[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pcc-list
Subject:    Re: Status update
From:       Anders Magnusson <ragge () ludd ! ltu ! se>
Date:       2010-05-13 9:50:09
Message-ID: 4BEBCB51.8050005 () ludd ! ltu ! se
[Download RAW message or body]

Peter Kuschnerus wrote:
> Anders Magnusson <ragge <at> ludd.ltu.se> writes:
>
>   
>> Hi,
>>
>> I know I haven't written a status update for a long time (if ever?) on 
>> this list, but it might
>> be a good time to do it.  Lack of status updates do not mean that 
>> nothing has happened,
>> so here is a brief summary:
>> Targets:
>> - i386 should work mostly fine and be quite bug-free
>> - amd64 have issues with PIC code, otherwise should work fine.
>>
>> C Frontend:
>> - There are one hard-to-find bug in cpp, but it only reveals itself in 
>> test programs.
>> - C99-style code should be parsed just fine.
>> - All common gcc extensions are handled (in some way).
>>
>> Backend stuff:
>> - All optimizer steps seem to do what they should; but after SSA 
>> conversion the register
>>   allocator creates unneccessary many moves (but still correct code).  I 
>> have some ideas
>>   about how to fix this, and will try it the nearest days.
>> - Missing nice optimizations:
>>     - Constant propagation (easy to fix)
>>     - Strength reduction (more difficult)
>>
>> After the above backend stuff fixed the compiler will be in a very good 
>> shape from a
>> code-generation point-of-view and be really usable for doing things 
>>
>> Also, thanks to people that helps keeping the pcc wiki clean from junk.  
>> We should need
>> a webmaster, anyone interested? 
>>
>>     
>
> Hi
>
> Thank You for that information.
> Great !!
>
> Additionally I am interested to know, what is planned next and later.
> I would like to be able to look forward.
>
> Specially I am interested in the feature "common subexpression optimization".
> Is it already handeled, or is it planned ?
>   
It's not already handled, but as soon as SSA works as it should then it 
will be somewhat
easier to add that functionality also.

-- Ragge

[prev in list] [next in list] [prev in thread] [next in thread]