On Fri, 27 Sep 2002, Yuri Per wrote: > disconnect() and probe() should be simulated only when > usb_reset_device() result was successful, as it is done now. > > The only change is that scsi layer now allowed to make later one more > attempt to perform the request.
It doesn't seem right to return a code indicating success when in fact we don't know whether the device reset worked or not. Perhaps the bus_reset() routine in scsiglue.c should wait for a short period and then try to determine if the device was properly reset (how?) before returning or continuing. If something is really wrong with the device, and we keep telling the scsi layer that it has been reset successfully when in fact it has not, then we run the risk of getting caught in an endless INQUIRE - fail - reset - succeed loop. The only reason I can think of for doing as you suggest would be if we are truly unable to tell whether the device was reset, regardless of the return code from usb_reset_device(). Alan ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
