Re: [PATCH] Driver 'sd' needs updating

[Hannes Reinecke]

Hi Boaz,

Boaz Harrosh wrote:
> Hannes Reinecke wrote:
> > If a driver sets blk_queue_prep_rq() it should clean up itself
> > and not rely on the bus callbacks to handle this. This removes
> > the need to hook into bus->remove() as these should not be used
> > at the same time as driver->remove().
> >
> > Signed-off-by: Hannes Reinecke <hare@xxxxxxx>
> > Signed-off-by: Kay Sievers <kay.sievers@xxxxxxxx>
>
> Hi Hannes, please I do not understand this patch.
>
> I was always under the impression that the scsi ULD
> remove() function is called from inside scsi_bus_remove()
> at the very end, please see below.
>
> > ---
> >  drivers/scsi/scsi_lib.c    |    6 ++++++
> >  drivers/scsi/scsi_sysfs.c  |   17 -----------------
> >  drivers/scsi/sd.c          |    2 ++
> >  drivers/scsi/sr.c          |    1 +
> >  include/scsi/scsi_driver.h |    1 +
> >  5 files changed, 10 insertions(+), 17 deletions(-)
> >
> > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> > index 4b13e36..73df41b 100644
> > --- a/drivers/scsi/scsi_lib.c
> > +++ b/drivers/scsi/scsi_lib.c
> > @@ -1222,6 +1222,12 @@ int scsi_prep_fn(struct request_queue *q, struct request *req)
> >  	return scsi_prep_return(q, req, ret);
> >  }
> >  
> > +void scsi_reset_prep_fn(struct request_queue *q)
> > +{
> > +	blk_queue_prep_rq(q, scsi_prep_fn);
> > +}
> > +EXPORT_SYMBOL(scsi_reset_prep_fn);
> > +
> >  /*
> >   * scsi_dev_queue_ready: if we can send requests to sdev, return 1 else
> >   * return 0.
> > diff --git a/drivers/scsi/scsi_sysfs.c b/drivers/scsi/scsi_sysfs.c
> > index fa4711d..91482f2 100644
> > --- a/drivers/scsi/scsi_sysfs.c
> > +++ b/drivers/scsi/scsi_sysfs.c
> > @@ -420,29 +420,12 @@ static int scsi_bus_resume(struct device * dev)
> >  	return err;
> >  }
> >  
> > -static int scsi_bus_remove(struct device *dev)
> > -{
> > -	struct device_driver *drv = dev->driver;
> > -	struct scsi_device *sdev = to_scsi_device(dev);
> > -	int err = 0;
> > -
> > -	/* reset the prep_fn back to the default since the
> > -	 * driver may have altered it and it's being removed */
> > -	blk_queue_prep_rq(sdev->request_queue, scsi_prep_fn);
> > -
> > -	if (drv && drv->remove)
> > -		err = drv->remove(dev);
>
> This here is where my osd_uld.c::osd_remove() is called.
>
> If this vector is not used will the drv-core call drv->remove(dev)
> in behalf of scsi?
>
> From drivers/base/dd.c:__device_release_driver()

		if (dev->bus && dev->bus->remove)
			dev->bus->remove(dev);
		else if (drv->remove)
			drv->remove(dev);

So the answer will be yes.

> > -
> > -	return 0;
> > -}
> > -
> >  struct bus_type scsi_bus_type = {
> >          .name		= "scsi",
> >          .match		= scsi_bus_match,
> >  	.uevent		= scsi_bus_uevent,
> >  	.suspend	= scsi_bus_suspend,
> >  	.resume		= scsi_bus_resume,
> > -	.remove		= scsi_bus_remove,
> >  };
> >  EXPORT_SYMBOL_GPL(scsi_bus_type);
> >  
> > diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
> > index aeab5d9..64e88e2 100644
> > --- a/drivers/scsi/sd.c
> > +++ b/drivers/scsi/sd.c
> > @@ -2068,6 +2068,8 @@ static int sd_remove(struct device *dev)
>
> Until today this was called from within scsi_bus_remove
> what happens after your patch?
Driver core calls it (see above).

> >  {
> >  	struct scsi_disk *sdkp = dev_get_drvdata(dev);
> >  
> > +	scsi_reset_prep_fn(sdkp->device->request_queue);
> > +
> >  	device_del(&sdkp->dev);
> >  	del_gendisk(sdkp->disk);
> >  	sd_shutdown(dev);
> > diff --git a/drivers/scsi/sr.c b/drivers/scsi/sr.c
> > index e7fa3ca..914733a 100644
> > --- a/drivers/scsi/sr.c
> > +++ b/drivers/scsi/sr.c
> > @@ -889,6 +889,7 @@ static int sr_remove(struct device *dev)
> >  {
> >  	struct scsi_cd *cd = dev_get_drvdata(dev);
> >  
> > +	scsi_reset_prep_fn(cd->device->request_queue);
> >  	del_gendisk(cd->disk);
> >  
> >  	mutex_lock(&sr_ref_mutex);
> > diff --git a/include/scsi/scsi_driver.h b/include/scsi/scsi_driver.h
> > index 1f5ca7f..2e22929 100644
> > --- a/include/scsi/scsi_driver.h
> > +++ b/include/scsi/scsi_driver.h
> > @@ -32,5 +32,6 @@ int scsi_setup_blk_pc_cmnd(struct scsi_device *sdev, struct request *req);
> >  int scsi_setup_fs_cmnd(struct scsi_device *sdev, struct request *req);
> >  int scsi_prep_state_check(struct scsi_device *sdev, struct request *req);
> >  int scsi_prep_return(struct request_queue *q, struct request *req, int ret);
> > +int scsi_reset_prep_fn(struct request_queue *);
> >  
> >  #endif /* _SCSI_SCSI_DRIVER_H */
>
> I did not test this patch with osd_uld yet, but I will tomorrow. If my assumption
> is right then drv-core should call my osd_remove just the same, right?
Yes, correct.

> Are there any different timing issue for example the presence of scsi_device
> is still true?
No. Or rather, none you'll notice. (We're effectively saving on indirection here,
but I doubt it'll be measureable)

Cheers,

Hannes
--
Dr. Hannes Reinecke		      zSeries & Storage
hare@xxxxxxx			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 NÃrnberg
GF: Markus Rex, HRB 16746 (AG NÃrnberg)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/