Re: Regressions on x86

[Markus Deuling <deuling@de.ibm.com>]

Daniel Jacobowitz wrote:
> On Tue, Jun 19, 2007 at 03:18:22PM +0200, Markus Deuling wrote:
>> Daniel Jacobowitz wrote:
>>> On Tue, Jun 19, 2007 at 02:37:35PM +0200, Markus Deuling wrote:
>>>>> Was elfread.o linked in?  If not, what does config.log say about the
>>>>> -lbfd test?
>>>> bfd/elf.o was compiled. elfread.o not.
>>> Then see the other part of my question :-)
>>
>> configure:21807: checking for ELF support in BFD
>> configure:21831: gcc -o conftest -g -O2 -I/home/deuling/gdb/dev/gdb/../include 
>> -I/../bfd -I/home/deuling/gdb/dev/gdb/../bfd   -L../bfd -L../libiberty 
>> conftest.c -lncurses -lm  -lbfd -liberty >&5
>> In file included from /home/deuling/gdb/dev/gdb/../bfd/elf-bfd.h:27,
>>                 from conftest.c:133:
>> /home/deuling/gdb/dev/gdb/../include/elf/external.h:248: error: syntax error 
>> before "Elf_External_Versym"
>> /home/deuling/gdb/dev/gdb/../include/elf/external.h:248: warning: data 
>> definition has no type or storage class
>> In file included from conftest.c:133:
>> /home/deuling/gdb/dev/gdb/../bfd/elf-bfd.h:1459: error: syntax error before '*' 
>> token
>> /home/deuling/gdb/dev/gdb/../bfd/elf-bfd.h:1461: error: syntax error before 
>> "Elf_External_Versym"
> 
> That is ATTRIBUTE_PACKED, which is always defined by ansidecl.h, which
> is always included by bfd.h.  "-I/../bfd" is not a good sign.  It
> looks like I messed up.  Please try this patch?
> 
configure:21807: checking for ELF support in BFD
configure:21831: gcc -o conftest -g -O2 -I../../gdb/../include -I../bfd -I../../gdb/../bfd   -L../bfd -L../libiberty conftest.c -lncurses -lm  -lbfd -liberty >&5

# of expected passes            11552
# of unexpected failures        25

Looks good :-)


-- 
Markus Deuling
GNU Toolchain for Linux on Cell BE
deuling@de.ibm.com