Re: sort out the ->eh_timed_out mess

[Bart Van Assche <Bart.VanAssche@sandisk.com>]

On Mon, 2017-01-30 at 13:18 +0100, Christoph Hellwig wrote:
> We originally only supported the methods in the transport_template,
> which led to problems with drivers needing it and having to fake
> up a template.  Then we added it to the host to avoid that issue,
> but because the transport template takes precedence we still have
> various placed that need to hack around it.
> 
> This series removes the transport_template instance and switches
> everyone to use the host method, similar to how we supply generic
> helpers elsewhere.
> 
> In addition it removes a bit of cruft in the same area.

Hello Christoph,

Patches 1, 2 and 4 of this series are a real improvement in my opinion.
However, I'm not sure whether patch 3 is the best way to avoid that drivers
have to re-override the transport template timeout handler. Have you
considered to modify the SCSI EH such that the host timeout handler has
priority over the transport timeout handler? I think that would result in
a much smaller patch 3/4.

Thanks,

Bart.