MITE transfer error with the NI PCIe card

[Éric Piel]

Hi,

We have several systems which experience the same error from time to time. During an acquisition (analog input), the read fails. Then, any following commands to the board fails with "input/output error" (5). The only way to restore the functionality is to power off the computer, wait a few seconds, and power it on again.

On the time of the error, I see such message on the kernel side:

[ 2843.893823] comedi comedi0: mite: transfer error ffffffff
[ 2843.893856] comedi comedi0: ni_tio_handle_interrupt: Gi_Gate_Error detected.
[ 2843.893861] comedi comedi0: ni_tio_handle_interrupt: Gi_DRQ_Error detected.
[ 2843.893867] comedi comedi0: ni_tio_handle_interrupt: Gi_Gate_Error detected.
[ 2843.893870] comedi comedi0: ni_tio_handle_interrupt: Gi_DRQ_Error detected.
[ 2845.107044] comedi comedi0: timeout loading channel/gain list
[ 2845.109086] comedi comedi0: timed out waiting for AO_TMRDACWRs_In_Progress_St to clear
[ 2845.615575] comedi comedi0: timeout loading channel/gain list
[ 2845.617689] comedi comedi0: timed out waiting for AO_TMRDACWRs_In_Progress_St to clear
[ 2846.123539] comedi comedi0: timeout loading channel/gain list
[ 2846.125651] comedi comedi0: timed out waiting for AO_TMRDACWRs_In_Progress_St to clear
[ 2846.631557] comedi comedi0: timeout loading channel/gain list
[ 2846.633673] comedi comedi0: timed out waiting for AO_TMRDACWRs_In_Progress_St to clear

Our systems are x86-64 with Ubuntu 18.04 (Linux 5.4) or Ubuntu 20.04 (Linux 5.13). The errors are quite hard to reproduce. On some systems, it doesn't seem to ever happen, on some others, it's only once in a month, and few systems experience it almost daily. All the boards are NI PCIe-6251.

Any idea of what a "MITE transfer error" is? Is there any way we could recover from it in an easier way? Could it be a sign of a hardware issue? Any hint is welcome!

Best,

Éric Piel

[Ian Abbott]

Hi 'Éric,

On 16/05/2022 15:51, 'Éric Piel' via Comedi: Linux Control and

The "MITE transfer error" is a side effect of reading an unexpected
value from the MITE status register. Reading ffffffff unexpectedly from
a PCI device register is often caused by an unreliable PCI connection,
but in this case I'm not so sure because it is affecting several of your
systems.

--
-=( Ian Abbott <abb...@mev.co.uk> || MEV Ltd. is a company )=-
-=( registered in England & Wales. Regd. number: 02862268. )=-
-=( Regd. addr.: S11 & 12 Building 67, Europa Business Park, )=-
-=( Bird Hall Lane, STOCKPORT, SK3 0XA, UK. || www.mev.co.uk )=-

[Éric Piel]

Hi,

As a follow up, we've struggled, but eventually pinpointed the issue to the PCIe slots of the motherboard. All our systems use the same motherboard. In some cases, one slot doesn't work well with the NI DAQ board (we are still not really sure why), and just changing the slot seems to make the DAQ board reliable again.

Thanks Ian for pointing us towards a hardware issue!

Cheers,

Éric