| To: | Christoph Hellwig <hch@xxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [PATCH 0/5] do not take the iolock in inode reclaim context |
| From: | Sage Weil <sage@xxxxxxxxxxx> |
| Date: | Thu, 5 Jul 2012 17:05:56 -0700 (PDT) |
| Cc: | xfs@xxxxxxxxxxx, sage@xxxxxxxxxxxx |
| In-reply-to: | <20120704151328.928543446@xxxxxxxxxxxxxxxxxxxxxx> |
| References: | <20120704151328.928543446@xxxxxxxxxxxxxxxxxxxxxx> |
Hi Christoph, On Wed, 4 Jul 2012, Christoph Hellwig wrote: > This series should fix the (false-positive) lockdep warnings Sage > has seen while testing ceph workloads with heavy attr usage. I tried this out (minus the last patch, which didn't apply to linus/master) and it does make the warnings go away, thanks! I'll run it through a larger test run to make sure nothing else comes up. > Btw, you probably should create the XFS filesystems for Ceph usage > with large inodes to avoid going out of line for the attributes. Oh, right. The qa stuff is still just doing mkfs.xfs defaults. Thanks! sage |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: 3.5.0-rc5: inconsistent lock state, Christian Kujau |
|---|---|
| Next by Date: | Re: [PATCH] xfs: implement ->update_time, Dave Chinner |
| Previous by Thread: | Re: [PATCH 1/5] xfs: clean up xfs_inactive, Rich Johnston |
| Next by Thread: | Re: [PATCH 0/5] do not take the iolock in inode reclaim context, Sage Weil |
| Indexes: | [Date] [Thread] [Top] [All Lists] |