Mail Archives: djgpp/1997/01/26/11:35:55
On Sun, 26 Jan 1997, Mark S. Teel wrote:
> Thanks, that seemed to work. But this raises some other questions...
> 1) Replacing the -g with -O2 causes optimization and no debug information
> generated.
> Does this mean the libraries were built without debug information? Will I
> not be able to
> step into the library routines?
The only library that's built when you build gcc is libgcc.a. If you
need to step inside it, just delete libgcc.a and the .o files which are
put into it (you can use `ar tv' to see what these are), and say thus:
make CFLAGS='-O2 -g'
and it should only rebuild libgcc.a. Note that in order to step into a
library function and see its source code (not the disassembled machine
instructions), you will need to also have the source of that library
function handy, which usually isn't the case, so the whole point might be
a moot one.
> 2) There was a plethora of EXEs generated in addition to cc1.exe - what
> files need to be
> moved into the bin directory, lib directory, etc?
As far as I can remember, you only need cpp.exe, gcc.exe, cc1.exe and
libgcc.a, but you can verify this using (1) the contents of gcc2721b.zip
archive; and (2) the install: target of the Makefile in the gcc
distribution, which should show you exactly which files it moves to which
directories.
> An interesting observation is that while I was building gcc, I received
> mail via M$ Internet
> Mail and it caused the compiler to crash!?!
If you still got the crash traceback (or can reproduce the crash), you
could run symify on the compiler and see where does the crash happen. It
might be some real problem in DJGPP.
- Raw text -