Re: [Qemu-devel] [PATCH v2 1/8] target-arm: A64: add support for ld/st pair

[Peter Maydell]

On 12 December 2013 11:43, Alex Bennée <address@hidden> wrote:
>
> address@hidden writes:
>
>> Hi,
>>
>> I saw a missing return below:
> <snip>
>>> +    default: /* Failed decoder tree? */
>>> +        unallocated_encoding(s);
>>> +        break;
>>> +    }
>>
>> This doesn't seem right (break instead of return):
>>
>> default:
>>     unallocated_encoding(s);
>>     return;
>> }
>
> Good catch. I suspect it never hits though (or risu doesn't generate
> enough unallocated versions).

Oh, I saw that and then forgot to mention it.
Either the comment or the code is wrong -- if
the decoder tree prevents us getting to this point
with that value of index then we should be asserting,
not calling unallocated_encoding(). If the encoding
is really supposed to be unallocated then it's not
a failure of the decoder tree that we got here,
it's just another case we need to check. Which is it?

thanks
-- PMM