Re: [Xen-devel] [PATCH 1/3] x86/mmuext: tighten TLB flush address checks

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 20.01.16 at 15:23, <andrew.cooper3@xxxxxxxxxx> wrote:
> On 20/01/16 14:05, Jan Beulich wrote:
>> Addresses passed by PV guests should be subjected to __addr_ok(),
>> avoiding undue TLB flushes; .
>>
>> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
>>
>> --- a/xen/arch/x86/mm.c
>> +++ b/xen/arch/x86/mm.c
>> @@ -3268,8 +3268,9 @@ long do_mmuext_op(
>>          case MMUEXT_INVLPG_LOCAL:
>>              if ( unlikely(d != pg_owner) )
>>                  rc = -EPERM;
>> -            else if ( !paging_mode_enabled(d) ||
>> -                      paging_invlpg(curr, op.arg1.linear_addr) != 0 )
>> +            else if ( !paging_mode_enabled(d)
>> +                      ? __addr_ok(op.arg1.linear_addr)
>> +                      : paging_invlpg(curr, op.arg1.linear_addr) )
> 
> paging_mode_enabled() changes depending on whether the guest has
> logdirty currently enabled.
> 
> As you have also patched paging_invlpg() to DTRT with __addr_ok(), I
> think this hunk can be dropped.

How that? Without the change there's no address validation at
all when !paging_mode_enabled(d).

> Everything else, Reviewed-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>

Please clarify.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel