Next-in-Thread Next Message

None Some context would help 

Forum: PCMCIA Wireless Network Adapter Issues
Re: Question Receive interrupt does not seem to be incrementing properly (Richard Smith)
Date: 2000, Aug 15
From: David Hinds <dhinds@pcmcia.sourceforge.org>

That isn't the exact message, and the rest of the output would be
useful, but my guess is what it means is that you are using a wireless
network card in a PCI host adapter in a desktop system and you haven't
specified the right startup options to configure the adapter to use
PCI interrupts.

But without more context, that's just a guess.

-- Dave

Next-in-Thread Next Message

Messages Inline: 0 1

None Re: Some context would help

Re: None Some context would help (David Hinds)
Date: 2000, Aug 15
From: Richard A. Smith smithbone

On Tue, 15 Aug 2000 09:59:48 -0700, David Hinds wrote:

>That isn't the exact message, and the rest of the output would be

Yeah.. I tried to wing it from memory. Should have mentioned that. Sorry.

>useful, but my guess is what it means is that you are using a wireless
>network card in a PCI host adapter in a desktop system and you haven't
>specified the right startup options to configure the adapter to use
>PCI interrupts.
>
>But without more context, that's just a guess.

My bad... I should know better. I was in the middle messing with various PNP bios 
settings and didn't prepare well enough.
 
I am using a RayLink wireless card with the adapter that came with it. But it's 
not PCI. That would probally work.  It's an ISA PNP adapter using the Ricoh 
RF5C296 bridge.

All appears to go without problems until I actually try to use the cards.  They 
both start networks when one should start and one should join.  (FWIW I can't get 
them to work under windows either)  Data on Cory's page and other web sources say 
that this is either a card failure or some sort of receive problem where the card 
dosen't get any data thus it starts a new network.

So far nothing in the debug level 5 output from ray_cs shows anything to be 
amiss.

When I found the network script and it mentioned RX seeming to have a problem I 
started to look at my IRQs'.  (I am in the middle of installing isapnptools now 
to see if I can find any more info)

Here's the output from test_network:  (more or less I am retyping)

----------

PCMCIA network settings are coming from /etc/pcmcia/network.opts

Checking network itnterface eth1 (ray_cs driver)

Kernel messages for eth1:
eth1:  RayLink, irq 5, hw_addr 00:00:8f:48:e3:03

current ifconfig setting
all pretty much std..
IP:   10.0.1.1
Bcast:  10.0.1.255
mask:  255.255.255.0
zeros on all the  RX, TX, etc stats:
Interrupt 5

The interface is configured and running
There hasn't been any traffic on this interface

There is a route to a gateway at 10.0.1.1

Pinging 10.0.1.1.... response received!
The device interrupt does not seem to be incrementing normally.

----------





--
Richard A. Smith                         Bitworks, Inc.               
rsmith@bitworks.com               501.846.5777                        
Sr. Design Engineer        http://www.bitworks.com   

None Not sure if it is an interrupt issue; gateway set wrong?

Re: None Re: Some context would help (Richard A. Smith)
Date: 2000, Aug 15
From: David Hinds <dhinds@pcmcia.sourceforge.org>

Does /proc/interrupts show a non-zero interrupt count for the card?
If the count is 0, then try using a different irq (by excluding things
in /etc/pcmcia/config.opts).

> There is a route to a gateway at 10.0.1.1

There's something wrong with your network settings and this is
confusing the script.  Your gateway address seems to be the same as
your own IP address, which doesn't make any sense.

> Pinging 10.0.1.1.... response received!
> The device interrupt does not seem to be incrementing normally.

In this context, I think this doesn't mean anything: it thinks it is
pinging your gateway, but it is pinging the box itself, which doesn't
cause any packets to go out through the card.  So it won't produce any
interrupts.

I'll add a test to the script to complain if the gateway address is
the same as the IP address.

-- Dave

None Re: Not sure if it is an interrupt issue; gateway set wrong?

Re: None Not sure if it is an interrupt issue; gateway set wrong? (David Hinds)
Date: 2000, Aug 15
From: Richard A. Smith smithbone

On Tue, 15 Aug 2000 11:53:06 -0700, David Hinds wrote:

>Does /proc/interrupts show a non-zero interrupt count for the card?
>If the count is 0, then try using a different irq (by excluding things
>in /etc/pcmcia/config.opts).

Yeah it shows 107 interrupts.

