Interpretting Post Crash Logs

I have a BeagleBone Angstom image that basically polls MODBUS/TCP data on the network from a PLC using a Python script. This image has been run in both the BBB and BBW and we are seeing a periodic crash after a few days of operation if the flash card is branded by Sandisk. If the exact same image is run on a flash card from Transcend we do not see crashes. The kernel is 3.8.13.

The journal includes some lines like shown below about the time of the crash. I would like some help/recommendations about what this means.

Also based on your experience are Transcend cards better? They claim to have ECC features and I cannot find any such claim on the Sandisk cards.

Is ECC a good thing for these flash cards?

Mar 14 14:15:09 beaglebone kernel: gadget: sending command-failure status
Mar 14 14:15:09 beaglebone kernel: gadget: REQUEST SENSE is buggy! Expected length 6 but we got 12
Mar 14 14:15:09 beaglebone kernel: gadget: sending command-failure status
Mar 14 14:15:09 beaglebone kernel[123]: [272306.161158] gadget: sending command-failure status
Mar 14 14:15:09 beaglebone kernel[123]: [272306.161524] gadget: REQUEST SENSE is buggy! Expected length 6 but we got 12
Mar 14 14:15:09 beaglebone kernel[123]: [272306.162365] gadget: sending command-failure status
Mar 14 14:15:09 beaglebone kernel[123]: [272306.162932] gadget: REQUEST SENSE is buggy! Expected length 6 but we got 12
Mar 14 14:15:09 beaglebone kernel[123]: [272306.164480] gadget: sending command-failure status
Mar 14 14:15:09 beaglebone kernel[123]: [272306.165168] gadget: REQUEST SENSE is buggy! Expected length 6 but we got 12

Thanks,

Randall