Re: [PATCH v3 2/2][BNX2]: Add iSCSI support to BNX2 devices.

[FUJITA Tomonori <tomof@acm.org>]

On Tue, 25 Sep 2007 10:39:17 +0200
Hannes Reinecke <hare@suse.de> wrote:

> Hi Tomo,
> 
> FUJITA Tomonori wrote:
> > On Sat, 8 Sep 2007 13:00:36 +0100
> > Christoph Hellwig <hch@infradead.org> wrote:
> > 
> >> On Sat, Sep 08, 2007 at 07:32:27AM -0400, Jeff Garzik wrote:
> >>> FUJITA Tomonori wrote:
> >>>> Yeah, iommu code ignores the lld limitations (the problem is that the
> >>>> lld limitations are in request_queue and iommu code can't access to
> >>>> request_queue). There is no way to tell iommu code about the lld
> >>>> limitations.
> >>>
> >>> This fact very much wants fixing.
> >>
> >> Absolutely.  Unfortunately everyone wastes their time on creating workarounds
> >> instead of fixing the underlying problem.
> > 
> > Any ideas on how to fix this?
> > 
> > I chatted to Jens and James on this last week.
> > 
> > - we could just copies the lld limitations to device structure. it's
> > hacky but device structure already has hacky stuff.
> > 
> > - we could just link device structure to request_queue structure so
> > that iommu code can see request_queue structure.
> > 
> > - we could remove the lld limitations in request_queue strucutre and
> > have a new strucutre (something like struct io_restrictions). then
> > somehow we could link the new structure with request_queue and device
> > strucutres.
> > 
> I'd prefer the latter. These struct io_restrictions could then be used
> by dm (which has it's own version right now) to merge queue capabilities.

Yeah, we could nicely handle lld's restrictions (especially with
stacking devices). But iommu code needs only max_segment_size and
seg_boundary_mask, right? If so, the first simple approach to add two
values to device structure is not so bad, I think.