On Wed, 8 Sept 2021 at 01:09, Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote: > Hi all, > > On Wed, 8 Sep 2021 09:07:20 +1000 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote: > > > > Building Linus' tree, today's linux-next build (mips > > allmodconfig) failed like this: > > > > In file included from /kisskb/src/include/linux/compiler_types.h:65:0, > > from <command-line>:0: > > include/linux/compiler_attributes.h:29:29: error: "__GCC4_has_attribute___no_sanitize_coverage__" is not defined [-Werror=undef] > > # define __has_attribute(x) __GCC4_has_attribute_##x > > ^ > > include/linux/compiler-gcc.h:125:29: note: in expansion of macro '__has_attribute' > > #if defined(CONFIG_KCOV) && __has_attribute(__no_sanitize_coverage__) > > ^ > > cc1: all warnings being treated as errors > > > > Caused by commit > > > > 540540d06e9d ("kcov: add __no_sanitize_coverage to fix noinstr for all architectures") > > > > This ia a gcc 4.9 build, so presumably this? > > > > diff -ru a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h > > --- a/include/linux/compiler_attributes.h 2021-09-08 09:03:35.169225661 +1000 > > +++ b/include/linux/compiler_attributes.h 2021-09-08 09:05:47.790907780 +1000 > > @@ -36,6 +36,7 @@ > > # define __GCC4_has_attribute___no_profile_instrument_function__ 0 > > # define __GCC4_has_attribute___nonstring__ 0 > > # define __GCC4_has_attribute___no_sanitize_address__ 1 > > +# define __GCC4_has_attribute___no_sanitize_coverage__ 0 > > # define __GCC4_has_attribute___no_sanitize_undefined__ 1 > > # define __GCC4_has_attribute___fallthrough__ 0 > > #endif > > Just to be clear, I haven't tested the above in any way. I see this in next-20210907: "Compiler Attributes: fix __has_attribute(__no_sanitize_coverage__) for GCC 4" Which does the same fix. Not sure what happened to it. I would have also expected this to be merged as a fix into mainline by now? Miguel? Thanks, -- Marco