11-20-2018 01:31 PM
Random suggestion - have you tried putting a 'get queue status' in front of or in parallel to the dequeue? Very strange..
11-21-2018 12:36 AM
Ty, I haven't but I'll try when as soon as get the chance.
I realize though that I haven't posted about one other attempt to understand the problem:
When the Dequeue was frozen, I ran a VI that obtained this queue by name and enqueued one element. That then released the frozen dequeue.
11-21-2018 03:55 AM
Strange indeed. It behaves like you've wired -1 to timeout. Have we found a bug?
/Y
11-21-2018 01:41 PM
What version of LabVIEW are you using? The behavior sounds a lot like CAR 677003 which was fixed in a 2017 SP1 patch.
http://www.ni.com/product-documentation/54421/en/
11-22-2018 12:55 AM
I'm using LabVIEW 2017 32-bit (17.0.0) so that as well may be the issue. I'll do an experiment with SP1.
Thank you!
11-22-2018 07:01 AM - edited 11-22-2018 07:01 AM
Interesting. We've observed the opposite.
When reading from a DMA channel on RT systems, we wire in timeout -1, non-zero requested number of elements, the node has no data, returns immediately with no data, no error. This is a combination of performance and parameters which should not co-exist.
11-26-2018 09:37 AM
@Intaris wrote:
Interesting. We've observed the opposite.
When reading from a DMA channel on RT systems, we wire in timeout -1, non-zero requested number of elements, the node has no data, returns immediately with no data, no error. This is a combination of performance and parameters which should not co-exist.
Possibly related to CAR 611315? Sounds like similar symptoms but I'm not sure whether DMA FIFOs and Queues share implementation details so you would have to call in to support and have them ask R&D if it's important for your project.
http://www.ni.com/product-documentation/53294/en/
11-27-2018 04:43 AM
I don't think they're linked to that CAR. We only have a single timeout ever wired to our DMA read. IIRC, it was a bug in the RIO drivers which apparently got patched after we reported a way to reproduce the issue.