From: Stanislav Kinsburskiy
There might be a situation, when task is expected to be killed, but it's
execution already passed this check in __rpc_execute() and the task is not yet
sleeping. In this case it will fall into sleep. I.e. it's a race.
This patch add check for per-net "kill_tasks" is en
The remaining part of part9 will be sent later, after dependency
've/printk: per-CT printk helpers introduced' gets in.
Stanislav Kinsburskiy (14):
ve/kthreadd: emulate kthreadd by kworker thread, attached to CT
ve/kthreadd: add per-ve kthreads creation interface
ve/kthreadd: create kthreadd
From: Stanislav Kinsburskiy
This will allow to use a different one for containers.
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 commit 041bcd5475d5 ("ve/umh: make queue
function one of the subprocess_info parameters"))
Signed-off-by: Nikita Yushchenko
---
include/linux/umh.h
From: Stanislav Kinsburskiy
This patch aborts SUNRPC tasks upon raise of per-net "kill_tasks" toggle.
This is needed to abort pernding tasks, which are waiting for timeout.
There could be a situation, when RPC task is not yet sleeping, but already
passed the check for "kill_tasks" in __rpc_execu
From: Stanislav Kinsburskiy
Patch introduces new function kthread_create_on_node_ve_flags and a few helpers:
1) kthread_create_on_node_ve and
1) kthread_run_ve and
2) kthread_create_ve
Note: we need to able to provide fork flags to kthread creation to be able to
fork one kthread from another as
From: Stanislav Kinsburskiy
We had this logic in vz6. But it was applied to any stop there.
The intention was to solve 2 tasks:
1) allow to stop container when network is unaccessible
2) allow to suspend container with blocked network traffic.
In vz7 we need the traffic to flow on suspend. Thus
From: Stanislav Kinsburskiy
So, rationale...
UHM request is just a work, attached to some queue, which is handled by
per-cpu kernel threads (this is a simple explanation, but true in general).
We want to execute this work in container environment because:
1) we need to set container root to exec
From: Stanislav Kinsburskiy
Does:
1) virtualize rpc_pipefs file system
2) virtualize nfs file system
3) allows to mount from CTs initial user ns
4) add VE_FEATURE_NFS check during nfsd mount
Signed-off-by: Stanislav Kinsbursky
+++
ve/fs/nfs: NFS containerization - allow to mount NFS in CT root
From: Stanislav Kinsburskiy
Signed-off-by: Stanislav Kinsburskiy
+++
ve, kthread: Remove some useless crud.
Simplify the code by removing useless macroses.
Signed-off-by: Andrey Ryabinin
(cherry-picked from vz8 commit 890476830003 ("ve/umh: add per-container
usermodehelper thread"))
Signed
From: Stanislav Kinsburskiy
Boolean per-net SUNPRC proc entry. Will be used to abort SUNRPC traffic when
necessary.
https://jira.sw.ru/browse/PSBM-66510
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 commit 22ca0165a75b ("sunrpc: add "kill-tasks"
proc entry"))
Follow file_opera
From: Stanislav Kinsburskiy
By default UHM requests are handled by a workqueue, i.e. they need "struct
work" parameter.
It will be different in containers, so move this work init and usage out of
generic subprocess handling code.
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 com
From: Stanislav Kinsburskiy
This is the first version of kthreadd in a container (there will some
improvements on top).
With this patch kthreadd won't be visible in a container, but its children
will be.
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 commit 1cc3a639c3fb ("ve/kthr
From: Stanislav Kinsburskiy
The idea is simple: create another kworker in the one, attached to CT.
The latter has containers pid ns for children, so the former will appear in
containers pid ns.
The latter then can be destroyed, thus leaving us "kthreadd" kernel thread not
just attached, but also
From: Stanislav Kinsburskiy
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 commit 27d2bf69ee2a ("ve/umh: enable user
mode helper in container"))
Signed-off-by: Nikita Yushchenko
---
kernel/umh.c | 3 ---
1 file changed, 3 deletions(-)
diff --git a/kernel/umh.c b/kernel/umh.c
i
From: Stanislav Kinsburskiy
The main and only difference with generic usermode helper is that
subprocess_info is queued via additional kthread_work member to per-container
usermode helper worker.
Signed-off-by: Stanislav Kinsburskiy
(cherry-picked from vz8 commit 871dd2e11512 ("ve/umh: introdu
Can we also fixup/merge 2e319b6173a1 ("ve/sunrpc: Enable rpc_pipefs
mounts in Container root user namespace") commit from vz8 to this patch?
On 27.09.2021 09:30, Nikita Yushchenko wrote:
From: Stanislav Kinsburskiy
Does:
1) virtualize rpc_pipefs file system
2) virtualize nfs file system
3) al
And also it looks like you miss in your gitconfig
[sendemail]
suppresscc = all
else you send patches to everybody mentioned in patch, which probably
not what you want.
___
Devel mailing list
Devel@openvz.org
https://lists.openvz.org/mailman/
27.09.2021 10:23, Pavel Tikhomirov wrote:
Can we also fixup/merge 2e319b6173a1 ("ve/sunrpc: Enable rpc_pipefs mounts in Container root user
namespace") commit from vz8 to this patch?
I guess that "ve/sunrpc: Enable rpc_pipefs mounts in Container root user namespace' depends on it's
predecessor
On 27.09.2021 11:05, Nikita Yushchenko wrote:
27.09.2021 10:23, Pavel Tikhomirov wrote:
Can we also fixup/merge 2e319b6173a1 ("ve/sunrpc: Enable rpc_pipefs
mounts in Container root user namespace") commit from vz8 to this patch?
I guess that "ve/sunrpc: Enable rpc_pipefs mounts in Container
On 23.09.2021 15:49, Konstantin Khorenko wrote:
> ms commit b7d6c3033323 ("block: fix use-after-free on cached
> last_lookup partition") has changed the semantic of
> disk_map_sector_rcu() function: previously it only returned appropriate
> partition but after the patch it additionally gets a refco
It's overkill.
Signed-off-by: Kirill Tkhai
---
drivers/md/dm-push-backup.c |3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/md/dm-push-backup.c b/drivers/md/dm-push-backup.c
index 16e8cf27211f..75f080fe34cf 100644
--- a/drivers/md/dm-push-backup.c
+++ b/drivers/md
21 matches
Mail list logo
