On 06/03/2016 09:01, Xiao Guangrong wrote:
>
>
> On 03/05/2016 07:27 PM, Huaitong Han wrote:
>> This patch adds pkeys support for gva_to_gpa funcions.
>
> It is not needed if you follow the approach i raised in the previous patch.
It's not needed anyway, I think.
PFERR_PK_MASK is set to CR4.PKE. However, patch 3 sets pku to false if
CR4.PKE is clear, and then the corresponding bit is always clear in
mmu->permissions[byte].
Thus I think permission_fault() operates in just the same way even if
PFERR_PK_MASK is always set; CR4.PKE=0 is caught through the permissions
bitmask (as it's meant to be).
Instead of setting PFERR_PK_MASK, we can instead handle it in
permission_fault like:
pkru = is_long_mode(vcpu) ? read_pkru() : 0;
if (unlikely(pkru)) {
...
/* Set PK bit of pfec if pkru_bits is non-zero */
pfec |= pkru_bits ? PFERR_PK_MASK : 0;
}
I'll get to the topic of branches in another message.
Paolo
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
