long name in .file bug
Joel Sherrill
joel.sherrill@OARcorp.com
Mon Nov 6 15:35:00 GMT 2000
Thanks to DJ Delorie, I now have proper filenames in the .o's
and final executables. Unfortunately, gdb is not able to
grok the directory part of this information enough to find the file.
I prowled a little in gdb and got to "open_source_file()" It is passed
a "struct symtab *" and the full name is not there -- the directory
part is missing. Prowling further this target has a simulator and
the load command is simulator specific.
So this leads to a 1/2 gdb, 1/2 binutils question. What specific
BFD routine should I set a break point on to watch whether this
information is ever being loaded correctly from the file?
And if it is a problem with the simulator specific load, any pointers
to the best example.
Thanks.
Joel Sherrill wrote:
>
> I am trying to track down a bug in the (unsubmitted) TI C4x
> gas target and think I may have stumbled across a generic COFF
> bug. The test case is:
>
> .version 32
> .file
> "../../../../../../../rtems-current/c/src/tests/samples/hello/longfilename.c"
> .text
>
> c4x-rtems-as -o l.o l.s
> strings l.o
> # shows filename is corrupt.
>
> c4x-rtems is actually c4x-coff which according to gas/configure.in is
> "fmt=coff bfd_gas=yes". I want to try my test case on another target
> that is also "fmt=coff bfd_gas=yes" but all the fmt=coff ones I see are
> either some type of emulation (mostly pe) or do not set bfd_gas=yes.
>
> The other coff toolsets I have tried (sh and i960) do not set
> bfd_gas=yes.
> But they pass this simple test.
>
> Any ideas on another target to see whether this is generic or target
> specific?
>
> --
> Joel Sherrill, Ph.D. Director of Research & Development
> joel@OARcorp.com On-Line Applications Research
> Ask me about RTEMS: a free RTOS Huntsville AL 35805
> Support Available (256) 722-9985
--
Joel Sherrill, Ph.D. Director of Research & Development
joel@OARcorp.com On-Line Applications Research
Ask me about RTEMS: a free RTOS Huntsville AL 35805
Support Available (256) 722-9985
More information about the Binutils
mailing list