[Nut-upsuser] libusb_get_interrupt: error reading from interrupt endpoint /dev/ugen0.1

[Charles Lepple]

On Mar 24, 2011, at 7:46 AM, John Bayly wrote:

> On 24/03/2011 02:17, Charles Lepple wrote:
>> On Mar 23, 2011, at 12:09 PM, John Bayly wrote:
>>
>>> Using an Eaton Evolution S 3000 (usbhid-ups) 2.6.0 on FreeBSD 7.3.  
>>> After checking the debug log for an unrelated reason I saw that it  
>>> was being swamped by the following message:
>>
>> The USB stack in FreeBSD prior to 8.0 did not lend itself to being  
>> precisely controlled from userspace by libusb. It sounds like you  
>> are seeing slightly different problems than I saw while trying to  
>> make tripplite_usb work under FreeBSD, but the common symptom was  
>> the interrupt endpoints. (FreeBSD prior to 8.0 seems to poll the  
>> endpoints more frequently than userspace code reads from them, and  
>> I often saw partial packets dropped when the buffer filled up.)
> That was kind of what I'd read. Unfortunately I'm stuck with a pre  
> 8.0 release for the moment due to ISDN.

Understood.

>>
>> How do things work if you add the "pollonly" option to the  
>> configuration file?
> I've just tried this, and the USB debug messages stopped on restart.  
> Really appreciate the help.

Be sure that the measured values shown in upsc are updating as well.  
With the "pollonly" flag, they should be explicitly requested via  
control messages, and you should see error messages if those control  
messages fail, but that's making a lot of assumptions about the USB  
stack.