Hello,
I am trying to upgrade my labtop to 2.6.8.1. I have ArcNet COM20020
PCMCIA card. After editing /etc/pcmcia/config to make it know about the
module, it finds the com20020 with no problems but as soon as I try to
start the network device the ifconfig process crashes.
I have looked a bit at the code and compared it to 2.4.20 which I used
before (with my own patches which I have sent here to - among other things
- to make com20020_cs work).
Somebody (how do I find out whom???) have made an open function pointer
instead of the old open_close function pointer. The generic ArcNet
framework calls this function pointer into the more specialized drivers
when you start your device. But none of the specialized drivers do never
initialize the function pointer! I.e. when you start your arcnet device
you get a call to 0x0!
I fixed it by changing
lp->hw.open(dev);
to
if(lp->hw.open) {
lp->hw.open(dev);
}
Esben
Hello.
In article <[email protected]> (at Thu, 23 Sep 2004 17:08:47 +0200 (METDST)), Esben Nielsen <[email protected]> says:
> I am trying to upgrade my labtop to 2.6.8.1. I have ArcNet COM20020
> PCMCIA card. After editing /etc/pcmcia/config to make it know about the
> module, it finds the com20020 with no problems but as soon as I try to
> start the network device the ifconfig process crashes.
:
> I fixed it by changing
> lp->hw.open(dev);
> to
> if(lp->hw.open) {
> lp->hw.open(dev);
> }
Thanks.
Signed-off-by: Hideaki YOSHIFUJI <[email protected]>
===== drivers/net/arcnet/arcnet.c 1.18 vs edited =====
--- 1.18/drivers/net/arcnet/arcnet.c 2004-07-01 13:18:14 +09:00
+++ edited/drivers/net/arcnet/arcnet.c 2004-09-24 00:11:35 +09:00
@@ -401,7 +401,8 @@
lp->rfc1201.sequence = 1;
/* bring up the hardware driver */
- lp->hw.open(dev);
+ if (lp->hw.open)
+ lp->hw.open(dev);
if (dev->dev_addr[0] == 0)
BUGMSG(D_NORMAL, "WARNING! Station address 00 is reserved "
--
Hideaki YOSHIFUJI @ USAGI Project <[email protected]>
GPG FP: 9022 65EB 1ECF 3AD1 0BDF 80D8 4807 F894 E062 0EEA
After I got the arcnet device running labtop computer froze up. I have
turned off preemtion and SMP. It seems to make it more stable but I can't
be conclusive.
Esben
On Fri, 24 Sep 2004, YOSHIFUJI Hideaki / [iso-2022-jp] 吉藤英明 wrote:
> Hello.
>
> In article <[email protected]> (at Thu, 23 Sep 2004 17:08:47 +0200 (METDST)), Esben Nielsen <[email protected]> says:
>
> > I am trying to upgrade my labtop to 2.6.8.1. I have ArcNet COM20020
> > PCMCIA card. After editing /etc/pcmcia/config to make it know about the
> > module, it finds the com20020 with no problems but as soon as I try to
> > start the network device the ifconfig process crashes.
> :
> > I fixed it by changing
> > lp->hw.open(dev);
> > to
> > if(lp->hw.open) {
> > lp->hw.open(dev);
> > }
>
> Thanks.
>
> Signed-off-by: Hideaki YOSHIFUJI <[email protected]>
>
> ===== drivers/net/arcnet/arcnet.c 1.18 vs edited =====
> --- 1.18/drivers/net/arcnet/arcnet.c 2004-07-01 13:18:14 +09:00
> +++ edited/drivers/net/arcnet/arcnet.c 2004-09-24 00:11:35 +09:00
> @@ -401,7 +401,8 @@
> lp->rfc1201.sequence = 1;
>
> /* bring up the hardware driver */
> - lp->hw.open(dev);
> + if (lp->hw.open)
> + lp->hw.open(dev);
>
> if (dev->dev_addr[0] == 0)
> BUGMSG(D_NORMAL, "WARNING! Station address 00 is reserved "
