> > On 2010/07/11 10:11, Danny Backx <danny.ba...@scarlet.be> wrote:
> >> I can give you and other interested parties write access to the cegcc
> >> svn if you like.
> >
> > Hi Danny & Pavel,
> >
> > I also thought about updating cegcc, but decided against it because I
> > found merging/rebasing with Subversion too cumbersome.  Thought about
> > doing it cleanly with git (import gcc svn with all its history, copy
> > cegcc on top, rebase on latest gcc) - this way, we could easily
> > extract separate patches from cegcc, and submit them to gcc.
> 
> We are still waiting that Pedro Alves merge the cegcc changes upstream...
> 
> Vincent
> 



By the way, if anybody will do that, I think we should consider some changes 
that I had to apply to cegcc: 
gcc\config\arm\coff.h :  (see 
http://readlist.com/lists/gcc.gnu.org/gcc-help/3/19086.html )


/* COFF targets use constant pool instead of MOVW/MOVT.  */
#define TARGET_USE_MOVT 0
This makes sure that gcc doesn't try to generate MOVW/MOVT sequence for relocs 
when you build for armv7.

It's probably binutils limitation, the error is:
....s: Assembler messages:
....s:10: Error: cannot represent BFD_RELOC_ARM_MOVW relocation in this object 
file format.


Still, that kinds of relocs are probably supported by the kernel 
(IMAGE_REL_BASED_HIGH and IMAGE_REL_BASED_LOW fron winnt.h)



The other important change is in gcc\config\arm\wince-pe.h:
#define BIGGEST_ALIGNMENT 64 has to be changed to 128. Many project use 
attribute aligned of 16 and gcc for that kind of attributes issues a warning 
that max platform alignment is 8 and aligns to 8 bytes only (which may or may 
not happen to be 8 bytes).

The reason for "has to be" instead of "should be" is because some neon 
instructions require 16 byte alignment. (vst.128)

------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Cegcc-devel mailing list
Cegcc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cegcc-devel

Reply via email to