Difference between revisions of "Download/kernel/rhel6/042stab111.11/changes"

From OpenVZ Virtuozzo Containers Wiki
Jump to: navigation, search
(111.11 changes)
 
(fix formatting)
 
(4 intermediate revisions by 3 users not shown)
Line 3: Line 3:
 
Since {{kernel link|rhel6|042stab108.8}}:
 
Since {{kernel link|rhel6|042stab108.8}}:
  
−
<!-- 111.11 -->
+
* Rebase on 504.30.3.el6 RHEL 6.6 kernel (non-security bug fixes, see link below)
−
* Fix for {{CVE|2015-5366}} and {{CVE|2015-5364}} extracted from RHEL6 kernel 2.6.32-573.3.1.el6 backported version of mainline commit beb39db59d1 "udp: fix behavior of wrong checksums"
+
: (all security fixes from new RHEL6 kernels were included into previous stable kernel 042stab108.8)
−
* fixed host hang where ploop stuck inside ploop_make_request() (PSBM-29066, {{OVZ|6293}}, {{OVZ|6392}})
+
* ploop could get stuck inside ploop_make_request() causing Node to hang (PSBM-29066, {{OVZ|6293}}, {{OVZ|6392}})
−
* cbt: add missed mutex_unlock (PSBM-39257)
+
* Under certain network workload, Container checkpoint/restore operations could result in conntrack duplication which could prevent online migration of said Container (PSBM-32411)
−
<!-- 111.10 -->
+
* Soft lockup in 'cgroup_kernel_close()' (PSBM-33300)
−
* dm-thin: do not optimize requeued bios (PSBM-39205)
+
* Kernel could crash in 'attach_task_by_pid()' when tasks exited. (PSBM-33381)
−
<!-- 111.9 -->
+
* CPT image compatibility was broken in 042stab108.x kernels. As a result, received but nondelivered TCP packets saved into CPT image could, in certain circumstances, lose TCP flags during restore on 042stab108.x kernels. This resulted in incorrect processing of restored packets and could cause unexpected kernel warnings or lead to issues in userspace applications (PSBM-33549, {{OVZ|6278}})
−
* s/PTR_ERR/PTR_RET, otherwise we always go to the err path (PSBM-35380)
+
* Fixed the reason of invalid warning "ida_remove called for id=0 which is not allocated." (PSBM-33789, {{OVZ|6161}})
−
<!-- 111.6 -->
+
* ploop journal could get stuck in 'block_write_full_page()' causing Node to hang. (PSBM-35095)
−
* cpt: do not dump packet sock vma to pram (PSBM-39014)
+
* In certain circumstances, Hardware Nodes with 108.x kernels could crash. (PSBM-37612)
−
* sched: Fix schedule_tail() to disable preemption (PSBM-35082)
+
* rsyslogd was not allowed to dump Container-related kernel messages inside Container. ({{OVZ|5774}})
−
* device_cgroup: fake allowing all devices for docker inside VZCT (PSBM-38691)
 
−
* cbt: fix bytes to block conversion bug (PSBM-39050)
 
−
<!-- 111.5 -->
 
−
* tty: Fix recursive deadlock in tty_perform_flush() (PSBM-37612)
 
−
* netfilter/ip6table_raw: Add permission check in ip6table_raw pernet (PSBM-35380)
 
−
* ploop: dio_fastmap() must refresh bvec_merge_data
 
−
* ploop: fix race on prealloc request (PSBM-35095)
 
−
<!-- 110.1 -->
 
−
* This is a rebase of 042stab109.12 to RHEL6.6 2.6.32-504.23.4.el6 kernel
 
−
<!-- 110.2 -->
 
−
* fix name_to_handle_at() for empty pathname ({{B|3280}})
 
−
* AT_EMPTY_PATH support for fstatat() with empty relative ({{B|3280|}})
 
−
* Double rtnl_unlock in netlink_dump() ({{B|3270}})
 
−
* diff-rst-iterative-adjust-percent-free-swap-pages ({{B|3231}})
 
−
* cbt: ignore device shrink
 
−
* ipv4: take rtnl_lock and mark mrt table as freed on namespace
 
−
* extra rtnl_unlock in sky2_resume()
 
−
* missing rtnl_unlock in i40evf_resume()
 
−
* missing rtnl_unlock in igb_sriov_reinit()
 
−
* rtnl_lock called twice in i40e_pci_error_resume()
 
−
* rtnl_lock in ip6mr_net_exit()
 
  
−
<!-- 109.12 -->
+
=== See also ===
−
* block: blk_cbt_update_size add block_dev sanity check (PSBM-34151, {{B|3240}}, {{B|3257}})
 
−
* rebase of 042stab109.12 to RHEL6.6 2.6.32-504.23.4.el6 kernel
 
  
−
<!-- 042stab109.11 -->
+
* {{RHSA|2015-1221}}
−
* block: blk_cbt_update_size add block_dev sanity check (PSBM-34151)
 
