DragonFly commits List (threaded) for 2009-08
DragonFly BSD
DragonFly commits List (threaded) for 2009-08
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: git: disk/fd - fix panics


From: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
Date: Mon, 17 Aug 2009 08:29:18 -0700 (PDT)

:Matt,
:
:as you are calling setdiskinfo from inside the attach routine, when
:device_busy is called, the device isn't attached yet.
:While the probing is asynchronous, maybe it still occurs too early,
:before finalizing the attach?
:
:Sincerely,
:Alex Hornung

    Hm..  I was calling it at the end of the attach code, but I think
    I made a fatal assumption.  I assumed that the probe code would
    open the base device.  But it doesn't, does it!  I think it just
    uses the raw device passed to disk_create()!

    Most of the floppy driver's real probe code is in its open()
    function.  If the function does not get called then it will
    blow up.

    Hmm.  We don't want to do the open synchronously in the probe code.
    What if we added a DSO_ option to tell the devfs/disk subsystem that
    it must dev_dopen() the raw device for the probe operation?

						-Matt




[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]