[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: powerpc-rtems-4.7-gcc-4.0 patch commited



Ralf Corsepius wrote:
Hi,

FYI: I have committed my powerpc RTEMS-gcc multilib patches to GCC's CVS
trunk (aka. GCC-4.0).

This patch introduces a completely new layout of powerpc multilibs:

# powerpc-rtems4.7-gcc --print-multi-lib
.;@mrelocatable-lib@mno-eabi@mstrict-align
m403;@mcpu=403@mrelocatable-lib@mno-eabi@mstrict-align
m505;@mcpu=505@mrelocatable-lib@mno-eabi@mstrict-align
m601;@mcpu=601@mrelocatable-lib@mno-eabi@mstrict-align
m603e;@mcpu=603e@mrelocatable-lib@mno-eabi@mstrict-align
m604;@mcpu=604@mrelocatable-lib@mno-eabi@mstrict-align
m860;@mcpu=860@mrelocatable-lib@mno-eabi@mstrict-align
m7400;@mcpu=7400@mrelocatable-lib@mno-eabi@mstrict-align
nof;@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
m601/nof;@mcpu=601@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
m603e/mpc8260;@mcpu=603e@Dmpc8260@mrelocatable-lib@mno-eabi@mstrict-align
m603e/nof;@mcpu=603e@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
m603e/mpc8260/nof;@mcpu=603e@Dmpc8260@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
m604/nof;@mcpu=604@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
m7400/nof;@mcpu=7400@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align
roe/m403;@D_OLD_EXCEPTIONS@mcpu=403@mrelocatable-lib@mno-eabi@mstrict-align
roe/m603e;@D_OLD_EXCEPTIONS@mcpu=603e@mrelocatable-lib@mno-eabi@mstrict-align
roe/m603e/nof;@D_OLD_EXCEPTIONS@mcpu=603e@msoft-float@mrelocatable-lib@mno-eabi@mstrict-align

It reduces the number of multilibs from 30 variants down to 18 and has a
-mcpu=7400 (altivec) multilib variant added. The default is -mcpu=750.
Many CPU variants, which had separate multilib variants before, have
been mapped ("multilib matched") to other cpus.
Old-exception multilib variants which are not used by BSPs in the public
source-tree have been removed.

Until now, * this GCC version has only been tested with RTEMS-CVS (aka. 4.7) and
has _not_ been tested with rtems-4.6.x or before.
* I am not sure this new multilib scheme is worth back-porting to
gcc-3.x. Until now, I haven't tried and am not planning to do.

Thanks for doing this Ralf.

My hope/goal is to have GCC 4.x be the toolset for RTEMS 4.7. GCC 4.x appears to support Ada at least as well as gcc 3.2 which is the primary
toolset for the 4.6 series. Gcc 3.3 and 3.4 were broken for cross Ada.
Certainly gcc 3.3.x is usable for RTEMS 4.6 for C/C++ but not Ada.
I agree with you that these are very likely not worth backporting to
older gcc's.


In a perfect world, gcc 4.x and RTEMS 4.7 would closely coincide, but there are still some item son the table left to address.

--joel

Ralf