Messages in this thread | | | Date | Thu, 12 Jan 2006 08:35:03 +0100 | From | Jens Axboe <> | Subject | Re: 2.6.15-mm2 |
| |
On Thu, Jan 12 2006, Neil Brown wrote: > On Wednesday January 11, axboe@suse.de wrote: > > > > Then the barrier changes from git2 -> git3 should not have anything to > > do with it. Strange... I guess you should try the git bisect method to > > narrow it down. > > Not true, though you seem to have already figured that out. > > md uses barrier writes when writing the superblock. This is partly > because it seems like a good idea, but largely to test if barrier > writes are going to work on the component devices. If any device > claims not to support barriers, then raid1 will claim not to support > barriers. > > And the strange hang happens while md is trying to update the > superblock.
Yeah that's what I found out later on in the thread, indeed killing that barrier write made the problem disappear.
-- Jens Axboe
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |