[PATCH] make sr_mod report more accurate drive status after closing the tray.

[Peter Jones]

Right now, when using sr_mod and issuing the CDROM_DRIVE_STATUS ioctl,
there's no way to differentiate between when you've just closed the
drive tray, but the media is not yet loaded, and when there's no media.
This seems to be accidental.

Here's a patch that seems to fix this behaviour:

Signed-off-by: Peter Jones

diff --git a/drivers/scsi/sr_ioctl.c b/drivers/scsi/sr_ioctl.c
index ae87d08..43a084b 100644
--- a/drivers/scsi/sr_ioctl.c
+++ b/drivers/scsi/sr_ioctl.c
@@ -306,10 +306,9 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
int slot)
    	/* we have no changer support */
    	return -EINVAL;
    }
-	if (0 == sr_test_unit_ready(cd->device, &sshdr))
-		return CDS_DISC_OK;
-
-	if (!cdrom_get_media_event(cdi, &med)) {
+	if (0 == sr_test_unit_ready(cd->device, &sshdr)
+			&& sshdr.sense_key == 0
+			&& !cdrom_get_media_event(cdi, &med)) {
    	if (med.media_present)
    		return CDS_DISC_OK;
    	else if (med.door_open)
@@ -319,10 +318,27 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
int slot)
    }

/*
-	 * 0x04 is format in progress .. but there must be a disc present!
+	 * ASC: 0x04: "logical unit is not ready"
+	 * ASCQ: 0x01: cause not reportable
+	 *       0x02: in process of becoming ready
+	 *       0x03: initializing command required
+	 *       0x04: format in progress .. but there must be a disc present!
+	 *       0x07: operation in progress
+	 *       0x08: long write in progress
     */
-	if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04)
-		return CDS_DISC_OK;
+	if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04) {
+		switch (sshdr.ascq) {
+			case 0x01:
+			case 0x02:
+				return CDS_DRIVE_NOT_READY;
+			case 0x03:
+				return CDS_TRAY_OPEN;
+			case 0x04:
+			case 0x07:
+			case 0x08:
+				return CDS_DRIVE_NOT_READY;
+		}
+	}

/*
     * If not using Mt Fuji extended media tray reports,

[James Bottomley]

On Thu, 2008-07-10 at 17:24 -0400, Peter Jones wrote:
> Right now, when using sr_mod and issuing the CDROM_DRIVE_STATUS ioctl,
> there's no way to differentiate between when you've just closed the
> drive tray, but the media is not yet loaded, and when there's no media.
> This seems to be accidental.
>
> Here's a patch that seems to fix this behaviour:

I'm very wary of doing something like this because it took us months to
fix up all the breakage the last time ...
> Signed-off-by: Peter Jones
>
> diff --git a/drivers/scsi/sr_ioctl.c b/drivers/scsi/sr_ioctl.c
> index ae87d08..43a084b 100644
> --- a/drivers/scsi/sr_ioctl.c
> +++ b/drivers/scsi/sr_ioctl.c
> @@ -306,10 +306,9 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
> int slot)
>    	/* we have no changer support */
>    	return -EINVAL;
>    }
> if (0 == sr_test_unit_ready(cd->device, &sshdr))
> return CDS_DISC_OK;
> if (!cdrom_get_media_event(cdi, &med)) {
> if (0 == sr_test_unit_ready(cd->device, &sshdr)
> 	&& sshdr.sense_key == 0

This can't be right; if the return is zero, the sense_key must also be
zero (as in to have valid sense, it must have returned with at least
DRIVER_SENSE)
> 	&& !cdrom_get_media_event(cdi, &med)) {

And this really doesn't look right either.  Now you're only calling the
media event if the test unit ready succeeded.  A drive can be open (thus
returning not ready and hence non zero) and still give you a valid media
event.
> if (med.media_present)
> 	return CDS_DISC_OK;
> else if (med.door_open)
>
> @@ -319,10 +318,27 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
> int slot)
>    }
>
> /*
> * 0x04 is format in progress .. but there must be a disc present!
> * ASC: 0x04: "logical unit is not ready"
> * ASCQ: 0x01: cause not reportable
> *       0x02: in process of becoming ready
> *       0x03: initializing command required
> *       0x04: format in progress .. but there must be a disc present!
> *       0x07: operation in progress
> *       0x08: long write in progress
> */
> if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04)
> return CDS_DISC_OK;

This seems to be a nasty historical lie.  It seems to play into the new
media changed stuff by stalling the change until the media is ready.
Convince me that if we actually tell the truth here we're not going to
fire a slew of spurious events at hal.
> if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04) {
> switch (sshdr.ascq) {
> 	case 0x01:
> 	case 0x02:
> 		return CDS_DRIVE_NOT_READY;
> 	case 0x03:
> 		return CDS_TRAY_OPEN;

That's a stretch ... the initialising command can be start motor, but
the tray can be closed just fine.
> 	case 0x04:
> 	case 0x07:
> 	case 0x08:
> 		return CDS_DRIVE_NOT_READY;
> }
> }

James

[Peter Jones]

James Bottomley wrote:
> On Thu, 2008-07-10 at 17:24 -0400, Peter Jones wrote:
> > Right now, when using sr_mod and issuing the CDROM_DRIVE_STATUS ioctl,
> > there's no way to differentiate between when you've just closed the
> > drive tray, but the media is not yet loaded, and when there's no media.
> > This seems to be accidental.
> >
> > Here's a patch that seems to fix this behaviour:
>
> I'm very wary of doing something like this because it took us months to
> fix up all the breakage the last time ...

