Greg - was able to do some additional testing here on Tuesday, but wasn't able to post until now. Glad to see you guys were able to reproduce the issues in some respects. My findings below in context of yours.
Quote:
Originally Posted by Greg McKaskle
If the Cypress is plugged into an external hub, it will fail to enumerate about half the time -- red light still on. Usually, moving the Cypress to be plugged directly into the Classmate immediately fixed the problem -- red light goes off -- with no need for reboot or DS restart.
|
This is in line with our observations as well, more or less.
Quote:
If the Cypress is plugged directly into the Classmate port, the I/O board would occasionally fail to be recognized on a cold boot -- approximately 1 in 8 would fail.
When this failed, the CyMiniProg Service was always running, but restarting it always fixed the issue.
|
While we were unable to run a significantly large number of tests due to time constraints, I can report that moving the device to being directly plugged in also has greatly increased success rate here - we didn't have any failures in the handful of cold boots that we did.
Quote:
|
If the Cypress is plugged directly into the Classmate and in a recognized state, we saw no failures when suspended and reawakened.
|
To further this... we did a lot of mixed testing of suspend (as this is the primary use case of concern for us). That is, we did not cold boot when changing the configuration. Simply moved where the device was plugged in, waited for it to be recognize, and then started doing suspend/wake tests.
When directly plugged into the left-side classmate port we didn't see any failures suspending/awakening. When plugged into the hub, we saw intermittent failures.
We've since modified our physical enclosure to permit direct connection.
Unfortunately I was unable to gather any data on behavior with no loads connected to the Cypress because unhooking all the devices would've taken more time than we had to test.