This is the mail archive of the cygwin-developers@cygwin.com mailing list for the Cygwin 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: Problems compiling cygwin from CVS - Undefined symbol


On Tue, Feb 10, 2004 at 05:06:48PM -0500, Brian Keener wrote:
>Brian Keener wrote:
>> Brian Keener Wrote:
>> >> Cannot export NOSIGFE: symbol not defined
>> >> Cannot export longjmp: symbol not defined
>> >> Cannot export longjmp: symbol not defined
>> <snip>
>> Christopher Faylor wrote:
>> > So, this being a mailing list called "cygwin-developers", the question for
>> > you is "What is incorrectly generating the cygwin.def file?"  You should
>> > look at the makefile and figure this out.
>> >
>> Well that is a good question
>>
>Welp, still trying to figure this out and it is really hard for me to believe 
>that I am the only person on W2k trying to compile Cygwin straight from the cvs 
>that is having this problem but I've still been digging hoping I'll stumble 
>over something. 

And, yet, you can see evidence of a few people building cygwin from CVS
on the cygwin mailing list.

>DEF_FILE:=cygwin.def
><snip>
>Makefile: cygwin.din
>
>$(DEF_FILE): gendef cygwin.din $(srcdir)/tlsoffsets.h
>        $^ $@ sigfe.s
>
>$(srcdir)/tlsoffsets.h: gentls_offsets cygtls.h
>        $^ $@ $(COMPILE_CXX)
>
>sigfe.s: $(DEF_FILE)
>        @touch $@
>
>sigfe.o: sigfe.s
>        $(CC) -c -o $@ $?
>
>the output from gendef ultimately builds SIGFE.s and SIGFE.o and I am
>guessing based on the longjmp reference as NOSIGFE that it is not going
>to be part of SIGFE.  But if it is not part of SIGFE and all the other
>longjmp references in the source seen it be for other processors and
>machines - where does the i686 longjmp come from.

gendef creates sigfe.s.  sigfe.s contains the longjmp definition.  It is
trivial to see this by INSPECTING sigfe.s rather than assuming that it
can't possibly be there because you see something called NOSIGFE
following a line in cygwin.din.

>Today I updated from cvs again since I saw several changes had been made 
>(hoping maybe my problem got fixed)

I am making changes to gendef but it has been correctly building sigfe.s
for some time.  I'm not trying to fix your problem since I'm pretty sure
your problem is a cockpit problem.


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