Re: [PATCH 1/1] Drivers: scsi: Derive the FLUSH_TIMEOUT from the basic I/O timeout

[James Bottomley <jbottomley@parallels.com>]

On Fri, 2013-10-04 at 22:01 +0000, KY Srinivasan wrote:
> 
> > -----Original Message-----
> > From: James Bottomley [mailto:jbottomley@parallels.com]
> > Sent: Friday, October 04, 2013 2:42 PM
> > To: KY Srinivasan
> > Cc: gregkh@linuxfoundation.org; linux-kernel@vger.kernel.org;
> > devel@linuxdriverproject.org; ohering@suse.com; hch@infradead.org;
> > eric@purestorage.com; nab@linux-iscsi.org; linux-scsi@vger.kernel.org
> > Subject: Re: [PATCH 1/1] Drivers: scsi: Derive the FLUSH_TIMEOUT from the basic
> > I/O timeout
> > 
> > On Fri, 2013-10-04 at 12:38 -0700, K. Y. Srinivasan wrote:
> > > Rather than having a separate constant for specifying the timeout on FLUSH
> > > operations, use the basic I/O timeout value that is already configurable
> > > on a per target basis to derive the FLUSH timeout. Looking at the current
> > > definitions of these timeout values, the FLUSH operation is supposed to have
> > > a value that is twice the normal timeout value. This patch preserves this
> > > relationship while leveraging the flexibility of specifying the I/O timeout.
> > >
> > > I would like to thank Eric Seppanen <eric@purestorage.com> and
> > > Nicholas A. Bellinger <nab@linux-iscsi.org> for their help in resolving
> > > this issue.
> > >
> > > Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
> > > ---
> > >  drivers/scsi/sd.c |    5 +++--
> > >  1 files changed, 3 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
> > > index e62d17d..8aff306 100644
> > > --- a/drivers/scsi/sd.c
> > > +++ b/drivers/scsi/sd.c
> > > @@ -829,7 +829,7 @@ static int sd_setup_write_same_cmnd(struct
> > scsi_device *sdp, struct request *rq)
> > >
> > >  static int scsi_setup_flush_cmnd(struct scsi_device *sdp, struct request *rq)
> > >  {
> > > -	rq->timeout = SD_FLUSH_TIMEOUT;
> > > +	rq->timeout *= 2;
> > >  	rq->retries = SD_MAX_RETRIES;
> > >  	rq->cmd[0] = SYNCHRONIZE_CACHE;
> > >  	rq->cmd_len = 10;
> > > @@ -1433,6 +1433,7 @@ static int sd_sync_cache(struct scsi_disk *sdkp)
> > >  {
> > >  	int retries, res;
> > >  	struct scsi_device *sdp = sdkp->device;
> > > +	unsigned int timeout = sdp->request_queue->rq_timeout;
> > 
> > The timeout is signed in the function prototype
> > 
> > >  	struct scsi_sense_hdr sshdr;
> > >
> > >  	if (!scsi_device_online(sdp))
> > > @@ -1448,7 +1449,7 @@ static int sd_sync_cache(struct scsi_disk *sdkp)
> > >  		 * flush everything.
> > >  		 */
> > >  		res = scsi_execute_req_flags(sdp, cmd, DMA_NONE, NULL, 0,
> > > -					     &sshdr, SD_FLUSH_TIMEOUT,
> > > +					     &sshdr, timeout * 2,
> > >  					     SD_MAX_RETRIES, NULL, REQ_PM);
> > >  		if (res == 0)
> > >  			break;
> > 
> > Not like this, please: you now leave us with a dangling #define whose
> > name makes you think it should be related to flushing and a couple of
> > curious magic constants.  It's almost hand crafted to confuse people
> > reading the code.
> > 
> > Please do it like this instead: with a comment explaining what we're
> > doing above the #define and a name that clearly relates to the actions.
> > 
> > James
> > 
> > ---
> > 
> > diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
> > index e62d17d..5c9496d 100644
> > --- a/drivers/scsi/sd.c
> > +++ b/drivers/scsi/sd.c
> > @@ -829,7 +829,7 @@ static int sd_setup_write_same_cmnd(struct scsi_device
> > *sdp, struct request *rq)
> > 
> >  static int scsi_setup_flush_cmnd(struct scsi_device *sdp, struct request *rq)
> >  {
> > -	rq->timeout = SD_FLUSH_TIMEOUT;
> > +	rq->timeout *= SD_FLUSH_TIMEOUT_MULTIPLIER;
> >  	rq->retries = SD_MAX_RETRIES;
> >  	rq->cmd[0] = SYNCHRONIZE_CACHE;
> >  	rq->cmd_len = 10;
> > @@ -1433,6 +1433,8 @@ static int sd_sync_cache(struct scsi_disk *sdkp)
> >  {
> >  	int retries, res;
> >  	struct scsi_device *sdp = sdkp->device;
> > +	const int timeout = sdp->request_queue->rq_timeout
> > +		* SD_FLUSH_TIMEOUT_MULTIPLIER;
> >  	struct scsi_sense_hdr sshdr;
> > 
> >  	if (!scsi_device_online(sdp))
> > @@ -1448,8 +1450,8 @@ static int sd_sync_cache(struct scsi_disk *sdkp)
> >  		 * flush everything.
> >  		 */
> >  		res = scsi_execute_req_flags(sdp, cmd, DMA_NONE, NULL, 0,
> > -					     &sshdr, SD_FLUSH_TIMEOUT,
> > -					     SD_MAX_RETRIES, NULL, REQ_PM);
> > +					     &sshdr, timeout, SD_MAX_RETRIES,
> > +					     NULL, REQ_PM);
> >  		if (res == 0)
> >  			break;
> >  	}
> > diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h
> > index 7a049de..7f7999c 100644
> > --- a/drivers/scsi/sd.h
> > +++ b/drivers/scsi/sd.h
> > @@ -13,7 +13,11 @@
> >   */
> >  #define SD_TIMEOUT		(30 * HZ)
> >  #define SD_MOD_TIMEOUT		(75 * HZ)
> > -#define SD_FLUSH_TIMEOUT	(60 * HZ)
> > +/*
> > + * Flush timeout is a multiplier over the standard device timeout which is
> > + * user modifiable via sysfs but initially set to SD_TIMEOUT
> > + */
> > +#define SD_FLUSH_TIMEOUT_MULTIPLIER	2
> >  #define SD_WRITE_SAME_TIMEOUT	(120 * HZ)
> > 
> >  /*
> 
> Thanks James. Given that you have already made the changes you want done (in the
> Patch above), do you still want me to send you a fixed up patch?

Either way is fine.  I just need the patch to be on the mailing list for
my workflow (which it is now).

James