kern/15608: acd0 / cd0 give inconsistent errors on empty tray open()

[Dan Strick]

>>
> Synopsis: acd0 / cd0 give inconsistent errors on empty tray open()
>
> State-Changed-From-To: feedback->closed
> State-Changed-By: ken
> State-Changed-When: Mon Jan 26 11:59:35 PST 2004
> State-Changed-Why:
> Bruce said that ENXIO was the correct error to return, and we do return
> that when media isn't present.  Also, I believe that the issues that Bruce
> brought up about the CDIOCCLOSE ioctl, etc., have been fixed.
>
>
> Responsible-Changed-From-To: scsi->ken
> Responsible-Changed-By: ken
> Responsible-Changed-When: Mon Jan 26 11:59:35 PST 2004
> Responsible-Changed-Why:
> Bruce said that ENXIO was the correct error to return, and we do return
> that when media isn't present.  Also, I believe that the issues that Bruce
> brought up about the CDIOCCLOSE ioctl, etc., have been fixed.
>
> http://www.freebsd.org/cgi/query-pr.cgi?pr=15608
>>

ENXIO would be historically the correct error to return for nonexistant
disk media, including an attempt to begin a read/write one or more sectors
beyond the end of a disk.  This should also be the error returned for an
attempt to begin a read one or more sectors beyond the last written sector
of a write-once medium (e.g. CD-R) though this might be difficult situation
to recognize.  It might be the correct error to return whenever a read
produces a "blank check" though one might sometimes make a good argument
for EIO in that situation.

Of course, a read that begins at or before the end of the medium but extends
past the end should simply result in a short read.  I mention this because
it works on FreeBSD with CD-ROM but not with CD-R.  I don't think I have
tested this on the most recent FreeBSD releases so it might have been fixed.
(This glitch, present in both scsi and atapi cd drivers, also triggered a
bug in ata/atapi dma code that the ata driver eventually resolved by timing
out waiting for dma to complete and disabling dma for the device.)

Dan Strick
strick at covad.net