Messages in this thread | | | Date | Tue, 27 May 2014 12:03:29 +0800 | Subject | Re: [: [bio] kernel BUG at drivers/block/virtio_blk.c:166!] | From | Ming Lei <> |
| |
On Tue, May 27, 2014 at 3:43 AM, Maurizio Lombardi <mlombard@redhat.com> wrote: > Hi Jens, > > looks like that commit 3979ef4dcf3d1de55a560a3a4016c30a835df44d ("bio-modify-__bio_add_page-to-accept-pages-that-dont-start-a-new-segment-v3") > introduces a regression, as reported by Jet Chan. > > Do you have any idea about the possible problem with this patch? > > it is the one that performs a recount of the segments in case of failure in __bio_add_page() > > http://www.spinics.net/lists/mm-commits/msg103684.html > > I would not be surprised if the bug was introduced by fceb38f36f, because it > contained a mystake that commit 3979ef4dcf supposedly fixed. > But learning that commit 3979ef4dcf is introducing a regression leaves > me quite puzzled.
From code of __blk_recalc_rq_segments(), looks it won't check if recounted physical segment number is bigger than queue_max_segments(), so wondering if blk_recount_segments() can always decrease physical segment number.
Thanks, -- Ming Lei
| |