Re: [PATCH] x86: work around clang IAS bug referencing __force_order

[Sedat Dilek]

GCC toolchain simply ignores if kaslr_64.o has __force_order means the
build ends up successfully whereas LLVM toolchain and IAS breaks and
the build stops and needs explicitly commit
df6d4f9db79c1a5d6f48b59db35ccd1e9ff9adfc ("x86/boot/compressed: Don't
declare __force_order in kaslr_64.c") reverted to fix this.
With the revert GCC toolchain is also fine.

Maybe it is good to revert that commit?

This is with [1]:

diff --git a/arch/x86/include/asm/special_insns.h
b/arch/x86/include/asm/special_insns.h
index 59a3e13204c3..e1c19c5ecd5e 100644
--- a/arch/x86/include/asm/special_insns.h
+++ b/arch/x86/include/asm/special_insns.h
@@ -17,7 +17,7 @@
  * all loads stores around it, which can hurt performance. Solution is to
  * use a variable and mimic reads and writes to it to enforce serialization
  */
-extern unsigned long __force_order;
+extern unsigned long __force_order __weak;

 void native_write_cr0(unsigned long val);

...and the patchset of "x86/boot: Remove run-time relocations from
compressed kernel" applied [3].

More details in [4].

- Sedat -

References:
[1] https://github.com/ClangBuiltLinux/linux/issues/1120#issuecomment-674182703
[2] https://git.kernel.org/linus/df6d4f9db79c1a5d6f48b59db35ccd1e9ff9adfc
[3] https://lore.kernel.org/patchwork/project/lkml/list/?series=456251
[4] https://github.com/ClangBuiltLinux/linux/issues/1120#issuecomment-674502114