[Devel] [PATCH RH9] ve/devtmpfs: lightweight virtualization

From: Stanislav Kinsburskiy Due to changes in RH8.4 we need to rewrork it, actually the logic becomes much more simple, we mount/umount single tmpts per ve on cgroup creation/removal, all actual devtmpfs mount calls only increase a refcount on corresponding ve's mount like with hosts devtmps. Or

[Devel] [PATCH RH9 3/3] commit 22b5a8a84548

From: Vladimir Davydov ve/fs/inotify: do not impose limit on the number of instances by default In Vz7 we haven't switched to user ns yet. As a result, all containers use the same user_struct for the same user id. This leads to hitting fs.inotify.max_user_instances sysctl limit q

[Devel] [PATCH RH9 0/3] part23 part

https://jira.sw.ru/browse/PSBM-134015 --- Alexander Mikhalitsyn (1): shm: skip shm_destroy if task IPC namespace was changed Pavel Tikhomirov (1): cgroup/net_prio: virtualize ifpriomap per-ve Vladimir Davydov (1): commit 22b5a8a84548 fs/notify/inotify/inotify_user.c |4

[Devel] [PATCH RH9 2/3] cgroup/net_prio: virtualize ifpriomap per-ve

From: Pavel Tikhomirov Ifpriomap is a map of net_prio cgroup id to device prio. Each process is in some netprio cgroup and all sockets of this process have prio cgroup id of this cgroup. When packet from such socket goes through network stack we choose priority for packet on each device we go thr

[Devel] [PATCH RH9 1/3] shm: skip shm_destroy if task IPC namespace was changed

From: Alexander Mikhalitsyn Fixes: ab602f79915 ("shm: make exit_shm work proportional to task activity") (ms commit) https://jira.sw.ru/browse/PSBM-131142 Signed-off-by: Alexander Mikhalitsyn Signed-off-by: Kirill Tkhai --- ipc/shm.c | 10 +- 1 file changed, 9 insertions(+), 1 dele