[hal-info patch] Reintroduce GCC-4240N CD-ROM polling blacklisting

[Sjoerd Simons]

On Wed, Nov 14, 2007 at 11:52:22PM +0100, Martin Pitt wrote:
> Hi Danny,
> 
> Danny Kukawka [2007-11-14 19:30 +0100]:
> > Wouldn't it make more sence to do this only on Dell Laptops and independent 
> > from the blockdevice naming?
> 
> The patch that I attached, is the one that already had been in
> hal/hal-info a while ago. Since I do not have the hardware, I didn't
> change it.
> 
> However, if the bug is in the CD-ROM's firmware, it doesn't matter
> whether it is in a Dell or other machine.
> 
> As for the check for hdb/hdd, I can only speculate. Maybe the issue
> only happens if the drive is a slave? I don't object at all to
> dropping the device name check, that'll make the rule easier.

This thread seems to have died out without good conclusion. Iirc the reason of
the problem isn't strictly that the cdrom drive is a slave, but that the
harddisk and the cdrom drive are on the same bus. Meaning that because of the
issue in the cd drive, the harddisk is prevented from communicating over the
ide bus for some time everytime we poll, hence it gets really slow.

Now one would hope only Dell is so silly/cheap to not have seperate ide busses
for both harddisk and cdrom. But otoh, i wouldn't count on it. So imho we
should either keep it on the safe side and always disable polling for this
cdrom driver or somebody needs to come up with a nice way to express that there
is another drive on the same bus :)

If there are no complaints, i'll commit a patch for this somewhere next week..

  Sjoerd
-- 
Federal grants are offered for... research into the recreation
potential of interplanetary space travel for the culturally disadvantaged.