This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: GCC mainline does not bootstrap on PowerPC


Andreas Jaeger <aj@suse.de> writes:

> On 64-bit PowerPC I see the following bootstrap failure:
>  
> gcc -c   -g  -DIN_GCC   -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-p
> rototypes  -fno-common   -DHAVE_CONFIG_H -DGENERATOR_FILE    -I. -I. -I../../gcc
>  -I../../gcc/. -I../../gcc/../include -I../../gcc/../libcpp/include -I../../gcc/
> ../libbanshee/libcompat -I../../gcc/../libbanshee -I../../gcc/../libbanshee/poin
> ts-to  insn-conditions.c
> insn-conditions.c:187: error: `flag_unsafe_math_optimizations' undeclared here (not in a function)
> insn-conditions.c:187: error: `flag_unsafe_math_optimizations' undeclared here (not in a function)
> insn-conditions.c:187: error: initializer element is not constant
> insn-conditions.c:187: error: (near initialization for `insn_conditions[38].value')

This would, I suspect, be a consequence of using the new Var() .opt-
file notation for that flag, rather than defining it explicitly in a
header file.  The cure would be to have insn-conditions.c include and
depend on the generated header file, whose name I forget.

(I'm open to suggestions for how to do insn-conditions.c better.  It
performs a valuable space optimization - can reduce the size of cc1 by
hundreds of K - but it is *very* clever, and not in a good way.)

zw


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]