[PATCH] x86/nospec: Fixup array_index_nospec_mask() asm constraint

[PATCH] x86/nospec: Fixup array_index_nospec_mask() asm constraint

[Dan Williams]

Allow the compiler to handle @size as an immediate value rather than
allocating a register.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 arch/x86/include/asm/barrier.h |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/barrier.h b/arch/x86/include/asm/barrier.h
index 30d406146016..9e9d6f27825a 100644
--- a/arch/x86/include/asm/barrier.h
+++ b/arch/x86/include/asm/barrier.h
@@ -40,7 +40,7 @@ static inline unsigned long array_index_mask_nospec(unsigned long index,
 
 	asm ("cmp %1,%2; sbb %0,%0;"
 			:"=r" (mask)
-			:"r"(size),"r" (index)
+			:"ir"(size),"r" (index)
 			:"cc");
 	return mask;
 }

Re: [PATCH] x86/nospec: Fixup array_index_nospec_mask() asm constraint

[Linus Torvalds]

On Tue, Feb 6, 2018 at 5:33 PM, Dan Williams <dan.j.williams@intel.com> wrote:
> Allow the compiler to handle @size as an immediate value rather than
> allocating a register.

Actually, maybe that "ir" should be "g".

Because it's fine if it's a memory location too. "cmp" takes pretty
much anything, as long as the thing we compare _to_ is a register.

              Linus

Re: [PATCH] x86/nospec: Fixup array_index_nospec_mask() asm constraint

[Dan Williams]

On Tue, Feb 6, 2018 at 5:50 PM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> On Tue, Feb 6, 2018 at 5:33 PM, Dan Williams <dan.j.williams@intel.com> wrote:
>> Allow the compiler to handle @size as an immediate value rather than
>> allocating a register.
>
> Actually, maybe that "ir" should be "g".
>
> Because it's fine if it's a memory location too. "cmp" takes pretty
> much anything, as long as the thing we compare _to_ is a register.

Ok, no worries I'll do a v2. In fact you suggested 'g' in your initial
version and I lost that along the way while wrestling with why the
compiler miscompiled it until I put it in a static inline rather than
a macro.