Glibc-2.14

[Matthew Burgess]

New version. Release announcement at ​http://sources.redhat.com/ml/libc-alpha/2011-05/msg00103.html.

[bdubbs@…]

It looks like someone already found a build bug:

}}} In file included from libc-start.c:35:0: ../sysdeps/unix/sysv/linux/dl-osinfo.h: In function '_dl_setup_stack_chk_guard': ../sysdeps/unix/sysv/linux/dl-osinfo.h:84:19: error: 'ssize_' undeclared (first use in this function) }}}

Probably should be ssize_t.

[houhongxun]

Replying to bdubbs@…:

It looks like someone already found a build bug:

}}} In file included from libc-start.c:35:0: ../sysdeps/unix/sysv/linux/dl-osinfo.h: In function '_dl_setup_stack_chk_guard': ../sysdeps/unix/sysv/linux/dl-osinfo.h:84:19: error: 'ssize_' undeclared (first use in this function) }}}

Probably should be ssize_t.

It's already been fixed:​http://sourceware.org/ml/glibc-cvs/2011-q2/msg00222.html

I upgrated to glibc-2.14 yesterday, and found that Binutils needs to be patched to successfully build glibc-2.14 at least on x86_64 architecture:​http://sourceware.org/bugzilla/show_bug.cgi?id=12366 See ​http://sourceware.org/bugzilla/show_bug.cgi?id=12801 and ​http://sourceware.org/bugzilla/show_bug.cgi?id=12819 I haven't run into any problems so far, except that MPlayer started to crash.

[bdubbs@…]

Hopefully binutils-2.21.1 will fix the problems building this.

[bdubbs@…]

Just an update. I found that configure would not complete in the LFS Chapter 5 environment because it could not compile with cpuid.h. I created a patch to skip that test:

diff -Naur glibc-2.14/sysdeps/i386/configure glibc-2.14.new/sysdeps/i386/configure
--- glibc-2.14/sysdeps/i386/configure   2011-05-30 23:12:33.000000000 -0500
+++ glibc-2.14.new/sysdeps/i386/configure       2011-06-27 17:20:58.000000000 -0500
@@ -636,7 +636,8 @@
 if test "x$ac_cv_header_cpuid_h" = x""yes; then :

 else
-  as_fn_error $? "gcc must provide the <cpuid.h> header" "$LINENO" 5
+  :
+  #as_fn_error $? "gcc must provide the <cpuid.h> header" "$LINENO" 5
 fi

binutils-2.21.1 seems to have fixed the linking problem.

[Bryan Kadzban]

I like my patch better, since it actually fixes the problem, but hey whatever. :-)

​http://linuxfromscratch.org/pipermail/lfs-dev/2011-June/064803.html

Matt said something about a different version in the followup, but I don't see a patch, so ... eh? Either way. :-)

​http://linuxfromscratch.org/pipermail/lfs-dev/2011-June/064804.html

[bdubbs@…]

Both patches are pretty minimal. I don't think there is ever a need to change configure.in because the problem is lfs chapter 5 only.

My test system is x86_64.

I don't remember the original conversation. I was pretty busy at the time. I've reviewed it though. I'll probably go with my patch, just because it is slightly shorter.

[Matthew Burgess]

Replying to bryan@…:

Matt said something about a different version in the followup, but I don't see a patch, so ... eh? Either way. :-)

Yeah, I did, didn't I :-) From what I remember, my approach there didn't work out so well, and I've since lost my working area that patch was in anyway! I've a slight preference for Bryan's patch but without the configure.in hunk, but not overly bothered either way really.

[(none)]

Milestone 6.9 deleted