Project Home
Project Home
Wiki
Wiki
Discussion Forums
Discussions
Project Information
Project Info
Forum Topic - Lockup problem with Net.e1000 Net.r6040: (23 Items)
   
Lockup problem with Net.e1000 Net.r6040  
HI there

I am currently attempting to evaluate QNX 4.25 on a Vortex DX3.  The SBC has two ethernet interfaces.  As such I am 
having issues with both of them.

First point to note I am only ever running one driver at any given time.

Net.e1000 

I bring it all up happily and then begin my test by ftping a file then uploading it again.  When it runs at 100 MBps/
Full Duplex after about 60 to 100 operations the driver stops responding even to ping.  

I cannot ping out of the board either.

I run trace log and get (55) NET raw app could not buffer pkt.

Netinfo shows no errors of any kind and neither does netstat.  

The only way to bring it back up and working is to slay and relaunch everything including Net.

I have verbose mode turned on for each and the driver -vvv

Neither prints anything to console upon the error.

I have already switched cables and removed the switch etc. 

Net.r6040

This one is slightly more problematic.   Doing the same test as above with ping also running I suffer a lot of timeouts 
on ping.  I then run out test application which just echoe UDP packets back to a client.  And it times out as well.  

I have tried the released driver and a patched one from these forums in the 6040 post.

Get the same issues on both.

The r6040 is the onboard SOC one for the Vortex and the i210 is on the southbridge connected to it.

Hoping someone may have insight

Thanks in advance

Paul
Re: Lockup problem with Net.e1000 Net.r6040  
Please can you post the output from "sin ver" on your system while running the e1000 driver?

Thanks, Hugh.

On 2019-02-04, 11:04 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    HI there
    
    I am currently attempting to evaluate QNX 4.25 on a Vortex DX3.  The SBC has two ethernet interfaces.  As such I am 
having issues with both of them.
    
    First point to note I am only ever running one driver at any given time.
    
    Net.e1000 
    
    I bring it all up happily and then begin my test by ftping a file then uploading it again.  When it runs at 100 MBps
/Full Duplex after about 60 to 100 operations the driver stops responding even to ping.  
    
    I cannot ping out of the board either.
    
    I run trace log and get (55) NET raw app could not buffer pkt.
    
    Netinfo shows no errors of any kind and neither does netstat.  
    
    The only way to bring it back up and working is to slay and relaunch everything including Net.
    
    I have verbose mode turned on for each and the driver -vvv
    
    Neither prints anything to console upon the error.
    
    I have already switched cables and removed the switch etc. 
    
    Net.r6040
    
    This one is slightly more problematic.   Doing the same test as above with ping also running I suffer a lot of 
timeouts on ping.  I then run out test application which just echoe UDP packets back to a client.  And it times out as 
well.  
    
    I have tried the released driver and a patched one from these forums in the 6040 post.
    
    Get the same issues on both.
    
    The r6040 is the onboard SOC one for the Vortex and the i210 is on the southbridge connected to it.
    
    Hoping someone may have insight
    
    Thanks in advance
    
    Paul
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119468
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
See below.  I am aware there is the newer driver in 2015 but I get the same issue with it also which is why i downgraded
 by 1 version to the one listed below.

I ran test overnight with it running at 10 Mpbs FDX and did not encounter an issue after 2000 tests.  But at 100 Mpbs 
FDX it fails generally within 100 even with a bigger slug between each iteration of test (from 1 to 10 seconds as it is 
now)

Everything else like Net I think is up to date.


