Re: [patch] kernel events layer

From: Robert Love
Date: Fri Jul 23 2004 - 21:18:57 EST

On Fri, 2004-07-23 at 20:03 -0700, Andrew Morton wrote:

> OK. Can you give us a ballpark estimate of how many send_kmessage() calls
> we're likely to have in two years time?

Predicting the future is hard, but I suspect this number to be small.
Let's say 10 in core kernel code?

If this takes off as a solution for error reporting, that number will be
much larger in drivers.

> - The GFP_ATOMIC page allocation is unfortunate. Please pass in the
> gfp_flags, or change it to GFP_KERNEL and provide a separate
> send_kmessage_atomic()?

I like the latter.

> - Methinks the kernel won't build if the user set CONFIG_NETLINK_DEV=n

I will test and fix.

> - When fixing that up, please add CONFIG_KERNEL_EVENTS or whatever,
> provide the appropriate do-nothing stubs if it's disabled. For the tiny
> systems.


> - send_kmessage() is racy against kmessage_exit(). I'm not sure that's
> worth fixing - if you agree then it would set minds at ease to simply
> remove kmessage_exit().

The race is definitely not worth fixing. If it bothers you, then
removing kmessage_exit() makes sense. I will do that.

> - This code will never work as a module, so why include the

Can be removed.

> - What led to the decision to export send_kmessage() to only GPL modules?

I am a fanatic about freedom? Seriously, I will talk to Arjan about
changing it. I do not care either way.

Updated patch forthcoming.


Robert Love

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at