>> There is a route to a gateway at 10.0.1.1
>
>There's something wrong with your network settings and this is
>confusing the script.  Your gateway address seems to be the same as
>your own IP address, which doesn't make any sense.
>

Oh... DUH!.. ID10T error. Of course...Man I can't believe I missed that.. yesh.. 
rookie mistake.

I didn't scrutinize the settings hard enough.  I guess I should note that I was 
just taking the default settings. Everything but the IP is preset in the 
network.opts so I just made the IP match the already pre-configured settings.  For 
my app (a private wireless net) at first glance they appeared to be perfect.

>In this context, I think this doesn't mean anything: it thinks it is
>pinging your gateway, but it is pinging the box itself, which doesn't
>cause any packets to go out through the card.  So it won't produce any
>interrupts.

Yeah.. thats exactly whats happening...  Crap that means I still dont have a clue 
why the cards don't talk..

>I'll add a test to the script to complain if the gateway address is
>the same as the IP address.

Glad to help *grin*  Whats that quote? Something about... no amount of idiot 
proofing is good enough for a sufficiently powered idiot.






--
Richard A. Smith                         Bitworks, Inc.               
rsmith@bitworks.com               501.846.5777                        
Sr. Design Engineer        http://www.bitworks.com   

None Raylink problem solved..

Re: None Not sure if it is an interrupt issue; gateway set wrong? (David Hinds)
Date: 2000, Aug 15
From: Richard A. Smith smithbone

I'm a talking now...

Turned out to be a good old fashioned hardware failure.  I got 2 new cards in 
about an hour ago and it was truly plug and play.

Further investagation led to finding that one of the orginal pair I have just 
dosn't work.  Looks like it works... the drivers think it works.. But something 
in the chain is broke..

This is kinda scary for me because the card worked when we originally received 
them. I am useing the card in a custom OEM application of mine where I am talking 
to it via a TI 5402 DSP.  Thus I am in the process of writeing a custom driver 
for it.  I am worried that in that process I may have done something that gacked 
the card.  If so then it was a serious gack because I re-programmed the firmware 
on the card to make sure I hadn't done that.

Oh well back to hacking.. Perhaps I can keep from breaking my new cards.

Thanks Dave for your your help.

One quick question.. In the process of all this I installed like 3 different 
pcmcia-cs versions and a couple of ray_cs driver versions all multiple times.  

Whenever I do a make install it seems that all the configuration files have write 
permissions cleared.  Doing an install over an existing install will silently 
fail to update all the settings in /etc/init.d (I use debian) and /etc/pcmcia.  

This can lead to some very confusing problems.  As such I had to do a chmod +w 
every time I changed things and then double check all the settings to make sure I 
didn't miss one.  It was quite annoying.

Are the files marked read only for a reason or is something strange about my 
setup?



--
Richard A. Smith                         Bitworks, Inc.               
rsmith@bitworks.com               501.846.5777                        
Sr. Design Engineer        http://www.bitworks.com   

None Configuration files

Re: None Raylink problem solved.. (Richard A. Smith)
Date: 2000, Aug 15
From: David Hinds <dhinds@pcmcia.sourceforge.org>

On Tue, Aug 15, 2000 at 01:51:46PM -0700, Richard A. Smith wrote:
> 
> Whenever I do a make install it seems that all the configuration
> files have write permissions cleared.  Doing an install over an
> existing install will silently fail to update all the settings in
> /etc/init.d (I use debian) and /etc/pcmcia.

I'm not sure what you mean here.  Files in /etc/pcmcia like config,
network, etc are installed read-only and are generally not intended to
be modified by the user.  If you reinstall, and the new versions
differ from what is already installed, the old files are renamed to
*.O before the new versions are installed.  The *.opts files are never
overwritten if they already exist.

The scripts in /etc/init.d will generally not be overwritten because
some distributions use startup scripts that are not directly
compatible with the startup script included in the PCMCIA package.

> This can lead to some very confusing problems.  As such I had to do
> a chmod +w every time I changed things and then double check all the
> settings to make sure I didn't miss one.  It was quite annoying.

I don't think I understand.  What things do you need to change after
doing a reinstall?  I do "make install" dozens of times per day and
never need to touch permissions or edit files.

-- Dave
Some context would help


Add Message to: "Some context would help"

Members Subscribe Admin Mode Show Frames Help for HyperNews at pcmcia-cs.sourceforge.net 1.10