[LSF/MM TOPIC] Integrated Unit Attention handling

[LSF/MM TOPIC] Integrated Unit Attention handling

[Hannes Reinecke]

Hi all,

I'd like to discusss 'Integrated Unit Attention handling' at LSF/MM
2015.

Currently we're interpreting some Unit Attention codes and send out
uevents, with the hope that some userspace application will make use
of this.
However, some UAs are easily handled within the kernel directly, eg

- REPORT LUN DATA CHANGED: this could easily trigger a rescan of the
  scsi host driving that LUN.
- INQUIRY DATA CHANGED: Rescan the VPD pages and INQUIRY data itself

When implementing such an integrated UA handling, however, we would
need to revisit the design choice of having the inquiry data fixed
to struct scsi_device. Currently any modification to the inquiry
data requires a complete device removal; this is not really feasible
for integrated event handling.

I'd like to discuss if such an integrated UA handling is desirable
and, if so, if devices should become flexible to have the inquiry
data changed on the fly.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		               zSeries & Storage
hare@suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [Lsf-pc] [LSF/MM TOPIC] Integrated Unit Attention handling

[Christoph Hellwig]

Sounds like a good topic, altough I hope we'll see considerable
progress before LSF on this item..

Re: [Lsf-pc] [LSF/MM TOPIC] Integrated Unit Attention handling

[Hannes Reinecke]

On 12/16/2014 09:32 AM, Christoph Hellwig wrote:
> Sounds like a good topic, altough I hope we'll see considerable
> progress before LSF on this item..
> 
Full integration requires a change of the SCSI device model;
currently struct scsi_device has a fixed type and inquiry data, so
rescan is essentially a no-op.
So the current model requires the complete removal of the device and
a rescan on the host.
Doing that from within the SCSI stack will open quite some
interesting race conditions, let alone interesting userspace
interactions.
Alternatively we can change the model to allow for a change of the
inquiry data, but that should be discussed.

And then there is the Power-On/Reset handling, which probably would
warrant a discussion on its own ...

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		               zSeries & Storage
hare@suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [Lsf-pc] [LSF/MM TOPIC] Integrated Unit Attention handling

[Christoph Hellwig]

On Tue, Dec 16, 2014 at 04:30:34PM +0100, Hannes Reinecke wrote:
> On 12/16/2014 09:32 AM, Christoph Hellwig wrote:
> > Sounds like a good topic, altough I hope we'll see considerable
> > progress before LSF on this item..
> > 
> Full integration requires a change of the SCSI device model;
> currently struct scsi_device has a fixed type and inquiry data, so
> rescan is essentially a no-op.
> So the current model requires the complete removal of the device and
> a rescan on the host.
> Doing that from within the SCSI stack will open quite some
> interesting race conditions, let alone interesting userspace
> interactions.
> Alternatively we can change the model to allow for a change of the
> inquiry data, but that should be discussed.

For now we can just react to new LUNs or size changes, which would be
what most people want, and offline devices for anything else.

That alone would be a gigantic improvement.

Re: [Lsf-pc] [LSF/MM TOPIC] Integrated Unit Attention handling

[Bart Van Assche]

On 12/16/14 16:30, Hannes Reinecke wrote:
> On 12/16/2014 09:32 AM, Christoph Hellwig wrote:
>> Sounds like a good topic, altough I hope we'll see considerable
>> progress before LSF on this item..
>
> Full integration requires a change of the SCSI device model;
> currently struct scsi_device has a fixed type and inquiry data, so
> rescan is essentially a no-op.
> So the current model requires the complete removal of the device and
> a rescan on the host.
> Doing that from within the SCSI stack will open quite some
> interesting race conditions, let alone interesting userspace
> interactions.
> Alternatively we can change the model to allow for a change of the
> inquiry data, but that should be discussed.
> 
> And then there is the Power-On/Reset handling, which probably would
> warrant a discussion on its own ...

Hello Hannes and Christoph,

If any of the topics raised by Hannes have not been resolved before the
LSF/MM starts I'm definitely interested in attending this discussion.

Bart.