NI_PCIMIO interrupts triggering ehci_hcd:usb2 interrupts when performing AI acquisition

[Christopher Braga]

Hello,

When commanding two PXI-6229 cards to perform asynchronous AI acquisition it has been observed that the interrupts sent by the ni_pcimio driver are causing the ehci_hcd:usb2 interrupts to trigger as well. 

When monitoring the interrupts with Tuna (a tool that allows adjusting of scheduler policy, RT priority, and CPI affinity) it was observed that the ehci_hcd:usb2 interrupts would rapidly trigger at a rate of 20000 events per second when AI acquisition was started, and eventually cap out around 200 000 or 300 000 interrupts. As the NI cards in use are connected to our machine through a PXI chassis, it is not clear why the USB interrupts are triggering at all. Using Comedi without AI acquisition doesn't cause the ehci_hcd:usb2 interrupts to trigger, so the issue is definitely linked to the AI functionality.

The Comedi command structure I am sending to each card specifies that all 32 channels on a card should be sampled until comedi_cancel is called, and the TRIG_WAKE_EOS flag is set in order to allow samples to be quickly retrieved from the card. I currently only have access to a PXI-6229 card, so I am unable to determine if this behaviour is unique to this driver and/or card.

Any help on this matter would be greatly appreciated.

Thanks,

Christopher Braga

[Arron McDaniel]

5

[Arron McDaniel]

5

On Friday, May 12, 2017 at 9:09:40 AM UTC-6, Christopher Braga wrote: