(Follow up)
Hi, Pavel. I took a good look at the terminal windows and suspect the lines I typed the first time were in pseudo 
terminals instead of the actual photon session. So just in case I moved aside any pterm windows and typed inside the 
actual photon session. Starting the drivers in there froze the windows completely.

For instance, typing:  Input -d /dev/photon kbd fd -d/dev/kbd ps2 -r kb -2 &
froze everything.  I cannot change any consoles or move anything, and I suspect the whole phditto session ended.

I also cannot start a new Phindows session (so I am going to the lab to reboot the PC right now). I don't know it 
mattered for your test whether I had typed in a pterm or a photon session, but I tried both ways now.  Thanks.

(Follow up #2)

Pavel, the more I think about the clues the more I think there is an IRQ conflict whenever the USB mouse and keyboard 
are used.

1.  When installing QNX4, if I use diagnostics mode and tell it to check the USB plug and play the installation 
completely freezes.  If I don't have it check the USB plug and play it installs fine but treats the USB mouse and 
keyboard as PS/2.

2.  When I start the mouse as USB QNX freezes and requires a reboot.  If it starts in the initialization files it never 
comes up all the way and halts without allowing login or connection.

3.  I have seen signs in the field of USB conflict with other hardware which uses the same IRQ.  Before we started using
 USB mouse and keyboard we never had problems, but we figured out a pattern in which all systems which started using the
 USB mouse and keyboard had mysterious cases where the hardware would fail to respond. It only started with USB mouse 
and keyboard.

Early into trouble shooting I checked the bios to see if I could control IRQ's like I could in the old legacy days of 
ISA, but I can't anymore.

I saw in some of the recent patch notes that there are fixes for USB conflicts, but I think there are still problems 
with it.  And I'm using the latest QNX4 installation CD (June 2011).

I look forward to any more suggestions or solutions you can offer.

Thanks very much.

Regards,

Guillermo Lande

Hi Guillermo,

It seems I misguide you. You should run Photon (if it is not started)
from a console directly on your QNX4 system, not from phwindows session
and pterm.

The Input manager terminates because the device specified by -d option
(/dev/photon) doesn't not exist. By default when you run the Photon GUI
from a console by the ph script the Photon server creates the
/dev/photon prefix.

I've checked all my recommendations with Phindows on my system and all
starts correctly. But in my opinion it will be better to use telnet for
these purposes.

As for USB, yes the io-usb manager in QNX4 has some problems with
interrupts sharing. But from logs you've sent me I see that io-usb works
and detects USB devices. As a next step run devu-mouse and devu-kbd drivers.
From remote pterm or from remote console (telnet) run following commands:
# io-usb
# devu-mouse
# devu-kbd
# ls /dev | grep usb

Check for usbmouse0 and usbkbd0 prefixes.
If prefixes are created and system doesn't freeze. Then run:
# slay -f Input
# Input -d /dev/photon msoft fd -d /dev/usbmouse0 &
# Input -d /dev/photon kbd fd -d /dev/usbkbd0 &


Regards,
Pavel

> Hi, Pavel.
>
> In the sixteen years I've ween working with QNX and the thirteen with QNX4 I never thought of starting a photon 
session within a pterm itself. That was a neat test.  Per your suggestion I used Phindows to spawn a pterm connection 
and then started a photon session within it (by typing "ph").
>
> First I performed:
> # slay -f Input
> # Input -d /dev/photon kbd fd -d/dev/kbd ps2 -r kb -2 &
>
> The Input program crashed right away.  I repeated the test without the "-r" and had the same result.
>
> Next I performed:
> # slay -f Input  (even though Input was already crashed and down)
> # Input -d /dev/photon kbd kb &  (Input crashed right away)
> # Input -d /dev/photon ps2 -r kb -2 &  (Input crashed right away)
>
> So this also failed to spawn the Input drivers.  Hitting "enter" after each command came back with the message that 
the process had died.
>
> For the outputs you requested I zipped them into "usb_test.zip" (hopefully attached).
>
> It contains the following files:
> - usb_show_pci_vvv.txt  (output of show_pci -vvv)
> - usb_sin.txt (output of sin)
> - usb_sin_ar.txt (output of sin args)
> - usb_sin_ir.txt (output of sin irq)
> - usb_sin_ver.txt (output of sin version just in case you want that)
> - usb_usb_vvv.txt (output of usb -vvv)
>
> I did a textto -c on each of the files in case you read them in Windows.
>
> Thanks very much.
>
> Regards,
>
> Guillermo Lande
> (Side note, we just put an order in for several of these PC's so will be buying QNX4 licenses for them soon. Hopefully
 this all works out. Thanks again.)
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96186
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com

