On 26/06/2020 15:26, Jan Beulich wrote:
> On 26.06.2020 15:59, Ross Lagerwall wrote:
>> On 2020-06-26 13:24, Andrew Cooper wrote:
>>> @@ -56,18 +57,48 @@ int arch_livepatch_safety_check(void)
>>> return -EBUSY;
>>> }
>>>
>>> -int arch_livepatch_quiesce(void)
>>> +int noinline arch_livepatch_quiesce(void)
>>> {
>>> + /* If Shadow Stacks are in use, disable CR4.CET so we can modify
>>> CR0.WP. */
>>> + if ( cpu_has_xen_shstk )
>> Should this be:
>> if ( IS_ENABLED(CONFIG_XEN_SHSTK) && cpu_has_xen_shstk )
>>
>> to match arch_livepatch_revive?
> While it may look a little asymmetric, I think it's preferable
> to is IS_ENABLED() only where really needed, i.e. here it
> guarding code that otherwise may not build.
The reason for the asymmetry is because of the asm() block, which needs
compiling out when we detect that we don't have a compatible assembler.
I was wondering whether I should make cpu_has_xen_shstk be false for
!CONFIG_XEN_SHSTK, but that would be 4.15 work at this point.
~Andrew
