Re: [PATCH 3/9] configure/meson: Only check sys/signal.h on non-Linux

[Peter Maydell]

On Wed, 13 Jan 2021 at 07:06, Thomas Huth <thuth@redhat.com> wrote:
>
> On 21/12/2020 01.53, Jiaxun Yang wrote:
> > signal.h is equlevant of sys/signal.h on Linux, musl would complain
> > wrong usage of sys/signal.h.
> >
> > In file included from /builds/FlyGoat/qemu/include/qemu/osdep.h:108,
> >                   from ../tests/qemu-iotests/socket_scm_helper.c:13:
> > /usr/include/sys/signal.h:1:2: error: #warning redirecting incorrect 
> > #include <sys/signal.h> to <signal.h> [-Werror=cpp]
> >      1 | #warning redirecting incorrect #include <sys/signal.h> to 
> > <signal.h>
> >        |  ^~~~~~~
> >
> > Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
> > ---
> >   meson.build | 5 ++++-
> >   1 file changed, 4 insertions(+), 1 deletion(-)
> >
> > diff --git a/meson.build b/meson.build
> > index 372576f82c..1ef8722b3a 100644
> > --- a/meson.build
> > +++ b/meson.build
> > @@ -841,7 +841,10 @@ config_host_data.set('HAVE_DRM_H', 
> > cc.has_header('libdrm/drm.h'))
> >   config_host_data.set('HAVE_PTY_H', cc.has_header('pty.h'))
> >   config_host_data.set('HAVE_SYS_IOCCOM_H', cc.has_header('sys/ioccom.h'))
> >   config_host_data.set('HAVE_SYS_KCOV_H', cc.has_header('sys/kcov.h'))
> > -config_host_data.set('HAVE_SYS_SIGNAL_H', cc.has_header('sys/signal.h'))
> > +if targetos != 'linux'
> > +  # signal.h is equlevant of sys/signal.h on Linux
> > +  config_host_data.set('HAVE_SYS_SIGNAL_H', cc.has_header('sys/signal.h'))
> > +endif
>
> Seems like it sys/signal.h was introduced for OpenBSD once (see commit
> 128ab2ff50a), so this new check should be fine.

The right way to fix this would be to apply
https://patchew.org/QEMU/20201027003848.10920-1-mforney@mforney.org/
which got reviewed but never applied (it might need a respin to work
with meson) which simply removes the sys/signal.h include and the
"does the header exist" test entirely -- there is no platform which
needs sys/signal.h.

thanks
-- PMM