**Receipt of Message**
Thanks, Pavel. To make sure I do not misunderstand I will repeat the tests you mentioned today three ways:
- pterm from Phindows
- console from Phindows (ditto -kqn1 to reach $con)
- telnet -8 [ip address]

**Follow Up**

This is interesting.  Before wiping the system and starting over I remembered I had a copy of the original devu-mouse 
driver from the install CD.  I put the original driver back, rebooted the syste and repeated the test with the original 
driver.

I had to use telnet, of course, because Phindows would not let me connect.

# io-usb (worked fine)
# devu-mouse (worked fine)
# devu-kbd (worked fine)
# ls /dev/ |grep usb  (found the prefixes)

# slay -f Input  (slew the two Inputs)
# Input -d /dev/photon msoft fd -d /dev/usbmouse0 &

Here I hit enter a couple of times to make sure the system was responsive, and it was. So I continued.

# Input -d /dev/photon kbd fd -d /dev/usbkbd0 &

And the system froze.  I'll reboot and try again, and if I get a different result I'll let you know. But we may have to 
stick with your latest devu-mouse driver.

Hi, Pavel.  Well, your last test indicates the issue is with the way QNX4 installs and not with the drivers itself 
(though I did do the test with the mouse driver you gave Pete recently and not the one from the install CD). I will 
repeat the test later with a fresh install and the 2011 driver.

First I performed a clean reboot and tried typing io-usb from a console. That froze me.

So I rebooted again, started photon (ph) and tried to Phindow into the system.  Phindows could not establish the 
connection (it saw it as IP but could not start up).

Next I went to a QNX4 system and typed telnet [IP address] to connect, and there I typed:
# io-usb
# devu-mouse  (your recent devu-mouse)
# devu-kbd
# ls /dev |grep usb

The /dev/usbmouse0 and /dev/usbkbd0 prefixes were present, so I typed:

# slay -f Input
# Input -d /dev/photon msoft fd -d /dev/usbmouse0 &
# Input -d /dev/photon kbd fd -d /dev/usbkbd0 &

Next I tried to open Phindows again, and it connected right away.  So I went to the lab and used the mouse and keyboard 
on Photon directly, and they worked fine.

The challenge, then, is going to be figuring out a procedure I can give our install technicians to install QNX4 in a way
 that can see the USB mouse and keyboard because the 2011 install CD can't do it automatically.  It freezes if we tell 
it to search for USB plug and play during the installation, and without selecting that it sets up the system as PS/2.

I'm going to go in the computer room and wipe the system and see what I can do, but I am not optimistic as I have 
already tried re-installation many times (at least 6 but probably 10 or 12).

I'll check here later to see if you have any more suggestions.

Thanks, Pavel.

Regards,

Guillermo Lande

*Confirmation*

I repeated the test a third time with the original devu-mouse driver from the 2011 installation CD, and the system froze
 again.  The issue is definitely with the devu-mouse driver, so it looks like if I can make an install CD that uses 
Pavel's recent devu-mouse driver instead of the old one the system should in theory be able to install normally with USB
 plug and play support.

Hi Guillermo,

For now there is no way to use USB keyboards in text mode if the io-usb
is running. As I wrote previously  when you start the io-usb manager it
initializes all USB ports and from that moment your USB mouse and USB
keyboard will not
work in PS/2 legacy mode and it will be possible to use them only in
Photon with devu-mouse and devu-kbd.

