[patch] init_array/fini_array support

[Peter S. Mazinger]

On Thu, 19 Jan 2006, Peter S. Mazinger wrote:

> On Thu, 19 Jan 2006, Joseph S. Myers wrote:
> 
> > On Thu, 19 Jan 2006, Peter S. Mazinger wrote:
> > 
> > > Does it break compatibility w/ old compiled apps? If yes we should try to 
> > > allow an update path from .28 even if all the apps have to be recompiled.
> > > The compat stuff is generally removed after the next release.
> > 
> > There should be no compatibility problems with old compiled apps.  They'll 
> > continue to pass _init and _fini to __uClibc_main, which will continue to 
> > call the functions passed.
> 
> ok, then it is compatible as of runtime

another thought. Do we need to keep _init/_fini after the next release? If 
not that should become a compatibility option that will be removed after 
.29 is released. We do not have to keep compatibility until 1.0, we try 
only to allow an update path (even if rebuilding all apps is necessary)

I am not sure if you read my other comments regarding jump relocations and 
visible symbols. Will you please redo the patches accordingly so that 
readelf -s *.o[s] does show up only __GI_x internal versions of uClibc 
provided ones (like memset,...)

Thanks, Peter

-- 
Peter S. Mazinger <ps dot m at gmx dot net>           ID: 0xA5F059F2
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08  BB6E C389 975E A5F0 59F2