Re: [PATCH] x86/asm/entry/32: simplify pushes of zeroed pt_regs->REGs

[Denys Vlasenko]

On 03/12/2016 07:05 PM, Andy Lutomirski wrote:
> On Sat, Mar 12, 2016 at 9:53 AM, Denys Vlasenko <dvlasenk@xxxxxxxxxx> wrote:
>> On 03/12/2016 04:38 PM, Ingo Molnar wrote:
>>>
>>> * Denys Vlasenko <dvlasenk@xxxxxxxxxx> wrote:
>>>
>>>> Use of a temporary R8 register here seems to be unnecessary.
>>>>
>>>> "push %r8" is a two-byte insn (it needs REX prefix to specify R8),
>>>> "push $0" is two-byte too. It seems just using the latter would be
>>>> no worse.
>>>>
>>>> Thus, code had an unnecessary "xorq %r8,%r8" insn.
>>>
>>> Neat!
>>>
>>>> It probably costs nothing in execution time here since we are probably
>>>> limited by store bandwidth at this point, but still.
>>>>
>>>> Run-tested under QEMU: 32-bit calls still work:
>>>>
>>>> / # ./test_syscall_vdso32
>>>
>>> Did you manage to test all 3 compat variants:
>>>
>>>> @@ -72,24 +72,23 @@ ENTRY(entry_SYSENTER_compat)
>>>> @@ -205,17 +204,16 @@ ENTRY(entry_SYSCALL_compat)
>>>> @@ -316,11 +314,10 @@ ENTRY(entry_INT80_compat)
>>
>> Yes.
>>
>> test_syscall_vdso32 checks vdso syscall (if available)
>> and direct int80 syscall.
>> Booting two times, with different qemu flags:
>>
>>         qemu-system-x86_64 -cpu Opteron_G4
>>         qemu-system-x86_64 -cpu SandyBridge
>>
>> makes kernel choose either SYSCALL or SYSENTER vdso.
>> So it's all covered.
> 
> How carefully did you check the latter bit?

To double-check, I built a kernel with intentionally crippled
SYSENTER handling (infinite loop).

qemu-system-x86_64 -cpu Opteron_G4 - works
qemu-system-x86_64 -cpu SandyBridge - ./test_syscall_vdso_32 hung

This proves that -cpu SandyBridge does cause SYSENTER path to be used.