CVI real-time hangs on reboot

[tedoyle]

I am using CVI real-time 9.0 and I'm having an issue with my system hanging on system reboot.  The hardware (PC/104 ATOM based CPU) passes all of the platform checks and runs my application with no problems.  However, if I issue a reboot command (e.g. from MAX) the board receives the command but hangs.  Can someone explain the low-level mechanism being used to trigger the board reset?  I have tested this code on two other boards (not ATOM based) with no problems.

[nickb]

Hello tedoyle,

I think what you've run into is a known issue with the CVI 9.0 RT module that will happen under some circumstances.  We have a patch created for this issue, and I have attached it below.  I thought I had created a KB for it, but it's apparently slipped off my to do list.  I will get the KB created in the next couple days and post a link to it here for anyone who runs into this issue in the future.

Included in the attached zip is a readme describing how to install the patch.  Please don't hesitate to let me know if you have any questions.

NickB

National Instruments  

Message Edited by nickb on 10-23-2009 11:07 AM

[tedoyle]

Nick,

Thanks for the quick response!  I have a quick question regarding your feedback.  Has this patch been applied to the CVI real-time engine 9.1 (in beta)?  Also, under CVI-RT 9.0 why does the reboot work correctly on other CPU's?

[nickb]

The fix was made before the release of the beta and so should be included in the beta version of cvi_lvrt.  If you have time to test, I'd like to know if you see something to the contrary.

The primary reason it appears on some machines and not others is that a race condition existed that could result in deadlock on some machines.  This deadlock occurs in some of the thread synchronization as the threads of the RT process exit.

NickB

National Instruments  

[tedoyle]

Nick,

Sorry for the late reply but I have been waiting on a vendor to return the PC/104 board that I am testing.  I reran the test with CVI-RT 9.1 on the ATOM based PC/104 board.  When I issue a reboot from MAX, the board sees the command.  However, the board does not reset.

Any suggestions?

[nickb]

Sorry for taking so long to get back to you...

Will a very simple RT application (basically just sleeping in a for loop) reboot when issued the command from MAX?  We don't currently have access to the hardware you're using, so I don't really have anything more helpful to offer in the meantime.  However , I have seen reboot attempts foiled in the past because the application left open various system objects (locks, thread handles, etc).  Could this be a possibility (the resource tracking can be invaluable here...)?

NickB

National Instruments 

[tedoyle]

Nick,

The problem here does not seem to be related to software.  I have created a very simple app that reproduces the problem on the ATOM based board.  However, the same app runs fine (reboot works) on other processors.

The problem may be with the Poulsbo chipset.  This chipset is expecting the reset call to control register 0CF9h (according to Intel docs).  Is this the register (with the correct bit sets) being used by CVI-RT for the reboot action?  If no, would it be possible to get a patch to work around this issue?

Tom