Re: [2.5.43] (DAC960 compile failure)

[Kevin Brosius]

Hi Dave,
  I'll give it a shot tonight.  Did you post the patch to the
linux-kernel mailing list?  I can't find it in the archives.  If not,
would you mail a copy to my home address (compuserve, or have you posted
it on a web location?)  If you'd like me to put up copies of the patch
for general consumption I'll make space available on a web page.

Kevin

> 
> I just posted a DAC960 patch for 2.5.42.  Tomorrow, I'll post
> a 2.5.43 patch. There are some minor changes needed (about three
> lines) for 2.5.43.  See the patch posting regarding ACPI
> interactions.
> 
> I'd love to hear any good or bad results you have with these
> patches.
> 
> Thanks!
> 
> Dave Olien
> Open Source Developement Lab
> 
> On Wed, Oct 16, 2002 at 07:28:07PM -0400, Kevin Brosius wrote:
> > > 
> > > looking at that i realise that DAC960 code in 2.5.43
> > > is not supposed to be tested:
> > > ======
> > > #error I am a non-portable driver, please convert me to use the \
> > > Documentation/DMA-mapping.txt interfaces ======
> > > am i right?
> > > 
> > > the following weirdo appears in both gcc-3.1 and 3.2 (also in 2.5.42)
> > > ======
> > > drivers/block/DAC960.c: In function `DAC960_DetectControllers':
> > > drivers/block/DAC960.c:2465: `Controller' undeclared (first use in this function)
> > > drivers/block/DAC960.c:2465: (Each undeclared identifier is reported only once
> > > drivers/block/DAC960.c:2465: for each function it appears in.)
> > > 
> > 
> > Yes, 2.5.42 did this also.  It looks like gcc 3.2 doesn't like goto's
> > which reference variables outside their native scope.  You can move the
> > Controller definition to full function scope to fix that error.
> > 
> > The DAC960 doesn't seem usable out of the stock kernel build though. 
> > You'll need to try patches previously posted to the list.  (Which don't
> > fully work for me either...)
> > 
> > -- 
> > Kevin