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: A bug in 2.9.1 SH linker



  In message <199901042312.SAA13958@qnx.com>you write:
  > 
  > --%--multipart-mixed-boundary-1.264.915491563--%
  > Content-Type: text/plain; charset=us-ascii
  > Content-Transfer-Encoding: 7bit
  > 
  > Hi there,
  > 
  > I found a bug in gcc linker for Hitachi SH4 of release 2.9.1:
  > 
  > When object files are linked to an relocateable executable file,
  > and then the relocateable is linked to a fix-address executable, the
  > address of calls to local functions (in the same module) are wrong.
  > They are called to wrong places. 
  > 
  > However, if object files are linked directly to a fix-address
  > excutable file, there is no problem.
  > 
  > An example are attached with one makefile, two C files and a link
  > script file.
  > When the example is maked, two final executables are generated --
  > "wrong" and "right". "right" is generated directly from object files;
  > "wrong" is generated from a relocateable "mid", and "mid" is
  > generated by link the object files relocateably.
  > When the file "wrong" is examined by "objdump", we could see that two
  > calls in function "main" (to local funcations in the module main.c)
  > are wrong(calling into wrong address).
  > 
  > Quick fixes will be greatly appreciated as we are blocked by this
  > bug.
  > 
  > Thanks and regards,
Problems with the linker, assembler and other binary utilities should
probably go to bug-gnu-utils@gnu.org or something like that.  They are not
part of egcs.

jeff


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