Re: [PATCH] powerpc/e500: move qemu machine spec together with the rest

[Scott Wood <scottwood@freescale.com>]

On Mon, 2015-09-14 at 16:17 +0300, Laurentiu Tudor wrote:
> On 09/10/2015 02:01 AM, Scott Wood wrote:
> > On Fri, 2015-09-04 at 15:46 +0300, Laurentiu Tudor wrote:
> > > This way we get rid of an entire file with mostly
> > > duplicated code plus a Kconfig option that you always
> > > had to take care to check it in order for kvm to work.
> > > 
> > > Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com>
> > > ---
> > >  arch/powerpc/platforms/85xx/Kconfig           | 15 -----
> > >  arch/powerpc/platforms/85xx/Makefile          |  1 -
> > >  arch/powerpc/platforms/85xx/corenet_generic.c |  1 +
> > >  arch/powerpc/platforms/85xx/qemu_e500.c       | 85 --------------------
> > > ----
> > 
> > 
> > qemu_e500 is not only for corenet chips.  
> 
> That's too bad. :-(
> I remember discussions on dropping the e500v2 support at some point in time?
> 
> > We can add it to the defconfig (in fact I've been meaning to do so).
> 
> Or maybe just drop de KConfig option and
> wrap the file in an #ifdef CONFIG_KVM or something along these lines?
>  
> > > -static void __init qemu_e500_setup_arch(void)
> > > -{
> > > -     ppc_md.progress("qemu_e500_setup_arch()", 0);
> > > -
> > > -     fsl_pci_assign_primary();
> > > -     swiotlb_detect_4g();
> > 
> > Where is fsl_pci_assign_primary() in corenet_generic.c?  
> 
> This commit claims it's not needed and drops it:
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id}4d595dad30328bc6153e235d90f54c918fc127

That commit has nothing to do with QEMU.

> > At one point this 
> > was needed for QEMU's PCI implementation -- have you tested QEMU PCI 
> > without 
> > it?
> 
> Well, somehow i've (embarrassingly) messed up my initial tests.
> I've retested after seeing your comment and indeed this breaks pci under 
> qemu.
> Adding to the confusion, the commit above made me think that the removal 
> was safe.
> Why pci qemu doesn't work without the call to fsl_pci_assign_primary() is
> an interesting subject.

IIRC it has to do with QEMU not liking a BAR set to zero.

-Scott