[Qemu-devel] [PATCH] build: disable Xen on ARM

[Qemu-devel] [PATCH] build: disable Xen on ARM

[Paolo Bonzini]

While ARM could present the xenpv machine, it does not and trying to enable
it breaks compilation.

Fixes: 3b6b75506de44c5070639943c30a0ad5850f5d02
Reported-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 configure | 1 -
 1 file changed, 1 deletion(-)

diff --git a/configure b/configure
index a2bec60a97..97b02244fe 100755
--- a/configure
+++ b/configure
@@ -203,7 +203,6 @@ supported_xen_target() {
     test "$xen" = "yes" || return 1
     glob "$1" "*-softmmu" || return 1
     case "${1%-softmmu}:$cpu" in
-        arm:arm | aarch64:aarch64 | \
         i386:i386 | i386:x86_64 | x86_64:i386 | x86_64:x86_64)
             return 0
         ;;
-- 
2.13.0

Re: [Qemu-devel] [PATCH] build: disable Xen on ARM

[Peter Maydell]

On 11 July 2017 at 09:23, Paolo Bonzini <pbonzini@redhat.com> wrote:
> While ARM could present the xenpv machine, it does not and trying to enable
> it breaks compilation.
>
> Fixes: 3b6b75506de44c5070639943c30a0ad5850f5d02
> Reported-by: Alex Bennée <alex.bennee@linaro.org>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>  configure | 1 -
>  1 file changed, 1 deletion(-)
>
> diff --git a/configure b/configure
> index a2bec60a97..97b02244fe 100755
> --- a/configure
> +++ b/configure
> @@ -203,7 +203,6 @@ supported_xen_target() {
>      test "$xen" = "yes" || return 1
>      glob "$1" "*-softmmu" || return 1
>      case "${1%-softmmu}:$cpu" in
> -        arm:arm | aarch64:aarch64 | \
>          i386:i386 | i386:x86_64 | x86_64:i386 | x86_64:x86_64)
>              return 0
>          ;;

Does this actually do the right thing? It's still testing
target-cpu:guest-cpu, if I'm reading it correctly, whereas
previously we only looked at target-cpu to decide whether
to set CONFIG_XEN.

In particular, I thought that for aarch64/arm Xen setups
we would end up building an i386-softmmu target on an
arm/aarch64 host and wanted CONFIG_XEN to be set in that
setup ?  (I could be wrong there -- cc'ing Stefan and
Anthony to check.)

thanks
-- PMM

Re: [Qemu-devel] [PATCH] build: disable Xen on ARM

[Paolo Bonzini]

On 11/07/2017 11:53, Peter Maydell wrote:
> On 11 July 2017 at 09:23, Paolo Bonzini <pbonzini@redhat.com> wrote:
>> While ARM could present the xenpv machine, it does not and trying to enable
>> it breaks compilation.
>>
>> Fixes: 3b6b75506de44c5070639943c30a0ad5850f5d02
>> Reported-by: Alex Bennée <alex.bennee@linaro.org>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>>  configure | 1 -
>>  1 file changed, 1 deletion(-)
>>
>> diff --git a/configure b/configure
>> index a2bec60a97..97b02244fe 100755
>> --- a/configure
>> +++ b/configure
>> @@ -203,7 +203,6 @@ supported_xen_target() {
>>      test "$xen" = "yes" || return 1
>>      glob "$1" "*-softmmu" || return 1
>>      case "${1%-softmmu}:$cpu" in
>> -        arm:arm | aarch64:aarch64 | \
>>          i386:i386 | i386:x86_64 | x86_64:i386 | x86_64:x86_64)
>>              return 0
>>          ;;
> 
> Does this actually do the right thing? It's still testing
> target-cpu:guest-cpu, if I'm reading it correctly, whereas
> previously we only looked at target-cpu to decide whether
> to set CONFIG_XEN.
> 
> In particular, I thought that for aarch64/arm Xen setups
> we would end up building an i386-softmmu target on an
> arm/aarch64 host and wanted CONFIG_XEN to be set in that
> setup ?

Yes, that's correct.

Paolo

Re: [Qemu-devel] [PATCH] build: disable Xen on ARM

[Stefano Stabellini]

On Tue, 11 Jul 2017, Paolo Bonzini wrote:
> On 11/07/2017 11:53, Peter Maydell wrote:
> > On 11 July 2017 at 09:23, Paolo Bonzini <pbonzini@redhat.com> wrote:
> >> While ARM could present the xenpv machine, it does not and trying to enable
> >> it breaks compilation.
> >>
> >> Fixes: 3b6b75506de44c5070639943c30a0ad5850f5d02
> >> Reported-by: Alex Bennée <alex.bennee@linaro.org>
> >> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> >> ---
> >>  configure | 1 -
> >>  1 file changed, 1 deletion(-)
> >>
> >> diff --git a/configure b/configure
> >> index a2bec60a97..97b02244fe 100755
> >> --- a/configure
> >> +++ b/configure
> >> @@ -203,7 +203,6 @@ supported_xen_target() {
> >>      test "$xen" = "yes" || return 1
> >>      glob "$1" "*-softmmu" || return 1
> >>      case "${1%-softmmu}:$cpu" in
> >> -        arm:arm | aarch64:aarch64 | \
> >>          i386:i386 | i386:x86_64 | x86_64:i386 | x86_64:x86_64)
> >>              return 0
> >>          ;;
> > 
> > Does this actually do the right thing? It's still testing
> > target-cpu:guest-cpu, if I'm reading it correctly, whereas
> > previously we only looked at target-cpu to decide whether
> > to set CONFIG_XEN.
> > 
> > In particular, I thought that for aarch64/arm Xen setups
> > we would end up building an i386-softmmu target on an
> > arm/aarch64 host and wanted CONFIG_XEN to be set in that
> > setup ?
> 
> Yes, that's correct.

Yes, you are right. The patch is good, thanks!