Intermittent Malibu Problem

[eglaser]

Hello,

I have been working on the sample Quixote applications as well as my own app. I've noticed that sometimes the messages being sent from the target to the host are corrupted, with the TypeCode set to 0 and the data set quite randomly. I'm using TiBusmasterStream for sending both messages and data packets.

I'll get the error in the CpuOutRate sample application. After I download the COFF, then click start, it'll print out a 'Invalid Io message received'. I notice similar behavior in my app, in that all the messages being sent have the typecode set to 0, and contain random data.

This doesn't seem to always be the case. Sometimes if I reboot, it works fine. Other times not. I never had these kind of problems with the Armada toolset. I do have both Armada and Malibu drivers installed on my development system. Will this cause a problem? I have the correct drivers loaded (according to Windows Hardware Manager), and I'm using the EXO file that came with Malibu. I have the same problem with the 2M and 6M FPGA.

Any help is greatly appreciated.

Eli

[jhenderson]

Eli -

Please enable data integrity checking within the supplied CpuBmIn and CpuBmOut examples, and stream for an extended period. Does the error count increment? These examples include a data integrity checking facility to weed out the type of error you're describing.

Also, try plugging the suspect Quixote into a standard desktop PC, using a simple cPCI to PMC adapter. Does the problem track the board?

[eglaser]

Jim,

When it gets in this mode, where all messages from the Quixote have a TypeCode of 0, CpuOutRate doesn't run. I download the COFF, it prints out the normal Loading <...> messages, then "Download Complete", then "Invalid Io message received".

At this point, clicking the Start button doesn't seem to do anything but lock up the app. Whether Data Integrity Checking is on or off, it appears that no data is being transmitted (at least all the status bar indicator are 0).

Then, I have to go into task manager to kill the app.

This is on a develop machine using the cPCI to PCI adapter. I haven't tried looking at whether this happens on other boards; I will as others become available.

Thanks,

Eli

[eglaser]

Jim,

It appears that the problem has to do with reloading the logic (exo).

If I turn on the system, then load the logic, everything works fine. If I then reload the logic, I get the message-typecode-0 problem I described before, and nothing works until I shutdown the computer and reboot.

A simple Windows restart doesn't work, since I suppose the Quixote has power throughout the reset process. I have to actually shutdown the system and then restart it.

Re-reloading the logic doesn't help. It seems that under Quixote-Malibu, the logic file must only be loaded once. Is this the desired behavior? Under the older pre-Malibu setup, you could reload the logic without any problems.

Is there a fix for this? Is there a way to programmatically check whether or not the logic is loaded?

Thanks,

Eli