Re: [Devel] [PATCH 3/3] proc-net: virtualize all the network proc entries

2017-07-20 Thread Stanislav Kinsburskiy
20 июля 2017 г. 23:34 пользователь Andrey Vagin написал: On Thu, Jul 20, 2017 at 08:47:58PM +0400, Stanislav Kinsburskiy wrote: > Use proc_net_create() and proc_net_create_data() for all the per-net dentries. > We need this to expose per-net sysfs dentries once commit > e5990221bde54d757a5b5d85e

[Devel] [PATCH vz7] ploop: push_backup: ploop_pb_get_pending should wait again instead of ENOENT

2017-07-20 Thread Maxim Patlasov
The patch fixes a race when ploop_pb_get_pending was rightly woken up to pass an extent to userspace, but before it re-acquire pbd->ppb_lock another thread of vz_backup_agent reports exactly this extent as processed. This effectively steals the extent from ploop_pb_get_pending, so it fails to get

Re: [Devel] [PATCH 3/3] proc-net: virtualize all the network proc entries

2017-07-20 Thread Andrei Vagin
On Thu, Jul 20, 2017 at 08:47:58PM +0400, Stanislav Kinsburskiy wrote: > Use proc_net_create() and proc_net_create_data() for all the per-net dentries. > We need this to expose per-net sysfs dentries once commit > e5990221bde54d757a5b5d85e81b2535064570ad is reverted Could you give more details? Do

Re: [Devel] [PATCH 0/3] proc-net: rework containerization

2017-07-20 Thread Dmitry Safonov
On 07/20/2017 07:47 PM, Stanislav Kinsburskiy wrote: This is precursor series in scope of https://jira.sw.ru/browse/PSBM-58574 S_ISGID is set for procfs only in proc_net.c, so this looks good to me, Reviewed-by: Dmitry Safonov --- Stanislav Kinsburskiy (3): proc-net: revert e599022

[Devel] [PATCH 3/3] proc-net: virtualize all the network proc entries

2017-07-20 Thread Stanislav Kinsburskiy
Use proc_net_create() and proc_net_create_data() for all the per-net dentries. We need this to expose per-net sysfs dentries once commit e5990221bde54d757a5b5d85e81b2535064570ad is reverted Signed-off-by: Stanislav Kinsburskiy --- drivers/net/ppp/pppoe.c|2 +- net

[Devel] [PATCH 1/3] proc-net: revert e5990221bde54d757a5b5d85e81b2535064570ad

2017-07-20 Thread Stanislav Kinsburskiy
Revert "proc-net: mark all files in proc-net directories as virtualized". This tirck doesn't allow to define any non-visible per-net dentry, while we will need it it to support container migration with disabled netfilter. Signed-off-by: Stanislav Kinsburskiy --- fs/proc/generic.c |3 -

[Devel] [PATCH 0/3] proc-net: rework containerization

2017-07-20 Thread Stanislav Kinsburskiy
This is precursor series in scope of https://jira.sw.ru/browse/PSBM-58574 --- Stanislav Kinsburskiy (3): proc-net: revert e5990221bde54d757a5b5d85e81b2535064570ad proc-net: proc_net_create() helper introduced proc-net: virtualize all the network proc entries drivers/net/ppp/

[Devel] [PATCH 2/3] proc-net: proc_net_create() helper introduced

2017-07-20 Thread Stanislav Kinsburskiy
Signed-off-by: Stanislav Kinsburskiy --- include/linux/proc_fs.h |7 +++ 1 file changed, 7 insertions(+) diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h index 4660327..1b516bc 100644 --- a/include/linux/proc_fs.h +++ b/include/linux/proc_fs.h @@ -67,6 +67,13 @@ static inli

Re: [Devel] [PATCH rh7] kobj, uevent: allow to broadcast event to ve0 from any ve

2017-07-20 Thread Cyrill Gorcunov
On Thu, Jul 20, 2017 at 06:49:29PM +0300, Andrey Ryabinin wrote: > Commit 732eb5c2ac0a ("ve/kobj: Send events per VE instead of > all net-namespaces broadcasting") forbid to send uevents to any another > ve. So when container stumps on FS-corruption and sends uevent about that, > the host uevents w

Re: [Devel] [PATCH criu] page-xfer: handle a case when splice returns zero

2017-07-20 Thread Dmitry Safonov
On 07/20/2017 05:33 AM, Andrei Vagin wrote: From: Andrei Vagin A return value of 0 means end of input, so we need to stop reading from this descriptor. https://jira.sw.ru/browse/PSBM-68732 Cc: Dmitry Safonov Signed-off-by: Andrei Vagin Applied to vz-criu, released in criu-3.0.0.24-1.vz7

[Devel] [PATCH rh7] kobj, uevent: allow to broadcast event to ve0 from any ve

2017-07-20 Thread Andrey Ryabinin
Commit 732eb5c2ac0a ("ve/kobj: Send events per VE instead of all net-namespaces broadcasting") forbid to send uevents to any another ve. So when container stumps on FS-corruption and sends uevent about that, the host uevents won't see it. Let's allow to broadcast any events to the ve0 from other ve

