On Sun, 2010-11-07 at 16:21 -0500, Ted Ts'o wrote: > OK, I think I have a new version of the patch that should fix both the > original and the second BUG_ON which you reported. > Subject: [PATCH -v2] ext4: handle writeback of inodes which are being freed Tested-By: Tony Vroon <tony@xxxxxxxx> This survived a 'cvs up' in /cvs/gentoo-x86 with only the CVS directory in place; a workload that died at roughly 90% before. I did my best to recreate the heavy I/O conditions that existed when it died there before, with Evolution being opened just as cvs starts the first writes. I opened Transmission-gtk with a few LiveDVD ISOs just for good measure. The system was loaded enough for "su -" to stall for 20 seconds at some points but ext4 survived now. I do believe you have found the culprit, many thanks :) Regards, -- Tony Vroon UNIX systems administrator London Internet Exchange Ltd, Trinity Court, Trinity Street, Peterborough, PE1 1DA Registered in England number 3137929 E-Mail: tony@xxxxxxxx
Attachment:
signature.asc
Description: This is a digitally signed message part