That's certainly understandable.  What I'm hitting is that I can't tell 
the difference any more between not having media and it simply not being 
ready yet.  And the amount of time it takes to become ready seems to 
vary wildly (1s on one drive I've got here, ~20s on another fairly new 
drive), so a timeout isn't even a very effective kludge.
> > Signed-off-by: Peter Jones
> >
> > diff --git a/drivers/scsi/sr_ioctl.c b/drivers/scsi/sr_ioctl.c
> > index ae87d08..43a084b 100644
> > --- a/drivers/scsi/sr_ioctl.c
> > +++ b/drivers/scsi/sr_ioctl.c
> > @@ -306,10 +306,9 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
> > int slot)
> >    	/* we have no changer support */
> >    	return -EINVAL;
> >    }
> > if (0 == sr_test_unit_ready(cd->device, &sshdr))
> > return CDS_DISC_OK;
> > if (!cdrom_get_media_event(cdi, &med)) {
> > if (0 == sr_test_unit_ready(cd->device, &sshdr)
> > 	&& sshdr.sense_key == 0
>
> This can't be right; if the return is zero, the sense_key must also be
> zero (as in to have valid sense, it must have returned with at least
> DRIVER_SENSE)
> > 	&& !cdrom_get_media_event(cdi, &med)) {
>
> And this really doesn't look right either.  Now you're only calling the
> media event if the test unit ready succeeded.  A drive can be open (thus
> returning not ready and hence non zero) and still give you a valid media
> event.

Ok, that's a fair point.  That part I'm really trying to avoid here is 
returning the CDS_NO_DSIC case when the sense key says we don't yet know 
(NOT_READY).  So probably that case simply shouldn't be returned until 
after we do the SK/ASC/ASCQ check.
> > if (med.media_present)
> > 	return CDS_DISC_OK;
> > else if (med.door_open)
> >
> > @@ -319,10 +318,27 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
> > int slot)
> >    }
> >
> > /*
> > * 0x04 is format in progress .. but there must be a disc present!
> > * ASC: 0x04: "logical unit is not ready"
> > * ASCQ: 0x01: cause not reportable
> > *       0x02: in process of becoming ready
> > *       0x03: initializing command required
> > *       0x04: format in progress .. but there must be a disc present!
> > *       0x07: operation in progress
> > *       0x08: long write in progress
> > */
> > if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04)
> > return CDS_DISC_OK;
>
> This seems to be a nasty historical lie.  It seems to play into the new
> media changed stuff by stalling the change until the media is ready.
> Convince me that if we actually tell the truth here we're not going to
> fire a slew of spurious events at hal.

Would you be ok with a more minimal approach, where I only change the 
behaviour of NOT_READY/4/4 ?  I understand if you still want more study 
of the effects WRT udev/hal.
> > if (sshdr.sense_key == NOT_READY && sshdr.asc == 0x04) {
> > switch (sshdr.ascq) {
> > 	case 0x01:
> > 	case 0x02:
> > 		return CDS_DRIVE_NOT_READY;
> > 	case 0x03:
> > 		return CDS_TRAY_OPEN;
>
> That's a stretch ... the initialising command can be start motor, but
> the tray can be closed just fine.

Yes, it definitely is.  I don't feel strongly about this particular 
case, but the rationale was that as long as we're not handling that in 
the driver, then this case essentially means user intervention is 
required (though there's really no way for the user to know what that 
would be).  CDS_TRAY_OPEN is, AFAICT, the only return value that means 
that in other cases (i.e., on slot-loaders).  This part isn't critical 
to the functionality I need; it just seemed wrong when I was reading it.

[Peter Jones]

So, what's happening here is that the drive is reporting a sense of 
2/4/1 ("logical unit is becoming ready") from sr_test_unit_ready(), and 
then we ask for the media event notification before checking that result 
at all.  The check_media_event_descriptor() call isn't getting a check 
condition, but it's also reporting that the tray is closed and that 
there's no media.  In actuality it doesn't yet know if there's media or 
not, but there's no way to express that in the media event status field.

My current thought is that if it told us the device isn't yet ready, we 
should return that immediately, since there's nothing that'll tell us 
any more data than that reliably:

diff --git a/drivers/scsi/sr_ioctl.c b/drivers/scsi/sr_ioctl.c
index ae87d08..25f2008 100644
--- a/drivers/scsi/sr_ioctl.c
+++ b/drivers/scsi/sr_ioctl.c
@@ -309,6 +309,11 @@ int sr_drive_status(struct cdrom_device_info *cdi, 
int slot)
         if (0 == sr_test_unit_ready(cd->device, &sshdr))
                 return CDS_DISC_OK;

+       /* SK/ASC/ASCQ of 2/4/1 means "unit is becoming ready" */
+       if (scsi_sense_valid(&sshdr) && sshdr.sense_key == NOT_READY
+                       && sshdr.asc == 0x04 && sshdr.ascq == 0x01)
+               return CDS_DRIVE_NOT_READY;
+
         if (!cdrom_get_media_event(cdi, &med)) {
                 if (med.media_present)
                         return CDS_DISC_OK;