Message ID | 20210518181247.GA10062@hyeyoo (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | mm, slub: Fix support for clang 10 | expand |
Nathan, Thank you for suggesting the idea and allowing me to write the patch. it compiled and booted without problem when compiled with gcc-10.2.0, clang-10.0.1, clang-11.0.0 with x86_64 default config. Thanks, Hyeonggon
And Naresh, Thank you for reporting the problem! If there's any problem with it or I can improve anything, any feedback will be welcomed. and I'll reply soon. Thanks, Hyeonggon
On Tue, May 18, 2021 at 11:12 AM Hyeonggon Yoo <42.hyeyoo@gmail.com> wrote: > > Previously in 'commit ff3daafe3fd3 ("mm, slub: change run-time assertion > in kmalloc_index() to compile-time")', changed kmalloc_index's run-time > assertion to compile-time assertion. > > But clang 10 has a bug misevaluating __builtin_constant_p() as true, > making it unable to compile. This bug was fixed in clang 11. > > To support clang 10, introduce a macro to do run-time assertion if clang > version is less than 11, even if the size is constant. Might revert this > commit later if we choose not to support clang 10. > > Fixes: ff3daafe3fd3 ("mm, slub: change run-time assertion in kmalloc_index() to compile-time") > Link: https://lore.kernel.org/r/CA+G9fYvYxqVhUTkertjZjcrUq8LWPnO7qC==Wum3gYCwWF9D6Q@mail.gmail.com/ > Link: https://lkml.org/lkml/2021/5/11/872 > Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> > Suggested-by: Nathan Chancellor <nathan@kernel.org> > Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Thanks for the patch; hopefully this isn't too much burden in order to help us support clang-10 a little bit longer. We will rip it out when dropping clang-10 someday; grepping for CLANG_VERSION should turn this up easily. Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> > --- > include/linux/slab.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/include/linux/slab.h b/include/linux/slab.h > index 9d316aac0aba..8d8dd8571261 100644 > --- a/include/linux/slab.h > +++ b/include/linux/slab.h > @@ -413,7 +413,7 @@ static __always_inline unsigned int __kmalloc_index(size_t size, > if (size <= 16 * 1024 * 1024) return 24; > if (size <= 32 * 1024 * 1024) return 25; > > - if (size_is_constant) > + if ((IS_ENABLED(CONFIG_CC_IS_GCC) || CONFIG_CLANG_VERSION >= 110000) && size_is_constant) > BUILD_BUG_ON_MSG(1, "unexpected size in kmalloc_index()"); > else > BUG(); > -- > 2.25.1 >
diff --git a/include/linux/slab.h b/include/linux/slab.h index 9d316aac0aba..8d8dd8571261 100644 --- a/include/linux/slab.h +++ b/include/linux/slab.h @@ -413,7 +413,7 @@ static __always_inline unsigned int __kmalloc_index(size_t size, if (size <= 16 * 1024 * 1024) return 24; if (size <= 32 * 1024 * 1024) return 25; - if (size_is_constant) + if ((IS_ENABLED(CONFIG_CC_IS_GCC) || CONFIG_CLANG_VERSION >= 110000) && size_is_constant) BUILD_BUG_ON_MSG(1, "unexpected size in kmalloc_index()"); else BUG();
Previously in 'commit ff3daafe3fd3 ("mm, slub: change run-time assertion in kmalloc_index() to compile-time")', changed kmalloc_index's run-time assertion to compile-time assertion. But clang 10 has a bug misevaluating __builtin_constant_p() as true, making it unable to compile. This bug was fixed in clang 11. To support clang 10, introduce a macro to do run-time assertion if clang version is less than 11, even if the size is constant. Might revert this commit later if we choose not to support clang 10. Fixes: ff3daafe3fd3 ("mm, slub: change run-time assertion in kmalloc_index() to compile-time") Link: https://lore.kernel.org/r/CA+G9fYvYxqVhUTkertjZjcrUq8LWPnO7qC==Wum3gYCwWF9D6Q@mail.gmail.com/ Link: https://lkml.org/lkml/2021/5/11/872 Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Suggested-by: Nathan Chancellor <nathan@kernel.org> Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> --- include/linux/slab.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)