If it is suitable for you to boot directly into the Photon then you can
try following:
1. During installation choose to boot directly into the Photon.
2. Copy latest USB stack components.
3. Add to the sysinit.<node>:
io-usb
devu-mouse
devu-kbd
4. Modify your /etc/config/trap/input.<node>:
msoft fd -d/dev/usbmouse0
kbd fd -d/dev/usbkbd0

Don't forget to prepare .altboot and altsysinit for recovering if
something goes wrong.

Regards,
Pavel

> Hi, Pavel.  Well, your last test indicates the issue is with the way QNX4 installs and not with the drivers itself 
(though I did do the test with the mouse driver you gave Pete recently and not the one from the install CD). I will 
repeat the test later with a fresh install and the 2011 driver.
>
> First I performed a clean reboot and tried typing io-usb from a console. That froze me.
>
> So I rebooted again, started photon (ph) and tried to Phindow into the system.  Phindows could not establish the 
connection (it saw it as IP but could not start up).
>
> Next I went to a QNX4 system and typed telnet [IP address] to connect, and there I typed:
> # io-usb
> # devu-mouse  (your recent devu-mouse)
> # devu-kbd
> # ls /dev |grep usb
>
> The /dev/usbmouse0 and /dev/usbkbd0 prefixes were present, so I typed:
>
> # slay -f Input
> # Input -d /dev/photon msoft fd -d /dev/usbmouse0 &
> # Input -d /dev/photon kbd fd -d /dev/usbkbd0 &
>
> Next I tried to open Phindows again, and it connected right away.  So I went to the lab and used the mouse and 
keyboard on Photon directly, and they worked fine.
>
> The challenge, then, is going to be figuring out a procedure I can give our install technicians to install QNX4 in a 
way that can see the USB mouse and keyboard because the 2011 install CD can't do it automatically.  It freezes if we 
tell it to search for USB plug and play during the installation, and without selecting that it sets up the system as PS/
2.
>
> I'm going to go in the computer room and wipe the system and see what I can do, but I am not optimistic as I have 
already tried re-installation many times (at least 6 but probably 10 or 12).
>
> I'll check here later to see if you have any more suggestions.
>
> Thanks, Pavel.
>
> Regards,
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96231
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
> .
>

Thanks, Pavel. I'll re-install cleanly right now and try that as you describe.

I'm not sure if you'll see this e-mail today, but is there any particular place in the sysinit file those lines need to 
go.  I tried putting them several times in various places yesterday with and without keeping the "kbd [environment 
variable" line in it, but it always blocked the system from coming up.

So I'll try it on a clean install which boots directly into Photon, but any tips on placement you can give will be 
greatly appreciated.  I'll also let you know how the test goes.

Thanks again.

Guillermo Lande

The only restriction: io-usb and drivers should be started before Photon
script. Also delay (e.g. sleep 1) could be required.

Regards,
Pavel


> Thanks, Pavel. I'll re-install cleanly right now and try that as you describe.
>
> I'm not sure if you'll see this e-mail today, but is there any particular place in the sysinit file those lines need 
to go.  I tried putting them several times in various places yesterday with and without keeping the "kbd [environment 
variable" line in it, but it always blocked the system from coming up.
>
> So I'll try it on a clean install which boots directly into Photon, but any tips on placement you can give will be 
greatly appreciated.  I'll also let you know how the test goes.
>
> Thanks again.
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96286
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
>

Hi, Pavel. This is just an incremental update for your information. Re-installing QNX4 to boot directly into photon and 
then replacing the devu-mouse driver with your recent one and putting the new lines into sysinit.1 and input.1 did not 
work. Photon would just hang on waiting for input driver.

You said something about making sure the latest stack is in place. I used the 2011 install CD plus your devu-mouse 
driver, but my next step will be to see if I can find a newer stack than that. Previous searches did not find newer, but
 I'll look again.

In case you're curious, the attached are the sysinit.1 and input.1 files which did not work. Thanks.

Hi, Pavel.  Well, I'm officially stumped. I can't find a newer USB stack than June 2, 2011 (what comes in the 2011 
install CD) with the exception of your September 2012 devu-mouse.

And anything I do boots up into a photon for which the mouse is erratic and dangerous with random clicks and drags of 
both mouse buttons.  I can't find a good installation combination which would allow photon to use, and the only way to 
even use QNX4 at all is to not boot into photon and let it install the USB mouse and keyboard as PS/2 so I can use 
Phindows.