PROGRAM                 NAME         VERSION DATE
sys/Proc32              Proc         4.25Q   Jul 18 2005
sys/Slib32              Slib32       4.24B   Aug 12 1997
/bin/Fsys               Fsys32       4.24Z   Jul 06 2009
/bin/Fsys.atapi         atapi        4.25G   Jun 10 2010
//1/bin/Dev32           Dev32        4.23G   Oct 04 1996
//1/bin/Pipe            Pipe         4.23A   Feb 26 1996
//1/bin/Dev32.ansi      Dev32.ansi   4.23H   Nov 21 1996
//1/bin/Dev32.ser       Dev.ser      4.25A   Feb 14 2003
//1/bin/Net             Net          4.25G   Jul 06 2010
//1/bin/Net.e1000       Net.e1000    4.25D   Dec 19 2013
//1/usr/ucb/Socket      Socket       4.25K   Feb 12 2003
Re: Lockup problem with Net.e1000 Net.r6040  
I doubt very much that this is a driver problem, as the e1000 driver has been running at many sites for years. This 
rather sounds like a board/driver combination problem.
I have been running tests on the Net.e1000 4.25E for several hours at 100Mb/s without any problems. There isn't much we 
can do without hardware.


On 2019-02-05, 5:37 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    See below.  I am aware there is the newer driver in 2015 but I get the same issue with it also which is why i 
downgraded by 1 version to the one listed below.
    
    I ran test overnight with it running at 10 Mpbs FDX and did not encounter an issue after 2000 tests.  But at 100 
Mpbs FDX it fails generally within 100 even with a bigger slug between each iteration of test (from 1 to 10 seconds as 
it is now)
    
    Everything else like Net I think is up to date.
    
    
    PROGRAM                 NAME         VERSION DATE
    sys/Proc32              Proc         4.25Q   Jul 18 2005
    sys/Slib32              Slib32       4.24B   Aug 12 1997
    /bin/Fsys               Fsys32       4.24Z   Jul 06 2009
    /bin/Fsys.atapi         atapi        4.25G   Jun 10 2010
    //1/bin/Dev32           Dev32        4.23G   Oct 04 1996
    //1/bin/Pipe            Pipe         4.23A   Feb 26 1996
    //1/bin/Dev32.ansi      Dev32.ansi   4.23H   Nov 21 1996
    //1/bin/Dev32.ser       Dev.ser      4.25A   Feb 14 2003
    //1/bin/Net             Net          4.25G   Jul 06 2010
    //1/bin/Net.e1000       Net.e1000    4.25D   Dec 19 2013
    //1/usr/ucb/Socket      Socket       4.25K   Feb 12 2003
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119472
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
I have switched to another board and it runs better for still has the same issue eventually

The board is a Adlink CM1-86DX3.  It has two interfaces one i210 and one R6040 (Vortex SoC).  They both share the same 
IRQ line.

I dont know if this would factor into their being an issue,  Just done a test with both drivers running and e1000 
configured with address.  Still fails.

In your setups are you launching the Net.e1000 with any particular settings
Re: Lockup problem with Net.e1000 Net.r6040  
I'm just running "Net.e1000 -l2 -d1533" as it is the second interface on my machine. Can you send me the steps you run 
to reproduce the problem?
If the i210 and Vortex are sharing the same interrupt, are any other device using the same interrupt?

On 2019-02-05, 8:39 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    I have switched to another board and it runs better for still has the same issue eventually
    
    The board is a Adlink CM1-86DX3.  It has two interfaces one i210 and one R6040 (Vortex SoC).  They both share the 
same IRQ line.
    
    I dont know if this would factor into their being an issue,  Just done a test with both drivers running and e1000 
configured with address.  Still fails.
    
    In your setups are you launching the Net.e1000 with any particular settings
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119475
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Yeah sure

So my system comes up with no network configured.  Everything is down.

I then manually run

Net &
Net.e1000 -l1 &

nameloc &

Socket localhost &

ifconfig lo0 localhost up
ifconfig en1 144.179.1.199 netmask 255.255.255.0 up

inetd

I wait a few seconds between each step.  Speed of PC and board us 100 Mpbs FDX.  No switches in the way.

Also for the last two iterations I have been running the second driver also Net.r6040 -l2 & launched after Net.e1000

I then start a script running on PC which connects to FTP, does a 3 meg download, then uploads the same file and then 
disconnects.  Waits 1 second then repeats.

