Re: 4.15.14 crash with iscsi target and dvd

[Wakko Warner]

Ming Lei wrote:
> On Tue, Apr 10, 2018 at 08:45:25PM -0400, Wakko Warner wrote:
> > Sorry for the delay.  I reverted my change, added this one.  I didn't
> > reboot, I just unloaded and loaded this one.
> > Note: /dev/sr1 as seen from the initiator is /dev/sr0 (physical disc) on the
> > target.
> > 
> > Doesn't crash, however on the initiator I see this:
> > [9273849.707777] ISO 9660 Extensions: RRIP_1991A
> > [9273863.359718] scsi_io_completion: 13 callbacks suppressed
> > [9273863.359788] sr 26:0:0:0: [sr1] tag#1 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x08
> > [9273863.359909] sr 26:0:0:0: [sr1] tag#1 Sense Key : 0x2 [current] 
> > [9273863.359974] sr 26:0:0:0: [sr1] tag#1 ASC=0x8 ASCQ=0x0 
> > [9273863.360036] sr 26:0:0:0: [sr1] tag#1 CDB: opcode=0x28 28 00 00 22 f6 96 00 00 80 00
> > [9273863.360116] blk_update_request: 13 callbacks suppressed
> > [9273863.360177] blk_update_request: I/O error, dev sr1, sector 9165400
> > [9273875.864648] sr 26:0:0:0: [sr1] tag#1 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x08
> > [9273875.864738] sr 26:0:0:0: [sr1] tag#1 Sense Key : 0x2 [current] 
> > [9273875.864801] sr 26:0:0:0: [sr1] tag#1 ASC=0x8 ASCQ=0x0 
> > [9273875.864890] sr 26:0:0:0: [sr1] tag#1 CDB: opcode=0x28 28 00 00 22 f7 16 00 00 80 00
> > [9273875.864971] blk_update_request: I/O error, dev sr1, sector 9165912
> > 
> > To cause this, I mounted the dvd as seen in the first line and ran this
> > command: find /cdrom2 -type f | xargs -tn1 cat > /dev/null
> > I did some various tests.  Each test was done after umount and mount to
> > clear the cache.
> > cat <file> > /dev/null causes the message.
> > dd if=<file> of=/dev/null bs=2048 doesn't
> > 	using bs=4096 doesn't
> > 	using bs=64k doesn't
> > 	using bs=128k does
> > cat uses a blocksize of 128k.
> > 
> > The following was done without being mounted.
> > ddrescue -f -f /dev/sr1 /dev/null 
> > 	doesn't cause the message
> > dd if=/dev/sr1 of=/dev/null bs=128k
> > 	doesn't cause the message
> > using bs=256k causes the message once:
> > [9275916.857409] sr 27:0:0:0: [sr1] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x08
> > [9275916.857482] sr 27:0:0:0: [sr1] tag#0 Sense Key : 0x2 [current] 
> > [9275916.857520] sr 27:0:0:0: [sr1] tag#0 ASC=0x8 ASCQ=0x0 
> > [9275916.857556] sr 27:0:0:0: [sr1] tag#0 CDB: opcode=0x28 28 00 00 00 00 00 00 00 80 00
> > [9275916.857614] blk_update_request: I/O error, dev sr1, sector 0
> > 
> > If I access the disc from the target natively either by mounting and
> > accessing files or working with the device directly (ie dd) no errors are
> > logged on the target.
> 
> OK, thanks for your test.
> 
> Could you test the following patch and see if there is still the failure
> message?
> 
> diff --git a/drivers/target/target_core_pscsi.c b/drivers/target/target_core_pscsi.c
> index 0d99b242e82e..6137287b52fb 100644
> --- a/drivers/target/target_core_pscsi.c
> +++ b/drivers/target/target_core_pscsi.c
> @@ -913,9 +913,11 @@ pscsi_map_sg(struct se_cmd *cmd, struct scatterlist *sgl, u32 sgl_nents,
>  
>  			rc = bio_add_pc_page(pdv->pdv_sd->request_queue,
>  					bio, page, bytes, off);
> +			if (rc != bytes)
> +				goto fail;
>  			pr_debug("PSCSI: bio->bi_vcnt: %d nr_vecs: %d\n",
>  				bio_segments(bio), nr_vecs);
> -			if (rc != bytes) {
> +			if (/*rc != bytes*/0) {
>  				pr_debug("PSCSI: Reached bio->bi_vcnt max:"
>  					" %d i: %d bio: %p, allocating another"
>  					" bio\n", bio->bi_vcnt, i, bio);

Target doesn't crash but the errors on the initiator are still there.

Seems that if I do large transfers, I see this in the initiator's logs.
With the previous patch, I burned 3 dvds at the same time, compared the
files to the originals and I have a script that catalogs the files.  The
files consist of debian packages and source files.  The 3 operations did not
show any errors in the kernel log on either end.

I did this test:
initiator: dd if=/dev/sr1 bs=512k count=1024 | md5sum
target:    dd if=/dev/sr0 bs=512k count=1024 | md5sum

Result: the same.  It's OK even with the i/o errors shown on the initiator.

The above patch was added on top of the one you gave me before, but I don't
believe that that would be an issue.

...  Now if someone could help me with a kvm virtualization problem I'm
having with 4.16 that wasn't there with 4.15...

-- 
 Microsoft has beaten Volkswagen's world record.  Volkswagen only created 22
 million bugs.