| To: | Bob Gilligan <gilligan@xxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [linux-iscsi-devel] Crash running XFS over iSCSI volume |
| From: | Christoph Hellwig <hch@xxxxxx> |
| Date: | Tue, 24 Aug 2004 12:38:24 +0200 |
| Cc: | linux-iscsi-devel@xxxxxxxxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, linux-xfs@xxxxxxxxxxx |
| In-reply-to: | <412A40D6.3070408@intransa.com> |
| Mail-followup-to: | Christoph Hellwig <hch>, Bob Gilligan <gilligan@xxxxxxxxxxxx>, linux-iscsi-devel@xxxxxxxxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, linux-xfs@xxxxxxxxxxx |
| References: | <412A40D6.3070408@intransa.com> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
| User-agent: | Mutt/1.3.28i |
On Mon, Aug 23, 2004 at 12:09:10PM -0700, Bob Gilligan wrote: > Hi Folks -- I'm running Linux kernel 2.6.8.1 and version 4.0.1.8 of the > Linux initiator. I mounted an XFS filesystem over a 900 GB iSCSI > volume, then ran "bonnie" with the args "-v 5 -o_direct". Within about > 30 seconds, the kernel started spewing printks from bad_page(). The > first few were: Kown issues. XFS uses slab pages (as you noticed below), networking code doesn't like that. DRBD has been tripping the same issue. We need to come up with a consensus about what type of pages can be used in block I/O requests. The drbd folks promised to get this discussed on lkml but I haven't seen a discussion yet - so let's do it now. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | TAKE 914196 - upgrade kdb, Nathan Scott |
|---|---|
| Next by Date: | Write Barriers in 2.6.?, Joshua Schmidlkofer |
| Previous by Thread: | TAKE 914196 - upgrade kdb, Nathan Scott |
| Next by Thread: | Write Barriers in 2.6.?, Joshua Schmidlkofer |
| Indexes: | [Date] [Thread] [Top] [All Lists] |