[uClinux-dev] GCC tool chain for uClinux

From: Schipper, Martijn (Contractor) (martijn.schipper@intersil.com)
Date: Mon Jan 22 2001 - 10:42:08 EST


Hello,

I'm fairly new to uClinux, and try to build a stable tool chain. We discovered a
bug in GCC 2.96 that can be seen with do_fork:
(http://www.uclinux.org/pub/uClinux/archive/0740.html)

The GCC Steering Committee says it's undesirable to use this version
(http://gcc.gnu.org/gcc-2.96.html), so I wanted to use GCC 2.95.2 as starting
point.

If I compile the kernel with this version the bug disappeared.

I have found patches from Philip Blundell to change the Position Independent
Code generation for ARM:
(http://gcc.gnu.org/ml/gcc-patches/1999-07n/msg00573.html) and the patches from
Vadim Lebedev
(http://gcc.gnu.org/ml/gcc-patches/2000-01/msg00216.html) to make PIC without
Global Offset Table (GOT). This option is enabled with -mno-got while the
version of GCC 2.96 we use from Aplio, has the option -mdisable-got.

The patches from Philip Blundell are NOT against GCC 2.95.2 sources, so I can't
apply these. The patches from Vadim Lebedev need the patches from Philip
Blundell.

My questions are:
- Why is everybody using 2.96?
- What version of 2.96 is this then? (what patches are applied?)
- Which of these patches are going to be incorporated in GCC 3.0?
- Why can't I use the GCC -fpic or -fPIC options to create Position Independent
code?
- Why do I need the patch from Philip Blundell? (To what source tree should I
apply this patch?)
- Why do I need the patch from Vadim Lebedev? (To what source tree should I
apply this patch?)
- Is there another patch that implements the -mdisable-got option?

Thanks in advance and best regards,
Martijn Schipper.
This message resent by the uclinux-dev@uclinux.org list server http://www.uClinux.org/


This archive was generated by hypermail 2.1.4 : Thu Sep 19 2002 - 13:19:27 EDT