Re: [PATCH] x86/boot/compressed: Disable jump tables for clang

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Jun 24, 2026 at 11:36:46AM +0200, Ingo Molnar wrote:
> 
> * Nathan Chancellor <nathan@kernel.org> wrote:
> 
> > After a recent upstream LLVM change to start generating jump and lookup
> > tables in switch statements in more instances [1], linking the
> > compressed x86 boot image when CONFIG_KERNEL_ZSTD is enabled fails with:
> > 
> >   ld.lld: error: Unexpected run-time relocations (.rela) detected!
> > 
> > Dumping the relocations in misc.o, which is the only file influenced by
> > CONFIG_KERNEL_ZSTD in the decompressor, shows dynamic relocations to
> > some string constants, which correspond to the string literals in the
> > switch statement in handle_zstd_error():
> > 
> >   Relocation section '.rela.data.rel.ro' at offset 0x277b0 contains 31 entries:
> >       Offset             Info             Type               Symbol's Value  Symbol's Name + Addend
> >   0000000000000000  0000006600000001 R_X86_64_64            0000000000000000 .rodata.str1.1 + 73a
> >   0000000000000008  0000006600000001 R_X86_64_64            0000000000000000 .rodata.str1.1 + 78e
> >   0000000000000010  0000006600000001 R_X86_64_64            0000000000000000 .rodata.str1.1 + 78e
> >   0000000000000018  0000006600000001 R_X86_64_64            0000000000000000 .rodata.str1.1 + 78e
> >   ...
> > 
> > This optimization is problematic for the decompressor environment, as it
> > is built as -fPIE without any explicit absolute references (as described
> > at the top of misc.c) while not applying any dynamic relocations, hence
> > the linker assertion. To opt out of this optimization, which is of
> > little value in this special early boot code, disable jump tables in the
> > decompressor when building with clang. This mirrors the other x86
> > startup code in arch/x86/boot/startup.
> > 
> > Cc: stable@vger.kernel.org
> > Closes: https://github.com/ClangBuiltLinux/linux/issues/2165
> > Link: https://github.com/llvm/llvm-project/commit/fa02a6ed66b1700c996b49c96c6bc0eb014c9518 [1]
> > Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> > ---
> >  arch/x86/boot/compressed/Makefile | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/arch/x86/boot/compressed/Makefile b/arch/x86/boot/compressed/Makefile
> > index 07e0e64b9a98..1c0d29e3eeba 100644
> > --- a/arch/x86/boot/compressed/Makefile
> > +++ b/arch/x86/boot/compressed/Makefile
> > @@ -31,6 +31,7 @@ KBUILD_CFLAGS += -Wundef
> >  KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING
> >  cflags-$(CONFIG_X86_32) := -march=i386
> >  cflags-$(CONFIG_X86_64) := -mcmodel=small -mno-red-zone
> > +cflags-$(CONFIG_CC_IS_CLANG) += -fno-jump-tables
> 
> So, shouldn't we just use -fno-jump-tables for *all* compilers,
> like we do in arch/x86/boot/startup/Makefile?
> 
> The point wouldn't be to just work around any Clang
> jump-table optimization complications alone, but also
> to synchronize the build options of very early code and such.

I'm sitting on a patch to unconditionally disable jump-tables for
x86_64:

  https://web.git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git/log/?h=x86/syscall

I need to fix the robot fallout and then actually post this.