On 07/01/2009 01:42 PM, Frédéric Boiteux wrote:
Le Sat, 27 Jun 2009 06:48:03 +0000, owner@bugs.debian.org (Debian Bug Tracking System) a écrit :Differences are no bug, but usually bugfixes. BastianLe Mon, 29 Jun 2009 22:17:54 +0100, Ben Hutchings<ben@decadent.org.uk> a écrit :libata is a library module, not a driver. Please explain what the actual problem is rather than pointing at changes that may or may not be related.Hello, This problem of some PATA hardware driven by piix driver instead of ata_piix is in Debian since a long time (see also bug # 444182) : in the SVN web interface, I see it first 8499, 2 years ago, in Etch : http://svn.debian.org/wsvn/kernel/dists/trunk/linux-2.6/debian/patches/debian/drivers-ata-ata_piix-postpone-pata.patch?op=log&rev=0&sc=0&isdir=0 But I didn't find why it have been added, which bug it fixes : does anybody have info about this ? Is it reported at kernel.org ? Perhaps it has been fixed since the addition of this patch ? Thanks for any tip, Fred.
I can think of one reason to exclude these PCI IDs - if the libata piix driver is used it changes the device name.
I recall the device name changing more than once so there was probably some time when the driver was built, then people complained it changes drive names and so it was disabled.
However, the right way to disable it is to not include the driver in the Debian kernel config, not to break it in the source so that it cannot work.
Also we have UUID support by now so this issue is moot. It is also inconsistent with other libata based pata drivers which do not seem to be crippled this way.There is one other possible issue which might have caused this patch. In the early days of libata there was a warning that compiling drivers for the same device from both ide and libata subsystems may cause trouble but this too should be resolved by now.
Either way the patch is incorrect and it breaks functionality available in the upstream kernel for no good reason so it should be reverted.
Thanks Michal