2006-08-22 23:49:33

by Bjo Breiskoll

[permalink] [raw]
Subject: Relay Subsystem the 2nd.

Hello,
back to my question last week. I would like to implement a new kernel-modul
with the ability to write data from kernelspace to userspace via the new
relay subsystem. The less info from block/blktrace.c is insufficient
unfortunately. The userspace-program should be able to "listen" to the
kernel-modul if there is new data written. My first Implementation with data
exchange over copy_to_user() and a char-device is insufficient. So i need an
hint for something like blocked-IO for relay-files. Is this possible with
the new relay-subsystem?

Thanx in advance
BJO


2006-08-23 00:57:57

by Paul Mundt

[permalink] [raw]
Subject: Re: Relay Subsystem the 2nd.

On Wed, Aug 23, 2006 at 01:49:29AM +0200, Bjo Breiskoll wrote:
> back to my question last week. I would like to implement a new kernel-modul
> with the ability to write data from kernelspace to userspace via the new
> relay subsystem. The less info from block/blktrace.c is insufficient
> unfortunately. The userspace-program should be able to "listen" to the
> kernel-modul if there is new data written. My first Implementation with data
> exchange over copy_to_user() and a char-device is insufficient. So i need an
> hint for something like blocked-IO for relay-files. Is this possible with
> the new relay-subsystem?
>
Is there something insufficient with poll() on relay files?

2006-08-26 08:28:53

by Bjo Breiskoll

[permalink] [raw]
Subject: AW: Relay Subsystem the 2nd.

> Is there something insufficient with poll() on relay files?

Sometimes the solution is easy like that.
Thanks .. It works :)

BJO