[ubuntu/utopic-proposed] crash 7.0.5-1ubuntu1 (Accepted)
Matthias Klose
doko at ubuntu.com
Mon May 19 13:53:18 UTC 2014
crash (7.0.5-1ubuntu1) utopic; urgency=medium
* Merge with Debian; remaining changes:
- Build for armhf, arm64.
- Minor fixes for live autopkgtest.
- debian/tests/live: Redirect gpg stderr to stdout, as that's the only
known and expected stderr source.
* Add ppc64el support (Bharata Rao, Mauricio Faria de Oliveira).
crash (7.0.5-1) unstable; urgency=low
* New upstream version 7.0.5
* Fix for the "runq -g" option for kernels that are configured with
CONFIG_FAIR_GROUP_SCHED, but not CONFIG_CFS_BANDWIDTH. Without the patch,
the command fails with the message "runq: invalid structure member offset:
cfs_rq_throttled".
* Add support for Xen PVH guest types introduced in Xen 4.4. Without the
patch, running against a Xen 4.4 hypervisor binary would fail during
session initialization with the error message "crash: invalid structure
member offset: domain_is_hvm". In addition, the PVH guest type is being
registered internally as an HVM guest type, the debug "help -X ofs"
command's display of the domain_domain_flags offset has been fixed to show
it in decimal, and the setting of the internal dc->domain_flags has been
fixed to contain all flags set, not just the first one found.
* Fix for the "kmem -S" command on Linux 3.1 and later kernels that are
configured with CONFIG_SLUB. Because the the page structure's inuse and
objects fields used by SLUB were changed from discrete u16 types to
bit-fields within an unsigned int, the display of per-node partial slab
statistics are incorrect. Without the patch, the TOTAL and ALLOCATED
values are incorrectly shown as equal values, and therefore the FREE value
is always zero.
* Fix for the "kmem -S" command for kernels that are configured with
CONFIG_SLUB. Eash per-cpu slab object dump may show incorrect ALLOCATED
and FREE values; and as seen on Linux 3.5 and later kernels, the TOTAL
value and the number of individual objects dumped may also be incorrect
(too small).
* When executing the commands from an input file specified by the "-i
<file>" command line option, or when accepting input from a file as a set
of commands or as a set of command arguments using the "<" redirection
character, unconditionally cease the operation if CTRL-c is entered.
Without the patch, depending upon the command that was running when the
SIGINT was received, the operation may continue uninterruptibly until the
file contents are consumed.
* Enhanced the "bt -F" option such that if "-F" is entered twice, and if the
stack frame contents reference a slab cache object, both the slab cache
name and the stack contents will be displayed within brackets.
* Enhanced the "rd -S" option such that if "-S" is entered twice, and if the
memory contents reference a slab cache object, both the slab cache name
and the memory contents will be displayed within brackets.
* Fix for the X86_64 "bt" command to prevent an unwarranted message
indicating "WARNING: possibly bogus exception frame" generated from a
blocked kernel thread that was in the process of exec'ing a user process
via the call_usermodehelper() facility.
* Fix for the X86_64 "bt" command to more correctly determine the function
frame that called into an interrupted function. Without the patch, the
first frame just above an IRQ exception frame register dump may show an
invalid/stale function.
* Fix for the X86_64 "bt" command if a page fault exception was generated by
the invalid contents of the RIP register. Without the patch, the
exception frame register dump is not displayed above the "page_fault"
stack frame; and in a related issue, the "bt -e" option will not find and
display the exception frame.
* When invoking a crash session with a compressed vmlinux file, make the
same host-machine/vmlinux endian verification that is done with
uncompressed vmlinx files.
* Reduce the number of CTRL-c entries required to unconditionally terminate
any manually-entered command from three to one.
* Fix for the X86_64 "bt" command if an async page fault exception occurred
in a KVM guest running a Linux 2.6.38 or later kernel. Without the patch,
the exception frame register dump is not displayed above the
"async_page_fault" stack frame.
Date: Thu, 15 May 2014 18:22:35 +0200
Changed-By: Matthias Klose <doko at ubuntu.com>
Maintainer: Ubuntu Developers <ubuntu-devel-discuss at lists.ubuntu.com>
https://launchpad.net/ubuntu/utopic/+source/crash/7.0.5-1ubuntu1
-------------- next part --------------
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Format: 1.8
Date: Thu, 15 May 2014 18:22:35 +0200
Source: crash
Binary: crash
Architecture: source
Version: 7.0.5-1ubuntu1
Distribution: utopic
Urgency: medium
Maintainer: Ubuntu Developers <ubuntu-devel-discuss at lists.ubuntu.com>
Changed-By: Matthias Klose <doko at ubuntu.com>
Description:
crash - kernel debugging utility, allowing gdb like syntax
Changes:
crash (7.0.5-1ubuntu1) utopic; urgency=medium
.
* Merge with Debian; remaining changes:
- Build for armhf, arm64.
- Minor fixes for live autopkgtest.
- debian/tests/live: Redirect gpg stderr to stdout, as that's the only
known and expected stderr source.
* Add ppc64el support (Bharata Rao, Mauricio Faria de Oliveira).
