Security update for Linux kernel
(Last modified: 31AUG2006)
solutions Security update for Linux kernel SuSE Linux Maintenance Web (e2ef926c5f7d109a84807c89d6446736)
Product(s): SUSE CORE 9 for x86
Novell Linux Desktop 9 for x86
Novell Linux POS 9
Open Enterprise Server
- CVE-2006-3745: A double userspace copy in a SCTP ioctl allows local attackers to overflow a buffer in the kernel, potentially allowing code execution and privilege escalation. [#199441]
- CVE-2006-4093: Local attackers were able to crash PowerPC systems with PPC970 processor using a not correctly disabled privileged instruction ("attn"). [#197810]
- CVE-2006-3468: Remote attackers able to access an NFS of a ext2 or ext3 filesystem can cause a denial of service (file system panic) via a crafted UDP packet with a V2 lookup procedure that specifies a bad file handle (inode number), which triggers an error and causes an exported directory to be remounted read-only. [#192988]
- disable reading from /proc/sys/vm/toss_page_cache_nodes [#199686]
- patches.suse/delayed-atime-3: fixed a typo [#197710]
- Enable context-dependent symlinks in OCFS2 [#195081]
- Enable watchdog on IBM x460 for i386/x86-64 [#194555]
- Fix knfsd intermittently returns ENOENT on GETATTR of valid file or directory [#193934]
- fix race on link [#191713]
- A set of NFS fixes: [#189132]
- VFS: Ensure LOOKUP_CONTINUE flag is preserved by link_path_walk()
- Optimise away an excessive GETATTR call when a file is symlinked
- Cleanup initialisation of struct nfs_fattr
- Convert cache_change_attribute into a jiffy-based value
- Make nfs_fattr record the time at which the rpc call started
- Optimise inode attribute cache updates
- Remove redundant 'verifier' argument to nfs_update_inode()
- Don't let nfs_end_data_update() clobber attribute update information
- Optimise attribute revalidation on close().
- forcedeth: fix random memory scribbling bug [#188013]
- Fix problem with /proc/scsi remove-single-device [#186236]
- Fix possible work queue corruption resulting in oops. [#183391]
- sun disk label fix: use unsigned rather than signed int for sector count. [#183366]
- Updated: Page cache locking fix [#183048]
- CKRM: Free argv stuff at exec if it has been allocated earlier [#182973]
- Fix for kernel panic during power-on sequence. [#175925]
- Updated the double completion fix for fusion driver [#172910]
- Fix race in pagevec_strip [#158289]
- Fix reiserfs latencies caused by data=ordered [#154374]
- Fixed: sysctl_overcommit_memory is set to OVERCOMMIT_NEVER [#153342]
- Fix return value for 'mt tell' ioctl [#145383]
- reiserfs: fix transaction overflowing [#145070]
- ignore multiple COMM_START_WRITE_NETDUMP_ACK packets [#138321]
- Make portmap requests 'hard' when nfs filesystem is mounted 'hard' [#95374]
- Altix: amend patches.fixes/tioce-bringup-fixes. [#187261]
- Altix: keep tioce_reserve_m32() from walking off the end of the CE ATE array. [#183472]
- Added latest patchset from IBM: Patchcluster 39
- Problem-ID: 25393 - xpram: module parameter parsing.
- Problem-ID: 24645 - qeth: qethconf not adding ipa entries
- Problem-ID: 25507 - cio: 5 min timeout after setting chpid offline.
- Problem-ID: 25511 - cio: Fix some path grouping and path verification related problems.
- Problem-ID: 25564 - qeth: race during setup of qeth device
- Problem-ID: 25599 - zfcp: ERP "deadlock" when registering a scsi device
- Problem-ID: 22069 - zfcp: duplicate usage of zfcp's scsi_er_timer
- Problem-ID: 25799 - iucv: multiple interfaces with same peer established
- Problem-ID: 25801 - cio: permanent subchannel busy conditions may cause I/O stall
- Problem-ID: 25802 - cio: Disallow ccwgroup devices containing non-unique ccw devices.
- Problem-ID: 26016 - qeth: race when reboot and recovery run concurrently
- Problem-ID: 26068 - qeth: kernel panic under heavy UDP workload
- Problem-ID: 26014 - qeth: stack trace with msg "inconsistent lock state"
- Problem-ID: 26103 - cio: I/O stall due to lost interupt after CHPID vary off/on cycle
- Problem-ID: 26118 - dasd: kernel BUG when setting a DASD device offline.
- Problem-ID: 19628 - zfcp: reopen adapter on do_QDIO error
- Problem-ID: 26144 - qeth: Setrouting for ipv6 invalid on hipersockets.
- Problem-ID: 23427, 24855 - cio: Inconsistent values in channel measurement facility.
- In the former Kernel the HZ_TIMER was switched on by default. This is now switched off. (see
cat /proc/sys/kernel/hz_timeron the system)
First, find out which kernel package to download and use, for example with
rpm -qf /boot/vmlinuz
Download the kernel image fitting your setup and the kernel symbols (kernel-syms*.rpm) and install it with either:
- rpm -Fvh kernel-syms*.rpm kernel-default*.rpm for the default kernel image, or
- rpm -Fhv kernel-syms*.rpm kernel-smp*.rpm for the SMP kernel image with support for up to 64 GB, or
- rpm -Fhv kernel-syms*.rpm kernel-bigsmp*.rpm for the SMP kernel image with support for up to 64 GB
In case you are using LILO as bootmanager, please make sure that you also execute the command
after installing the update for the system to remain bootable.
Finally, reboot the system with
shutdown -r now
to load the new kernel (replace "now" with the appropriate amount of time to allow local users to cleanly log out, for example "+5" for five minutes.)
Download Source Packages
Download the source code of the patches for maintained products.
The Origin of this information may be internal or external to Novell. Novell makes all reasonable efforts to verify this information. However, the information provided in this document is for your information only. Novell makes no explicit or implied claims to the validity of this information.
Any trademarks referenced in this document are the property of their respective owners. Consult your product manuals for complete trademark information.