And I'm definitely concerned about the periodic symptoms that appear to be IRQ conflicts ever since I started using PC's
 that have USB ports. Hardware using the same IRQ (5, such as the encryption card) get periodical failures to respond, 
and this never ever happened in the PS/2 systems.

If you have any more suggestions I would be grateful.  Thanks very much.

Regards,

Guillermo Lande

Hi Guillermo,

The latest USB stack is available form the Latest Software Updates Page:
http://community.qnx.com/sf/wiki/do/viewPage/projects.qnx4/wiki/LatestSoftwareUpdates

This version should work more stable with interrupts sharing.

Also you should remove from your /etc/config/trap/input.1 file following
line:
#kbd fd -d/dev/kbd ps2 -r kb -2
Commenting is not supported in this file.

Regards,
Pavel


> Hi, Pavel. This is just an incremental update for your information. Re-installing QNX4 to boot directly into photon 
and then replacing the devu-mouse driver with your recent one and putting the new lines into sysinit.1 and input.1 did 
not work. Photon would just hang on waiting for input driver.
>
> You said something about making sure the latest stack is in place. I used the 2011 install CD plus your devu-mouse 
driver, but my next step will be to see if I can find a newer stack than that. Previous searches did not find newer, but
 I'll look again.
>
> In case you're curious, the attached are the sysinit.1 and input.1 files which did not work. Thanks.
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96293
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com

**Receipt of E-Mail**
Thanks for the suggestions, Pavel.  It may be a while before I can try them, but I see at a glance that the link you 
gave has at least three patches that look like they related to things I use in our systems. I will try them later today 
and submit another update when it's done.  Thanks again.

Guillermo

Hi, Pavel.

I downloaded the new QNX425 USB stack from the link you provided, and then I installed it into the clean system 
verifying the check sum values, permissions and executable attributes.  The instructions that came with the download 
were very thorough.

But when the system booted up it still said waiting for input on the photon screen which should be the sign in.  So 
booted into the alternate sysinit and repeated the old test of manually slaying all the Inputs and starting them up 
manually.  The whole system froze on bringing up the keyboard.

(Verify no commented out line in /etc/config/trap/input.1)
# slay -f Input
# io-usb
# devu-mouse
# devu-kbd
# ls /dev/ |grep usb
(usb, usbmouse0 and usbkbd0 were present)
# Input -d /dev/photon msoft fd -d /dev/usbmouse0 &
# [Enter] (to verify no freeze)
# Input -d /dev/photon kbd fd -d /dev/usbkbd0 &

The system never returned. It froze on that.

So even with the new stack it still has the same problem trying to bring up the USB keyboard and mouse.

Thanks very much.

Regards,

Guillermo Lande

Hi Guillermo,

I forgot to mention that when you are booting directly to the Photon the
/etc/config/bin/ph.<node> script is searching for the
/etc/config/bin/Input.<node> file and starting the Input manager as
described in this file.

So you should modify the /etc/config/bin/Input.<node> file to use USB
drivers e.g:
Input msoft fd -d /dev/usbmouse0 kbd fd -d /dev/usbkbd0 &

Or you can remove the /etc/config/bin/Input.<node> file. In this case
configuration form the /etc/config/trap/Input.<node> will be used.

Also increase delay (set sleep 5) after usb drivers in
/etc/config/sysinit.<node>.

I hope these will help.

Regards,
Pavel

> Hi, Pavel.
>
> I downloaded the new QNX425 USB stack from the link you provided, and then I installed it into the clean system 
verifying the check sum values, permissions and executable attributes.  The instructions that came with the download 
were very thorough.
>
> But when the system booted up it still said waiting for input on the photon screen which should be the sign in.  So 
booted into the alternate sysinit and repeated the old test of manually slaying all the Inputs and starting them up 
manually.  The whole system froze on bringing up the keyboard.
>
> (Verify no commented out line in /etc/config/trap/input.1)
> # slay -f Input
> # io-usb
> # devu-mouse
> # devu-kbd
> # ls /dev/ |grep usb
> (usb, usbmouse0 and usbkbd0 were present)
> # Input -d /dev/photon msoft fd -d /dev/usbmouse0 &
> # [Enter] (to verify no freeze)
> # Input -d /dev/photon kbd fd -d /dev/usbkbd0 &
>
> The system never returned. It froze on that.
>
> So even with the new stack it still has the same problem trying to bring up the USB keyboard and mouse.
>
> Thanks very much.
>
> Regards,
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96338
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
>

