Testing gnulib on riscv64 - replacement modules (update)

[Richard W.M. Jones]

Back in March I was trying to test gnulib on RISC-V using qemu:

https://lists.gnu.org/archive/html/bug-gnulib/2018-03/threads.html#00022

We now have real hardware so I am trying to do the tests again,

I still get stuck on the "replacement modules" step in the
instructions.

  "Verify that Gnulib has built no replacement/workaround code
  (gllib/*.o files) - if so, this indicates problems in the libc."

and yes we do indeed have these files, many of them:

$ ls gllib/*.o
gllib/allocator.o         gllib/fpurge.o            gllib/printf-frexpl.o
gllib/areadlinkat.o       gllib/freading.o          gllib/printf-frexp.o
gllib/areadlink.o         gllib/fseek.o             gllib/printf-parse.o
gllib/asnprintf.o         gllib/fseeko.o            gllib/pt_chown.o
gllib/basename-lgpl.o     gllib/fseterr.o           gllib/renameat2.o
gllib/careadlinkat.o      gllib/full-write.o        gllib/safe-write.o
gllib/c-ctype.o           gllib/getprogname.o       gllib/save-cwd.o
gllib/chdir-long.o        gllib/gettime.o           gllib/sig-handler.o
gllib/chmodat.o           gllib/hard-locale.o       gllib/sockets.o
gllib/chownat.o           gllib/integer_length.o    gllib/statat.o
gllib/cloexec.o           gllib/ioctl.o             gllib/stat-time.o
gllib/c-strcasecmp.o      gllib/localcharset.o      gllib/strerror_r.o
gllib/c-strncasecmp.o     gllib/localename.o        gllib/stripslash.o
gllib/dirname-lgpl.o      gllib/localtime-buffer.o  gllib/strnlen1.o
gllib/dup-safer-flag.o    gllib/malloca.o           gllib/sys_socket.o
gllib/dup-safer.o         gllib/math.o              gllib/tempname.o
gllib/exitfail.o          gllib/mbrlen.o            gllib/timespec.o
gllib/fclose.o            gllib/mbrtowc.o           gllib/tmpdir.o
gllib/fcntl.o             gllib/nanosleep.o         gllib/unistd.o
gllib/fd-hook.o           gllib/nonblocking.o       gllib/utimens.o
gllib/fd-safer-flag.o     gllib/openat-die.o        gllib/vasnprintf.o
gllib/fd-safer.o          gllib/openat-proc.o       gllib/wctype-h.o
gllib/fflush.o            gllib/pipe-safer.o        gllib/xsize.o
gllib/filenamecat-lgpl.o  gllib/printf-args.o

$ (cd gllib; ls -1 $(${GNULIB_CHECKOUT}/posix-modules | sed -e 's|-posix$||' | 
sort -u | grep -v 'nonblocking' | sed -e 's|$|.o|') 2>/dev/null )
fclose.o
fcntl.o
fflush.o
fseek.o
fseeko.o
ioctl.o
math.o
mbrlen.o
mbrtowc.o
nanosleep.o
strerror_r.o
sys_socket.o
unistd.o
wctype-h.o

Full build log:

  http://oirase.annexia.org/tmp/build.log

We're using glibc 2.27.9000-7.fc28.riscv64 from Fedora, and the latest
gnulib from git.

It looks like the glob.o bug has been fixed but there is still a lot
of investigation to be done.  BTW the real hardware has working
floating point.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-p2v converts physical machines to virtual machines.  Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v