[PATCH v2 19/23] travis/gitlab/azure: Drop the -E flag

Tom Rini trini at konsulko.com
Tue Mar 17 17:42:51 CET 2020


On Tue, Mar 17, 2020 at 10:32:32AM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Mar 2020 at 10:22, Tom Rini <trini at konsulko.com> wrote:
> >
> > On Tue, Mar 17, 2020 at 10:11:52AM -0600, Simon Glass wrote:
> > > Hi Tom,
> > >
> > > On Mon, 16 Mar 2020 at 15:21, Tom Rini <trini at konsulko.com> wrote:
> > > >
> > > > On Sun, Mar 15, 2020 at 05:42:59PM -0600, Simon Glass wrote:
> > > >
> > > > > It doesn't seem to make sense to tell buildman to report warning as errors
> > > > > (thus ensuring there are no warnings) and then ignore the warnings.
> > > > >
> > > > > The simplist thing is to just drop the -E flag. This allows us to drop the
> > > > > check for exit code 129.
> > > > >
> > > > > Dropping -E is not enough to cover all warnings though. For example this
> > > > > warning:
> > > > >
> > > > >     ===================== WARNING ======================
> > > > >     This board does not use CONFIG_DM. CONFIG_DM will be
> > > > >     compulsory starting with the v2020.01 release.
> > > > >     Failure to update may result in board removal.
> > > > >     See doc/driver-model/migration.rst for more info.
> > > > >
> > > > > also causes buildman to return an exit code of 129. So use -W to suppress
> > > > > that, since otherwise the build will fail.
> > > > >
> > > > > Signed-off-by: Simon Glass <sjg at chromium.org>
> > > > > Fixes: 329f5ef51d2 (travis.yml: run buildman with option -E)
> > > >
> > > > Ah, we have something funny going on, or at least not clear enough.  We
> > > > need and want -E here as that causes us to build with -Werror and so
> > > > warnings become errors and the build fails.  We still ignore warnings
> > > > such as "go convert X to DM" (which is its own issue to deal with) and
> > > > also dtc warnings (which is its own issue to deal with).
> > >
> > > OK I see. So should we add both -E and -W? Perhaps the is the behaviour we want?
> >
> > I'm not sure.  We need -E, yes.  I guess I'm ambivalent on if we should
> > make buildman have an option to return 0 still here, or just have CI
> > know that specific error-code is OK as we've otherwise turned code
> > warnings to errors.
> >
> > Having typed that, I see the problem I have now.  We need to make it
> > clearer, in both the -W and -E cases what kinds of warnings we're
> > talking about.  -E enables -Werror to the C compiler.  -W suppresses the
> > kinds of WARNINGS we echo to console but I'm not sure how to best
> > characterize.
> 
> OK shall I beef up the help text in buildman for those two?

And the commit message for adding -W, yes, thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200317/6cdec314/attachment.sig>


More information about the U-Boot mailing list