**Receipt of E-Mail** Hi, Pavel. Thanks for the new suggestions. I'm sorry I was away all day, but I will try these new 
steps tomorrow and look forward to seeing their results and letting you know how it goes. Thank you very much.

Regards,

Guillermo Lande

Hi, Pavel.  I tried the new tests, and they didn't work at all.  Changing the Input.1 to use the USB driver made it so 
even the alternate boot would hang waiting for input.  At least the alternate boot still let me access the system with "
telnet -8 [IP]" even though Phindows would not work.

I really think the USB driver is hanging with an IP conflict, but it's hard to prove because the whole system is locking
 up.

Just in case you want it I put a zip archive (attached) with three files converted to DOS carriage returns with "textto 
-c".

- sysinit.1 from /etc/config
- Input.1 from /etc/config/bin
- input.1 from /etc/config/trap

Thanks,

Guillermo Lande

*Edit* Sorry, I meant to say I think it's an IRQ conflict, not an IP conflict. Thanks.

Hi Guillermo,

Please try attached USB stack. It is an experimental version and it is
possible that it can help. Anyway it will be interesting to check this
version on your hardware.

Regards,
Pavel

> Hi, Pavel.  I tried the new tests, and they didn't work at all.  Changing the Input.1 to use the USB driver made it so
 even the alternate boot would hang waiting for input.  At least the alternate boot still let me access the system with 
"telnet -8 [IP]" even though Phindows would not work.
>
> I really think the USB driver is hanging with an IP conflict, but it's hard to prove because the whole system is 
locking up.
>
> Just in case you want it I put a zip archive (attached) with three files converted to DOS carriage returns with "
textto -c".
>
> - sysinit.1 from /etc/config
> - Input.1 from /etc/config/bin
> - input.1 from /etc/config/trap
>
> Thanks,
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96408
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com

**Receipt of Message**

Thanks, Pavel.  I will try it right away and let you know how it goes.

Regards,

Guillermo Lande

Hi, Pavel.

I downloaded the experimental USB stack, verified the check sum values and executable status and installed them.  Using 
the same sysinit.1, Input.1 and input.1 files as yesterday but with the experimental USB stack still has the problem in 
which Photon says: "Waiting for input driver."

However *drumroll* the system is not completely locked up like the other USB stacks had. I can telnet into it, and I can
 Phindows into it.

Thus I do not have a functional photon, but I also am not locked out of the whole system requiring an alternate reboot.

Also, and this is new, when I type "ls /dev/usb*" I see that usb, usbkbd0 and usbmouse0 have successfully registered.  I
 could not see this before because the system would be locked up.

So the good news is this seems to have gotten past the IRQ conflict and registered the USB drivers, but Photon does not 
see the input drivers.

Progress?

Regards,

Guillermo Lande

Hi Guillermo,

Yes. It is a good result.
As for Photon I guess the problem is in your configuration. Please check
carefully all setup files.

For example, if you are using usb drivers the Input.1 file you've
provided previously is incorrect. It contains:

Input kbd fd -d/dev/kbd ps2 -r kb -2 &

instead of

Input msoft fd -d /dev/usbmouse0 kbd fd -d /dev/usbkbd0 &


Regards,
Pavel

> Hi, Pavel.
>
> I downloaded the experimental USB stack, verified the check sum values and executable status and installed them.  
Using the same sysinit.1, Input.1 and input.1 files as yesterday but with the experimental USB stack still has the 
problem in which Photon says: "Waiting for input driver."
>
> However *drumroll* the system is not completely locked up like the other USB stacks had. I can telnet into it, and I 
can Phindows into it.
>
> Thus I do not have a functional photon, but I also am not locked out of the whole system requiring an alternate reboot
.
>
> Also, and this is new, when I type "ls /dev/usb*" I see that usb, usbkbd0 and usbmouse0 have successfully registered. 
 I could not see this before because the system would be locked up.
>
> So the good news is this seems to have gotten past the IRQ conflict and registered the USB drivers, but Photon does 
not see the input drivers.
>
> Progress?
>
> Regards,
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96435
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
>

Hi, Pavel.  I am encouraged that you think this should be able to work now.

