Re: [akpm-mm:mm-unstable 340/601] include/linux/compiler_types.h:702:38: error: call to '__compiletime_assert_501' declared with attribute error: BUILD_BUG failed

Kiryl Shutsemau <[email protected]> Mon, 27 Jul 2026 10:25:42 +0100
Newsgroups dev.linux.lists.oe-kbuild-all,org.kernel.vger.mm-commits,org.kvack.linux-mm
Message-ID <amcitKvUvFYr8W38@thinkstation>
On Sat, Jul 25, 2026 at 07:51:08PM -0700, Andrew Morton wrote:
> On Sat, 25 Jul 2026 08:56:41 +0800 kernel test robot <[email protected]> wrote:
> 
> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-unstable
> > head:   c04df384b55dd9dcf72c36b661becb5aaba371a9
> > commit: d2aed7c02fafacb9d055a2f683ac12e977ef16e0 [340/601] mm: preserve RWP marker across PTE rewrites
> > config: sparc64-randconfig-r134-20260725 (https://download.01.org/0day-ci/archive/20260725/[email protected]/config)
> > compiler: sparc64-linux-gcc (GCC) 8.5.0
> > sparse: v0.6.5-rc1
> > reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260725/[email protected]/reproduce)
> > 
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <[email protected]>
> > | Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
> > 
> > All errors (new ones prefixed by >>):
> > 
> >    In file included from <command-line>:
> >    mm/huge_memory.c: In function 'move_pages_huge_pmd':
> > >> include/linux/compiler_types.h:702:38: error: call to '__compiletime_assert_501' declared with attribute error: BUILD_BUG failed
> >      _compiletime_assert(condition, msg, __compiletime_assert_, __COUNTER__)
> >                                          ^
> >    include/linux/compiler_types.h:683:4: note: in definition of macro '__compiletime_assert'
> >        prefix ## suffix();    \
> 
> Well dang.
> 
> I can't reproduce with sparc64 gcc-15.2.0. 
> Documentation/process/changes.rst says we support 8,1, released in
> 2018.  Your 8.5 was released in 2021 so perhaps three other people are
> still using it :(
> 
> https://www.kernel.org/pub/tools/crosstool/ doesn't appear to have
> sparc64 gcc-8.5 so I gave up.
> 
> So I dunno.  I'm inclined to ignore and assume that if someone else hits
> this one day, they'll either fix it or provide us with enough details
> to figure it out.

I cannot reproduce it either. Maybe we can help compiler to derive that the
code is dead?

-- >8 --
From: "Kiryl Shutsemau (Meta)" <[email protected]>
Date: Mon, 27 Jul 2026 10:16:11 +0100
Subject: [PATCH] mm: fold userfaultfd_rwp() to false without
 CONFIG_ARCH_HAS_PTE_PROTNONE

RWP tracks accesses by installing PAGE_NONE (protnone) PTEs, so its code
paths are gated on userfaultfd_rwp(). Without CONFIG_ARCH_HAS_PTE_PROTNONE
there is no PAGE_NONE -- <linux/pgtable.h> defines it to a BUILD_BUG()
stub, relying on callers folding such paths to dead code via
IS_ENABLED(CONFIG_ARCH_HAS_PTE_PROTNONE).

userfaultfd_rwp() was not a compile-time constant, so the compiler could
not fold those paths. With an older compiler (gcc 8.5.0, sparc64) the
PAGE_NONE reference in move_pages_huge_pmd() survived to codegen:

  mm/huge_memory.c:2874: _dst_pmd = pmd_modify(_dst_pmd, PAGE_NONE);
  compiler_types.h:702: error: call to '__compiletime_assert_501'
    declared with attribute error: BUILD_BUG failed

RWP cannot exist without protnone, so return a compile-time false when
CONFIG_ARCH_HAS_PTE_PROTNONE is unset; every RWP path then folds away.

Reported-by: kernel test robot <[email protected]>
Signed-off-by: Kiryl Shutsemau (Meta) <[email protected]>
---
 include/linux/userfaultfd_k.h | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/include/linux/userfaultfd_k.h b/include/linux/userfaultfd_k.h
index bfbd6a59909f..0ce438ec294f 100644
--- a/include/linux/userfaultfd_k.h
+++ b/include/linux/userfaultfd_k.h
@@ -221,6 +221,12 @@ static inline bool userfaultfd_minor(struct vm_area_struct *vma)
 
 static inline bool userfaultfd_rwp(struct vm_area_struct *vma)
 {
+	/*
+	 * Callers gate PAGE_NONE usage on this; PAGE_NONE is a BUILD_BUG()
+	 * without CONFIG_ARCH_HAS_PTE_PROTNONE, so fold to false.
+	 */
+	if (!IS_ENABLED(CONFIG_ARCH_HAS_PTE_PROTNONE))
+		return false;
 	return vma_test_single_mask(vma, VMA_UFFD_RWP);
 }
 

base-commit: 505f6bc5c6c6db7d6390ea33c4e2ab44379e6f49
-- 
  Kiryl Shutsemau / Kirill A. Shutemov