re: CVS commit: src/usr.sbin/usbdevs
> On further thought, I think your patch won't work correctly in cases > where USB controllers are detached. For example, if you attach usb0 > and usb1, and then detach usb0, getusbcount_device() will return 1, > and usbdevs will try to list devices only on the detached usb0, and > not on the still attached usb1. good point. i'll re-think...including your suggestion about notifying the user about missing devices. thanks! .mrg.
re: CVS commit: src/usr.sbin/usbdevs
Matthew, On further thought, I think your patch won't work correctly in cases where USB controllers are detached. For example, if you attach usb0 and usb1, and then detach usb0, getusbcount_device() will return 1, and usbdevs will try to list devices only on the detached usb0, and not on the still attached usb1. -- Andreas Gustafsson, g...@netbsd.org
re: CVS commit: src/usr.sbin/usbdevs
Hi mrg You wrote: > > Look for up to 16 USB controllers, to match the number of device nodes > > now created by MAKEDEV. > > thanks for doing this. you inspired me to get usbdevs to get the > count from drvctl(8) if available, so next time we don't have to > patch it again. based upon the code in drvctl(8). > > what do you think? > >https://www.netbsd.org/~mrg/usbdevs.anycount.diff Thanks for improving on my commit. Your patch looks correct as far as I can see, but rather complicated for what it does. Couldn't we get the same end result simply by looping until opening /dev/usbN yields ENOENT? Or alternatively, if we do spend the effort to walk the device tree, we should use the information gained from that to full advantage by issuing warnings about missing /dev/usbN device nodes rather than just silently ignoring the affected controllers. -- Andreas Gustafsson, g...@netbsd.org
re: CVS commit: src/usr.sbin/usbdevs
"Andreas Gustafsson" writes: > Module Name: src > Committed By: gson > Date: Sat Sep 21 16:22:25 UTC 2019 > > Modified Files: > src/usr.sbin/usbdevs: usbdevs.c > > Log Message: > Look for up to 16 USB controllers, to match the number of device nodes > now created by MAKEDEV. thanks for doing this. you inspired me to get usbdevs to get the count from drvctl(8) if available, so next time we don't have to patch it again. based upon the code in drvctl(8). what do you think? https://www.netbsd.org/~mrg/usbdevs.anycount.diff thanks. .mrg.