From mboxrd@z Thu Jan 1 00:00:00 1970 From: Douglas Gilbert Subject: Re: [PATCH] Revert "scsi: core: return BLK_STS_OK for DID_OK in __scsi_error_from_host_byte()" Date: Wed, 4 Apr 2018 13:03:55 -0400 Message-ID: <7cda8ed9-0c8d-8c25-0d8e-ba889acddacc@interlog.com> References: <20180404163028.9826-1-bart.vanassche@wdc.com> Reply-To: dgilbert@interlog.com Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20180404163028.9826-1-bart.vanassche@wdc.com> Content-Language: en-CA Sender: stable-owner@vger.kernel.org To: Bart Van Assche , "Martin K . Petersen" , "James E . J . Bottomley" Cc: linux-scsi@vger.kernel.org, Hannes Reinecke , Damien Le Moal , Christoph Hellwig , stable@vger.kernel.org List-Id: linux-scsi@vger.kernel.org On 2018-04-04 12:30 PM, Bart Van Assche wrote: > The description of commit e39a97353e53 is wrong: it mentions that > commit 2a842acab109 introduced a bug in __scsi_error_from_host_byte() > although that commit did not change the behavior of that function. > Additionally, that commit introduced a severe bug: it causes commands > that fail with hostbyte=DID_OK and driverbyte=DRIVER_SENSE to be > completed with BLK_STS_OK. Although that commit claims to fix a bug > it does not mention which bug it fixes. Hence revert that commit. That patch arose from the thread started by me titled: [PATCH v2] Make SCSI Status CONDITION MET equivalent to GOOD on 26 February this year. Hannes Reinecke made the case for reducing the helper scsi_io_completion_nz_result() to a single returned value (an idea that was applied and later reverted). He wrote: > Hmm. Can't we return blk_stat from this function, and adjusting the > 'result' value after it with an if-clause like > > if (blk_stat == BLK_STS_OK) > result = 0; > > That would cleanup up the function and avoid having (essentially) two > return values. > > The only problem here is that __scsi_error_from_hostbyte() will return > BLK_STS_IOERR if result == 0; doubt that is intended. > And I guess it'll affect this issue, too. > > Mind sending a separate patch for that? Actually Hannes had already sent a patch for that before I had time to react (on the same day). That is the patch you now plan to revert. In Hannes' defence (and I reviewed it) it is counter intuitive that for callers to work properly, this function supplied with DID_OK in the result argument should return BLK_STS_IOERR. Please find out why that is so and craft a comment so nobody else falls down this particular rabbit hole. Plus add 'case DID_OK:' before the 'default:' to emphasis the point. Doug Gilbert > Fixes: e39a97353e53 ("scsi: core: return BLK_STS_OK for DID_OK in __scsi_error_from_host_byte()") > Reported-by: Damien Le Moal > Signed-off-by: Bart Van Assche > Cc: Hannes Reinecke > Cc: Douglas Gilbert > Cc: Damien Le Moal > Cc: Christoph Hellwig > Cc: stable@vger.kernel.org > --- > drivers/scsi/scsi_lib.c | 2 -- > 1 file changed, 2 deletions(-) > > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c > index 74a39db57d49..71f5b010684c 100644 > --- a/drivers/scsi/scsi_lib.c > +++ b/drivers/scsi/scsi_lib.c > @@ -735,8 +735,6 @@ static blk_status_t __scsi_error_from_host_byte(struct scsi_cmnd *cmd, > int result) > { > switch (host_byte(result)) { > - case DID_OK: > - return BLK_STS_OK; > case DID_TRANSPORT_FAILFAST: > return BLK_STS_TRANSPORT; > case DID_TARGET_FAILURE: >