[libvirt] [PATCH] qemu: Support OVMF on aarch64 guests

[Cole Robinson]

On 11/19/2014 11:22 AM, Daniel P. Berrange wrote:
> On Wed, Nov 19, 2014 at 11:17:30AM -0500, Cole Robinson wrote:
>> On 11/19/2014 11:13 AM, Daniel P. Berrange wrote:
>>> On Wed, Nov 19, 2014 at 10:40:09AM -0500, Cole Robinson wrote:
>>>> On 11/19/2014 10:30 AM, Michal Privoznik wrote:
>>>>> Currently, we are whitelisting architectures, that we know how to run
>>>>> OVMF on. So far, only x86_64 was enabled. However, looking at qemu
>>>>> code, the same commandline can be used to enable OVMF for aarch64.
>>>>>
>>>>> Signed-off-by: Michal Privoznik <mprivozn at redhat.com>
>>>>> ---
>>>>>   src/qemu/qemu_command.c | 3 ++-
>>>>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c
>>>>> index d2e6991..ca57e35 100644
>>>>> --- a/src/qemu/qemu_command.c
>>>>> +++ b/src/qemu/qemu_command.c
>>>>> @@ -7749,7 +7749,8 @@ qemuBuildDomainLoaderCommandLine(virCommandPtr cmd,
>>>>>
>>>>>       case VIR_DOMAIN_LOADER_TYPE_PFLASH:
>>>>>           /* UEFI is supported only for x86_64 currently */
>>>>> -        if (def->os.arch != VIR_ARCH_X86_64) {
>>>>> +        if (def->os.arch != VIR_ARCH_X86_64 &&
>>>>> +            def->os.arch != VIR_ARCH_AARCH64) {
>>>>>               virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
>>>>>                              _("pflash is not supported for %s guest architecture"),
>>>>>                              virArchToString(def->os.arch));
>>>>>
>>>>
>>>> Please add armv7hl as well, it should work completely identically (if/when
>>>> we have an OS that supports it). ACK with that
>>>
>>> Really ? I thought ARM7 world was going to use its legacy BIOS
>>> approach forever, only AArch64 going for UEFI approach.
>>
>> There is arm32 support in UEFI, but I don't know if distros are ever going
>> to do the work of adopting it, because real hw is all u-boot based.
>>
>> But -M virt is very similar regardless of aarch64 or arm32, so _if_ anyone
>> ever produces an arm32 disk image with uefi boot support, the qemu command
>> line should be identical to the aarch64 WRT uefi/nvram/pflash. That's my
>> understanding anyways
>
> Ok, I guess it doesn't hurt to have it enabled for arm7 then, even if
> no one is likely to use it
>

Agreed

though frankly I don't really understand the point of restricting it in 
libvirt code to x86 in the first place. if we hadn't done that, we wouldn't 
need this patch for aarch64. Hence my original patch to just drop the arch 
check entirely

I understand sometimes detecting error conditions in libvirt before qemu can 
throw an error is important for improving error reporting. But we should be 
careful about trying to get into the game of predicting what will and won't 
work with qemu, it's just more code that needs to be maintained and kept up to 
date. Just my 2 cents

- Cole