Regression in 2.6.39-rc1 from commit 7eaceaccab5f40bbfda044629a6298616aeaed50

[Larry Finger <Larry.Finger@lwfinger.net>]

After I pulled 2.6.39-rc1, my HP laptop failed to boot by never advancing past 
the device detection stage on my x86_64 system. The problem was bisected, which 
led to:

7eaceaccab5f40bbfda044629a6298616aeaed50 is the first bad commit
commit 7eaceaccab5f40bbfda044629a6298616aeaed50
Author: Jens Axboe <jaxboe@fusionio.com>
Date:   Thu Mar 10 08:52:07 2011 +0100

     block: remove per-queue plugging

     Code has been converted over to the new explicit on-stack plugging,
     and delay users have been converted to use the new API for that.
     So lets kill off the old plugging along with aops->sync_page().

     Signed-off-by: Jens Axboe <jaxboe@fusionio.com>

As this patch cannot be reversed on current 2.6.39-rc1-git, I have not been able 
to do the final test of reverting it, but the bisection was straight forward.

When the system hangs, I can get it to finish booting if I type CTRL/C; however, 
the only disk partition that has been mounted is /. If it matters, I use a 
mixture of ext3 and ext4 file systems.

My disk controller is "nVidia Corporation MCP67 IDE Controller [10de:0560] (rev a1)"

Please let me know what additional info is needed.

Larry