Changelog
glibc (2.22-0experimental2) experimental; urgency=medium
[ Aurelien Jarno ]
* debian/patches/m68k/local-fpic.diff: drop, obsolete.
* debian/patches/m68k/local-mathinline_h.diff: drop, not applied for a
lot of time, obsolete.
* debian/patches/sparc/submitted-timing.diff: drop, obsolete, not
needed anymore since libc6-sparcv9 has been removed.
* debian/patches/any/local-no-malloc-backtrace.diff: drop, the real bug
has been fixed upstream.
* debian/patches/any/local-no-SOCK_NONBLOCK.diff: drop, both GNU/Hurd
and GNU/kFreeBSD now support SOCK_NONBLOCK.
* debian/patches/any/local-libgcc-compat-ports.diff: refresh.
* debian/patches/any/local-disable-test-tgmath2.diff: drop, 8 years have
passed, the build daemons should now have enough resources to build
and run this test.
* debian/patches/s390/submitted-nexttowardf.diff: drop, fixed upstream
another way.
* debian/patches/alpha/submitted-rtld-fPIC.diff: drop, fixed upstream
another way.
* debian/patches/any/unsubmitted-scanf-includes.diff: replace the hack by
the upstream fix patches/any/cvs-bits-libc-stdio-lock.diff. This also
stop installing <bits/libc-lock.h> and <bits/stdio-lock.h>. Closes:
#404532, #639697.
* debian/control, debian/rules.d/control.mk, debian/sysdeps/arm.mk: remove
support for the ARM old-ABI architecture, it is not supported upstream
anymore.
* debian/sysdeps/*.mk: cleanup and homogenize:
- define options common to all flavours using the non-flavour specific
variable.
- reorganize to not mix generic variables with flavour specific variables.
- move -march and -mtune options from extra_cflags to CC and CXX as they
might slightly change the ABI and the configure script needs to detect
that.
* debian/testsuite-checking/expected-results-sparc64-linux-gnu-libc: allow
new tests to fail, cleanup resolved issues.
* Replace debian/patches/i386/submitted-i586-ulps.diff by
debian/patches/i386/local-i386-ulps.diff. The actual math precision
depends on the CPU manufacturer and on the ability to use SSE2
instructions or not. Change the i386 ULPs to the lowest precision
produced by AMD and Intel CPU with and without SSE2 until upstream can
figure out a better solution.
* debian/testsuite-checking/expected-results-powerpc64le-linux-gnu-libc:
allow nptl/tst-cancel17 and nptl/tst-cancelx17, this is a know upstream
race condition.
* debian/testsuite-checking/*, debian/testsuite-xfail-debian.mk,
debian/rules.d/build.mk: rework testsuite system by extending upstream.
* debian/debhelper.in/libc{,-alt,otherbuild).install: stop shipping
testsuite results as it make the package build non-reproducible in case
of different kernels and race conditions in the testsuite.
* debian/testsuite-xfail-debian.mk: allow stdlib/tst-makecontext to fail
on armel and armhf, as it is a new test introduced with glibc 2.22 and
not a regression.
* debian/patches/git-updates.diff: update to the latest commit of the
2.22 branch.
* debian/patches/alpha/submitted-abilist.diff: drop, merged upstream.
* debian/patches/hppa/cvs-inline-syscall.diff: drop, merged upstream.
* debian/patches/hppa/cvs-start.diff: drop, merged upstream.
* debian/control.in/main: update VCS fields to the canonical location.
* debian/control.in/{libc,libnss-dns-udeb,libnss-files-udeb}: rename
XC-Package-Type into Package-Type.
* debian/control.in/libc: drop conflicts on prelink (<= 0.0.20090311-1) as
we don't support Lenny to Stretch upgrades.
* debian/debhelper.in/libc-bin.{postinst,lintian-overrides},
debian/rules.d/debhelper.mk, debian/local/sbin/ldconfig: remove ldconfig
wrapper as new debhelper versions use the trigger instead of an explicit
call to ldconfig.
* debian/testsuite-xfail-debian.mk: allow POSIX2008/netdb.h/linknamespace
and -XOPEN2K/netdb.h/linknamespace to fail on kfreebsd-* (new tests).
* debian/testsuite-xfail-debian.mk: allow tst-setcontext2 to fail on
kfreebsd-* (new test also failing on native FreeBSD).
[ Adam Conrad ]
* debian/patches/any/cvs-gawk-gensub.diff: Fix scary output from newer gawk.
* debian/debhelper.in/libc-bin.postinst: Call ldconfig during configure as
well, or major version upgrades will leave us without due to dep ordering.
-- Aurelien Jarno <email address hidden> Mon, 15 Feb 2016 21:48:56 +0100