>
> --
> Hideaki YOSHIFUJI @ USAGI Project <[email protected]>
> GPG FP: 9022 65EB 1ECF 3AD1 0BDF 80D8 4807 F894 E062 0EEA
>
ArcNet can't compete with ethernet as LAN: It can only run 10 Mbit whereas
100 Mbit ethernet can run roughly 50 Mbit in practice.
But ArcNet doesn't degrade in performance when you try to fill it up as
ethernet does. Thus ArcNet is very good for real-time applications - and
is used for such in the industry. But that is not an area where people
usually use Linux.
But don't worry, the windoze drivers aren't good either. The official
driver for the PCMCIA card crashed Windows XP - with no chance of fixing
them as I have with the Linux one :-)
Esben
On Thu, 23 Sep 2004, David S. Miller wrote:
> On Thu, 23 Sep 2004 22:59:58 +0200 (METDST)
> Esben Nielsen <[email protected]> wrote:
>
> > After I got the arcnet device running labtop computer froze up. I have
> > turned off preemtion and SMP. It seems to make it more stable but I can't
> > be conclusive.
>
> Based upon the fact that most Arcnet drivers set hw.open() to NULL,
> and you're the first person to report this, I doubt arcnet is
> getting any serious use or testing at all these days. Sorry :-/
> -
> To unsubscribe from this list: send the line "unsubscribe linux-net" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
On Thu, 23 Sep 2004 22:59:58 +0200 (METDST)
Esben Nielsen <[email protected]> wrote:
> After I got the arcnet device running labtop computer froze up. I have
> turned off preemtion and SMP. It seems to make it more stable but I can't
> be conclusive.
Based upon the fact that most Arcnet drivers set hw.open() to NULL,
and you're the first person to report this, I doubt arcnet is
getting any serious use or testing at all these days. Sorry :-/
I have not argued the technical pros and cons of arcnet.
I have merely stated that since you are the first person
reporting the OOPS, it is indicative that not many people
are testing arcnet in the 2.6.x kernel.
And I only want to explain why you don't see ArcNet used much with Linux
:-)
Oh well, I'll try to see if I can get time to make it stable with
preemption on. Is preemption safe enough to ensure SMP safe or would I
have to test on at least a hyperthreading machine as well?
Esben
On Thu, 23 Sep 2004, David S. Miller wrote:
>
> I have not argued the technical pros and cons of arcnet.
>
> I have merely stated that since you are the first person
> reporting the OOPS, it is indicative that not many people
> are testing arcnet in the 2.6.x kernel.
>
On Iau, 2004-09-23 at 22:46, Esben Nielsen wrote:
> ArcNet can't compete with ethernet as LAN: It can only run 10 Mbit whereas
> 100 Mbit ethernet can run roughly 50 Mbit in practice.
We can saturate 100Mbit ethernet full duplex at a steady 100Mbit, GBit
on decent boxes/cards.
> But ArcNet doesn't degrade in performance when you try to fill it up as
> ethernet does. Thus ArcNet is very good for real-time applications - and
> is used for such in the industry. But that is not an area where people
> usually use Linux.
Nod. Its fair to say you may well be "the arcnet user" by now however
8). Most real time people I've met use fieldbus nowdays for control
systems or ethernet and handshaking plus local timers
On Thu, 23 Sep 2004, Alan Cox wrote:
> [..]
> > But ArcNet doesn't degrade in performance when you try to fill it up as
> > ethernet does. Thus ArcNet is very good for real-time applications - and
> > is used for such in the industry. But that is not an area where people
> > usually use Linux.
>
> Nod. Its fair to say you may well be "the arcnet user" by now however
> 8). Most real time people I've met use fieldbus nowdays for control
> systems or ethernet and handshaking plus local timers
>
>
I don't think you are quite fair :-) The company I work for recently
merged with a competitor. The part I came from have been using ArcNet for
10 years as a backbone in a decentralized control system. We had been
talking about going to ethernet but things like backward combability
keeped us on ArcNet. The other part of the company, however, choose ArcNet
an I/O-network when they started to redisign their control system from
scratch just 3 years ago!
