Hi Akinobu,
On Wed, 2014-02-19 at 22:22 +0900, Akinobu Mita wrote:
> After commit 5837c80e870b ("bio-integrity: Fix bio_integrity_verify
> segment start bug"), loading scsi_debug with data integrity enabled
> (for example modprobe scsi_debug dif=1 dix=1) starts hitting BUG_ON()
> at bio_integrity_verify():
>
> BUG_ON(total > bio->bi_integrity->bip_iter.bi_size);
>
> Because bip_iter.bi_size has already been set to zero at this point
> as a result of consuming bip_iter.
>
> I can resolve this issue by removing the BUG_ON(). But I would like
> to ask whether we should preserve an alternative assert, or not
> before submitting the patch that simply removing the line.
So commit 5837c80e870b just allowed bio_integrity_verify() to actually
process protection information for READs again, which had been broken
since v3.10 code.
The bug your hitting is specific to v3.14-rc1, and was introduced with
the following change to support immutable biovecs:
commit d57a5f7c6605f15f3b5134837e68b448a7cea88e
Author: Kent Overstreet <kmo@xxxxxxxxxxxxx>
Date: Sat Nov 23 17:20:16 2013 -0800
bio-integrity: Convert to bvec_iter
Given that there is no (easy) way to ascertain what the original value
of bio_integrity->bip_iter.bi_size was post bio_integrity_advance(),
dropping this BUG_ON() probably makes the most sense.
MKP + Jens, care to ACK + pickup..?
--nab
>From 847ab029fe0d5ed9b54a1f1507c46025f6c3f9e0 Mon Sep 17 00:00:00 2001
From: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
Date: Thu, 20 Feb 2014 00:52:01 +0000
Subject: [PATCH] bio-integrity: Drop bio_integrity_verify BUG_ON in post
bip->bip_iter world
Given that bip->bip_iter.bi_size is decremented after bio_advance() ->
bio_integrity_advance() is called, the BUG_ON() in bio_integrity_verify()
ends up tripping in v3.14-rc1 code with the advent of immutable biovecs
in:
commit d57a5f7c6605f15f3b5134837e68b448a7cea88e
Author: Kent Overstreet <kmo@xxxxxxxxxxxxx>
Date: Sat Nov 23 17:20:16 2013 -0800
bio-integrity: Convert to bvec_iter
Given that there is no easy way to ascertain the original bi_size
value, go ahead and drop this BUG_ON().
Reported-by: Sagi Grimberg <sagig@xxxxxxxxxxxxxxxxxx>
Reported-by: Akinobu Mita <akinobu.mita@xxxxxxxxx>
Cc: Martin K. Petersen <martin.petersen@xxxxxxxxxx>
Cc: Jens Axboe <axboe@xxxxxxxxx>
Cc: Kent Overstreet <kmo@xxxxxxxxxxxxx>
Signed-off-by: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
---
fs/bio-integrity.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
index 0bad24d..47ed160 100644
--- a/fs/bio-integrity.c
+++ b/fs/bio-integrity.c
@@ -476,7 +476,6 @@ static int bio_integrity_verify(struct bio *bio)
sector += sectors;
prot_buf += sectors * bi->tuple_size;
total += sectors * bi->tuple_size;
- BUG_ON(total > bio->bi_integrity->bip_iter.bi_size);
kunmap_atomic(kaddr);
}
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