Eventually it just stops responding to FTP and only way to recover it is to slay everything including Net and relaunch 
as above.

When I ping from PC to card I can see from netinfo that packets are being recieved but the pings fail and board 
transmits nothing.

Re: Lockup problem with Net.e1000 Net.r6040  
Please can you post your script as well?

On 2019-02-05, 9:16 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Yeah sure
    
    So my system comes up with no network configured.  Everything is down.
    
    I then manually run
    
    Net &
    Net.e1000 -l1 &
    
    nameloc &
    
    Socket localhost &
    
    ifconfig lo0 localhost up
    ifconfig en1 144.179.1.199 netmask 255.255.255.0 up
    
    inetd
    
    I wait a few seconds between each step.  Speed of PC and board us 100 Mpbs FDX.  No switches in the way.
    
    Also for the last two iterations I have been running the second driver also Net.r6040 -l2 & launched after Net.e1000

    
    I then start a script running on PC which connects to FTP, does a 3 meg download, then uploads the same file and 
then disconnects.  Waits 1 second then repeats.
    
    Eventually it just stops responding to FTP and only way to recover it is to slay everything including Net and 
relaunch as above.
    
    When I ping from PC to card I can see from netinfo that packets are being recieved but the pings fail and board 
transmits nothing.
    
    
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119477
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Python 2.7

Downloads a 3 meg text file
Attachment: Text ftp.py 3.53 KB
Re: Lockup problem with Net.e1000 Net.r6040  
Are you sure that this isn't a memory problem? If you run "sin in" intermittently on another console while you are 
running your ftp test, does the memory available stay the same or does it decrease?


On 2019-02-05, 9:29 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Python 2.7
    
    Downloads a 3 meg text file
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119479
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
The Memory in the top section remains at about 1615/2031 M virtual sits at about  1219/2088 M

But I am running test again with switch back in and PC forced to 10 Mpbs and board running at 100.

Heapl 22928

Both it and virtual are fluctuating slightly at present but not by much.

So far it has passed the previous success limit of 444.

I will repeat the test on the faster speed again and run sin in.  But I have run it before and not noticed anything 
untoward?

Anything particular I should  look out for?

I am away to make a simple TCP server and client to run full packets at the board and echo them to test the interface 
more in line with the kind of tarf
Re: Lockup problem with Net.e1000 Net.r6040  
Hmm nope

I have python 2.7.15 installed on windows 7.  And do not set anything else except tick box in installer to run add to 
system PATH so I can run from command prompt.

Then from command line run: python ftp.py

Python fails immediately if it cannot do something as its runtime interpreted.

Re: Lockup problem with Net.e1000 Net.r6040  
The memory stays consistent.  

We have over 2 Gig of Ram on board.  And still have 1600 Meg available.

Virtual memory is at 1225/2000 odd Meg
 
HeapL is 22798

Nothing changes much.
Re: Lockup problem with Net.e1000 Net.r6040  
I have run your script from my Ubuntu Linux PC to my QNX4 PC at 1Gb/s for 100 loops without any problems. The file size 
is 195MB.
I am now running the same test via a 100Mb/s switch to see if anything changes.

On 2019-02-05, 11:28 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    The memory stays consistent.  
    
    We have over 2 Gig of Ram on board.  And still have 1600 Meg available.
    
    Virtual memory is at 1225/2000 odd Meg
     
    HeapL is 22798
    
    Nothing changes much.
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119485
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Attached,

Python 2.7
Using standard python libs.

Downloads a meg meg text file
Attachment: Text ftp.py 3.53 KB Text udplog2.txt 2.96 MB
Re: Lockup problem with Net.e1000 Net.r6040  
I'm not familiar with python, but do you set any environment variables? I get an "ImportError: No module name os".


On 2019-02-05, 9:32 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Attached,
    
    Python 2.7
    Using standard python libs.
    
    Downloads a meg meg text file
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119480
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Latest status.

Ran with new board and it gets to 400 before locking up.

