* Re: [PATCH] KVM: x86: Remove redundant and incorrect cpl check on task-switch
[not found] <1406703436-9920-1-git-send-email-namit@cs.technion.ac.il>
@ 2014-07-31 17:45 ` Nadav Amit
0 siblings, 0 replies; only message in thread
From: Nadav Amit @ 2014-07-31 17:45 UTC (permalink / raw)
To: Nadav Amit, pbonzini; +Cc: gleb, tglx, mingo, hpa, x86, linux-kernel, kvm
cc'ing the kvm mailing list that was mistakenly omitted.
On 7/30/14 9:57 AM, Nadav Amit wrote:
> Task-switch emulation checks the privilage level prior to performing the
> task-switch. This check is incorrect in the case of task-gates, in which the
> tss.dpl is ignored, and can cause superfluous exceptions. Moreover this check
> is unnecassary, since the CPU checks the privilage levels prior to exiting.
> Intel SDM 25.4.2 says "If CALL or JMP accesses a TSS descriptor directly
> outside IA-32e mode, privilege levels are checked on the TSS descriptor" prior
> to exiting. AMD 15.14.1 says "The intercept is checked before the task switch
> takes place but after the incoming TSS and task gate (if one was involved) have
> been checked for correctness."
>
> This patch removes the CPL checks for CALL and JMP.
>
> Signed-off-by: Nadav Amit <namit@cs.technion.ac.il>
> ---
> arch/x86/kvm/emulate.c | 8 ++------
> 1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
> index 56657b0..a9b2bd6 100644
> --- a/arch/x86/kvm/emulate.c
> +++ b/arch/x86/kvm/emulate.c
> @@ -2698,7 +2698,8 @@ static int emulator_do_task_switch(struct x86_emulate_ctxt *ctxt,
> *
> * 1. jmp/call/int to task gate: Check against DPL of the task gate
> * 2. Exception/IRQ/iret: No check is performed
> - * 3. jmp/call to TSS: Check against DPL of the TSS
> + * 3. jmp/call to TSS/task-gate: No check is performed since the
> + * hardware checks it before exiting.
> */
> if (reason == TASK_SWITCH_GATE) {
> if (idt_index != -1) {
> @@ -2715,13 +2716,8 @@ static int emulator_do_task_switch(struct x86_emulate_ctxt *ctxt,
> if ((tss_selector & 3) > dpl || ops->cpl(ctxt) > dpl)
> return emulate_gp(ctxt, (idt_index << 3) | 0x2);
> }
> - } else if (reason != TASK_SWITCH_IRET) {
> - int dpl = next_tss_desc.dpl;
> - if ((tss_selector & 3) > dpl || ops->cpl(ctxt) > dpl)
> - return emulate_gp(ctxt, tss_selector);
> }
>
> -
> desc_limit = desc_limit_scaled(&next_tss_desc);
> if (!next_tss_desc.p ||
> ((desc_limit < 0x67 && (next_tss_desc.type & 8)) ||
>
^ permalink raw reply [flat|nested] only message in thread