This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: m68k-unknown-coff cross failure
- To: law at cygnus dot com
- Subject: Re: m68k-unknown-coff cross failure
- From: Neal Becker <neal at ctd dot comsat dot com>
- Date: 05 Dec 1997 12:11:23 -0500
- Cc: egcs at cygnus dot com
- References: <24788.881340976@hurl.cygnus.com>
>>>>> "Jeffrey" == Jeffrey A Law <law@hurl.cygnus.com> writes:
Jeffrey> In message <u9btyypnnm.fsf@neal.ctd.comsat.com>you write:
>> I retried with 971201, same results. I suppose I could try again
>> explicitly telling configure where to find the target includes, but as
>> you see from TOOL_INCLUDE_DIR it seems to have found them without
>> help.
Jeffrey> Well, as Mike Neuhauser noted in a previous message, there's an
Jeffrey> inconsistency in where configure & Makefile expect to put/find the
Jeffrey> target include files that we need to resolve.
See my recent post that hasn't been sent out to list yet. I finally
got this working.
The problem was that libgloss.h was included in gcc build, in which we
have:
/* Don't set the target flags, this is done by the linker script */
#undef LIB_SPEC
#define LIB_SPEC ""
That's why we always failed linking libiberty for the target. I
simply edited the specs file in the build/gcc directory to say:
*lib:
%{!shared:%{g*:-lg} %{!p:%{!pg:-lc}}%{p:-lc_p}%{pg:-lc_p}}
It's also true that the correct includes were not used, but
fortunately this was harmless in this case.
After making this change to specs the build completed. I have just
recompiled an example app for the target and loaded it into the
hardware, which is an mpeg2 satellite receiver, and it is working
normally.