Processed: Re: Bug#779579: installer broken: multipath-udeb depends on non-existent libgcc1 udeb
Cyril Brulebois
kibi at debian.org
Mon Mar 2 22:36:21 UTC 2015
Control: reassign -1 multipath-udeb 0.5.0-5
Control: severity -1 serious
Debian Bug Tracking System <owner at bugs.debian.org> (2015-03-02):
> Processing control commands:
>
> > reassign -1 debian-installer
> Bug #779579 [src:multipath-tools] installer broken: multipath-udeb depends on non-existent libgcc1 udeb
> Bug reassigned from package 'src:multipath-tools' to 'debian-installer'.
> No longer marked as found in versions multipath-tools/0.5.0-5.
> Ignoring request to alter fixed versions of bug #779579 to the same values previously set
Ritesh, this is all we got about your bug report. Please always cc the
maintainers of the package you're reassigning bug reports to…
After a local build one can see in the build tree:
| $ objdump -x debian/multipath-udeb/lib/libmultipath.so.0
| …
| Dynamic Section:
| …
| NEEDED libgcc_s.so.1
| …
| Version References:
| required from libgcc_s.so.1:
| 0x0b792650 0x00 12 GCC_3.0
| 0x09265f61 0x00 11 GCC_3.3.1
| …
Cross checking (on amd64) the undefined symbols in libmultipath.so.0
that are not undefined in libgcc_s.so (using nm -D), one can see:
| __gcc_personality_v0
| pthread_mutex_lock
| pthread_mutex_unlock
| _Unwind_Resume
The other udeb depending on libgcc1 is espeakup, and you might find
the flags declared on top of debian/rules of some value to try and get
multipath-udeb to build a udeb that works:
| UDEB_CFLAGS ?= $(CFLAGS) -Os
| UDEB_LDLIBS ?= /usr/lib/$(DEB_HOST_MULTIARCH)/libespeak.a /usr/lib/$(DEB_HOST_MULTIARCH)/libsonic.a /usr/lib/$(DEB_HOST_MULTIARCH)/libportaudio.a /usr/lib/$(DEB_HOST_MULTIARCH)/libjack.a -lm -lpthread -lasound -lrt
| UDEB_LDFLAGS += -u _Unwind_Resume -u __gcc_personality_v0 -u _Unwind_ForcedUnwind -u _Unwind_GetCFA -u _Unwind_GetBSP -lgcc_s
You'll probably need two separate builds anyway: one for the regular
deb packages, and one of the udeb.
Reassigning back, and bumping severity to serious.
Mraw,
KiBi.
PS: The fact the libgcc1 package can't be installed by the resolver
isn't fatal; it's actually ignored as one can see in the reporter's
log; the reference to a missing library at runtime is the culprit.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/pkg-lvm-maintainers/attachments/20150302/43c023b3/attachment-0001.sig>
More information about the pkg-lvm-maintainers
mailing list