[PATCH] ppbus/ppc locking

[Bruce M. Simpson]

Bruce M. Simpson wrote:
> John Baldwin wrote:
>> ...
>> Hmm you don't see /dev/lpt0 and /dev/lpt0.ctl via ls(1)?  The attach 
>> routine creates the devices via make_dev() right after the 
>> "interrupt-driven port" message.
>>
>>   
>
> I see /dev/lpt0.ctl, but not /dev/lpt0.

As per IRC convo:
* I instrumented lpt_attach() to clean up if it can't create the cdevs.
* If I then kldload lpt after boot in single-user, and ls /dev, the ls 
process hangs inside devfs_populate_loop().

Something really fishy is going on.

I don't see this issue with devfs if I load lpt from the loader, 
however, lpt0 doesn't appear there either. I still see the 
destroy_devl() panic if I kldunload lpt.

This probably shouldn't be MFCed until this can get sorted out... in the 
meantime, this is my home repo box / smb / print server, need it up, and 
need to fix the total RAID meltdown I experienced yesterday after first 
trying this patch (I don't think this patch is to blame here), so, back 
to work :/