On Fri, Jul 19, 2019 at 01:41:08PM -0700, Sean Christopherson wrote:
> @@ -68,8 +67,22 @@ static __always_inline unsigned long __vmcs_readl(unsigned long field)
> {
> unsigned long value;
>
> - asm volatile (__ex_clear("vmread %1, %0", "%k0")
> - : "=r"(value) : "r"(field));
> + asm volatile("1: vmread %2, %1\n\t"
> + ".byte 0x3e\n\t" /* branch taken hint */
> + "ja 3f\n\t"
> + "mov %2, %%" _ASM_ARG1 "\n\t"
> + "xor %%" _ASM_ARG2 ", %%" _ASM_ARG2 "\n\t"
> + "2: call vmread_error\n\t"
> + "xor %k1, %k1\n\t"
> + "3:\n\t"
> +
> + ".pushsection .fixup, \"ax\"\n\t"
> + "4: mov %2, %%" _ASM_ARG1 "\n\t"
> + "mov $1, %%" _ASM_ARG2 "\n\t"
> + "jmp 2b\n\t"
> + ".popsection\n\t"
> + _ASM_EXTABLE(1b, 4b)
> + : ASM_CALL_CONSTRAINT, "=r"(value) : "r"(field) : "cc");
Was there a reason you didn't do the asm goto thing here like you did
for the previous patch? That seemed cleaner, and needs less asm.
I think the branch hints aren't needed -- they're ignored on modern
processors. Ditto for the previous patch.
Also please use named asm operands whereever you can, like "%[field]"
instead of "%2". It helps a lot with readability.
--
Josh