As for the current configurations, I am looking at it now and can't see what is wrong.

The /dev directory does have /dev/usb, /dev/usbkbd0 and /dev/usbmouse0

The /etc/config/bin/Input.1 has only the line:

Input msoft fd -d /dev/usbmouse0 kbd fd -d /dev/usbkbd0 &

The /etc/config/bin/input.1 has only the two lines:

msoft fd -d/dev/usbmouse0
kbd fd -d/dev/usbkbd0

And the /etc/config/sysinit.1 has the lines before calling tcpip.1 and ph.1:

io-usb
devu-mouse
devu-kbd
sleep 5

Should I try your September 2012 devu-mouse you made for Peter? Or shuold this experimental stack work as is?

Thanks very much.

Regards,

Guillermo Lande

**Follow Up**

I've tried rebooting a few more times today (partly to test increasing the sleep to 10 seconds instead of 5).  The non-
alternate boots (which use the USB settings) are always locking up the system this morning, and the alternate boot 
(which doesn't have the USB in the sysinit.1) is not locking up but doesn't have the /dev/usb* files.

I'm sure I did not boot as alternate boot yesterday because I had the /dev/usb* files, but I can't get that to come up 
without freezing this morning.  I'll change it back to 5 seconds and try rebooting a few more times.

Thanks.

Guillermo Lande

*Follow Up #2 10/19/2012*

I figured out the pattern and why it worked yesterday.  If I reboot with USB  the system comes up frozen (no telnet or 
Phindows capability). If I wait a long time (more than 2 minutes) it eventually unfreezes enough to allow telnet and 
Phindows.  It never sees the input driver for Photon, but I can get in the other remote ways.  Yesterday I must have 
tried more than 2 minutes after a reboot.

So my findings from yesterday are repeatable and remain true:

1.  If I use your experimental stack, the system will register the /dev/usb* files and will eventually allow telnet and 
Phindows, but it will not allow Photon, and

2.  If I use the other USB stacks the system completely locks up (as if it were an IRQ conflict) and never allows 
anything at all.

I'm not sure what else to try from here.

Thanks very much.

Regards,

Guillermo Lande

*Follow Up #3 10/19/2012*

Hi, Pavel.  To get the exact pattern I rebooted a few more times with a stop watch and measured how long it is from the 
Photon message "waiting for input driver" to telnet/Phindows being available.  It's consistently about 2 minutes and 58 
seconds which implies it's about 3 minutes from the spawning of the USB stack.

I also tested changing sysinit to wait 205 seconds instead of 5 seconds after spawning the USB stack, but that was 
disastrous and made me reboot into alternate image to change it back to 5 seconds.

I don't know why it takes 3 minutes for the system to unfreeze after spawning the USB stack, but maybe this is a clue.

Thanks very much.

Regards,

Guillermo Lande

It sounds to me like the USB stack is doing a number of retries (initializing/detecting something?)  before timing out..
. then it continues.

Martin Walter
QNX Software Systems Limited.
martin@qnx.com
613-271-9208


-----Original Message-----
From: Guillermo Lande [mailto:community-noreply@qnx.com] 
Sent: October-19-12 10:41 AM
To: general-qnx4
Subject: Re: USB Driver Fault

*Follow Up #3 10/19/2012*

Hi, Pavel.  To get the exact pattern I rebooted a few more times with a stop watch and measured how long it is from the 
Photon message "waiting for input driver" to telnet/Phindows being available.  It's consistently about 2 minutes and 58 
seconds which implies it's about 3 minutes from the spawning of the USB stack.

I also tested changing sysinit to wait 205 seconds instead of 5 seconds after spawning the USB stack, but that was 
disastrous and made me reboot into alternate image to change it back to 5 seconds.

