Re: [Qemu-devel] qemu 1.0 breaks PCI on NetBSD (bisected)

[Avi Kivity]

On 04/14/2012 05:48 PM, Michael Tokarev wrote:
> On 14.04.2012 18:32, Michael Tokarev wrote:
> > On 14.04.2012 18:17, Michael Tokarev wrote:
> >> Something in between 0.15 and 1.0 releases of qemu broke
> >> all NetBSD PCI functions.  This is visible in the lack of
> >> any network devices in a NetBSD guest, but actually there's
> >> no PCI devices at all -- e.g. disk is being assessed using
> >> legacy ways.
> >>
> >> Bisection shows this commit as guilty:
> >>
> >> commit d0ed8076cbdc26138a7e33fed5e45a35d019a103
> >> Author: Avi Kivity <address@hidden>
> >> Date:   Sun Jul 24 17:47:18 2011 +0300
> >>
> >>     pci_host: convert conf index and data ports to memory API
> >>
> >>     Reviewed-by: Richard Henderson  <address@hidden>
> >>     Signed-off-by: Avi Kivity <address@hidden>
> >>
> >> Is there something wrong with this commit or maybe
> >> with NetBSD?  The interesting thing here is that so
> >> far, apparently, only NetBSD is affected, even other
> >> variants of *BSD works fine...
> > 
> > Okay, this has been fixed in 1.0.1, as discussed in this
> > thread:
> > 
> >  http://thread.gmane.org/gmane.comp.emulators.qemu/130695
> > 
> > but there's still an error somewhere...  Current git
> > version does not work again, with exactly the same
> > sympthoms...  Digging further...
>
> Because neither current git nor stable 1.0.1 picked
> up this patch apparently.  So we've broken netbsd since
> 1.0.... Oh well... :(
>
> Should it be included, finally?
>

I'll send a pull request.

-- 
error compiling committee.c: too many arguments to function