Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux

selftests: kvm: Use a shorter encoding to clear RAX

If debug_regs.c is built with newer binutils, the resulting binary is "optimized"
by the assembler:

asm volatile("ss_start: "
"xor %%rax,%%rax\n\t"
"cpuid\n\t"
"movl $0x1a0,%%ecx\n\t"
"rdmsr\n\t"
: : : "rax", "ecx");

is translated to :

000000000040194e <ss_start>:
40194e: 31 c0 xor %eax,%eax <----- rax->eax?
401950: 0f a2 cpuid
401952: b9 a0 01 00 00 mov $0x1a0,%ecx
401957: 0f 32 rdmsr

As you can see rax is replaced with eax in target binary code.
This causes a difference is the length of xor instruction (2 Byte vs 3 Byte),
and makes the hard-coded instruction length check fail:

/* Instruction lengths starting at ss_start */
int ss_size[4] = {
3, /* xor */ <-------- 2 or 3?
2, /* cpuid */
5, /* mov */
2, /* rdmsr */
};

Encode the shorter version directly and, while at it, fix the "clobbers"
of the asm.

Cc: stable@vger.kernel.org
Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

authored by

Yang Weijiang and committed by
Paolo Bonzini
98b0bf02 e792415c

+2 -2
+2 -2
tools/testing/selftests/kvm/x86_64/debug_regs.c
··· 40 40 41 41 /* Single step test, covers 2 basic instructions and 2 emulated */ 42 42 asm volatile("ss_start: " 43 - "xor %%rax,%%rax\n\t" 43 + "xor %%eax,%%eax\n\t" 44 44 "cpuid\n\t" 45 45 "movl $0x1a0,%%ecx\n\t" 46 46 "rdmsr\n\t" 47 - : : : "rax", "ecx"); 47 + : : : "eax", "ebx", "ecx", "edx"); 48 48 49 49 /* DR6.BD test */ 50 50 asm volatile("bd_start: mov %%dr0, %%rax" : : : "rax");