Roland McGrath, le Mon 03 Mar 2014 15:36:52 -0800, a écrit : > > Roland McGrath, le Fri 20 Dec 2013 13:05:42 -0800, a ,Aicrit : > > > The linker is supposed to define those symbols automagically. > > > Did it break? > > > > We have been having > > > > $,1r| > > checking for sufficient default -shared layout... no > > What does this have to do with anything?
It is what makes the Makefile choose between using the empty shlib-lds/shlib-lds-flags or using the explicit shlib.lds file. > Are you saying that the linker does not define __start_* and __stop_* > symbols automagically when it's given a custom linker script? It seems libc's __start_* and __stop_* symbols get defined automagically indeed. This however doesn't happen for libpthread apparently: /usr/src/eglibc-2.18/build-tree/hurd-i386-libc/libpthread/libpthread_pic.a(pt-initialize.os): In function `__pthread_init': /usr/src/eglibc-2.18/libpthread/./pthread/pt-initialize.c:77: undefined reference to `__stop___pthread_init' /usr/bin/ld.bfd.real: /usr/src/eglibc-2.18/build-tree/hurd-i386-libc/libpthread/libpthread_pic.a(pt-initialize.os): relocation R_386_GOTOFF against undefined hidden symbol `__stop___pthread_init' can not be used when making a shared object Now I'm wondering why for instance __start/stop___libc_atexit are specified. This comes from Ulrich's “Place __libc_subfreeres, __libc_atexit, and __libc_thread_subfreeres sections after .jcr section.”. I don't know what this is really useful for. Samuel
