OpenVZ-legacy
  1. OpenVZ-legacy

linux-2.6.18-openvz

Public
AuthorCommitMessageCommit dateIssues
OpenVZ teamPavelOpenVZ team
e1b310daa23linux-2.6.18-028test007 released
Pavel EmelianovPavel Emelianov
8dd709613ecAdd asm/timex.h to linux/ve_task.hThis makes x86_64 compile.
Dmitry MishinPavel EmelianovDmitry Mishin
529e2b23b3eFixed capabilities check for tun device in VE.http://bugzilla.openvz.org/show_bug.cgi?id=342
Pavel EmelianovPavel Emelianov
84bcff981abOuput virtual pid not real as first field in /proc/$PID/stat(Bad pid confuses lsof and maybe some other tools) http://bugzilla.openvz.org/show_bug.cgi?id=382
Pavel EmelianovPavel Emelianov
c4a253d527c[CPT] Make CPT modules compile w/o beancounters.http://bugzilla.openvz.org/show_bug.cgi?id=377
Pavel EmelianovPavel Emelianov
d9d7d4ffb92Make kernel compile w/o beancounters.http://bugzilla.openvz.org/show_bug.cgi?id=377
Andrey MirkinPavel EmelianovAndrey Mirkin
1a29552fb78Fix compat_sys_mount OOPsdata_page pointer can be NULL, so we should skip data conversion in this case.
Denis V. LunevPavel EmelianovDenis V. Lunev
42ec7722767Hide the content of /proc/net/sockstat rather than the file itselfThis keeps sysstat from crashing (#72587)
Denis V. LunevPavel EmelianovDenis V. Lunev
bc8f9f2ebf6Fix dcache leak on race from dcache_charge[_forced] and dcache_unchangeThe idea: do not trust dentry_bc after count state change (#72051)
Alexey DobriyanPavel EmelianovAlexey Dobriyan
8117159cc95Clear stats on VCPU additionWhen VCPU is removed then added back, all system/idle/... statistics magically reappears and is not zero. This is a fix for previous patch.
Alexey DobriyanPavel EmelianovAlexey Dobriyan
965e7934881Fix VCPU user=system=nice=idle=0% bugVE has simple idle time collection logic (per VCPU ->strt_idle_time, ->idle_time). For ->idle_time incrementing ->strt_idle_time must not be 0. This happens when the very first task is scheduled on VCPU. Before that all VCPU statistics is zeroed out because of ve = kzalloc(sizeof(struct ve_struct)); including ->strt_idle_time. All this leads to suprising /proc/stat and, as a consequence, t...
Pavel EmelianovPavel Emelianov
c06a11e5140Fixed numiptent recharging in xt_replace_tableCharge that only difference between previous number of entries and new number is charged. Previous code tried to charge new number before and uncharge old value than. This leaded to failures even if the resulting value was below limits. (#72527) Based on patch from Dmitry <dim@openvz.org>
Pavel EmelianovPavel Emelianov
ea61ccc1f75Change context to ve0/ub0 on irq_enter automatically
Pavel EmelianovPavel Emelianov
e0297976ebbFix scheduler deadlock in idle_balance.load_balance_newidle() could spin around redo: mark infinitely because find_busiest_queue() didn't take into account the fact that it cold be called more than once. http://bugzilla.openvz.org/show_bug.cgi?id=379
Pavel EmelianovPavel Emelianov
8b11b2b1bcfAdd missed #include <linux/ve_proto.h> to fs/char_dev.cThis makes it comple w/o warnings and with CONFIG_VE=n.
Alexey DobriyanPavel EmelianovAlexey Dobriyan
c3f9ed9e2dcSimplify ve_nr_running_inc()/dec()Incrementing is incrementing. You don't need to pass cycles. Garbage-collect "cycles" variables.
Denis V. LunevPavel EmelianovDenis V. Lunev
7071fe31412Fixed "Oops in connect init" warnings during IPv6 connecthttp://bugzilla.openvz.org/show_bug.cgi?id=336
Dmitry MishinPavel EmelianovDmitry Mishin
2e3217d5344[NETFILTER] Fixed operation of IPT_SO_GET_REVISION_* getsockopts in compat mode
Dmitry MishinPavel EmelianovDmitry Mishin
9fc8aee38dc[NETFILTER] Targets/matches cleanupIptables reorganization. Now targets and matches are common for all VEs. Thus VEs may use majority of existent targets/matches without necessity to virtualize them.
