Re: [Qemu-devel] [PATCH v2] piix: fix 32bit pci hole

[Michael Roth <mdroth@linux.vnet.ibm.com>]

Quoting Michael S. Tsirkin (2014-01-12 00:50:13)
> On Fri, Jan 10, 2014 at 09:28:39PM +0100, Laszlo Ersek wrote:
> > On 11/28/13 17:03, Laszlo Ersek wrote:
> > > Mike,
> > > 
> > > On 11/27/13 12:57, Gerd Hoffmann wrote:
> > >> Make the 32bit pci hole start at end of ram, so all possible address
> > >> space is covered.  Of course the firmware can use less than that.
> > >> Leaving space unused is no problem, mapping pci bars outside the
> > >> hole causes problems though.
> > >>
> > >> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
> > >> ---
> > >>  hw/pci-host/piix.c | 10 +---------
> > >>  1 file changed, 1 insertion(+), 9 deletions(-)
> > >>
> > >> diff --git a/hw/pci-host/piix.c b/hw/pci-host/piix.c
> > >> index edc974e..8e41ac1 100644
> > >> --- a/hw/pci-host/piix.c
> > >> +++ b/hw/pci-host/piix.c
> > >> @@ -345,15 +345,7 @@ PCIBus *i440fx_init(PCII440FXState **pi440fx_state,
> > >>      f->ram_memory = ram_memory;
> > >>  
> > >>      i440fx = I440FX_PCI_HOST_BRIDGE(dev);
> > >> -    /* Set PCI window size the way seabios has always done it. */
> > >> -    /* Power of 2 so bios can cover it with a single MTRR */
> > >> -    if (ram_size <= 0x80000000) {
> > >> -        i440fx->pci_info.w32.begin = 0x80000000;
> > >> -    } else if (ram_size <= 0xc0000000) {
> > >> -        i440fx->pci_info.w32.begin = 0xc0000000;
> > >> -    } else {
> > >> -        i440fx->pci_info.w32.begin = 0xe0000000;
> > >> -    }
> > >> +    i440fx->pci_info.w32.begin = pci_hole_start;
> > >>  
> > >>      memory_region_init_alias(&f->pci_hole, OBJECT(d), "pci-hole", f->pci_address_space,
> > >>                               pci_hole_start, pci_hole_size);
> > >>
> > > 
> > > please pick this up for 1.7.1.
> > > 
> > > 1.7.0 has been released without this patch, also without etc/pci-info,
> > > but with etc/acpi/tables.
> > > 
> > > For OVMF to work with "etc/acpi/tables" correctly, with eg. a guest RAM
> > > size of 2560MB, OVMF needs:
> > > - either this patch in qemu, or
> > > - etc/pci-info (which won't come back), or
> > > - a hack in OVMF that mimicks the same 0x80000000/0xc0000000/0xe0000000
> > >   logic (which I won't add).
> > 
> > Nominating this for v1.7.1 again.
> > 
> > The qemu-2.0 version (ie. a forward-port) of this patch has been merged as
> > 
> >   ddaaefb piix: fix 32bit pci hole
> > 
> > If necessary I can resubmit the v1.7.1 patch.
> > 
> > Thanks,
> > Laszlo
> 
> 
> It seems cleaner to backport this into stable.
> Mike, can you pick this up please?

Sure, will queue this up for 1.7.1 and backport from upstream

> 
> -- 
> MST