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

2013-11-27 Thread Laszlo Ersek
On 11/27/13 07:46, Gerd Hoffmann wrote: > >>> +i440fx->pci_info.w32.begin = ram_size; > >> But this patch also obliterates the high bound, 0xe000, which can >> lead to: >> - w32.end - w32.begin <= 512M, or >> - a special case of the former, w32.end < w32.begin. > > ram_size is not the to

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

2013-11-26 Thread Gerd Hoffmann
> > +i440fx->pci_info.w32.begin = ram_size; > But this patch also obliterates the high bound, 0xe000, which can > lead to: > - w32.end - w32.begin <= 512M, or > - a special case of the former, w32.end < w32.begin. ram_size is not the total amount of memory, it is low memory only. There i

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

2013-11-26 Thread Jordan Justen
On Tue, Nov 26, 2013 at 8:16 AM, 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. > > Sig

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

2013-11-26 Thread Laszlo Ersek
On 11/26/13 17:16, 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: Ge

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

2013-11-26 Thread Gerd Hoffmann
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 --- hw/pci-host/piix.c | 10 +-