I don't know why it takes 3 minutes for the system to unfreeze after spawning the USB stack, but maybe this is a clue.

Thanks very much.

Regards,

Guillermo Lande



_______________________________________________

General
http://community.qnx.com/sf/go/post96466
To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com

Hi Guillermo,

Thanks for the detailed description.
For now I can't say for sure why and where the USB stack is waiting and
I am afraid that without hardware we can't fix this problem. I've tried
to reproduce your problem on different motherboards but even with shared
interrupt (USB and Fsys.atapi) stack works quite well.

One more thing that I recommend you is to try the previous version of
Input (form the Product suite 2008). I've attached it. Try it with your
USB mouse and  USB keyboard in PS/2 emulation mode.

Regards,
Pavel

> *Follow Up #3 10/19/2012*
>
> Hi, Pavel.  To get the exact pattern I rebooted a few more times with a stop watch and measured how long it is from 
the Photon message "waiting for input driver" to telnet/Phindows being available.  It's consistently about 2 minutes and
 58 seconds which implies it's about 3 minutes from the spawning of the USB stack.
>
> I also tested changing sysinit to wait 205 seconds instead of 5 seconds after spawning the USB stack, but that was 
disastrous and made me reboot into alternate image to change it back to 5 seconds.
>
> I don't know why it takes 3 minutes for the system to unfreeze after spawning the USB stack, but maybe this is a clue.

>
> Thanks very much.
>
> Regards,
>
> Guillermo Lande
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96466
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
>

Thanks, Pavel.  I will try the 2008 versions and let you know whether they work better.

On a side note, and I haven't checked with the INTERPRO owners yet how they would feel about it, but is it an available 
option to ship one of the computers to QNX for internal testing there? I don't know whether they would even want to do 
this, but I would like to check with you whether you might, too.

Thanks very much, and I will message later today with the results of the next test.

Regards,

Guillermo Lande

Hi, Pavel.

For your information, I put the 2008 Input binary in the /bin directory and repeated the test twice more measuring how 
long it takes for the system to unfreeze from the time Photon says waiting for input.  It still takes 3 minutes 
(approximately 2 minutes and 58 seconds).

So there is no difference in behavior between the 2008 version of Input and the 2010.

I know when it  unfreezes by starting a "telnet -8 [IP of problem box]" from another QNX node.  When I get the login 
prompt I know it is unfrozen and can allow Phindows and telnet (but still not Photon).

Thanks very much.

Regards,

Guillermo Lande

Hi, Pavel.  Is there an e-mail address I could send some follow up information. I'm seeing more signs of interrupt 
problems in the field, but I prefer not to describe them in public forums.

Thanks very much.

Regards,

Guillermo Lande
development@interpro-tech.com
586-850-4243

p.s. I will probably be out of the office tomorrow, October 25, 2012.

Hi Guillermo,

OK, we can continue our discussion by e-mail.
I've sent you a mail with my address.

Regards,
Pavel

> Hi, Pavel.  Is there an e-mail address I could send some follow up information. I'm seeing more signs of interrupt 
problems in the field, but I prefer not to describe them in public forums.
>
> Thanks very much.
>
> Regards,
>
> Guillermo Lande
> development@interpro-tech.com
> 586-850-4243
>
> p.s. I will probably be out of the office tomorrow, October 25, 2012.
>
>
>
> _______________________________________________
>
> General
> http://community.qnx.com/sf/go/post96604
> To cancel your subscription to this discussion, please e-mail general-qnx4-unsubscribe@community.qnx.com
>

Hi, Pavel.  First, thanks again for working with me both on these Foundry forums and offline by e-mail to resolve the 
issue of the erratic USB mouse and the freezing of the whole operating system if we try to bring up Photon using the USB
 drivers.

The attached usb_experimental.zip has:
- the USB stack you gave me which works
- the usb-index utility you wrote and gave me to identify what configurations to use
- readme_usb_patch.txt which is a text summary of how to install and configure QNX 4.25 using the June 2011 CD on the 
USB PC's.

Thanks again,

Regards,

Guillermo Lande