Date: Tue, 28 Mar 1995 11:03:36 -0700 From: Nate Williams <nate@trout.sri.MT.net> To: Bruce Evans <bde@zeta.org.au>, jkh@freefall.cdrom.com Cc: CVS-commiters@time.cdrom.com, cvs-lib@time.cdrom.com Subject: Re: cvs commit: src/lib/libc/stdlib strhash.c Message-ID: <199503281803.LAA19382@trout.sri.MT.net> In-Reply-To: Bruce Evans <bde@zeta.org.au> "Re: cvs commit: src/lib/libc/stdlib strhash.c" (Mar 29, 3:46am)
next in thread | previous in thread | raw e-mail | index | archive | help
> >Generally speaking most people don't want to debug binaries w/out the > >debugging information in (the poor unwashed never needed to use > >non-symbolic debuggers :), so I propose that we strip out the static > >symbols from the standard shared libraries. > > We don't supply libraries compiled with -g so the static symbols are > more important than they should be. Except that the majority of the users can't debug them w/out the debugging information in them. If users need debuggable libraries, they will have to compile them w/debugging information. IMHO, most users assume that the libraries don't need to be debugged. > `ld -r -x' on libc/obj/*.so saves a whole 21K out of 535K. Many static > symbols with long names are not removed. In fact, they become longer: > for the object yplib.so, yplib.so.L179 becomes yplib.so.yplib.so.L179. > I think there is a linker bug or two here. I'll look into it. > Shared objects are > "stripped" using `ld -r -X'. See my recent commit message. This was a mistake I made when I brought in the strip changes from NetBSD. There may be bugs in the stripping function in ld, but at least we are depending on the same ones as the NetBSD folks. :-) > The label for yplib.so was originally > L179 and "stripping" turned it into yplib.so.L179 instead of removing > it like the -X flag says to. Another 'ld/strip' bug I suspect. Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503281803.LAA19382>