diff mbox series

[2/2] UAS: fix disconnect by unplugging a hub

Message ID 20200915134501.13947-3-oneukum@suse.com
State Superseded
Headers show
Series [1/2] UAS: use macro for reporting results | expand

Commit Message

Oliver Neukum Sept. 15, 2020, 1:45 p.m. UTC
The SCSI layer can go into an ugly loop if you ignore that a device
is gone. You need to report an error in the command rather than
in the return value of the queue method.
We need to specifically check for ENODEV..

Signed-off-by: Oliver Neukum <oneukum@suse.com>
---
 drivers/usb/storage/uas.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

Comments

Greg Kroah-Hartman Sept. 15, 2020, 2 p.m. UTC | #1
On Tue, Sep 15, 2020 at 03:45:01PM +0200, Oliver Neukum wrote:
> The SCSI layer can go into an ugly loop if you ignore that a device
> is gone. You need to report an error in the command rather than
> in the return value of the queue method.
> We need to specifically check for ENODEV..
> 
> Signed-off-by: Oliver Neukum <oneukum@suse.com>
> ---
>  drivers/usb/storage/uas.c | 14 ++++++++++++--
>  1 file changed, 12 insertions(+), 2 deletions(-)

Should this one go to the stable kernels?

thanks,

greg k-h
Oliver Neukum Sept. 16, 2020, 8:11 a.m. UTC | #2
Am Dienstag, den 15.09.2020, 16:00 +0200 schrieb Greg KH:
> On Tue, Sep 15, 2020 at 03:45:01PM +0200, Oliver Neukum wrote:

> > The SCSI layer can go into an ugly loop if you ignore that a device

> > is gone. You need to report an error in the command rather than

> > in the return value of the queue method.

> > We need to specifically check for ENODEV..

> > 

> > Signed-off-by: Oliver Neukum <oneukum@suse.com>

> > ---

> >  drivers/usb/storage/uas.c | 14 ++++++++++++--

> >  1 file changed, 12 insertions(+), 2 deletions(-)

> 

> Should this one go to the stable kernels?


Hi,

in theory yes, but it depends on the previous patch.
Should I submit a separate patch?

	Regards
		Oliver
Greg Kroah-Hartman Sept. 16, 2020, 8:37 a.m. UTC | #3
On Wed, Sep 16, 2020 at 10:11:25AM +0200, Oliver Neukum wrote:
> Am Dienstag, den 15.09.2020, 16:00 +0200 schrieb Greg KH:
> > On Tue, Sep 15, 2020 at 03:45:01PM +0200, Oliver Neukum wrote:
> > > The SCSI layer can go into an ugly loop if you ignore that a device
> > > is gone. You need to report an error in the command rather than
> > > in the return value of the queue method.
> > > We need to specifically check for ENODEV..
> > > 
> > > Signed-off-by: Oliver Neukum <oneukum@suse.com>
> > > ---
> > >  drivers/usb/storage/uas.c | 14 ++++++++++++--
> > >  1 file changed, 12 insertions(+), 2 deletions(-)
> > 
> > Should this one go to the stable kernels?
> 
> Hi,
> 
> in theory yes, but it depends on the previous patch.
> Should I submit a separate patch?

Reordering these would be best, thanks!

greg k-h
diff mbox series

Patch

diff --git a/drivers/usb/storage/uas.c b/drivers/usb/storage/uas.c
index 842ca5c82091..19e730f9ad19 100644
--- a/drivers/usb/storage/uas.c
+++ b/drivers/usb/storage/uas.c
@@ -662,8 +662,7 @@  static int uas_queuecommand_lck(struct scsi_cmnd *cmnd,
 	if (devinfo->resetting) {
 		set_host_byte(cmnd, DID_ERROR);
 		cmnd->scsi_done(cmnd);
-		spin_unlock_irqrestore(&devinfo->lock, flags);
-		return 0;
+		goto zombie;
 	}
 
 	/* Find a free uas-tag */
@@ -699,6 +698,16 @@  static int uas_queuecommand_lck(struct scsi_cmnd *cmnd,
 		cmdinfo->state &= ~(SUBMIT_DATA_IN_URB | SUBMIT_DATA_OUT_URB);
 
 	err = uas_submit_urbs(cmnd, devinfo);
+	/*
+	 * in case of fatal errors the SCSI layer is peculiar
+	 * a command that has finished is a success for the purpose
+	 * of queueing, no matter how fatal the error
+	 */
+	if (err == -ENODEV) {
+		set_host_byte(cmnd, DID_ERROR);
+		cmnd->scsi_done(cmnd);
+		goto zombie;
+	}
 	if (err) {
 		/* If we did nothing, give up now */
 		if (cmdinfo->state & SUBMIT_STATUS_URB) {
@@ -709,6 +718,7 @@  static int uas_queuecommand_lck(struct scsi_cmnd *cmnd,
 	}
 
 	devinfo->cmnd[idx] = cmnd;
+zombie:
 	spin_unlock_irqrestore(&devinfo->lock, flags);
 	return 0;
 }