On Wed, 27 Jul 2005, David Cummings wrote: > Alright, thanks. I had found that, but it didn't seem to apply to the > problem with socket, as that had already been applied when the socket > happened. I'm also now having a problem with n32 which is similar if socket.S shouldn't be needed as recv.o is expected to be built as a stub based on syscalls.list. Have you regenerated the configure script corresponding to the changed configure.in template? > not the same as the one Rolf had not too long ago. Something about a > __fork_block not being found. Anyplace else I should look? Thanks, Check if the symbol is defined where it's expected to. Source and binary RPM packages of glibc 2.3.5 built for o32 and (n)64 are available at my site; built with GCC 4.0.0 which is even pickier than 3.4.4. Availability of binary packages means successful builds are possibile. Figuring out how is left as an exercise for the reader; source packages are provided as an aid to that. Maciej