Today I asked one of them why they did that: Open standard, low
gaurantied latencies (down to 5ms depending on the number of nodes on the
network), relatively low drain on the CPU, high bandwidth (10Mbit/s),
possibility of mixing electrical and fiber lines, long wires (the 150m we
need is no problem) and the availability of relative cheap industry grade
controllers running in extended temperature range. They found that all the
other kind of "fieldbuses" around and ethernet didn't have all these
qualities.
I think some industrial, medico and military companies still design new
products with ArcNet today, but these kind of companies are not as open
about their choice of technology as telecom is so you simply don't hear
about it!
Esben
On Thu, 23 Sep 2004, Esben Nielsen wrote:
> ArcNet can't compete with ethernet as LAN: It can only run 10 Mbit whereas
> 100 Mbit ethernet can run roughly 50 Mbit in practice.
most 100Mb/s ethernet card/switch combo's will happily push 90+ Mb/s
bi-directionally between two hosts.
> But ArcNet doesn't degrade in performance when you try to fill it up as
> ethernet does. Thus ArcNet is very good for real-time applications - and
> is used for such in the industry. But that is not an area where people
> usually use Linux.
well maybe your "realtime" application doesn't run on linux.
if you want non-degrading ethernet performance buy a switch, if you want
bandwidth reservation, buy an l3 switch.
There just isn't anyone using arcnet or arcnet+ anymore, the writing was
on the wall for thomas conrad more than a decade ago, and my datapoint
mainframe days are far behind me, and I don't miss them.
> But don't worry, the windoze drivers aren't good either. The official
> driver for the PCMCIA card crashed Windows XP - with no chance of fixing
> them as I have with the Linux one :-)
>
> Esben
>
>
>
>
> On Thu, 23 Sep 2004, David S. Miller wrote:
>
>> On Thu, 23 Sep 2004 22:59:58 +0200 (METDST)
>> Esben Nielsen <[email protected]> wrote:
>>
>>> After I got the arcnet device running labtop computer froze up. I have
>>> turned off preemtion and SMP. It seems to make it more stable but I can't
>>> be conclusive.
>>
>> Based upon the fact that most Arcnet drivers set hw.open() to NULL,
>> and you're the first person to report this, I doubt arcnet is
>> getting any serious use or testing at all these days. Sorry :-/
>> -
>> To unsubscribe from this list: send the line "unsubscribe linux-net" in
>> the body of a message to [email protected]
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
--
--------------------------------------------------------------------------
Joel Jaeggli Unix Consulting [email protected]
GPG Key Fingerprint: 5C6E 0104 BAF0 40B0 5BD3 C38B F000 35AB B67F 56B2
Hello!
> And I only want to explain why you don't see ArcNet used much with Linux :-)
I still have some ArcNet hardware at home from the good old times when
I was one of the people maintaining the ArcNet driver, however, I have
only one ISA machine now and all my cards are ISA only.
If anybody would be willing to send me a PCI or PCMCIA ArcNet, I probably
would find time to make the driver work again.
Have a nice fortnight
--
Martin `MJ' Mares <[email protected]> http://atrey.karlin.mff.cuni.cz/~mj/
Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth
This line is umop apisdn.
On Thu, 2004-09-23 at 18:52, Esben Nielsen wrote:
> Oh well, I'll try to see if I can get time to make it stable with
> preemption on. Is preemption safe enough to ensure SMP safe or would I
> have to test on at least a hyperthreading machine as well?
>
In theory preempt safe should be SMP safe, but in practice it doesn't
always work this way. You need to test on an actual SMP system.
I am not sure that HT == SMP for your purposes either. ISTR some cases
with the VP patches where it broke on HT but not "real" SMP.
Lee