Re: [PATCH -tip perf/probes 0/2] x86 insn decoder bugfixes

[Masami Hiramatsu <mhiramat at redhat dot com>]

Masami Hiramatsu wrote:
> Ingo Molnar wrote:
>>
>> there's still build failures:
>>
>>     TEST    posttest
>>    Error: ffffffff8103b917:        c5 59 fc ff             vpaddb %xmm7,%xmm4,%xmm15
>>    Error: objdump says 4 bytes, but insn_get_length() says 3 (attr:8000)
>>    make[1]: *** [posttest] Error 2
>>    make: *** [bzImage] Error 2
> 
> Aah, that's the AVX code...
> I assume, random-data-byte-sequence in the kernel binary
> contains it.

Hmm, I've found the actual code.

ffffffff81043917 <xen_sysret32_reloc>:
ffffffff81043917:       c5 59 fc ff             vpaddb %xmm7,%xmm4,%xmm15

---
ENTRY(xen_sysret32)
        /*
         * We're already on the usermode stack at this point, but
         * still with the kernel gs, so we can easily switch back
         */
        movq %rsp, PER_CPU_VAR(old_rsp)
        movq PER_CPU_VAR(kernel_stack), %rsp

        pushq $__USER32_DS
        pushq PER_CPU_VAR(old_rsp)
        pushq %r11
        pushq $__USER32_CS
        pushq %rcx

        pushq $VGCF_in_syscall
1:      jmp hypercall_iret
ENDPATCH(xen_sysret32)
RELOC(xen_sysret32, 1b+1)
---
#define RELOC(x, v)     .globl x##_reloc; x##_reloc=v
---

OK, it's not a random, but a data in kernel text.
Anyway, I'll add AVX support then.

Thank you,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com