Re: cvs commit: src/sys/bus/cam cam_sim.c cam_sim.h

[Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>]

:  Do not destroy the device queue, it is needed by the peripheral code
:  (e.g. scsi/scsi_da.c, etc) even if the backend hardware has disappeared.
:  When cam_sim_free() is called, set up a dummy poll and action callback
:  and make the action callback generate an I/O error.
:  
:  This allows things like filesystem mounts of USB memory sticks to start
:  returning I/O errors instead of blocking forever if the stick is pulled.
:  
:  Revision  Changes    Path
:  1.9       +34 -5     src/sys/bus/cam/cam_sim.c
:  1.6       +0 -1      src/sys/bus/cam/cam_sim.h

    I'm still working on this.  umount works with the USB memory stick
    pulled, but the system will panic with 'dangling vnodes' under
    numerous conditions.  I am going to try to track some of them down.
    Theoretically we should be able to make the system robust enough to
    deal (as in not crash) with a pulled stick even if programs are CD'd
    into the mount.

    It would be really cool if the system could 'reconnect' to a USB memory
    stick if it is pulled and then put back in, but that's a project for
    another time (or for someone else to do!)... I don't even know if USB
    memory sticks have working serial numbers in order to figure out that
    the stick inserted is the same one that was ripped out.

						-Matt