RE: [avr-gcc-list] Re: [avr-libc-dev] Constant address to programmemory in C

[Weddington, Eric]

> -----Original Message-----
> From: Shaun Jackman [mailto:address@hidden 
> Sent: Friday, November 23, 2007 10:39 AM
> To: Weddington, Eric
> Cc: Joerg Wunsch; address@hidden
> Subject: Re: [avr-gcc-list] Re: [avr-libc-dev] Constant 
> address to programmemory in C
> 
> On Nov 23, 2007 10:21 AM, Weddington, Eric 
> <address@hidden> wrote:
> ...
> > > > This rather smells like a compiler bug to me, than 
> something where
> > > > library support would be a real help.  If the compiler 
> offers the
> > > > feature to jump to an arbitrary address using "goto", then it's
> > > > responsible to translate that code into whatever is necessary to
> > > > accomplish the goal.  The tools always and only count in byte
> > > > addresses.
> > >
> > > I agree. Unfortunately, there's a bunch of code out there now that
> > > assumes goto *x needs to be fixed up with a goto *(x/2).
> >
> > There is already a GCC bug filed for the above problem.
> 
> Great! Do you know the bug number? I searched GCC's bugzilla for `Host
> contains the string avr' and found `Zarro Boogs'. The search form
> seems a little busted, or perhaps just unusable.

GCC bug #27192:
<http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27192>

Note that this may not be exactly what you mean. The bug above is about
using a function pointer, adding an offset to the pointer, and then
trying to do an indirect call on the pointer: it goes to the wrong
address.

Also note that I keep track of all AVR bugs on a web page at the
avr-libc project:
<http://www.nongnu.org/avr-libc/bugs.html>

Eric Weddington