On Thu, Apr 29, 2004 at 09:14:15AM -0700, David Brownell wrote:
> That's going to be painful in 2.4 kernels; usbcore itself doesn't
> guarantee enough, plus there are similar issues in SCSI and other
> areas. On the other hand, 2.6 is a lot better already, and folk
> are trying to chase down the i
Glenn Maynard wrote:
We're trying to use USB memory sticks in a "hostile" environment, where
devices may be plugged and unplugged unexpectedly, and we need to be
sure that the kernel and our application won't crash as a result.
That's going to be painful in 2.4 kernels; usbcore itself doesn't
guar
At the cost of replying to myself, we found that this actually reproduces
very quickly and reproducably while running a simple sh script, busy
looping on "cat /proc/bus/usb/devices > /dev/null". This eliminates the
variables from the app running. This also happened off a clean boot, with
only usb
We're trying to use USB memory sticks in a "hostile" environment, where
devices may be plugged and unplugged unexpectedly, and we need to be
sure that the kernel and our application won't crash as a result. We've
hit crashes doing a straightforward test: plugging and unplugging one
or more memory
