[for-5.2] nsis: Fix build for 64 bit installer

Pass cpu instead of cpu_family to the NSIS installer script.

That script checks for "x86_64" which is the cpu value,
while cpu_family is "x86".

Signed-off-by: Stefan Weil <sw@weilnetz.de>
---
 meson.build | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

On Wed, 25 Nov 2020 at 19:23, Stefan Weil <sw@weilnetz.de> wrote:
>
> Pass cpu instead of cpu_family to the NSIS installer script.
>
> That script checks for "x86_64" which is the cpu value,
> while cpu_family is "x86".
>
> Signed-off-by: Stefan Weil <sw@weilnetz.de>
> ---
>  meson.build | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

Very unfortunate that this has only surfaced after rc3,
which should in theory be the cutoff point for 5.2 changes.
Is it possible to get something into the CI/make check
so that we find Windows installer issues sooner?

thanks
-- PMM

Am 25.11.20 um 20:28 schrieb Peter Maydell:

> On Wed, 25 Nov 2020 at 19:23, Stefan Weil <sw@weilnetz.de> wrote:
>> Pass cpu instead of cpu_family to the NSIS installer script.
>>
>> That script checks for "x86_64" which is the cpu value,
>> while cpu_family is "x86".
>>
>> Signed-off-by: Stefan Weil <sw@weilnetz.de>
>> ---
>>   meson.build | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
> Very unfortunate that this has only surfaced after rc3,
> which should in theory be the cutoff point for 5.2 changes.
> Is it possible to get something into the CI/make check
> so that we find Windows installer issues sooner?
>
> thanks
> -- PMM


I am afraid there are more open issues for QEMU on Windows. They can be 
fixed in 5.2.1, so don't hesitate to finish 5.2.0.

The Meson based build sets bindir=/qemu/. which does not work correctly 
in get_relocated_path().

Finding such issues would require a CI environment which not only builds 
QEMU for Windows, but also runs the results. Some basic tests could be 
done on Linux using Wine, more advanced tests would require a real 
Windows host.

Regards,
Stefan

On Wed, Nov 25, 2020 at 11:22 PM Stefan Weil <sw@weilnetz.de> wrote:

> Pass cpu instead of cpu_family to the NSIS installer script.
>
> That script checks for "x86_64" which is the cpu value,
> while cpu_family is "x86".
>
> Signed-off-by: Stefan Weil <sw@weilnetz.de>
>

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>

---
>  meson.build | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/meson.build b/meson.build
> index 5062407c70..8a99c948d3 100644
> --- a/meson.build
> +++ b/meson.build
> @@ -1956,7 +1956,7 @@ if host_machine.system() == 'windows'
>      '@OUTPUT@',
>      get_option('prefix'),
>      meson.current_source_dir(),
> -    host_machine.cpu_family(),
> +    host_machine.cpu(),
>      '--',
>      '-DDISPLAYVERSION=' + meson.project_version(),
>    ]
> --
> 2.29.2
>
>
>

On Wed, Nov 25, 2020 at 11:28 PM Peter Maydell <peter.maydell@linaro.org>
wrote:

> On Wed, 25 Nov 2020 at 19:23, Stefan Weil <sw@weilnetz.de> wrote:
> >
> > Pass cpu instead of cpu_family to the NSIS installer script.
> >
> > That script checks for "x86_64" which is the cpu value,
> > while cpu_family is "x86".
> >
> > Signed-off-by: Stefan Weil <sw@weilnetz.de>
> > ---
> >  meson.build | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
>
> Very unfortunate that this has only surfaced after rc3,
> which should in theory be the cutoff point for 5.2 changes.
>

There is also "[PATCH] qxl: fix segfault" that missed rc3.

On Wed, Nov 25, 2020 at 09:59:48PM +0100, Stefan Weil wrote:
> Am 25.11.20 um 20:28 schrieb Peter Maydell:
> 
> > On Wed, 25 Nov 2020 at 19:23, Stefan Weil <sw@weilnetz.de> wrote:
> > > Pass cpu instead of cpu_family to the NSIS installer script.
> > > 
> > > That script checks for "x86_64" which is the cpu value,
> > > while cpu_family is "x86".
> > > 
> > > Signed-off-by: Stefan Weil <sw@weilnetz.de>
> > > ---
> > >   meson.build | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > Very unfortunate that this has only surfaced after rc3,
> > which should in theory be the cutoff point for 5.2 changes.
> > Is it possible to get something into the CI/make check
> > so that we find Windows installer issues sooner?
> > 
> > thanks
> > -- PMM
> 
> 
> I am afraid there are more open issues for QEMU on Windows. They can be
> fixed in 5.2.1, so don't hesitate to finish 5.2.0.
> 
> The Meson based build sets bindir=/qemu/. which does not work correctly in
> get_relocated_path().
> 
> Finding such issues would require a CI environment which not only builds
> QEMU for Windows, but also runs the results. Some basic tests could be done
> on Linux using Wine, more advanced tests would require a real Windows host.

Even if we can't automate testing of the installer on a Windows machine,
we should definitely introduce a gitlab job that actually builds the
installer. This will ensure that we always have a valid installer exe
available, built from a well-defined set of distro packages that we
can easily download and test manually.

Regards,
Daniel