This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] gas: Add --with-optimization=OPTION


On Tue, Feb 27, 2018 at 4:54 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 27.02.18 at 12:55, <hjl.tools@gmail.com> wrote:
>> On Tue, Feb 27, 2018 at 12:13 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>>>> On 26.02.18 at 19:41, <hjl.tools@gmail.com> wrote:
>>>> On Mon, Feb 26, 2018 at 10:33 AM, Joseph Myers <joseph@codesourcery.com> wrote:
>>>>> On Mon, 26 Feb 2018, H.J. Lu wrote:
>>>>>
>>>>>> 1. Add --with-optimization=OPTION configure option for x86 gas.  This
>>>>>> allows the default optimization type to be adjusted at configure time.
>>>>>
>>>>> I don't think such a configure option is appropriate.  It's far outside
>>>>> the standard definition of either --enable or --with options.
>>>>> Command-line options and assembler directives should be fully sufficient
>>>>> without any such configure option, especially given that the assembler is
>>>>
>>>> So it is OK to enable encoding optimization by default.  But it is inappropriate
>>>> to control it at configure time.
>>>>
>>>>> mainly used for assembling compiler output.
>>>>>
>>>>
>>>> x86 assembler optimizes, immediate. displacement as well as VEX vs EVEX.
>>>> I am intending to enable this by default.   --with-optimization=OPTION provides
>>>> a way to test it for a while.
>>>
>>> I don't think optimizations should be enabled by default - that's
>>> simply incompatible with prior versions of gas, and may break
>>> existing code.
>>
>> If this is true, we shouldn't do it at all.
>
> So maybe I misunderstood: Did you mean the assembler to
> default to optimizing, or just the configure process to default
> to _allow_ optimizing to be enabled via command line option?
>

I'd like to change x86 assembler to turn -O2 on by default without
any configure-time option.  But if such encoding changes may break
existing code, I don't want to risk it to even support -On at all.

-- 
H.J.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]