Re: [PATCH v2 1/3] powerpc: sstep: Fix load and update emulation

[Michael Ellerman <mpe@ellerman.id.au>]

Sandipan Das <sandipan@linux.ibm.com> writes:
> On 03/02/21 3:19 pm, Naveen N. Rao wrote:
>> [...]
>> 
>> Wouldn't it be easier to just do the below at the end? Or, am I missing something?
>> 
>> diff --git a/arch/powerpc/lib/sstep.c b/arch/powerpc/lib/sstep.c
>> index ede093e9623472..a2d726d2a5e9d1 100644
>> --- a/arch/powerpc/lib/sstep.c
>> +++ b/arch/powerpc/lib/sstep.c
>> @@ -2980,6 +2980,10 @@ int analyse_instr(struct instruction_op *op, const struct pt_regs *regs,
>>         }
>>  #endif /* CONFIG_VSX */
>> 
>> +       if (GETTYPE(op->type) == LOAD && (op->type & UPDATE) &&
>> +                       (ra == 0 || ra == rd))
>> +               goto unknown_opcode;
>> +
>>         return 0;
>> 
>>   logical_done:
>> 
>
> This looks good?
>
> diff --git a/arch/powerpc/lib/sstep.c b/arch/powerpc/lib/sstep.c
> index e96cff845ef7..a9c149bfd2f5 100644
> --- a/arch/powerpc/lib/sstep.c
> +++ b/arch/powerpc/lib/sstep.c
> @@ -3017,6 +3017,21 @@ int analyse_instr(struct instruction_op *op, const struct pt_regs *regs,
>  
>         }
>  
> +       if (op->type & UPDATE) {
> +               if (ra == rd && GETTYPE(op->type) == LOAD)
> +                       goto unknown_opcode;
> +               else if (ra == 0)
> +                       switch(GETTYPE(op->type)) {
> +                       case LOAD:
> +                       case STORE:
> +#ifdef CONFIG_PPC_FPU
> +                       case LOAD_FP:
> +                       case STORE_FP:
> +#endif

Why make it conditional?

cheers