Re: Updated ColdFire toolchain sources available on my CVS server

[Peter Barada <peter at the-baradas dot com>]

>>I've patched my gcc-3.4.0 and gcc-3.4.1 trees to fix this.  The
>>problem was that the subsi3 pattern allows a label-ref to be
>>subtracced from a general memory operand, something that is illegal
>>for ColdFire, but legal for 68k.
>>
>>Could you try building gcc-3.4.0/gcc-3.4.1 from my CVS server and
>>testing them again?  If that passes, then I'll submit a patch to mainline.
>>  
>>
>The compilation passes for now, the subsi3 bug dissapeared.
>But when trying compile bigger project (output binary file has about 200kB)
>with optimalization enabled (CFLAGS += -O or CFLAGS += -O2),
>internal error occured (see attached compiler output report).
>Compilation process passes successfully without optimalization enabled.
>I am sorry not to attach project *.c and *.h source files, but I have
>no permission of author to do it (it is not my project).
>I hope compiler output report will be enough to detect the
>compilation fail.

Is there anyway to run C/Diagnostika_main.c throught the preprocess
(using -E) and then trime out everything except the fucntion that ends
at line 422, and any variable/type defintions used in the function?

>From the ouptut I see that someone has decided that it can mix mode5
source addressing with mode6 destination addression, something that is
illegal on ColdFire.  I need to be able to debug why that insn was
constructed.

-- 
Peter Barada
peter@the-baradas.com

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com