This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


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

Re: building egcs on LynxOs



  In message <34C6346A.BAE76633@hotmail.com>you write:
  > hi,
  > 
  > I tried building egcs-1.0.1 on lynxos-2.5.0.
This really belongs on one of the egcs lists -- you're more likely to get
help if you post to a list than send mail to an individual.

  > however I still had to do a little tweaking (virtualy nothing works the
  > first time on lynx).
Yup.  I'm all too aware of this issue.

  > At first I had to comment out the extern
  > declarations of mktemp(), as it is declared by a "#define mktemp()
  > lynx-mktemp()".
Seems to me that this should be autoconf'd.


  >  Next, I had to comment out the linker argument "P1000" in "lynx-ng.h".
I have no idea why this is in lynx-ng.  What did that argument used to do?

It almost sounds like you need another .h file for whatever specific
version of lynx you're running.  It would basically include lynx-ng.h
and just override some stuff (like the linker argument you mentioned).

  > I could not make that out, I found something remotely like 'Ltext0' in
  > the file 'dbxout.c'. The archive libgcc2.a was filled with unresolved
  > ('U') refernces to 'Ltext0'. Apparently, the assembly generated contains
  > a forward refernce to 'Ltext0', and it declares it a few lines down the
  > road. However, when checking with 'nm' - the symbol is 'Unresolved'. I
  > saw something about symbols begining with 'L' being local, but I don't
  > see why they are not resolved.
This sounds like an assembler bug.

jeff


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