Download/kernel/2.6.16/026test007.1/changes
Contents
- 1 Changes
- 1.1 Patches
- 1.1.1 diff-ms-files-fix-20060329
- 1.1.2 diff-ms-files-race-fix-20060329
- 1.1.3 diff-ubc-follow-page-k-20060321
- 1.1.4 diff-ubc-kmem-objuse-20060329
- 1.1.5 diff-ubc-net-skb-orphan-unch-20060401
- 1.1.6 diff-ubc-thread-info-20060329
- 1.1.7 diff-ve-compile-20060323
- 1.1.8 diff-ve-file-owner-leak-20060329
- 1.1.9 diff-ve-net-conntrack-oops-fix-20060331
- 1.1.10 diff-ve-net-ipv6-loopback-20060320
- 1.1.11 diff-ve-net-netfilter-ip-ct-enable-warn-20060327
- 1.1.12 diff-ve-net-netfilter-ub-200060330
- 1.1.13 diff-ve-net-nf-unconfirmed-20060331
- 1.1 Patches
Changes
- Critical files_struct refcounting and conntracks fixes
- Compilation fixes
- Lost UBC accounting.
For the complete list of changes in this release, see git changelog for kernel 026test007.
Patches
diff-ms-files-fix-20060329
Patch from Dmitry Mishin <dim@openvz.org>:
Fixed error path of dup_fd()
diff-ms-files-race-fix-20060329
Patch from Pavel Emelianov <xemul@openvz.org>:
Fixed race on put_files_struct on exec with proc.
Restoring files on current on error path may lead to proc holds pointer on already kfree-d files_struct.
diff-ubc-follow-page-k-20060321
Patch from Pavel Emelianov <xemul@openvz.org>:
Remove FOLL_KERN flag form follow_page. It was used by vmalloc_ub only and vmalloc_to_page is enough for it.
diff-ubc-kmem-objuse-20060329
Patch from Pavel Emelianov <xemul@openvz.org>:
Issue found by Alexey Kuznetsov (alexey@openvz.org).
In 2.6.16 some wrong numbers are counted in kmemsize. The code makes some efforts to calculate right number (objuse), but does not use the result. Instead it uses allocated size (obj_size), which is less than even the number shown in /proc/slabinfo (buffer_size).
diff-ubc-net-skb-orphan-unch-20060401
Patch from Alexey Kuznetsov <alexey@openvz.org>:
Fix leakage in UBC on datagram sockets
diff-ubc-thread-info-20060329
Patch from Alexey Kuznetsov <alexey@openvz.org>:
Kernel stacks are not accounted in ubc
diff-ve-compile-20060323
Patch from Dmitry Mishin <dim@openvz.org>:
Fixed compilation with CONFIG_VE_CALLS=y and CONFIG_PROC_FS=n.
Issue found by Benedict (chubby):
http://forum.openvz.org/index.php?t=tree&th=384&mid=2161
diff-ve-file-owner-leak-20060329
Patch from Alexey Kuznetsov <alexey@openvz.org>:
When numfiles ubc overflows, ve_struct leaks
diff-ve-net-conntrack-oops-fix-20060331
Patch from Dmitry Mishin <dim@openvz.org>:
Fix of oops in __ip_conntrack_find.
VPS could issue SO_GET_ORIG_DST get sockopt while they have no ip_conntracks enabled. This leads to oops. Added check to return -ENOPROTOOPT for such case.
diff-ve-net-ipv6-loopback-20060320
Patch from Dmitry Mishin <dim@openvz.org>:
Fixed compilation with CONFIG_IPV6 on, which was broken due to loopback_dev virtualization.
diff-ve-net-netfilter-ip-ct-enable-warn-20060327
Patch from Dmitry Mishin <dim@openvz.org>:
Added warn message for obsoleted ip_conntrack_enable_ve0 param.
diff-ve-net-netfilter-ub-200060330
Patch from Dmitry Mishin <dim@openvz.org>:
x_tables should use mem_ub() function instead of vmalloc_ub()
diff-ve-net-nf-unconfirmed-20060331
Patch from Dmitry Mishin <dim@openvz.org>: Virtualize 'unconfirmed' conntracks' list and thus fix soft lockup on conntracks cleanup.