Re: [Qemu-devel] [PATCH v2 0/5]: Convert pci_info() to QObject

[Blue Swirl]

On Thu, Jan 21, 2010 at 1:52 PM, Luiz Capitulino <address@hidden> wrote:
> On Wed, 20 Jan 2010 19:22:48 +0000
> Blue Swirl <address@hidden> wrote:
>
>> On Wed, Jan 20, 2010 at 6:11 PM, Luiz Capitulino <address@hidden> wrote:
>> > On Wed, 20 Jan 2010 18:57:56 +0100
>> > Markus Armbruster <address@hidden> wrote:
>> >
>> >> Luiz Capitulino <address@hidden> writes:
>> >>
>> >> >  Hi,
>> >> >
>> >> >  This new version addresses Markus's comments.
>> >> >
>> >> > changelog
>> >> > ---------
>> >> >
>> >> > V1 -> V2
>> >> >
>> >> > - Make class_info's key 'desc' optional
>> >> > - Better indentation
>> >> > - Doc fixes
>> >> >
>> >> > V0 -> V1
>> >> >
>> >> > - Coding style fixes
>> >> > - Make 'BAR' and 'IRQ' keys lowercase
>> >> > - Add 'irq' key to the documentation
>> >> >
>> >> >  Thanks.
>> >>
>> >> Looks good, although one comment still applies: PATCH 3/5 regresses info
>> >> pci, 4/5 and 5/5 fix it.  Do we care?  They're separate because they're
>> >> untested.
>> >
>> >  There are two problems here, which apply for those whom emulate
>> > the hardware:
>> >
>> > 1. 'info pci' output will brake with git bisect
>> >
>> > 2. As the code is untested, it might be broken
>> >
>> >  Only two 2 seems serious.
>> >
>> >  Michael, does the sparc image on qemu.org have the hardware
>> > in question (pci bridge)?
>>
>> Sparc64 has two Simba bridges, but currently they are broken so there
>> are no devices behind them. In addition there should be a DEC 21154
>> bridge.
>
>  Can the DEC one have devices attached to it?

It should, but sun4u machine does not even create it yet. There are
two DEC 21154 models, located in grackle_pci.c and unin_pci.c (PPC).
They should be unified and given a file of their own so they could be
used also for Sparc64.

>> There is no Sparc64 test image yet (very few Sparc32 machines did have
>> any PCI and we don't emulate them), but you can test the output
>> without any images:
>
>  Oh man, thanks a lot!
>
>  I got an abort() :)
>