Re: [PATCH 0/5] reorganize the use of weak symbols

[Jens Gustedt <jens.gustedt@inria.fr>]

[-- Attachment #1: Type: text/plain, Size: 1637 bytes --]

Hello Rich,
yes, please see these patches mainly as an attempt for documentation
and readability.  They shouldn't change much on the reality of musl,
but perhaps by adding some bytes to some objects. I arbitrarily chose
32 bytes for the dummies. So this will perhaps add some hundred bytes
in total, unless musl is compiled with separate sections for each
symbol.

For the weak symbols versus weak alias, the patches mostly use weak
aliases as before, they only add a level of "weakness" for the dummy
objects and functions. The way I organized it, you could easily change
that to "static" again. So if weak symbols wouldn't be provided by all
tool chains it wouldn't be difficult at all to use that. (Currently I
see that the code is very gcc centric, so I don't think that this a
real issue.)

Where I really like to insist is the only case where a rw array is
aliased to a ro object, patch 2. Here the "dangerous" thing is done by
the aliasing and it should be documented there. The "wrong" type in
the local declaration of the array and the fact to later cast the
const away distracted from that fact.

So in summary, don't hesitate to modify what I wrote, give me feedback
so I'd do that, or not to take them at all, whatever you think is
appropriate for the project.

Jens

-- 
:: INRIA Nancy Grand Est :: http://www.loria.fr/~gustedt/   ::
:: AlGorille ::::::::::::::: office Nancy : +33 383593090   ::
:: ICube :::::::::::::: office Strasbourg : +33 368854536   ::
:: ::::::::::::::::::::::::::: gsm France : +33 651400183   ::
:: :::::::::::::::::::: gsm international : +49 15737185122 ::



[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]