Skip to contentSkip to conversation

net: genetlink: Move initialization to core_initcall

PATCH In review Daniel Lezcano 4 participants 6 messages Jul 15, 2020 – Jul 20, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 net: genetlink: Move initialization to core_initcall

The generic netlink is initialized far after the netlink protocol itself at subsys_initcall. The devlink is initialized at the same level, but after, as shown by a disassembly of the vmlinux:

[ ... ]
374 ffff8000115f22c0 <__initcall_devlink_init4>:
375 ffff8000115f22c4 <__initcall_genl_init4>:
[ ... ]

The function devlink_init() calls genl_register_family() before the generic netlink subsystem is initialized.

As the generic netlink initcall level is set since 2005, it seems that was not a problem, but now we have the thermal framework initialized at the core_initcall level which creates the generic netlink family and sends a notification which leads to a subtle memory corruption only detectable when the CONFIG_INIT_ON_ALLOC_DEFAULT_ON option is set with the earlycon at init time.

The thermal framework needs to be initialized early in order to begin the mitigation as soon as possible. Moving it to postcore_initcall is acceptable.

This patch changes the initialization level for the generic netlink family to the core_initcall and comes after the netlink protocol initialization.

---
 net/netlink/genetlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netlink/genetlink.c +1 -1
index 55ee680e9db1..36b8a1909826 100644
@@ -1263,7 +1263,7 @@ static int __init genl_init(void)
 	panic("GENL: Cannot register controller: %d\n", err);
 }
 
-subsys_initcall(genl_init);
+core_initcall(genl_init);
 
 static int genlmsg_mcast(struct sk_buff *skb, u32 portid, unsigned long group,
 			 gfp_t flags)
Signature · 1 line
2.17.1
#1 lore raw

Hi Dave,

if you agree with this change, is it possible I merge it through the thermal tree in order to fix the issue ?

Thanks

-- Daniel

Quoted 46 lines · Daniel Lezcano

The generic netlink is initialized far after the netlink protocol itself at subsys_initcall. The devlink is initialized at the same level, but after, as shown by a disassembly of the vmlinux:

[ ... ]
374 ffff8000115f22c0 <__initcall_devlink_init4>:
375 ffff8000115f22c4 <__initcall_genl_init4>:
[ ... ]

The function devlink_init() calls genl_register_family() before the generic netlink subsystem is initialized.

As the generic netlink initcall level is set since 2005, it seems that was not a problem, but now we have the thermal framework initialized at the core_initcall level which creates the generic netlink family and sends a notification which leads to a subtle memory corruption only detectable when the CONFIG_INIT_ON_ALLOC_DEFAULT_ON option is set with the earlycon at init time.

The thermal framework needs to be initialized early in order to begin the mitigation as soon as possible. Moving it to postcore_initcall is acceptable.

This patch changes the initialization level for the generic netlink family to the core_initcall and comes after the netlink protocol initialization.

---
 net/netlink/genetlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netlink/genetlink.c +1 -1
index 55ee680e9db1..36b8a1909826 100644
@@ -1263,7 +1263,7 @@ static int __init genl_init(void)
 	panic("GENL: Cannot register controller: %d\n", err);
 }
 
-subsys_initcall(genl_init);
+core_initcall(genl_init);
 
 static int genlmsg_mcast(struct sk_buff *skb, u32 portid, unsigned long group,
 			 gfp_t flags)
Signature · 5 lines
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog
#3 lore raw
Quoted 9 lines · David Miller

From: Daniel Lezcano <[email protected]>
Date: Wed, 15 Jul 2020 09:43:00 +0200

Quoted 2 lines

if you agree with this change, is it possible I merge it through the thermal tree in order to fix the issue ?

No problem:

Thanks!

Signature · 5 lines
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog
Tested-by on v1 1/1
#4 lore raw
In reply to Daniel Lezcano

Hi

Quoted 23 lines · Daniel Lezcano

The generic netlink is initialized far after the netlink protocol itself at subsys_initcall. The devlink is initialized at the same level, but after, as shown by a disassembly of the vmlinux:

[ ... ]
374 ffff8000115f22c0 <__initcall_devlink_init4>:
375 ffff8000115f22c4 <__initcall_genl_init4>:
[ ... ]

The function devlink_init() calls genl_register_family() before the generic netlink subsystem is initialized.

As the generic netlink initcall level is set since 2005, it seems that was not a problem, but now we have the thermal framework initialized at the core_initcall level which creates the generic netlink family and sends a notification which leads to a subtle memory corruption only detectable when the CONFIG_INIT_ON_ALLOC_DEFAULT_ON option is set with the earlycon at init time.

The thermal framework needs to be initialized early in order to begin the mitigation as soon as possible. Moving it to postcore_initcall is acceptable.

Quoted 5 lines

This patch changes the initialization level for the generic netlink family to the core_initcall and comes after the netlink protocol initialization.

I confirm, that this change together with the thermal subsystem initcall change fixes the issue observed in linux-next for the last few days.

Quoted 9 lines
---
  net/netlink/genetlink.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
net/netlink/genetlink.c +0 -0
index 55ee680e9db1..36b8a1909826 100644

net/netlink/genetlink.c static int __init genl_init(void) · line 1269

Quoted 8 lines
net/netlink/genetlink.c +1 -1
@@ -1263,7 +1263,7 @@ static int __init genl_init(void)
  	panic("GENL: Cannot register controller: %d\n", err);
  }
 
-subsys_initcall(genl_init);
+core_initcall(genl_init);
 
  static int genlmsg_mcast(struct sk_buff *skb, u32 portid, unsigned long group,
  			 gfp_t flags)

Best regards

Signature · 2 lines
Marek Szyprowski, PhD
Samsung R&D Institute Poland
Reviewed-by on v1 1/1
#5 lore raw
In reply to Daniel Lezcano
Quoted 24 lines · Daniel Lezcano

The generic netlink is initialized far after the netlink protocol itself at subsys_initcall. The devlink is initialized at the same level, but after, as shown by a disassembly of the vmlinux:

[ ... ]
374 ffff8000115f22c0 <__initcall_devlink_init4>:
375 ffff8000115f22c4 <__initcall_genl_init4>:
[ ... ]

The function devlink_init() calls genl_register_family() before the generic netlink subsystem is initialized.

As the generic netlink initcall level is set since 2005, it seems that was not a problem, but now we have the thermal framework initialized at the core_initcall level which creates the generic netlink family and sends a notification which leads to a subtle memory corruption only detectable when the CONFIG_INIT_ON_ALLOC_DEFAULT_ON option is set with the earlycon at init time.

The thermal framework needs to be initialized early in order to begin the mitigation as soon as possible. Moving it to postcore_initcall is acceptable.

Quoted 5 lines

This patch changes the initialization level for the generic netlink family to the core_initcall and comes after the netlink protocol initialization.

Quoted 20 lines
---
 net/netlink/genetlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netlink/genetlink.c +1 -2
index 55ee680e9db1..36b8a1909826 100644
@@ -1263,7 +1263,7 @@ static int __init genl_init(void)
        panic("GENL: Cannot register controller: %d\n", err);
 }
 
-subsys_initcall(genl_init);
+core_initcall(genl_init);
 
 static int genlmsg_mcast(struct sk_buff *skb, u32 portid, unsigned long group,
                         gfp_t flags)
--

2.17.1