Re: [PATCH v2 13/14] x86/S3: Save and restore Shadow Stack configuration

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 29/05/2020 13:52, Jan Beulich wrote:
> On 27.05.2020 21:18, Andrew Cooper wrote:
>> See code for details
>>
>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
>> ---
>> CC: Jan Beulich <JBeulich@xxxxxxxx>
>> CC: Wei Liu <wl@xxxxxxx>
>> CC: Roger Pau Monné <roger.pau@xxxxxxxxxx>
>>
>> Semi-RFC - I can't actually test this path.  Currently attempting to arrange
>> for someone else to.
> Nevertheless
> Reviewed-by: Jan Beulich <jbeulich@xxxxxxxx>
> with one question, just for my understanding:
>
>> @@ -48,6 +58,51 @@ ENTRY(s3_resume)
>>          pushq   %rax
>>          lretq
>>  1:
>> +#ifdef CONFIG_XEN_SHSTK
>> +        /*
>> +         * Restoring SSP is a little complicated, because we are 
>> intercepting
>> +         * an in-use shadow stack.  Write a temporary token under the stack,
>> +         * so SETSSBSY will successfully load a value useful for us, then
>> +         * reset MSR_PL0_SSP to its usual value and pop the temporary token.
>> +         */
>> +        mov     saved_rsp(%rip), %rdi
>> +        cmpq    $1, %rdi
>> +        je      .L_shstk_done
>> +
>> +        /* Set up MSR_S_CET. */
>> +        mov     $MSR_S_CET, %ecx
>> +        xor     %edx, %edx
>> +        mov     $CET_SHSTK_EN | CET_WRSS_EN, %eax
>> +        wrmsr
>> +
>> +        /* Construct the temporary supervisor token under SSP. */
>> +        sub     $8, %rdi
>> +
>> +        /* Load it into MSR_PL0_SSP. */
>> +        mov     $MSR_PL0_SSP, %ecx
>> +        mov     %rdi, %rdx
>> +        shr     $32, %rdx
>> +        mov     %edi, %eax
>> +        wrmsr
>> +
>> +        /* Enable CET.  MSR_INTERRUPT_SSP_TABLE is set up later in 
>> load_system_tables(). */
>> +        mov     $XEN_MINIMAL_CR4 | X86_CR4_CET, %ebx
>> +        mov     %rbx, %cr4
> Does this imply NMI or #MC are fatal between here and there?

Yes, but that is always the case during CPU bringup.

Only a few instructions ago, we didn't have an IDT, and we don't have
yet have an established %tr, so can't get the regular IST pointer either.

~Andrew