On 5/7/20 9:38 PM, Konstantin Khorenko wrote:
On 05/06/2020 02:59 PM, Pavel Tikhomirov wrote:
diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
index a4ca54304220..45e2752a85c9 100644
--- a/include/uapi/linux/fs.h
+++ b/include/uapi/linux/fs.h
@@ -99,6 +99,12 @@ struct inodes_stat
We have problem cases in nf_nat_ipv{4,6}_fn() only for now,
so let's don't call do_chain() in case it is nft_nat_do_chain()
if we process a chain with inappropriate netns.
We cannot just check "do_chain" argument in nf_nat_ipv{4,6}_fn() because
in that case we have to export nft_nat_do_chain() fun
Brief info:
===
* at the moment NAT chains are linked into a single list - even if they are for
different netns
* only first NAT chain can be processed on every conntrack setup.
Even if the chain handling function returns error, the conntrack is
considered as "configured" (from NAT's
Brief info:
===
* at the moment NAT chains are linked into a single list - even if they are for
different netns
* only first NAT chain can be processed on every conntrack setup.
Even if the chain handling function returns error, the conntrack is
considered as "configured" (from NAT's
We have problem cases in nf_nat_ipv{4,6}_fn() only for now,
so let's don't call do_chain() in case it is nft_nat_do_chain()
if we process a chain with inappropriate netns.
We cannot just check "do_chain" argument in nf_nat_ipv{4,6}_fn() because
in that case we have to export nft_nat_do_chain() fun
The commit is pushed to "branch-rh7-3.10.0-1127.vz7.160.x-ovz" and will appear
at https://src.openvz.org/scm/ovz/vzkernel.git
after rh7-3.10.0-1127.vz7.150.9
-->
commit a85d8068227eaf473e58a2161e7dd449239b7c24
Author: Andrei Vagin
Date: Fri May 8 17:05:34 2020 +0300
mnt: allow to add a
The commit is pushed to "branch-rh8-4.18.0-80.1.2.vz8.3.x-ovz" and will appear
at https://src.openvz.org/scm/ovz/vzkernel.git
after rh8-4.18.0-80.1.2.vz8.3.9
-->
commit 6c0cfbbab61642b5ce6d4cefad810cdd84f9234d
Author: Andrei Vagin
Date: Fri May 8 17:43:40 2020 +0300
mnt: allow to add a
The following race could happen
Initial state: kmem = 0, cgroup is online
CPU0:
charge_kmem(1); //kmem = 1
memcg_uncharge_kmem()
{
...
kmem = page_counter_uncharge(&memcg->kmem, nr_pages); //kmem = 0
-
On 08.05.2020 19:10, Andrey Ryabinin wrote:
> The following race could happen
>
> Initial state: kmem = 0, cgroup is online
>
> CPU0:
>charge_kmem(1); //kmem = 1
>memcg_uncharge_kmem()
>{
>...
> kmem = page_counter_uncharge(&memcg->kmem, nr_pages); //kmem =
