---|
| Next by Date: | Re: [PATCH] xfs: fix buffer use after free on IO error, Christoph Hellwig |
| Previous by Thread: | Re: [PATCH 2/2] remove ancient sys_getdents code paths, Christoph Hellwig |
| Next by Thread: | [PATCH 0/2] xfs: Clean the EFI on errors series, Mark Tinguely |
| Indexes: | [Date] [Thread] [Top] [All Lists] |