[PATCH RFC next 18/18] buildman: set W flag to e to be consistent with CONFIG_WERROR

[Simon Glass]

Hi Quentin,

On Tue, 5 May 2026 at 05:29, Quentin Schulz <quentin.schulz at cherry.de> wrote:
>
> Hi Simon,
>
> On 5/4/26 10:38 PM, Simon Glass wrote:
> > Hi Quentin,
> >
> > On 2026-05-04T13:20:27, Quentin Schulz <foss+uboot at 0leil.net> wrote:
> >> buildman: set W flag to e to be consistent with CONFIG_WERROR
> >>
> >> Buildman currently only enables Werror for KBUILD_CFLAGS (via KCFLAGS)
> >> and KBUILD_HOSTCFLAGS (via HOSTCFLAGS) but nothing else.
> >>
> >> scripts/Makefile.extrawarn now supports --fatal-warnings for the linker
> >> and assembler, which are enabled whenever CONFIG_WERROR is set.
> >>
> >> Let's make buildman set this variable to enable even more checks.
> >> KBUILD_CFLAGS and KBUILD_HOSTCFLAGS are kept in case buildman is run on
> >> a git-range where W=e is not available, it's better than nothing.
> >>
> >> Signed-off-by: Quentin Schulz <quentin.schulz at cherry.de>
> >>
> >> tools/buildman/builderthread.py | 1 +
> >>   1 file changed, 1 insertion(+)
> >
> >> diff --git a/tools/buildman/builderthread.py b/tools/buildman/builderthread.py
> >> @@ -233,6 +233,7 @@ class BuilderThread(threading.Thread):
> >>           if self.builder.warnings_as_errors:
> >>               args.append('KCFLAGS=-Werror')
> >>               args.append('HOSTCFLAGS=-Werror')
> >> +            args.append('W=e')
> >
> > Please update test_warnings_as_errors() in tools/buildman/func_test.py
> > to assert W=e is present in the make args alongside the existing
> > KCFLAGS / HOSTCFLAGS checks. Otherwise the new behaviour is silently
> > untested.
> >
>
> This test is only available in your fork of U-Boot. I'll therefore
> ignore this comment.

Oh yes, I haven't upstreamed those extra tests yet. I'll take a look
when this series is in. I sent an email about buildman if you have any
thoughts on that.

Regards,
Simon