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

Re: Odd problems with FC6 kernel



On Fri, 2007-08-03 at 14:34 -0700, alan wrote:
> I am not certain if this is a problem with FC6 or my machine.
> 
> The machine is a 3ghz Intel hyperthread cpu on an Intel motherboard.
> 
> I have a dvd burner on /dev/hda.
> 
> After burning dvds for a while, burning will wedge.  There will be a 
> message in /var/log/messages about a soft lockup on cpu 1.  The system 
> still runs, but the dvd drive no longer works and I get tons of "drive not 
> ready" errors until I reboot.
> 
> Disabling hyperthread does not help.  i get the same problem on CPU 0 and 
> the system is MUCH harder to use until I reboot.
> 
> Not certain if this is the kernel being wonky or my machine.  Has anyone 
> seen this?
> 
> I have two S-ATA drives and 4 IDE drives on the machine.  The DVD drive is 
> an IDE drive.  (There were not any SATA burners when I bought the drive. 
> At least, not where I got it.)
> 
> I have also seen problems when there is a lot of drive activity.  Things 
> slow down a lot.  Much more than they did with earlier kernels.  Part of 
> this was Beagle.  (Which I removed.)  It wanted to index EVERYTHING.  The 
> index was 1.7gig.
> 
> Ideas?  Should this go into bugzilla?  What other details are needed?

Note that many CD and DVD burners have duty cycle limits in that the
burning process uses a significantly larger amount of power which causes
the units to heat up.  Most of them will cease functioning until they
cool down or get power cycled/reset.  During this time, they'll return
"not ready" status to the controller.  Most of the Linux burning tools
will query the drive many times a second and log sh*tpots of those
messages and will enter a "D" state (I/O wait state), making them
difficult (or impossible) to kill.

I worked for a company that made a 100 CD autochanger with up to 4
drives.  We ran into this when we swapped to CD burners to create a
SOHO CD/DVD production system.  Our work around was to put a huge great
wonking fan in the box that blew directly on the drives.  It solved the
problem as long as the ambient temperature was below 80 degrees
Farenheit.  Above that we didn't get as many lockups, but it still
would lock up on occasion.

I suggest that you wait a few minutes (maybe 5 or 10) between burning
sessions and see if the drive stops locking up.  If so, the odds are
that you're hitting that duty cycle and you just have to pace yourself
accordingly.

> "ANSI C says access to the padding fields of a struct is undefined.
> ANSI C also says that struct assignment is a memcpy. Therefore struct
> assignment in ANSI C is a violation of ANSI C..."
>                                    - Alan Cox

And Alan was wrong.  _USER_ access to the padding fields is undefined.
Internal code generation can pretty much do what it wants.  This is
called "an implementation detail".  (sorry, it's Friday and I'm punchy!)

----------------------------------------------------------------------
- Rick Stevens, Principal Engineer             rstevens internap com -
- VitalStream, Inc.                       http://www.vitalstream.com -
-                                                                    -
-              Death is nature's way of dropping carrier             -
----------------------------------------------------------------------


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