Setting PC to 10 Mbps FDX through switch with board set to 100 Mbps FDX.  Test appears to run without issue.  Have run 
sin proxies and sin timers and there does not appear to be a build up of either.
Re: Lockup problem with Net.e1000 Net.r6040  
I left your test running last night, and this morning it is still running with 1300 loops completed. This is running 
through a 100Mb/s switch.


On 2019-02-06, 4:19 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Latest status.
    
    Ran with new board and it gets to 400 before locking up.
    
    Setting PC to 10 Mbps FDX through switch with board set to 100 Mbps FDX.  Test appears to run without issue.  Have 
run sin proxies and sin timers and there does not appear to be a build up of either.
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119487
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Yeah I left a test running while off with PC at 10 Mbps FDX through a 100 Mbpos switch with board runnig at 100 Mbps FDX
.  

It got to 27900 completions before it locked up



> I left your test running last night, and this morning it is still running with
>  1300 loops completed. This is running through a 100Mb/s switch.
> 
> 
> On 2019-02-06, 4:19 AM, "Paul Graham" <community-noreply@qnx.com> wrote:
> 
>     Latest status.
>     
>     Ran with new board and it gets to 400 before locking up.
>     
>     Setting PC to 10 Mbps FDX through switch with board set to 100 Mbps FDX.  
> Test appears to run without issue.  Have run sin proxies and sin timers and 
> there does not appear to be a build up of either.
>     
>     
>     
>     _______________________________________________
>     
>     General
>     http://community.qnx.com/sf/go/post119487
>     To cancel your subscription to this discussion, please e-mail general-qnx4
> -unsubscribe@community.qnx.com
>     
> 


Re: Lockup problem with Net.e1000 Net.r6040  
At this point I think that your best bet would be to contact your QNX sales representative and make arrangements to send
 us your hardware, as we are unable to reproduce this problem with the hardware that we have.

Thanks, Hugh.

On 2019-02-15, 10:07 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Yeah I left a test running while off with PC at 10 Mbps FDX through a 100 Mbpos switch with board runnig at 100 Mbps
 FDX.  
    
    It got to 27900 completions before it locked up
    
    
    
    > I left your test running last night, and this morning it is still running with
    >  1300 loops completed. This is running through a 100Mb/s switch.
    > 
    > 
    > On 2019-02-06, 4:19 AM, "Paul Graham" <community-noreply@qnx.com> wrote:
    > 
    >     Latest status.
    >     
    >     Ran with new board and it gets to 400 before locking up.
    >     
    >     Setting PC to 10 Mbps FDX through switch with board set to 100 Mbps FDX.  
    > Test appears to run without issue.  Have run sin proxies and sin timers and 
    > there does not appear to be a build up of either.
    >     
    >     
    >     
    >     _______________________________________________
    >     
    >     General
    >     http://community.qnx.com/sf/go/post119487
    >     To cancel your subscription to this discussion, please e-mail general-qnx4
    > -unsubscribe@community.qnx.com
    >     
    > 
    
    
    
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119522
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Can you post the output from "sin" and "sin irq" when the lockup occurs?


On 2019-02-06, 4:19 AM, "Paul Graham" <community-noreply@qnx.com> wrote:

    Latest status.
    
    Ran with new board and it gets to 400 before locking up.
    
    Setting PC to 10 Mbps FDX through switch with board set to 100 Mbps FDX.  Test appears to run without issue.  Have 
run sin proxies and sin timers and there does not appear to be a build up of either.
    
    
    
    _______________________________________________
    
    General
    http://community.qnx.com/sf/go/post119487
    To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
    

Re: Lockup problem with Net.e1000 Net.r6040  
Yeah will do.  I'll be back at it today so will recreate the failure and get that info
Re: Lockup problem with Net.e1000 Net.r6040  
Hi

Sorry for the delay.

Here is a screen shot of the sin irq stuff.  Doesn't mean much to me I am afraid
Attachment: Image 20190215_113024.jpg 4.15 MB