Re: [PATCH] [1/2] kernel error reporting (revised)

[kuznet@xxxxxxxxxxxxx]

Hello!

> netlink_broadcast() does read_lock(&nl_table_lock).  But nl_table_lock is
> not an irq-safe lock.

Just as reminder, there are _no_ irq safe locks in net/*. A few of
local_irq_disable()s are segregated in interface to device drivers.


> Possibly netlink_broadcast() can be made callable from hardirq context, but
> it looks to be non trivial.

Trivial or non-trivial, before all this is highly not desired.
net/* is better to remain in the form free of knowledge of hardirqs.

Alexey