Bug#508834: Processed: change status

[Jason White]

Felix Zielcke <fzielcke at z-51.de> wrote:
> Well there's even a comment before the part of the code which uses this
> ioctl:
>   /* Linux has a bug that the disk cache for a whole disk is not consistent
>      with the one for a partition of the disk.  */
> 
> so I think mainly the kernel needs to be fixed or the ioctl needs to be
> otimized.
> Or maybe someone should find out if that bug is still there with 2.6.30
> That comment is from 2003.

In addition to this, we need to confirm whether these ioctl() calls are causing
the enormous CPU load experienced by those of us who have been affected by
this bug, and if so, find out why this is happening.

So far, I haven't seen enough detail from other reporters of the bug to
suggest any common cause. In my case, the problem affects my desktop system:
/dev/sda is fake-RAID using the Fusion MPT SAS Host driver, with XFS as the
file system.

On a laptop, also with an XFS file system but a SATA drive, and with two
rather than four CPU cores, grub-probe returns quickly during package upgrades
and without causing high load.

Both systems are running Debian 2.6.30-5 x86_64 kernel images.