qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH] target-arm: Fix "no 64-bit EL2" assumption in a


From: Edgar E. Iglesias
Subject: Re: [Qemu-devel] [PATCH] target-arm: Fix "no 64-bit EL2" assumption in arm_excp_unmasked()
Date: Wed, 14 Oct 2015 18:05:32 +0200
User-agent: Mutt/1.5.21 (2010-09-15)

On Thu, Oct 08, 2015 at 07:08:49PM +0100, Peter Maydell wrote:
> The code in arm_excp_unmasked() suppresses the ability of PSTATE.AIF
> to mask exceptions from a lower EL targeting EL2 or EL3 if the
> CPU is 64-bit. This is correct for a target of EL3, but not correct
> for targeting EL2. Further, we go to some effort to calculate
> scr and hcr values which are not used at all for the 64-bit CPU
> case.
> 
> Rearrange the code to correctly implement the 64-bit CPU logic
> and keep the hcr/scr calculations in the 32-bit CPU codepath.
> 
> Signed-off-by: Peter Maydell <address@hidden>

Looks good to me! I gave it a test run together with
Sergey's eb4ac7e08b342ae20661a4feac885b8e2e966418 from
target-arm.next.

Reviewed-by: Edgar E. Iglesias <address@hidden>
Tested-by: Edgar E. Iglesias <address@hidden>



> ---
> As promised, here's the patch that fixes the assumption about
> 64-bit implying no EL2 in arm_excp_unmasked(). This applies on
> top of my current target-arm.next tree, since it has Sergey's
> patch in it which touches the same code.
> 
> 
> 
>  target-arm/cpu.h | 82 
> +++++++++++++++++++++++++++++++++++---------------------
>  1 file changed, 52 insertions(+), 30 deletions(-)
> 
> diff --git a/target-arm/cpu.h b/target-arm/cpu.h
> index 99b34c4..6b6e4e0 100644
> --- a/target-arm/cpu.h
> +++ b/target-arm/cpu.h
> @@ -1524,8 +1524,6 @@ static inline bool arm_excp_unmasked(CPUState *cs, 
> unsigned int excp_idx,
>      CPUARMState *env = cs->env_ptr;
>      unsigned int cur_el = arm_current_el(env);
>      bool secure = arm_is_secure(env);
> -    bool scr;
> -    bool hcr;
>      bool pstate_unmasked;
>      int8_t unmasked = 0;
>  
> @@ -1539,31 +1537,10 @@ static inline bool arm_excp_unmasked(CPUState *cs, 
> unsigned int excp_idx,
>  
>      switch (excp_idx) {
>      case EXCP_FIQ:
> -        /* If FIQs are routed to EL3 or EL2 then there are cases where we
> -         * override the CPSR.F in determining if the exception is masked or
> -         * not.  If neither of these are set then we fall back to the CPSR.F
> -         * setting otherwise we further assess the state below.
> -         */
> -        hcr = (env->cp15.hcr_el2 & HCR_FMO);
> -        scr = (env->cp15.scr_el3 & SCR_FIQ);
> -
> -        /* When EL3 is 32-bit, the SCR.FW bit controls whether the CPSR.F bit
> -         * masks FIQ interrupts when taken in non-secure state.  If SCR.FW is
> -         * set then FIQs can be masked by CPSR.F when non-secure but only
> -         * when FIQs are only routed to EL3.
> -         */
> -        scr = scr && !((env->cp15.scr_el3 & SCR_FW) && !hcr);
>          pstate_unmasked = !(env->daif & PSTATE_F);
>          break;
>  
>      case EXCP_IRQ:
> -        /* When EL3 execution state is 32-bit, if HCR.IMO is set then we may
> -         * override the CPSR.I masking when in non-secure state.  The SCR.IRQ
> -         * setting has already been taken into consideration when setting the
> -         * target EL, so it does not have a further affect here.
> -         */
> -        hcr = (env->cp15.hcr_el2 & HCR_IMO);
> -        scr = false;
>          pstate_unmasked = !(env->daif & PSTATE_I);
>          break;
>  
> @@ -1588,13 +1565,58 @@ static inline bool arm_excp_unmasked(CPUState *cs, 
> unsigned int excp_idx,
>       * interrupt.
>       */
>      if ((target_el > cur_el) && (target_el != 1)) {
> -        /* ARM_FEATURE_AARCH64 enabled means the highest EL is AArch64.
> -         * This code currently assumes that EL2 is not implemented
> -         * (and so that highest EL will be 3 and the target_el also 3).
> -         */
> -        if (arm_feature(env, ARM_FEATURE_AARCH64) ||
> -            ((scr || hcr) && (!secure))) {
> -            unmasked = 1;
> +        /* Exceptions targeting a higher EL may not be maskable */
> +        if (arm_feature(env, ARM_FEATURE_AARCH64)) {
> +            /* 64-bit masking rules are simple: exceptions to EL3
> +             * can't be masked, and exceptions to EL2 can only be
> +             * masked from Secure state. The HCR and SCR settings
> +             * don't affect the masking logic, only the interrupt routing.
> +             */
> +            if (target_el == 3 || !secure) {
> +                unmasked = 1;
> +            }
> +        } else {
> +            /* The old 32-bit-only environment has a more complicated
> +             * masking setup. HCR and SCR bits not only affect interrupt
> +             * routing but also change the behaviour of masking.
> +             */
> +            bool hcr, scr;
> +
> +            switch (excp_idx) {
> +            case EXCP_FIQ:
> +                /* If FIQs are routed to EL3 or EL2 then there are cases 
> where
> +                 * we override the CPSR.F in determining if the exception is
> +                 * masked or not. If neither of these are set then we fall 
> back
> +                 * to the CPSR.F setting otherwise we further assess the 
> state
> +                 * below.
> +                 */
> +                hcr = (env->cp15.hcr_el2 & HCR_FMO);
> +                scr = (env->cp15.scr_el3 & SCR_FIQ);
> +
> +                /* When EL3 is 32-bit, the SCR.FW bit controls whether the
> +                 * CPSR.F bit masks FIQ interrupts when taken in non-secure
> +                 * state. If SCR.FW is set then FIQs can be masked by CPSR.F
> +                 * when non-secure but only when FIQs are only routed to EL3.
> +                 */
> +                scr = scr && !((env->cp15.scr_el3 & SCR_FW) && !hcr);
> +                break;
> +            case EXCP_IRQ:
> +                /* When EL3 execution state is 32-bit, if HCR.IMO is set then
> +                 * we may override the CPSR.I masking when in non-secure 
> state.
> +                 * The SCR.IRQ setting has already been taken into 
> consideration
> +                 * when setting the target EL, so it does not have a further
> +                 * affect here.
> +                 */
> +                hcr = (env->cp15.hcr_el2 & HCR_IMO);
> +                scr = false;
> +                break;
> +            default:
> +                g_assert_not_reached();
> +            }
> +
> +            if ((scr || hcr) && !secure) {
> +                unmasked = 1;
> +            }
>          }
>      }
>  
> -- 
> 1.9.1
> 



reply via email to

[Prev in Thread] Current Thread [Next in Thread]