4 msgints, floats and chars
5 msgProblem using Lauterbach Debugger with GCC-ARM
4 msgHelp with understanding strict aliasing rules
2 msgPPC SDA
2 msgTokens not displaying
2 msgwhich version of gcc starts to support '__threa...
4 msgIBM/Rational Purify (or an alternative) on FC3/FC5
2 msgOT: help
5 msgMigration of cross compiler to GCC 3.4.6
4 msgSections (.got .got2 .gcc_except_table .fixup),...
9 msgGcc+binutils+libc cross-compiling: path to libc...
1 msgHow to get the bottom of the stack
2 msgGCJ and JNI
1 msgABI changes version 1 -> 2?
2 msgstatic libgcc_s.a?
2 msgC++ grammar

building fortran
\ Vadim Gutnik (22 Aug 2006)
. \ Tim Prince (23 Aug 2006)
. . \ Vadim Gutnik (23 Aug 2006)
. . . \ Kai Ruottu (23 Aug 2006)
. . . . \ Vadim Gutnik (23 Aug 2006)
. . . . . \ Kai Ruottu (24 Aug 2006)

17 msg__builtin_constant_p and inline assembly constr...
2 msgconfigure gcc4.0.2
3 msgweird characters in compiler warnings output
Subject:Re: building fortran
Group:Gcc-help
From:Kai Ruottu
Date:24 Aug 2006


Vadim Gutnik kirjoitti:
> On Wed, Aug 23, 2006 at 01:25:45PM -0700, Kai Ruottu wrote:
>
>
>> Please look those executables, which should use these shared libs
>> in the '/impinj/uns/amd64_2.4/lib/', with the 'objdump -p'
>> command... Your use > of the '-rpath' option should have added a
>> RPATH entry seen in the output!
>>
>
> Well, that does seem to be the problem. I don't think the rpath
> options got passed down into the configuration.
>
> Here's what I see. (I include objdump from gfortran, f951, and
> octave-2.1.73, which I built with the same options. There is no RPATH
> set in the gfortran or f951 executables, but it is set in octave.)
>
> I can't tell if this is a bug in the build process or if I'm doing
> something wrong. I set LDFLAGS at the very top level, before configure
> and make of gcc.
>
My guess would be a bug in the configure system... Setting
CFLAGS*, LDFLAGS* etc. in the environment doesn't always
go to the resulting Makefiles created in configure :-( I'm not
even sure if the environment settings should be noted in any
way, what the GCC developers have thought the rule being...

So the workaround and the usual practice in cases like this
would be to manually check and edit the main GCC 'Makefile'
and the 'gcc/Makefile' for the given settings really being put
there!

I myself have seen environment settings like '*_FOR_TARGET'
NOT going into the resulting Makefiles!



© 2004-2008 readlist.com