crt0 formalization

[Joel Sherrill]

Jeff Johnston wrote:
> It's for the spu.  I am guessing that Joel also got an error message 
> regarding the patch size so it never got posted to the newlib msg 
> list, though it made it to my inbox.
>
That definitely looks like the case.
> I have tar'd up the patch and attached to this message.
>
Thanks. 

Since this is SPU specific, I assume it is correct for that.  But I 
wonder why crt[in].S are included
in newlib.  Normally they are in gcc.

--joel
> -- Jeff J.
>
> Joel Sherrill wrote:
>> Jeff Johnston wrote:
>>> <resent due to sources.redhat.com rejecting my original reply 
>>> message due to the fact it included the original patch and thereby 
>>> exceeded a size constraint>
>>>
>>> Is it fair to say no one has objections to this patch?  I will check 
>>> it in later today unless I hear otherwise.
>>>
>> I don't remember seeing this patch and I don't see any email from 
>> jschopp or a message with
>> "formalization" in the subject in the newlib mailing list archives 
>> for the past few months .
>> It may be fine but I don't think it has been seen on the newlib list.
>>
>> Which platform is the crt0 for?
>>
>> --joel
>>> -- Jeff J.
>>>
>>> jschopp wrote:
>>>> In our recent SDK 2.0 and in patches being submitting for FSF gcc 
>>>> there are some changes to crt0.  The attached patch has been 
>>>> extensively tested as part of our SDK.  Everything works as 
>>>> expected.  The only minor regression is that it pulls in atexit(), 
>>>> which in turn pulls in malloc().  This works fine but takes up 
>>>> space, which is precious on this platform.  Previous discussions on 
>>>> this list have sent some patches with resolve the problem of 
>>>> pulling in malloc.
>>>>
>>>> In my opinion the attached patch should be committed.
>>>>
>>>> ------------------------------------------------------------------------ 
>>>>
>>