[regression] CD-ROM polling blocks suspend on some machines (Re:[PATCH 1/2] cdrom: always check_disk_change() on open)

[Jonathan Nieder]

Hi,

Various people[1] have been noticing a race (or races?) in which the
cdrom_id and scsi_id programs from udev can get stuck in the D state
if udev doesn't sleep a little while before running them.  This
prevents the machine from suspending.  The problem was discovered with
Debian kernels

  2.6.39-1
  2.6.39-2
  3.0.0-1

and was not experienced with kernel

  2.6.38-5

(These kernels are closely based on v2.6.39, v2.6.39.1, v3.0, and
v2.6.38.5, respectively.)  Reverting commit bf2253a6f00e (cdrom:
always check_disk_change() on open, 2011-04-29) seems to avoid
trouble.

One common theme seems to be DVD drives.  Details at [1].

Known problem?  Any ideas for tracking it down?

Looking forward to your thoughts,
Jonathan

[1] <http://bugs.debian.org/628600>.  Submitters cc-ed.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/