On 11/21/2018 05:52 AM, Thomas Gleixner wrote:
> On Tue, 20 Nov 2018, Tim Chen wrote:
>> According to software guidance:
>>
>> "Setting ... STIBP ... on a logical processor prevents the predicted
>> targets of indirect branches on any logical processor of that core
>> from being controlled by software that executes (or executed
>> previously) on another logical processor of the same core."
>>
>> https://software.intel.com/security-software-guidance/insights/deep-dive-single-thread-indirect-branch-predictors
>>
>> Hence setting STIBP on a sandboxed task will prevent the task
>> from attacking other sibling threads or getting attacked.
>
> That's not sufficient, because you need an IBPB when the sandboxed task is
> switched out.
>
Right. Which means this change is also needed:
Tim
diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
index 616694c..de31438 100644
--- a/arch/x86/mm/tlb.c
+++ b/arch/x86/mm/tlb.c
@@ -202,7 +202,8 @@ static bool ibpb_needed(struct task_struct *tsk, u64
last_ctx_id)
*/
if (static_branch_unlikely(&spectre_v2_app_lite))
- return test_tsk_thread_flag(tsk, TIF_SPEC_INDIR_BRANCH);
+ return test_tsk_thread_flag(tsk, TIF_SPEC_INDIR_BRANCH) ||
+ test_tsk_thread_flag(current, TIF_SPEC_INDIR_BRANCH);
else
return ptrace_may_access_sched(tsk, PTRACE_MODE_SPEC_IBPB);
}
