Thanks to all who have responded!
It appears (from several messages) that we are not supposed to be doing any
dev_alloc_skb()'s if they are eventually calling kmalloc(). The funny thing
is that the machine has been running constantly for the past 3 days with no
errors at all, and has been dev_alloc'ing skb's 800 times per second. I
will look into changing this in our code, however.
Stuart Hughes wrote:
>The scheme is that you add an additional network card to your PC, and
>have a dedicated real-time network. This approach does not change or
>use the networking code in the Linux kernel, instead the network code
>has been put into an RT module.
>
>The programming interface is the defacto standard BSD socket API.
>Support for ICMP, ARP and UDP are included.
I really like what Zentropix is doing, and the efforts they have made
towards
RTL. My problem was that I needed it to work last weekend. :( Also our
application requires the use of raw ethernet. The remote device does not
have support for anything but raw ethernet encoded frames. Does your
socket interface support RAW encoding?
Also, could Zentropix propose a "standard" set of driver changes that
would allow the ethernet driver writers to incorporate these changes into
the base Linux distribution? The IOCTL with a SIOCDEVPRIVATE seemed
like a good idea to us, but we will open it up for discussion. Along with
hooking netif_rx and mark_bh, it appears that hooking whatever alloc_skb
call they use seems appropriate as well.
>As the network code is based on the Linux kernel code, it will be
>release with full source code.
Excellent!
>If it is possible, it would be good to merge the various effort to
>provide a single module for real time.
I agree. When I started on this project two weeks or so ago, I couldn't
find
anyone on the net except Zentropix talking about RT ethernet. There
shouldn't be many variants floating around.. What we came up with
has a specialized audience (raw ethernet communication).
What Zentropix is working on has much broader appeal, IMHO. However,
it has been a question here as to whether a real time application really
needs an IP encapsulation around their data, but that may be a topic for
a different discussion. Having the use of ICMP and ARP provides definite
advantages.
[EMAIL PROTECTED] Wrote:
>This is very nice. You might be interested in comparing notes with
>Axel Bernal ([EMAIL PROTECTED])
I'll do that. As I said above, I was not aware of anyone else working on
this.
>How hard would it be to make this driver work with the new POSIXIO
>interface we added?
Given the above discussion (with Zentropix) it is not clear to me whether
this should be pursued. Are there other opinions?
---
Michael M. Morrison
VP, Chief Technical Officer
Hyperion Technologies, Inc.
(970) 493-1900
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