.
crash (7.0.5-1) unstable; urgency=low
.
* New upstream version 7.0.5
* Fix for the "runq -g" option for kernels that are configured with
CONFIG_FAIR_GROUP_SCHED, but not CONFIG_CFS_BANDWIDTH. Without the patch,
the command fails with the message "runq: invalid structure member offset:
cfs_rq_throttled".
* Add support for Xen PVH guest types introduced in Xen 4.4. Without the
patch, running against a Xen 4.4 hypervisor binary would fail during
session initialization with the error message "crash: invalid structure
member offset: domain_is_hvm". In addition, the PVH guest type is being
registered internally as an HVM guest type, the debug "help -X ofs"
command's display of the domain_domain_flags offset has been fixed to show
it in decimal, and the setting of the internal dc->domain_flags has been
fixed to contain all flags set, not just the first one found.
* Fix for the "kmem -S" command on Linux 3.1 and later kernels that are
configured with CONFIG_SLUB. Because the the page structure's inuse and
objects fields used by SLUB were changed from discrete u16 types to
bit-fields within an unsigned int, the display of per-node partial slab
statistics are incorrect. Without the patch, the TOTAL and ALLOCATED
values are incorrectly shown as equal values, and therefore the FREE value
is always zero.
* Fix for the "kmem -S" command for kernels that are configured with
CONFIG_SLUB. Eash per-cpu slab object dump may show incorrect ALLOCATED
and FREE values; and as seen on Linux 3.5 and later kernels, the TOTAL
value and the number of individual objects dumped may also be incorrect
(too small).
* When executing the commands from an input file specified by the "-i
<file>" command line option, or when accepting input from a file as a set
of commands or as a set of command arguments using the "<" redirection
character, unconditionally cease the operation if CTRL-c is entered.
Without the patch, depending upon the command that was running when the
SIGINT was received, the operation may continue uninterruptibly until the
file contents are consumed.
* Enhanced the "bt -F" option such that if "-F" is entered twice, and if the
stack frame contents reference a slab cache object, both the slab cache
name and the stack contents will be displayed within brackets.
* Enhanced the "rd -S" option such that if "-S" is entered twice, and if the
memory contents reference a slab cache object, both the slab cache name
and the memory contents will be displayed within brackets.
* Fix for the X86_64 "bt" command to prevent an unwarranted message
indicating "WARNING: possibly bogus exception frame" generated from a
blocked kernel thread that was in the process of exec'ing a user process
via the call_usermodehelper() facility.
* Fix for the X86_64 "bt" command to more correctly determine the function
frame that called into an interrupted function. Without the patch, the
first frame just above an IRQ exception frame register dump may show an
invalid/stale function.
* Fix for the X86_64 "bt" command if a page fault exception was generated by
the invalid contents of the RIP register. Without the patch, the
exception frame register dump is not displayed above the "page_fault"
stack frame; and in a related issue, the "bt -e" option will not find and
display the exception frame.
* When invoking a crash session with a compressed vmlinux file, make the
same host-machine/vmlinux endian verification that is done with
uncompressed vmlinx files.
* Reduce the number of CTRL-c entries required to unconditionally terminate
any manually-entered command from three to one.
* Fix for the X86_64 "bt" command if an async page fault exception occurred
in a KVM guest running a Linux 2.6.38 or later kernel. Without the patch,
the exception frame register dump is not displayed above the
"async_page_fault" stack frame.
Checksums-Sha1:
4b8829db41d49d18f379891537c42cefcbdbdd4c 1326 crash_7.0.5-1ubuntu1.dsc
515214f54d111df835f3ca3a352e0602073edd3e 32028646 crash_7.0.5.orig.tar.gz
9277482ae6f2e6790944d48a1e07f434f27e562e 122348 crash_7.0.5-1ubuntu1.diff.gz
Checksums-Sha256:
f067acf289c37addd547e6096a3b868c61b41f27cf68c5d85a72d4dcadee1bd1 1326 crash_7.0.5-1ubuntu1.dsc
d5c5b2c8a5dc1e85d38d847cdfea1aa37a58ad38711bc3727c28f22618bfe64c 32028646 crash_7.0.5.orig.tar.gz
fa7b600ce1e0691a827710d5a81d5538da1924f82f4c4da3ef0069897b0c32fc 122348 crash_7.0.5-1ubuntu1.diff.gz
Files:
270fdfb0ad3052194a6fb8ab18b89f9f 1326 utils optional crash_7.0.5-1ubuntu1.dsc
3fba47055539e61ba3fb4d6e8a0e84b0 32028646 utils optional crash_7.0.5.orig.tar.gz
88db23a52f7c2f041ed174a90d163d5f 122348 utils optional crash_7.0.5-1ubuntu1.diff.gz
Original-Maintainer: Troy Heber <troyh at debian.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iEYEARECAAYFAlN6DG8ACgkQStlRaw+TLJxl/wCgo1D8hPhLtl1BDTE877DfDW31
xScAn0jKwW1YYl7fAYYaTJ+Ua5fZlpF0
=2qJP
-----END PGP SIGNATURE-----
More information about the Utopic-changes
mailing list