libnih has been removed from the archive.
** Changed in: libnih (Ubuntu)
Status: New => Invalid
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to libnih in Ubuntu.
https://bugs.launchpad.net/bugs/1580601
Title:
There is now a similar issue with getting __glib_assert_msg see bug
1689344.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to libnih in Ubuntu.
https://bugs.launchpad.net/bugs/1580601
Title:
__nih_abort_msg symbol cannot be
** Branch linked: lp:apport
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to libnih in Ubuntu.
https://bugs.launchpad.net/bugs/1580601
Title:
__nih_abort_msg symbol cannot be read any more
Status in gcc-5 package in
> i guess it's the bind now that links the no-op symbol, instead of the
real one.
But libnih has built with "-z now" for a long time already:
CFLAGS := -Wall -fstack-protector -fPIE $(shell dpkg-buildflags --get CFLAGS)
LDFLAGS := -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -pie $(shell
If i recall correctly there was something funny with weak symbols /
symbol overwriting with __nih_ things and i guess it's the bind now that
links the no-op symbol, instead of the real one.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is
gcc -E output is identical between xenial and yakkety, as expected (as
neither glibc nor libnih changed).
__nih_abort_msg (nor any __nih symbol) does not appear anywhere in
"objdump -x", but there are some differences in the objdump output other
than addresses, like in the header:
-architecture:
Downgrading the compiler to xenial in an otherwise identical yakkety
environment fixes this again, so it's not a change in some other
component. With some manual dpkg -i --force-depends I confirm it's the
gcc-5 binary itself, not any lib*, -base, or -dev.
--
You received this bug notification
A no-change rebuild of libnih with yakkety's gcc does not fix this.
Curiously, libnih's debian/rules already has
CFLAGS := -Wall -fstack-protector -fPIE $(shell dpkg-buildflags --get CFLAGS)
LDFLAGS := -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -pie $(shell dpkg-buildflags
--get LDFLAGS)
i. e. it
Neither libnih nor glibc changed in yakkety, but what did change is that
gcc now applies the PIE option by default. Maybe libnih needs to be
adjusted for PIE on amd64, or this needs to be fixed in gcc, thus adding
a task for both for now.
** Tags removed: xenial
** Tags added:
9 matches
Mail list logo
