Re: PATCH: PR target/39911: The 'z' suffix doesn't work with 16bit integer insn

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Mon, Apr 27, 2009 at 1:55 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
> H.J. Lu wrote:
>>
>> On Mon, Apr 27, 2009 at 1:26 PM, Uros Bizjak <ubizjak@gmail.com> wrote:
>>
>>>
>>> H.J. Lu wrote:
>>>
>>>
>>>>
>>>> I checked the history on i386.c. %z works on x87 insns as far as back
>>>> to the initial version on gcc.gnu.org in 1992 and %z never worked on
>>>> memory operand with integer insns. I don't believe we should break
>>>> %z now. We should use %Z to support integer insns.
>>>>
>>>>
>>>
>>> The problem is, that users want "%z" due to the comment in i386.md.
>>> Whether
>>> gcc uses "%Z" or "%z" internally does not matter at all. I propose that
>>> we
>>> go ahead with my patch to see if/what breaks. It is just a matter of
>>> changing the letter throughout a couple of source files to switch the
>>> modifiers.
>>>
>>>
>>
>> Since we won't change 4.3/4.4 branches, we should update comments
>> on 4.3/4.4 branches to reflect how 'z' is actually used in gcc. It should
>> have
>> zero impact.. OK for 4.3/4.4?
>>
>
> I suggest that we leave this mess on 4.3 and 4.4 branches as is. x87 insns
> with register operands don't get suffixes, and suffixes for FP and integer
> memory operands of the same size are different (i.e. "s" and "l" for 32bit
> and 64bit FP ops vs. "s", "l" and "ll" for 16bit, 32bit and 64bit integer
> ops). Having "b", "w", "l" and "q" suffixes for integer insns doesn't fit
> into this at all.
>

Here is the updated patch for 4.3/4.4. OK for 4.3/4.4? Otherwise, people
get wrong information if they are reading gcc 4.3/4,4 sources.

Thanks.

-- 
H.J.

Attachment: gcc-z-1.patch
Description: Text document