Re: [Qemu-devel] [PATCH RESEND v1 3/5] target-microblaze: Allow the stack protection to be disabled

[Peter Crosthwaite]

On Mon, May 25, 2015 at 6:55 PM, Alistair Francis
<address@hidden> wrote:
> On Mon, May 25, 2015 at 2:07 PM, Peter Crosthwaite
> <address@hidden> wrote:
>> On Mon, May 18, 2015 at 4:13 PM, Alistair Francis
>> <address@hidden> wrote:
>>> Microblaze stack protection is configurable and isn't always enabled.
>>> This patch allows the stack protection to be disabled from the CPU
>>> properties.
>>>
>>> Signed-off-by: Alistair Francis <address@hidden>
>>> ---
>>> Changes since RFC:
>>>  - Move the cfg.stackproc check into translate.c
>>>  - Set the PVR register
>>>
>>>  target-microblaze/cpu-qom.h   |    5 +++++
>>>  target-microblaze/cpu.c       |    5 +++++
>>>  target-microblaze/cpu.h       |    1 +
>>>  target-microblaze/translate.c |    2 +-
>>>  4 files changed, 12 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/target-microblaze/cpu-qom.h b/target-microblaze/cpu-qom.h
>>> index e3e0701..7bc5b81 100644
>>> --- a/target-microblaze/cpu-qom.h
>>> +++ b/target-microblaze/cpu-qom.h
>>> @@ -59,6 +59,11 @@ typedef struct MicroBlazeCPU {
>>>      uint32_t base_vectors;
>>>      /*< public >*/
>>>
>>> +    /* Microblaze Configuration Settings */
>>> +    struct {
>>> +        bool stackproc;
>>
>> stackprot? Although should we just verbatim match to the TRMs name for
>> these variables (dropping the redundant leading C_)? That would make
>> this "use_stack_protection" and match to QOM property name exactly.
>
> I think "use_stack_protection" is too long, it makes it harder to stay under
> the 80 characters for each line.
>
> I'll change it to 'stackprot' though.
>
>>
>>> +    } cfg;
>>> +
>>>      CPUMBState env;
>>>  } MicroBlazeCPU;
>>>
>>> diff --git a/target-microblaze/cpu.c b/target-microblaze/cpu.c
>>> index 555bc4c..4deb256 100644
>>> --- a/target-microblaze/cpu.c
>>> +++ b/target-microblaze/cpu.c
>>> @@ -117,6 +117,9 @@ static void mb_cpu_realizefn(DeviceState *dev, Error 
>>> **errp)
>>>                          | PVR2_USE_FPU2_MASK \
>>>                          | PVR2_FPU_EXC_MASK \
>>>                          | 0;
>>> +
>>> +    env->pvr.regs[0] |= (cpu->cfg.stackproc ? PVR0_SPROT_MASK : 0);
>>> +
>>
>> Parentheses not needed.
>
> True, but then they get re-added in the next patch as more
> configuration variables get added.
>
> I can remove them from this one.
>

Nah just leave them. I get it now.

Regards,
Peter

>>
>>>      env->pvr.regs[10] = 0x0c000000; /* Default to spartan 3a dsp family.  
>>> */
>>>      env->pvr.regs[11] = PVR11_USE_MMU | (16 << 17);
>>>
>>> @@ -159,6 +162,8 @@ static const VMStateDescription vmstate_mb_cpu = {
>>>
>>>  static Property mb_properties[] = {
>>>      DEFINE_PROP_UINT32("xlnx.base-vectors", MicroBlazeCPU, base_vectors, 
>>> 0),
>>> +    DEFINE_PROP_BOOL("use-stack-protection", MicroBlazeCPU, cfg.stackproc,
>>> +                     true),
>>>      DEFINE_PROP_END_OF_LIST(),
>>>  };
>>>
>>> diff --git a/target-microblaze/cpu.h b/target-microblaze/cpu.h
>>> index e4c1cde..481f463 100644
>>> --- a/target-microblaze/cpu.h
>>> +++ b/target-microblaze/cpu.h
>>> @@ -128,6 +128,7 @@ typedef struct CPUMBState CPUMBState;
>>>  #define PVR0_FAULT                     0x00100000
>>>  #define PVR0_VERSION_MASK               0x0000FF00
>>>  #define PVR0_USER1_MASK                 0x000000FF
>>> +#define PVR0_SPROT_MASK                 0x00000001
>>>
>>>  /* User 2 PVR mask */
>>>  #define PVR1_USER2_MASK                 0xFFFFFFFF
>>> diff --git a/target-microblaze/translate.c b/target-microblaze/translate.c
>>> index 4068946..19faf40 100644
>>> --- a/target-microblaze/translate.c
>>> +++ b/target-microblaze/translate.c
>>> @@ -862,7 +862,7 @@ static inline TCGv *compute_ldst_addr(DisasContext *dc, 
>>> TCGv *t)
>>>      int stackprot = 0;
>>>
>>>      /* All load/stores use ra.  */
>>> -    if (dc->ra == 1) {
>>> +    if (dc->ra == 1 && dc->cpu->cfg.stackproc) {
>>
>> There is a similar logic below for dc->rb:
>>
>>         if (dc->rb == 1) {
>>             stackprot = 1;
>>         }
>>
>> Should it have the same guard?
>
> Yes, it should. I just missed that one.
>
> Thanks,
>
> Alistair
>
>>
>> Regards,
>> Peter
>>
>>>          stackprot = 1;
>>>      }
>>>
>>> --
>>> 1.7.1
>>>
>>>
>>
>