Re: the x86 sysret_rip test fails on the Intel FRED architecture

[H. Peter Anvin]

On January 20, 2023 1:10:09 PM PST, Andrew Cooper <Andrew.Cooper3@xxxxxxxxxx> wrote:
>On 20/01/2023 8:50 pm, H. Peter Anvin wrote:
>> On January 20, 2023 10:52:02 AM PST, Andrew Cooper <Andrew.Cooper3@xxxxxxxxxx> wrote:
>>> On 20/01/2023 5:45 pm, Dave Hansen wrote:
>>>> On 1/19/23 23:49, Li, Xin3 wrote:
>>>>> The x86 sysret_rip test has the following assertion:
>>>>>
>>>>>         /* R11 and EFLAGS should already match. */
>>>>>         assert(ctx->uc_mcontext.gregs[REG_EFL] ==
>>>>>                ctx->uc_mcontext.gregs[REG_R11]);
>>>>>
>>>>> This is being tested to avoid kernel state leak due to sysret vs iret,
>>>>> but that on FRED r11 is *always* preserved, and the test just fails.
>>>> Let's figure out the reason that FRED acts differently, first.  Right
>>>> now, the SDM says:
>>>>
>>>> 	SYSCALL also saves RFLAGS into R11
>>>>
>>>> so that behavior of SYSCALL _looks_ architectural to me.  Was this
>>>> change in SYSCALL behavior with FRED intentional?
>>> FRED 3.0 Section 7.4 says the only changes for the SYSCALL and SYSENTER
>>> instructions are the enablement conditions.  Nowhere else is there
>>> mention of a FRED OS needing to emulate legacy syscall behaviour by
>>> adjusting %r11/%rcx
>>>
>>> However, ERETU does handle flags different to SYSRET (in particular, I
>>> think you can establish TF on the instruction boundary after SYSCALL
>>> now).  What are the raw values of REG_EFL and REG_R11 ?
>>>
>>> ~Andrew
>>>
>> Just to avoid any confusion:
>>
>> Syscall and sysenter in a FRED system are treated equivalently to software interrupts, e.g. INT 0x80. They do not modify any registers.
>
>In which case can Intel please publish a v4 spec which actually says this?
>
>I can't see anything in the v3 spec which mentions a change in register
>behaviour for SYSCALL.
>
>~Andrew
>

I'll make sure it makes it into the next update.