−
<!-- 042stab109.10 -->
 
−
* l2tp: Fix UDP socket reference count bugs in the pppol2tp ({{B|3233}})
 
−
* ploop: added into pvd header new close signatures (PSBM-33882)
 
−
* vmscan,tmpfs: treat used once pages on tmpfs as used once (PSBM-31450, PSBM-31450)
 
−
* blk: add changed block trace infrastructure (PSBM-32023)
 
−
* block: cbt fixup use-after-free inside __blk_cbt_set
 
−
* cbt: Use propper mem allocation context
 
−
* cbt: support blockdevice size update v2
 
−
 
 
−
<!-- 042stab109.9 -->
 
−
* cpt: restore tcp_flags for skb from receive and out-of-order (PSBM-33991)
 
−
* rst: remove incorrect warning "dirty skb cb" ({{B|3216}})
 
−
* rst: proper synhronisation with restored zombie processes (PSBM-25446, {{B|3085}}, {{B|3240}}, {{B|3085}})
 
−
<!-- 042stab109.8 -->
 
−
* cpt: dump struct tcp_skb_cb in old format (PSBM-33549, {{B|3237}})
 
−
* rst: initialization of new fields of struct tcp_skb_cb (PSBM-33549, {{B|3237}})
 
−
* mm: vmscan: stall page reclaim and writeback pages based on dirty/writepage pages encountered (PSBM-31450)
 
−
* vfs: Fix invalid ida_remove() call with mnt_release_group_id() (PSBM-33789, {{B|3117}})
 
−
<!-- 042stab109.6 -->
 
−
* ve net: Fix unbalanced double spin_unlock() in tcp_v4_kill_ve_sockets() (PSBM-33652)
 
−
<!-- 042stab109.4 -->
 
−
* syslog: distinguish between /proc/kmsg and syscalls ({{B|2693}})                     
 
−
* ve: Do not duplicate devmnt for a device (PSBM-33244)
 
−
* conntrack: Replace printk with ve_printk ({{B|2940}})
 
−
* RH6 tun: fix module unload (PSBM-33063)
 
−
* vzcpt: Fail restore if wrong tty device (PSBM-32795)
 
−
* sched: load_balance_fair: fix possible div/0 (PSBM-33347)                                             
 
−
* ub_dcache_unuse: call ub_dcache_remove after racing unmounts completed (PSBM-32700)
 
−
<!-- 042stab109.3 -->
 
−
* fix diff-ve-make-get_ve_by_id-use-RCU-instead-of-ve_list_lock (PSBM-33300)
 
−
* rebase to rh6 kernel 2.6.32-504.16.2.el6
 
−
<!-- 042stab109.2 -->
 
−
* srcu: update to v3.10 state (v2) (PSBM-32230)
 
−
* linkwatch: linkwatch_forget_dev() to speedup device dismantle (PSBM-32230)
 
−
 
 
−
 
 
−
== See also ==
 
−
 
 
−
* {{RHSA|2015-1081}}
 
 
* {{RHSA|2015-1081}}
 
* {{RHSA|2015-1081}}
−
* {{CVE|2015-5366}}
 
−
* {{CVE|2015-5364}}
 

Latest revision as of 18:41, 13 September 2016

Changes[edit]

Since 042stab108.8:

  • Rebase on 504.30.3.el6 RHEL 6.6 kernel (non-security bug fixes, see link below)
(all security fixes from new RHEL6 kernels were included into previous stable kernel 042stab108.8)
  • ploop could get stuck inside ploop_make_request() causing Node to hang (PSBM-29066, OVZ-6293, OVZ-6392)
  • Under certain network workload, Container checkpoint/restore operations could result in conntrack duplication which could prevent online migration of said Container (PSBM-32411)
  • Soft lockup in 'cgroup_kernel_close()' (PSBM-33300)
  • Kernel could crash in 'attach_task_by_pid()' when tasks exited. (PSBM-33381)
  • CPT image compatibility was broken in 042stab108.x kernels. As a result, received but nondelivered TCP packets saved into CPT image could, in certain circumstances, lose TCP flags during restore on 042stab108.x kernels. This resulted in incorrect processing of restored packets and could cause unexpected kernel warnings or lead to issues in userspace applications (PSBM-33549, OVZ-6278)
  • Fixed the reason of invalid warning "ida_remove called for id=0 which is not allocated." (PSBM-33789, OVZ-6161)
  • ploop journal could get stuck in 'block_write_full_page()' causing Node to hang. (PSBM-35095)
  • In certain circumstances, Hardware Nodes with 108.x kernels could crash. (PSBM-37612)
  • rsyslogd was not allowed to dump Container-related kernel messages inside Container. (OVZ-5774)

See also[edit]

Retrieved from "https://wiki.openvz.org/index.php?title=Download/kernel/rhel6/042stab111.11/changes&oldid=20980"