[PATCH RESEND v2] Aarch64: Allow explicit size specifier for predicate operand of {sq, uq, }{incp, decp}

[Richard Sandiford]

"dongbo (E)" <dongbo4@huawei.com> writes:
> Hi, Richard.
>
> On 2022/9/30 22:54, Richard Sandiford wrote:
>> "dongbo (E)" <dongbo4@huawei.com> writes:
>>>
>>> We tried to put `S_H` in front of `NIL`:
>>>
>>> ```
>>>
>>>       #define OP_SVE_Vv_HSD                        \
>>> {                                                               \
>>>         QLF2(S_H,S_H),                                      \
>>>         QLF2(S_S,S_S),                                       \
>>>         QLF2(S_D,S_D),                                      \
>>>         QLF2(S_H,NIL),                                       \
>>>         QLF2(S_S,NIL),                                       \
>>>         QLF2(S_D,NIL),                                       \
>>>       }
>>>
>>> ```
>> Yeah, good point.  It should be in this order, like you say.
>>
>> The fixes you mention look correct to me.
>>
>>> But assembler will fail in `match_operands_qualifier` :(.
>>>
>>> ```
>>>
>>>       match_operands_qualifier (aarch64_inst *inst, bool update_p)
>>>       {
>>>           ...
>>>           if (!aarch64_find_best_match (...))
>>>           ...
>>>           if (inst->opcode->flags & F_STRICT)
>>>           {
>>>               /* Require an exact qualifier match, even for NIL
>>> qualifiers.  */
>>>               nops = aarch64_num_of_operands (inst->opcode);
>>>               for (i = 0; i < nops; ++i)
>>>                   if (inst->operands[i].qualifier != qualifiers[i])
>>>                       return false;
>>>           }
>>>       }
>>>
>>> ```
>> I think that's a misfeature of the F_STRICT handling.  Does it work
>> with the patch below?
>
> We cannot find the `tree-data-ref.cc` in binutils, it is a file in GCC?
>
> Is `0001-data-ref-Fix-ranges_maybe_overlap_p-test.patch` the patch you 
> meant to send?

Gah, no, sorry.

> We also found a way to fix the F_STRICT matching failure. See patch 
> below. :)
>
> The main point is to give another `find_best_match` try when we get a 
> qualifier mismatch for HSD operands.

I think it's simpler than that.  Here's the patch I meant to send
(which is more -s than +s).

Richard

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-aarch64-Tweak-handling-of-F_STRICT.patch
Type: text/x-diff
Size: 2706 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/binutils/attachments/20221010/8875df09/attachment.bin>