Re: [avr-gcc-list] Re: Removal of unused functions

[Joerg Wunsch]

Bernd Trog <address@hidden> wrote:

> Yes, but the docs for 'used' say nothing about the visibility of
> that used function. OTH, the docs for -fwhole-program say that
> everything will be static (invisible, local).

And that's exactly what happens.  It can be easily verified by
compiling and linking a program with both attribute settings: when
only "used" is specified, the code will be emitted but the assembly
code misses the .global statement, so the linker doesn't see it as a
symbol to overload the weak interrupt vector symbol from the library.

So in effect, all three attributes are needed: signal (or interrupt),
used, and externally_visible.

Btw., hashing these things out is exactly the thing I've been
requesting with my comments in that bug tracker.  Once that is done
and we are reasonable sure it works out well (hopefully by a few
people confirming it's working in their real life applications), *and*
everything is documented well, I'm happy with publically announcing
that we are now supporting all these compiler features.

-- 
cheers, J"org               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/                        NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)