From: Marc Zyngier marc.zyngier@arm.com
commit eb7c11ee3c5ce6c45ac28a5015a8e60ed458b412 upstream.
AArch64 toolchains suffer from the following bug:
$ cat blah.S 1: .inst 0x01020304 .if ((. - 1b) != 4) .error "blah" .endif $ aarch64-linux-gnu-gcc -c blah.S blah.S: Assembler messages: blah.S:3: Error: non-constant expression in ".if" statement
which precludes the use of msr_s and co as part of alternatives.
We workaround this issue by not directly testing the labels themselves, but by moving the current output pointer by a value that should always be zero. If this value is not null, then we will trigger a backward move, which is expclicitely forbidden. This triggers the error we're after:
AS arch/arm64/kvm/hyp.o arch/arm64/kvm/hyp.S: Assembler messages: arch/arm64/kvm/hyp.S:1377: Error: attempt to move .org backwards scripts/Makefile.build:294: recipe for target 'arch/arm64/kvm/hyp.o' failed make[1]: *** [arch/arm64/kvm/hyp.o] Error 1 Makefile:946: recipe for target 'arch/arm64/kvm' failed
Not pretty, but at least works on the current toolchains.
Acked-by: Will Deacon will.deacon@arm.com Signed-off-by: Marc Zyngier marc.zyngier@arm.com Signed-off-by: Catalin Marinas catalin.marinas@arm.com Signed-off-by: David Brown david.brown@linaro.org --- arch/arm64/include/asm/alternative.h | 25 ++++++++++++++++++------- 1 file changed, 18 insertions(+), 7 deletions(-)
diff --git a/arch/arm64/include/asm/alternative.h b/arch/arm64/include/asm/alternative.h index 265b13e..c385a0c 100644 --- a/arch/arm64/include/asm/alternative.h +++ b/arch/arm64/include/asm/alternative.h @@ -26,7 +26,20 @@ void free_alternatives_memory(void); " .byte 662b-661b\n" /* source len */ \ " .byte 664f-663f\n" /* replacement len */
-/* alternative assembly primitive: */ +/* + * alternative assembly primitive: + * + * If any of these .org directive fail, it means that insn1 and insn2 + * don't have the same length. This used to be written as + * + * .if ((664b-663b) != (662b-661b)) + * .error "Alternatives instruction length mismatch" + * .endif + * + * but most assemblers die if insn1 or insn2 have a .inst. This should + * be fixed in a binutils release posterior to 2.25.51.0.2 (anything + * containing commit 4e4d08cf7399b606 or c1baaddf8861). + */ #define ALTERNATIVE(oldinstr, newinstr, feature) \ "661:\n\t" \ oldinstr "\n" \ @@ -39,9 +52,8 @@ void free_alternatives_memory(void); newinstr "\n" \ "664:\n\t" \ ".popsection\n\t" \ - ".if ((664b-663b) != (662b-661b))\n\t" \ - " .error "Alternatives instruction length mismatch"\n\t"\ - ".endif\n" + ".org . - (664b-663b) + (662b-661b)\n\t" \ + ".org . - (662b-661b) + (664b-663b)\n"
#else
@@ -61,9 +73,8 @@ void free_alternatives_memory(void); .pushsection .altinstr_replacement, "ax" 663: \insn2 664: .popsection - .if ((664b-663b) != (662b-661b)) - .error "Alternatives instruction length mismatch" - .endif + .org . - (664b-663b) + (662b-661b) + .org . - (662b-661b) + (664b-663b) .endm
#endif /* __ASSEMBLY__ */