update to "OpenBSD 5.0-current (GENERIC) #109: Wed Nov 30 17:27:38 MST 2011" to include several diffs to ucom(4).
now instead of flooding the message buffer with "ucomstart: null oxfer", I got the concise message "uvm_mapent_alloc: out of static map entries" and my program returns with "i2c: open(): Input/output error: /dev/cuaU0". running 'usbdevs -v' does not cause a panic and I can 'safely' unplug the usb device without another panic. I consider this an improvement! now when I unplug/plug the device I get the system message "uhub1: port 2, set config at addr 2 failed" \n "uhub1: device problem, disabling port 2" and 'usbdevs -v' does not detect anything on that port. rebooting the system brings me back to normal. it took about ~ 1 day to trigger the usb errors from within a shell script. the intent with this i2c program is to interact with some analog stimuli (once the bugs are worked out) but for now this just sits opening, writing, reading and then closing a filedesc to my usb port... if this is 'bad practice' I'd like to know so I can revisit the program again. thanks to all who replied! On Thursday, December 01, 2011 12:58 PM, "Stuart Henderson" <[email protected]> wrote: > On 2011-11-30, Daniel Gracia <[email protected]> wrote: > > At the very least you're seeing some errors. In my case, the USB/serial > > adapters -uticom, uftdi and uplcom- would fail without notice. Ports > > would open, but with no TX/RX. Detaching/reattaching won't bring them > > back to live; only rebooting. > > > > If your project has deadline, search for a PCI/ePCI serial board and enjoy! > > careful with some of the PCIe boards though, OXPCIE95x aren't supported > by > the current puc(4) driver (it's fairly easy to get them partially working > but > the port speeds will be wrong and you won't get access to certain > speeds). > > Byron Klippert [email protected] (867) 332-4184