Dmitry MishinPavel EmelianovDmitry Mishin
1ca3f0768cd[NETFILTER] Fixed xt_limit operations in compat mode
Dmitry MishinPavel EmelianovDmitry Mishin
df3e9cfecf0[NETFILTER] Fix of iptables checksIssue found by Patrick McHardy. After checks reordering target and matches checks that they could be used for this hook returns true always due to not initialized e->comefrom field. So, order restored, necessary checks moved in mark_source_chains(). For compats this issue exists from the beginning.
Pavel EmelianovPavel Emelianov
ba0b76f060dRemove struct pid from vpid hash on free_pid().This fixes a random memory corruption foung on 028test006 kernel.
Pavel EmelianovPavel Emelianov
dac590fff95Fixed a misprint in recent patch (find_vpid())
Pavel EmelianovPavel Emelianov
1ec05b3deec[CPT] Fix deadlock in CPT rst_signal_completeThe problem is that vpid_to_pid() which takes tasklist_lock for reading is called with this lock held for writing. Fix this by introducing find_vpid() which scans throught the pid hash once and obtains needed struct pid. (#72469)
OpenVZ teamPavelOpenVZ team
dc8c4d7ac46linux-2.6.18-028test006 released
Dmitry MishinPavel EmelianovDmitry Mishin
2d46cb1712de1000: fix initialization of irqsIn case of irqpoll boot option set, e1000 may oops due to: 1) e1000 register it's handler with e1000_request_irq; 2) spurious interrupt happens; 3) kernel tries to handle this interrupt with all available descs; 4) e1000_intr is called and oops due to not initialized clean_rx handler, because e1000_up is not called yet. Solution is to initialize driver before handler registration. Sig...
Alexey DobriyanPavel EmelianovAlexey Dobriyan
2467b51b2ffVirtualize /proc/$PID/mountstatsCurrently /proc/$PID/mountstats leaks information about host. Approach identical to /proc/$PID/mounts which leaves only something like that. Some refactoring was done to minimize duplicated code. However this made errno semantics less than perfect. http://bugzilla.openvz.org/show_bug.cgi?id=354 Some changes in errno semantics by Pavel Emelianov <xemul@openvz.org>
Pavel EmelianovPavel Emelianov
c3e2efbd338Make ctl_table-s with virtualized handlers workE.g. /proc/sys/kernel/hostname must have rw access in VE but change hostname of caller's VE only. http://bugzilla.openvz.org/show_bug.cgi?id=376
Alexey DobriyanPavel EmelianovAlexey Dobriyan
6e36c32f919Make /proc/sys/kernel/virt_osrelease workRead from /proc/sys/kernel/virt_osrelease reported EINVAL due to missed case in switch statement http://bugzilla.openvz.org/show_bug.cgi?id=361
Pavel EmelianovPavel Emelianov
000de19ed27Cleanup VE cleaning procedure - remove KSYMCALLS; - intruduce pget_ve/pput_ve to make code look cleaner.
Alexey KuznetsovPavel EmelianovAlexey Kuznetsov
562f41b4993[PATCH] port of VPIDs to 2.6.18The patch mostly simplifies the things: it contains more -'s than +'s. 1. We do not create separate structures vpid_mapping anymore. Core already introduced this overhead as "struct pid". 2. Logically, we do not need to add to task_struct vpid's, they can be obtained as tsk->pids[PIDTYPE_x].pid->vnr. However, we still add vpid,vtgid,vpgrp and vsession. The reasons are: A. tsk->pids...
Alexey KuznetsovPavel EmelianovAlexey Kuznetsov
aea09e2f1db[CPT] do not checkpoint/restore global process groupsThe patch is three-fold: 1. Do not try to allocate process groups/sessions, unless they are not virtual. This is fix for bug #71825. However, it is too late to detect failure. 2. Do not checkpoint VE, if it contains references to extenal process groups/session ids. It is _destructive_ part. It definitely will prevent migration of some commonly used configurations, when some def...
Alexey KuznetsovPavel EmelianovAlexey Kuznetsov
854787c6059[CPT] Fail when VE refers to an invisible fileCheckpointing used to ignore EINVAL returned by d_path(). It was workaround for tmpfs shmem files, which use detached mounts. But this means that real invisible paths are detected too late: checkpointing succeds and restore fails, which is not acceptable. When d_path() fails, check that it is shmem file. If it is not, fail immediately.
Andrey MirkinPavel EmelianovAndrey Mirkin
da4e5ca5797[CPT] Wrong structure with inode attributes was used in fixup_file_content()We must be sure that we won't break file system, thus we do not set inode mode attributes from S_IFMT mask. (#71135, #71161)
Alexey DobriyanPavel EmelianovAlexey Dobriyan
b97b68864cb[VZDQ] Debug patch for resolving OVZ bugs #341, #116, #177 * Extend ->origin into array for previous origin tracking. * Print last two origins in debugging message. * Also print i_mode, i_op, i_fop, ... of the offending inode.
Andrey MirkinPavel EmelianovAndrey Mirkin
3410797a9eeWhen we moving network device from VE0 we must check that VE doesn't havedevice with the same name
Andrey MirkinPavel EmelianovAndrey Mirkin
09a52586283Context should be set from device (dev->owner_env) even for devices from VE0.
Denis V. LunevPavel EmelianovDenis V. Lunev
8e7af019747If a process is being killed inside __alloc_pages by OOM, it can't exit till it frees some space, which is impossible for now.The patch allows to dig into reserves for such a process (#71604, #71179)
Chris WrightPavel EmelianovChris Wright
b3e18f6d33cbridge: fix possible overflow in get_fdb_entries (CVE-2006-5751)Make sure to properly clamp maxnum to avoid overflow (CVE-2006-5751). Signed-off-by: Chris Wright <chrisw@sous-sol.org>CVE-2006
OpenVZ teamPavel EmelianovOpenVZ team
a92e24faa29MMerged 2.6.18.3 from /linux/kernel/git/stable/linux-2.6.18.y
Pavel EmelianovPavel Emelianov
2d13ace4fc6Fixed file name in comment of kernel/ve/devperms.c
Alexey DobriyanPavel EmelianovAlexey Dobriyan
e8b2f264aefRecursive locking on Alt+SysRq+a Alt+SysRq+a shows some information about virtual cpus and printk itunder runqueue lock. printk() also can take runqueue lock via default_wake_function __wake_up_common __wake_up wake_up_klogd release_console_sem __vprintk vprintk printk So, gather information under runqueue lock, print it later. http://bugzilla.openvz.org/show_bug.cgi?id=364
Alexey DobriyanPavel EmelianovAlexey Dobriyan
c8dbb170b98Alt+SysRq+a crash if command line contains %n
Pavel EmelianovPavel Emelianov
ae278dc3353Rework/cleanup of VE device permissions subsystem - move to -y part to eliminate KSYMCALLS; - make RCU-based; - code cleanup.
Andrey MirkinPavel EmelianovAndrey Mirkin
c7f0426142e[CPT] Move arch ret_from_fork hooks to appropriate places in entry.SIn CPT we have our own hooks in i386/x86_64 for processes restoring. In these hooks we have explicit offsets from ret_from_fork address. Thus, any changes in ret_from_fork can lead to kernel crash. The proper way is to move this hooks to entry.S. http://bugzilla.openvz.org/show_bug.cgi?id=351
Pavel EmelianovPavel Emelianov
56c0ac39e97Make per-VE task list RCUhttp://bugzilla.openvz.org/show_bug.cgi?id=358
Pavel EmelianovPavel Emelianov
5fafdf81ddf[VZDQ] Add #include <linux/vmalloc.h> to make vzdq_uid.c compile cleanly
Pavel EmelianovPavel Emelianov
0434b3aef1bFixe sk->owner_env access in some places
Pavel EmelianovPavel Emelianov
41d0fb27959Removed #include <ve_owner.h> from codeForgot to do it when removed the file itself.
Dmitry MishinPavel EmelianovDmitry Mishin
ad5ac420fedFixed iptables get functionality(#72270) Signed-off-by: Dmitry Mishin <dim@openvz.org> Acked-by: Kirill Korotaev <dev@openvz.org>