[PATCH] shared libbusybox, option for IMA mode

Bernhard Fischer rep.nop at aon.at
Thu Jan 19 10:16:50 UTC 2006


On Wed, Jan 18, 2006 at 04:38:53PM -0600, Rob Landley wrote:
>On Wednesday 18 January 2006 10:08, Bernhard Fischer wrote:
>> Some applets can't be compiled with gcc-2.95, iirc. I, personally,
>> wouldn't knock 2.95 out forcefully, but i admit that i don't use it
>> anymore.
>
>I just got a test environment together for Red Hat 7.2 with the evil that is 
>gcc 2.96, but if that doesn't build against that I'm only likely to fix it if 
>it's a really, really easy fix.
>
>> >What is "naked-include"?
>>
>> http://sourceware.org/cgi-bin/cvsweb.cgi/src/include/?cvsroot=src
>> But apparently the ansidecl.h comes from gcc, fwiw..
>>
>> Attached patch was tested with gcc versions
>> -3.3 -3.4 -4.0 -4.1-HEAD -4.2-HEAD
>>
>> ok to apply?
>> Bernhard
>
>Two things:
>
>1) Does this work if it's on the left of the symbol (with the other type 
>information, since it's basically funky type information)?  Having something 
>other than the symbol being defined at the right edge makes it hard to read 
>what the actual symbol name being defined is.
>
>If we put this in a standard place, say after the other type information like 
>"unsigned long int" but before the actual symbol name, it would a lot be 
>easier for newbies to parse.  (Heck, it'd be a lot easier for me to parse 
>too.  As with most programmers I find reading source code to be at about as 
>much work as writing it...)

Can we agree on
extern type_t symbolname() ATTRIBUTE_X;
type_t ATTRIBUTE_X symbolname() {}

I think that all gcc versions can deal with this.
Will change the patch accordingly and ci after retesting with 3.x and
4.x. Please let me know if you have objections.
>
>2) Perhaps the attribute_thingy names should be ATTRIBUTE_THINGY so it's more 

In http://busybox.net/lists/busybox/2006-January/017879.html
you talked about attribute_inline (all in lowercase), so i did it that
way the second time around.. I'll change all of them to uppercase.

>obvious it's a macro?  (Looking at the patch, this markup is really confusing 
>if you don't know what it is.  An allcaps name at least says "macro, go look 
>in the header files"...)

Uppercase is my preference also, i'll change them.
>
>Other than that it looks good to me.

thanks.
>
>> PS: Rules.mak still talks about version 1.1.0-rc1, shouldn't we bump
>> this to something like 1.1.1-pre0 ?
>
>Yeah, time to update that.  Feel free. :)
done.



More information about the busybox mailing list