Re: [Devel] [PATCH RH7 v3 1/2] fence-watchdog: log wdog actions directly to special file

2017-07-20 Thread Kirill Tkhai
On 20.07.2017 17:58, Pavel Tikhomirov wrote: > If watchdog was triggered, that means that userspace is unreliable > and we can't expect kernel messages are logged by userspace logger, > thus we need ourselfs write info about watchdog triggered to special > file configured through sysfs /sys/kernel/

[Devel] [PATCH RH7 v3 1/2] fence-watchdog: log wdog actions directly to special file

2017-07-20 Thread Pavel Tikhomirov
If watchdog was triggered, that means that userspace is unreliable and we can't expect kernel messages are logged by userspace logger, thus we need ourselfs write info about watchdog triggered to special file configured through sysfs /sys/kernel/watchdog_log_path, so that users can see that watchdo

[Devel] [PATCH v3 2/2] fence-watchdog: panic after 30 sec of reboot/halt

2017-07-20 Thread Pavel Tikhomirov
As we do reboot and halt actions in scope of scheduled worker it can never happen if scheduling does not work properly, so panic in case that previous action was not successful. https://jira.sw.ru/browse/PSBM-54747 Signed-off-by: Pavel Tikhomirov --- kernel/fence-watchdog.c | 20

Re: [Devel] [PATCH] kvm: x86: hyperv: avoid livelock in oneshot SynIC timers

2017-07-20 Thread Roman Kagan
On Thu, Jul 20, 2017 at 11:38:19AM +0300, Konstantin Khorenko wrote: > Need it in update 5? That's up to Den to decide. The livelock is only triggerable with one-shot SynIC timers; none of our regular guests uses them (Windows uses periodic timers, Linux does use one-shot timers but needs to be s

[Devel] [PATCH RHEL7 COMMIT] mm/tcache: use lockless lookups in tcache's page tree.

2017-07-20 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-514.26.1.vz7.33.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-514.26.1.vz7.33.11 --> commit ba2ea161f25b4cc8c7de15be06f1001dd4d3ee24 Author: Andrey Ryabinin Date: Thu Jul 20 12:52:43 2017 +0400 mm/tcach

[Devel] [PATCH RHEL7 COMMIT] mm/tcache: drop useless replace logic during attach page.

2017-07-20 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-514.26.1.vz7.33.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-514.26.1.vz7.33.11 --> commit 3858830c76c912e9bf9253da045d108f3a537cb4 Author: Andrey Ryabinin Date: Thu Jul 20 12:52:42 2017 +0400 mm/tcach

Re: [Devel] [PATCH] kernel: call task_work_run() before exit_task_namespaces()

2017-07-20 Thread Konstantin Khorenko
Andrey, i've applied the patch in order not to block, but please extend the comment a little bit more: - ms commits (which are big to backport) - and what is the difference between those ms commits and our small commit. i'll edit the commit message during next rebase. Thank you. -- Best regards

[Devel] [PATCH RHEL7 COMMIT] kernel: call task_work_run() before exit_task_namespaces()

2017-07-20 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-514.26.1.vz7.33.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-514.26.1.vz7.33.11 --> commit 8ec11163ae371bdf08f3e7f5e2420b82ea71c32d Author: Andrei Vagin Date: Thu Jul 20 12:41:59 2017 +0400 kernel: cal

Re: [Devel] [PATCH] kvm: x86: hyperv: avoid livelock in oneshot SynIC timers

2017-07-20 Thread Konstantin Khorenko
Need it in update 5? Any bug id? -- Best regards, Konstantin Khorenko, Virtuozzo Linux Kernel Team On 07/19/2017 08:31 PM, Roman Kagan wrote: If the SynIC timer message delivery fails due to SINT message slot being busy, there's no point to attempt starting the timer again until we're notified

Re: [Devel] [PATCH RHEL7 COMMIT] fuse: do not call end_page_writeback earlier than fuse_flush_writepages

2017-07-20 Thread Konstantin Khorenko
Please consider to prepare a ReadyKernel patch for it. But after it passes a serious testing in the scope of vz7 Update 5. (may be even after update 5 release) https://readykernel.com/ -- Best regards, Konstantin Khorenko, Virtuozzo Linux Kernel Team On 07/20/2017 11:32 AM, Konstantin Khorenk

[Devel] [PATCH RHEL7 COMMIT] fuse: do not call end_page_writeback earlier than fuse_flush_writepages

2017-07-20 Thread Konstantin Khorenko
The commit is pushed to "branch-rh7-3.10.0-514.26.1.vz7.33.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-514.26.1.vz7.33.11 --> commit 8f06daee801e10df7d9e441379bc9c0a0a5b5661 Author: Maxim Patlasov Date: Thu Jul 20 12:32:48 2017 +0400 fuse: do

[Devel] [PATCH vz7] fuse: do not call end_page_writeback earlier than fuse_flush_writepages

2017-07-20 Thread Maxim Patlasov
Some operations (flush, fsync, fallocate, fiemap) want to flush dirty pages to fused before doing something. So they perform to steps: 1) filemap_write_and_wait 2) fuse_sync_writes The first of them must ensure that all dirty pages went to fuse-writeback and the second supposedly flushes pages un