Re: mpt on sparc64

[Cameron Berkenpas <cam%eleventhhourfx.com@localhost>]

Well, it's scsiconf.h that references the device struct, not mpt.h.

So shouldn't scsiconf.h include sys/device.h?

I did find it strange that a struct was being used directly rather than being 
wrapped around a typedef, your email explains that though...

On Thursday 22 May 2008 5:54:54 am Michael Lorenz wrote:
> Hello,
>
> On May 22, 2008, at 02:56, Cameron Berkenpas wrote:
> > I'm not sure if this is the correct place to address this but here
> > it goes...
> >
> > The mpt driver doesn't compile for sparc64 (and probably other
> > architectures
> > as well).
> >
> > The error is on line 65 of scsiconf.h: "struct device is an
> > incomplete type"
> > That just sounds to me like gcc is only seeing prototypes for the
> > device
> > struct but no actual implementations. I found the implementation
> > in /usr/include/sys/device.h and included this file in scsiconf.h. The
> > compile completed successfully. Unfortunately I do not have a scsi
> > card that the mpt driver supports yet (I ordered one and it hasn't
> > arrived
> > yet). I can't imagine this is the correct fix to this problem...
>
> Well, the driver itself should #include <sys/device.h>. struct device
> is supposed to be replaced by device_t anyway - this certainly
> doesn't look like a 64bit problem at all, more like an incomplete
> conversion to device_t usage.
>
> have fun
> Michael