Closed clepple closed 6 years ago
There is another instance of usb_set_altinterface() in drivers/richcomm_usb.c, but since that device has a suspect USB VID:PID, I am not changing that driver.
@clepple : hum, it seems that bug should have been closed 2 times... and is still opened. can we close it?
@clepple : hum, it seems that bug should have been closed 2 times... and is still opened. can we close it?
@aquette: Not merged yet, because nobody has tested it yet. Want to help? :-)=
tested branch 'issue_138_altinterface_value' and compared against 2.7.1 on Linux: no regression at startup, no upsc diff. This is also valid for #121 Context: Linux kernel 3.14-2-amd64 ; libusb 0.1.12-24 ; UPS Eaton Ellipse MAX I will also do some endurance testing. Other platforms (BSD, Solaris, Windows) would also be worth testing. @balooloo : would you be able to test on Windows? @jimklimov : same for Solaris? @clepple : same for OSX and BSD?
@aquette already tested on Ubuntu, Debian, FreeBSD 9.3 and 10.0, and of course, OS X 10.9 (which drove the need for the patch).
@clepple : you byte me ;) (pun intended)
@aquette I'm taking OpenBSD off the testing list because I only have a VM, and the general state of USB in OpenBSD seems experimental at best. We can test more later, but I'm not going to hold up 2.7.3 because of that.
@clepple : very valid comment ;) @balooloo : if you could just please acknowledge on the feasibility of this test, would be great! I'll try to check on my side for Solaris (FYI @jimklimov )
For future reference, the extra usb_set_altinterface() also prevents an Emerson/Liebert GXT3 from working on Debian Linux:
I'm merging this. Worst case, people can use the driver option to restore the old behavior.
Seems to be an issue on FreeBSD 10.2: http://article.gmane.org/gmane.comp.monitoring.nut.user/9286
Closing since FreeBSD 10.2 is EOL as of 2016-12-31, and this will probably be overtaken by #300 (and maybe related to #490).
This call to usb_set_altinterface() is redundant, and on OS X, prevents tripplite_usb from connecting to an OMNIVS1000:
https://github.com/networkupstools/nut/blob/d7a5c6e5e0f9c0bdbfe4029287dc45b50f2ab207/drivers/libusb.c#L224
I'd just comment it out, but I think we should have a parameter in case some other equipment is relying on this behavior.
Need testing of branch 'issue_138_altinterface_value' on: