Re: [PATCH] configure: Support modules for Windows

[LIU Zhiwei]

On 2021/11/5 下午8:22, Daniel P. Berrangé wrote:
> On Fri, Nov 05, 2021 at 08:14:29PM +0800, LIU Zhiwei wrote:
> > On 2021/11/5 下午5:30, Daniel P. Berrangé wrote:
> > > On Fri, Nov 05, 2021 at 11:20:02AM +0800, LIU Zhiwei wrote:
> > >
> > > Generally we expect a commit message describing what is being solved
> > > and any reasons for the approach taken.
> > This patch relaxes all constraints on modules for Windows.  It firstly
> > exports  all symbols of the
> > target executable to an import library. Then  all modules link with the
> > import library .
> >
> > > > Signed-off-by: LIU Zhiwei <zhiwei_liu@c-sky.com>
> > > > ---
> > > >    configure   |  7 +------
> > > >    meson.build | 34 ++++++++++++++++++++++++----------
> > > >    2 files changed, 25 insertions(+), 16 deletions(-)
> > > >
> > > > diff --git a/configure b/configure
> > > > index 33682cb971..6280e412a9 100755
> > > > --- a/configure
> > > > +++ b/configure
> > > > @@ -1422,7 +1422,7 @@ cat << EOF
> > > >      bsd-user        all BSD usermode emulation targets
> > > >      guest-agent     build the QEMU Guest Agent
> > > >      pie             Position Independent Executables
> > > > -  modules         modules support (non-Windows)
> > > > +  modules         modules support
> > > >      module-upgrades try to load modules from alternate paths for upgrades
> > > >      debug-tcg       TCG debugging (default is disabled)
> > > >      debug-info      debugging information
> > > > @@ -1698,11 +1698,6 @@ else
> > > >      QEMU_CFLAGS="$QEMU_CFLAGS -Wno-missing-braces"
> > > >    fi
> > > > -# Our module code doesn't support Windows
> > > > -if test "$modules" = "yes" && test "$mingw32" = "yes" ; then
> > > > -  error_exit "Modules are not available for Windows"
> > > > -fi
> > > > -
> > > >    # module_upgrades is only reasonable if modules are enabled
> > > >    if test "$modules" = "no" && test "$module_upgrades" = "yes" ; then
> > > >      error_exit "Can't enable module-upgrades as Modules are not enabled"
> > > > diff --git a/meson.build b/meson.build
> > > > index 47df10afc2..26195cb24f 100644
> > > > --- a/meson.build
> > > > +++ b/meson.build
> > > > @@ -2625,7 +2625,7 @@ block_mods = []
> > > >    softmmu_mods = []
> > > >    foreach d, list : modules
> > > >      foreach m, module_ss : list
> > > > -    if enable_modules and targetos != 'windows'
> > > > +    if enable_modules
> > > >          module_ss = module_ss.apply(config_all, strict: false)
> > > >          sl = static_library(d + '-' + m, [genh, module_ss.sources()],
> > > >                              dependencies: [modulecommon, 
> > > > module_ss.dependencies()], pic: true)
> > > > @@ -2658,7 +2658,7 @@ endforeach
> > > >    foreach d, list : target_modules
> > > >      foreach m, module_ss : list
> > > > -    if enable_modules and targetos != 'windows'
> > > > +    if enable_modules
> > > >          foreach target : target_dirs
> > > >            if target.endswith('-softmmu')
> > > >              config_target = config_target_mak[target]
> > > > @@ -2800,14 +2800,15 @@ common_ss.add(hwcore)
> > > >    ###########
> > > >    # Targets #
> > > >    ###########
> > > > -
> > > > -foreach m : block_mods + softmmu_mods
> > > > -  shared_module(m.name(),
> > > > -                name_prefix: '',
> > > > -                link_whole: m,
> > > > -                install: true,
> > > > -                install_dir: qemu_moddir)
> > > > -endforeach
> > > > +if targetos != 'windows'
> > > > +  foreach m : block_mods + softmmu_mods
> > > > +    shared_module(m.name(),
> > > > +                  name_prefix: '',
> > > > +                  link_whole: m,
> > > > +                  install: true,
> > > > +                  install_dir: qemu_moddir)
> > > > +  endforeach
> > > > +endif
> > > >    softmmu_ss.add(authz, blockdev, chardev, crypto, io, qmp)
> > > >    common_ss.add(qom, qemuutil)
> > > > @@ -2826,6 +2827,7 @@ common_all = static_library('common',
> > > >    feature_to_c = find_program('scripts/feature_to_c.sh')
> > > >    emulators = {}
> > > > +emulator = 0
> > > >    foreach target : target_dirs
> > > >      config_target = config_target_mak[target]
> > > >      target_name = config_target['TARGET_NAME']
> > > > @@ -2966,6 +2968,7 @@ foreach target : target_dirs
> > > >                   link_language: link_language,
> > > >                   link_depends: [block_syms, qemu_syms] + 
> > > > exe.get('link_depends', []),
> > > >                   link_args: link_args,
> > > > +               implib: true,
> > > >                   win_subsystem: exe['win_subsystem'])
> > > >        if targetos == 'darwin'
> > > > @@ -3023,6 +3026,17 @@ foreach target : target_dirs
> > > >        endif
> > > >      endforeach
> > > >    endforeach
> > > > +if (targetos == 'windows')
> > > > +  foreach m : block_mods + softmmu_mods
> > > > +    shared_module(m.name(),
> > > > +           name_prefix: '',
> > > > +           link_whole: m,
> > > > +           install: true,
> > > > +           dependencies: glib,
> > > > +           link_with: emulator,
> > > AFAIK from the docs meson wants "link_with" to be a list of
> > > libraries.
> > The emulator is an import library.
> >
> > There has been an executable variable named emulator in origin meson.build.
> > An implib keyword args has been  added in this patch when the executable
> > generates.
> > And I declare it to the out-most level  to make it visible for modules after
> > the build of targets.
> Oh, so the "emulator" variable is already present in the loop body
> from existing meson rules.
>
> This still feels odd though - QEMU has 30+ emulator binaries it
> can build. This code is linking modules to the implib of whichever
> happens to be the last emulator binary built.
>
> Does this really let the modules work against any of the emulator
> binaries  at runtime ?

Think about the Linux build. There don't exist different modules for 
different emulator binaries.
I guess it doesn't have much connection with the target.

Once again, may be it is not a right way. Hope to get for your guidance 
and help.

Thanks,
Zhiwei

> Regards,
> Daniel