* [PATCH v2 net-next] net: initialize init_net earlier
@ 2022-02-05 17:01 Eric Dumazet
2022-02-06 11:10 ` patchwork-bot+netdevbpf
0 siblings, 1 reply; 2+ messages in thread
From: Eric Dumazet @ 2022-02-05 17:01 UTC (permalink / raw)
To: David S . Miller, Jakub Kicinski; +Cc: netdev, Eric Dumazet, Eric Dumazet
From: Eric Dumazet <edumazet@google.com>
While testing a patch that will follow later
("net: add netns refcount tracker to struct nsproxy")
I found that devtmpfs_init() was called before init_net
was initialized.
This is a bug, because devtmpfs_setup() calls
ksys_unshare(CLONE_NEWNS);
This has the effect of increasing init_net refcount,
which will be later overwritten to 1, as part of setup_net(&init_net)
We had too many prior patches [1] trying to work around the root cause.
Really, make sure init_net is in BSS section, and that net_ns_init()
is called earlier at boot time.
Note that another patch ("vfs: add netns refcount tracker
to struct fs_context") also will need net_ns_init() being called
before vfs_caches_init()
As a bonus, this patch saves around 4KB in .data section.
[1]
f8c46cb39079 ("netns: do not call pernet ops for not yet set up init_net namespace")
b5082df8019a ("net: Initialise init_net.count to 1")
734b65417b24 ("net: Statically initialize init_net.dev_base_head")
v2: fixed a build error reported by kernel build bots (CONFIG_NET=n)
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/net_namespace.h | 6 ++++++
init/main.c | 2 ++
net/core/dev.c | 3 +--
net/core/net_namespace.c | 17 +++++------------
4 files changed, 14 insertions(+), 14 deletions(-)
diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
index 5b61c462e534be468c81d2b0f4ef586b209dd4b8..374cc7b260fcdf15a8fc2c709d0b0fc6d99e8c5c 100644
--- a/include/net/net_namespace.h
+++ b/include/net/net_namespace.h
@@ -513,4 +513,10 @@ static inline void fnhe_genid_bump(struct net *net)
atomic_inc(&net->fnhe_genid);
}
+#ifdef CONFIG_NET
+void net_ns_init(void);
+#else
+static inline void net_ns_init(void) {}
+#endif
+
#endif /* __NET_NET_NAMESPACE_H */
diff --git a/init/main.c b/init/main.c
index 65fa2e41a9c0904131525d504f3ec86add44f141..ada50f5a15e4397e45b0e5c06bab051b1ce193d9 100644
--- a/init/main.c
+++ b/init/main.c
@@ -99,6 +99,7 @@
#include <linux/kcsan.h>
#include <linux/init_syscalls.h>
#include <linux/stackdepot.h>
+#include <net/net_namespace.h>
#include <asm/io.h>
#include <asm/bugs.h>
@@ -1116,6 +1117,7 @@ asmlinkage __visible void __init __no_sanitize_address start_kernel(void)
key_init();
security_init();
dbg_late_init();
+ net_ns_init();
vfs_caches_init();
pagecache_init();
signals_init();
diff --git a/net/core/dev.c b/net/core/dev.c
index 1eaa0b88e3ba5d800484656f2c3420af57050294..f662c6a7d7b49b836a05efc74aeffc7fc9e4e147 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -10732,8 +10732,7 @@ static int __net_init netdev_init(struct net *net)
BUILD_BUG_ON(GRO_HASH_BUCKETS >
8 * sizeof_field(struct napi_struct, gro_bitmask));
- if (net != &init_net)
- INIT_LIST_HEAD(&net->dev_base_head);
+ INIT_LIST_HEAD(&net->dev_base_head);
net->dev_name_head = netdev_create_hash();
if (net->dev_name_head == NULL)
diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
index 8711350085d6b55a4f3da19bc69e521f9efb7861..0ec2f5906a27c7f930e832835682d69a32e3c8e1 100644
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -44,13 +44,7 @@ EXPORT_SYMBOL_GPL(net_rwsem);
static struct key_tag init_net_key_domain = { .usage = REFCOUNT_INIT(1) };
#endif
-struct net init_net = {
- .ns.count = REFCOUNT_INIT(1),
- .dev_base_head = LIST_HEAD_INIT(init_net.dev_base_head),
-#ifdef CONFIG_KEYS
- .key_domain = &init_net_key_domain,
-#endif
-};
+struct net init_net;
EXPORT_SYMBOL(init_net);
static bool init_net_initialized;
@@ -1087,7 +1081,7 @@ static void rtnl_net_notifyid(struct net *net, int cmd, int id, u32 portid,
rtnl_set_sk_err(net, RTNLGRP_NSID, err);
}
-static int __init net_ns_init(void)
+void __init net_ns_init(void)
{
struct net_generic *ng;
@@ -1108,6 +1102,9 @@ static int __init net_ns_init(void)
rcu_assign_pointer(init_net.gen, ng);
+#ifdef CONFIG_KEYS
+ init_net.key_domain = &init_net_key_domain;
+#endif
down_write(&pernet_ops_rwsem);
if (setup_net(&init_net, &init_user_ns))
panic("Could not setup the initial network namespace");
@@ -1122,12 +1119,8 @@ static int __init net_ns_init(void)
RTNL_FLAG_DOIT_UNLOCKED);
rtnl_register(PF_UNSPEC, RTM_GETNSID, rtnl_net_getid, rtnl_net_dumpid,
RTNL_FLAG_DOIT_UNLOCKED);
-
- return 0;
}
-pure_initcall(net_ns_init);
-
static void free_exit_list(struct pernet_operations *ops, struct list_head *net_exit_list)
{
ops_pre_exit_list(ops, net_exit_list);
--
2.35.0.263.gb82422642f-goog
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH v2 net-next] net: initialize init_net earlier
2022-02-05 17:01 [PATCH v2 net-next] net: initialize init_net earlier Eric Dumazet
@ 2022-02-06 11:10 ` patchwork-bot+netdevbpf
0 siblings, 0 replies; 2+ messages in thread
From: patchwork-bot+netdevbpf @ 2022-02-06 11:10 UTC (permalink / raw)
To: Eric Dumazet; +Cc: davem, kuba, netdev, edumazet
Hello:
This patch was applied to netdev/net-next.git (master)
by David S. Miller <davem@davemloft.net>:
On Sat, 5 Feb 2022 09:01:25 -0800 you wrote:
> From: Eric Dumazet <edumazet@google.com>
>
> While testing a patch that will follow later
> ("net: add netns refcount tracker to struct nsproxy")
> I found that devtmpfs_init() was called before init_net
> was initialized.
>
> [...]
Here is the summary with links:
- [v2,net-next] net: initialize init_net earlier
https://git.kernel.org/netdev/net-next/c/9c1be1935fb6
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-02-06 11:10 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-05 17:01 [PATCH v2 net-next] net: initialize init_net earlier Eric Dumazet
2022-02-06 11:10 ` patchwork-bot+netdevbpf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).