qemu-devel
[Top][All Lists]
Advanced

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

Re: [Qemu-devel] [PATCH v8 18/27] target-arm: make c2_mask and c2_base_m


From: Greg Bellows
Subject: Re: [Qemu-devel] [PATCH v8 18/27] target-arm: make c2_mask and c2_base_mask banked
Date: Tue, 4 Nov 2014 16:46:30 -0600



On 31 October 2014 10:26, Peter Maydell <address@hidden> wrote:
On 30 October 2014 21:28, Greg Bellows <address@hidden> wrote:
> From: Fabian Aggeler <address@hidden>
>
> Since TTBCR is banked we will bank c2_mask and c2_base_mask too. This
> avoids recalculating them on switches from secure to non-secure world.

These fields are part of our TTBCR internal representation; we
should bank the whole TTBCR in one patch, not split over two.

Squashed the TTBCR and C2 mask patches in v9.
 

> Signed-off-by: Fabian Aggeler <address@hidden>
> Signed-off-by: Greg Bellows <address@hidden>
>
> ---
>
> v5 -> v6
> - Switch to use distinct CPREG secure flags
>
> v4 -> v5
> - Changed c2_mask updates to use the TTBCR cpreg bank flag for selcting the
>   secure bank instead of the A32_BANKED_CURRENT macro.  This more accurately
>   chooses the correct bank matching that of the TTBCR being accessed.
> ---
>  target-arm/cpu.h    | 10 ++++++++--
>  target-arm/helper.c | 24 ++++++++++++++++++------
>  2 files changed, 26 insertions(+), 8 deletions(-)
>
> diff --git a/target-arm/cpu.h b/target-arm/cpu.h
> index f125bdd..6e9f1c3 100644
> --- a/target-arm/cpu.h
> +++ b/target-arm/cpu.h
> @@ -226,8 +226,14 @@ typedef struct CPUARMState {
>              };
>              uint64_t tcr_el[4];
>          };
> -        uint32_t c2_mask; /* MMU translation table base selection mask.  */
> -        uint32_t c2_base_mask; /* MMU translation table base 0 mask. */
> +        struct { /* MMU translation table base selection mask. */
> +            uint32_t c2_mask_ns;
> +            uint32_t c2_mask_s;
> +        };
> +        struct { /* MMU translation table base 0 mask. */
> +            uint32_t c2_base_mask_ns;
> +            uint32_t c2_base_mask_s;
> +        };

I think we should actually have:
    typedef struct {
        uint64_t raw_ttbcr;
        uint32_t mask;
        uint32_t base_mask;
    } TTBCR;

    and then have TTBCR ttbcr[2];

and not use the BANKED_REG_SET/GET macros here...

I like the cleanliness of this approach but it does not take into consideration the tcr_el[] fields.  I have instead changed the naming to TCR/tcr_el and have added 4 entries instead of 2.  This is more consistent with the other changes in the patch set.
 

>          uint32_t c2_data; /* MPU data cachable bits.  */
>          uint32_t c2_insn; /* MPU instruction cachable bits.  */
>          uint32_t c3; /* MMU domain access control register
> diff --git a/target-arm/helper.c b/target-arm/helper.c
> index 896b40d..27eaf9c 100644
> --- a/target-arm/helper.c
> +++ b/target-arm/helper.c
> @@ -1584,8 +1584,14 @@ static void vmsa_ttbcr_raw_write(CPUARMState *env, const ARMCPRegInfo *ri,
>       * and the c2_mask and c2_base_mask values are meaningless.
>       */
>      raw_write(env, ri, value);
> -    env->cp15.c2_mask = ~(((uint32_t)0xffffffffu) >> maskshift);
> -    env->cp15.c2_base_mask = ~((uint32_t)0x3fffu >> maskshift);
> +
> +    /* Update the masks corresponding to the the TTBCR bank being written */
> +    A32_BANKED_REG_SET(env, c2_mask,
> +                       ARM_CP_SECSTATE_TEST(ri, ARM_CP_SECSTATE_S),
> +                       ~(((uint32_t)0xffffffffu) >> maskshift));
> +    A32_BANKED_REG_SET(env, c2_base_mask,
> +                       ARM_CP_SECSTATE_TEST(ri, ARM_CP_SECSTATE_S),
> +                       ~((uint32_t)0x3fffu >> maskshift));

... so this turns into:

    TTBCR t = env->cp15.ttbcr[ri->secure];

    t->raw_ttbcr = value;
    t->mask = ~(((uint32_t)0xffffffffu) >> maskshift);
    t->base_mask = ~((uint32_t)0x3fffu >> maskshift);

(XXX did we make ri->secure be a 0/1 or is it 1/2 ? anyway you get the idea.)


Changed in v9.  ri->secure is 1/2 as both bitsor neither may be set at definition time.
 
>  }
>
>  static void vmsa_ttbcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
> @@ -1604,9 +1610,15 @@ static void vmsa_ttbcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
>
>  static void vmsa_ttbcr_reset(CPUARMState *env, const ARMCPRegInfo *ri)
>  {
> -    env->cp15.c2_base_mask = 0xffffc000u;
> +    /* Rest both the TTBCR as well as the masks corresponding to the bank of
> +     * the TTBCR being reset.
> +     */
> +    A32_BANKED_REG_SET(env, c2_base_mask,
> +                       ARM_CP_SECSTATE_TEST(ri, ARM_CP_SECSTATE_S),
> +                       0xffffc000u);
> +    A32_BANKED_REG_SET(env, c2_mask,
> +                       ARM_CP_SECSTATE_TEST(ri, ARM_CP_SECSTATE_S), 0);
>      raw_write(env, ri, 0);
> -    env->cp15.c2_mask = 0;

Similarly this will be much cleaner.

Changed in v9.  Question on reset.  We call raw_write() which also sets the masks, but we set the masks separately here too, but different values.  It seems we should only need to set them in raw_write() is this true?
 

>  }
>
>  static void vmsa_tcr_el1_write(CPUARMState *env, const ARMCPRegInfo *ri,
> @@ -4440,7 +4452,7 @@ static bool get_level1_table_address(CPUARMState *env, uint32_t *table,
>       * AArch32 there is a secure and non-secure instance of the translation
>       * table registers.
>       */
> -    if (address & env->cp15.c2_mask) {
> +    if (address & A32_BANKED_CURRENT_REG_GET(env, c2_mask)) {
>          if (A32_BANKED_CURRENT_REG_GET(env, ttbcr) & TTBCR_PD1) {
>              /* Translation table walk disabled for TTBR1 */
>              return false;
> @@ -4452,7 +4464,7 @@ static bool get_level1_table_address(CPUARMState *env, uint32_t *table,
>              return false;
>          }
>          *table = A32_BANKED_CURRENT_REG_GET(env, ttbr0) &
> -                 env->cp15.c2_base_mask;
> +                 A32_BANKED_CURRENT_REG_GET(env, c2_base_mask);
>      }

and again here you can get a pointer to the correct TTBCR
struct and just reference it.


Yes.  Changed all accesses in v9.
  

>      *table |= (address >> 18) & 0x3ffc;
>      return true;
> --

thanks
-- PMM


reply via email to

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