2.6.15rc kernel packages for FC4.

[Thorsten Leemhuis]

Am Donnerstag, den 29.12.2005, 20:59 +0100 schrieb Patrick:
> On Thu, 2005-12-29 at 18:31 +0100, Thorsten Leemhuis wrote:
> > Am Donnerstag, den 29.12.2005, 17:25 +0100 schrieb Patrick:
> > > Yup it works for me but only after a kernel recompile with the DSDT
> > > patch added so that during boot the kernel looks for a new (fixed) DSDT
> > > in the initrd and loads that one. Iirc Dave was not interested in adding
> > > the patch to the kernel for support reasons [...]
> > 
> > That's true afaik. But what IMHO is more important: Upstream doesn't
> > want it, too. See:
> > http://sourceforge.net/mailarchive/message.php?msg_id=9175150
> 
> I respectfully disagree with Len Brown that "DSDT overrides are for
> developers, not end-users, not customers".

Tell him -- a lot of users whine on the fedora-lists that Fedora does
not have the dsdt-in-initrd patch. But I see seldom someone whine ^w
request the inclusion of that patch where it should be done: upstream,
e.g. on acpi-devel and/or lkml.

>[...]
> > Those that have problems with ACPI/DSDT IMHO should fill bugs in the
> > kernel bugzilla to get the damn acpi interpreter fixed. Nothing is
> > getting better in the long term if everybody privately works around
> > problems and uses a patched firmware without fixing the underlying
> > problems. 
> 
> I thought that the DSDT lives in the board's bios. How would a fix to an
> interpreter solve problems if the DSDT itself contains the error(s)?

I thought this is pretty obvious: There is at least one Operating System
out there that works perfectly with the hardware. 

The acpi interpreter on that OS might be to lax and have lots of bugs,
but at least it works (somehow). The acpi interpreter on linux probably
needs to be as lax (with a special boot option if necessary) and
probably even bug-to-bug compatible. That's not nice, but at least it
gives all users (even newbies) working machines. And that should be the
goal for the long term IMHO. 

> E.g. the DSDT on my Acer 4005 with bios 3a23 needs "Z00I" replaced with
> "Z001" in two places. Don't see how an interpreter-fix could solve this
> typo but maybe I totally misunderstand something.

Well, that seems stupid and something like this probably needs to be
fixed by the hardware manufacturer. But the argument still holds: There
is probably at least one Operating System out there that works perfectly
with the hardware, even with this bug. 

> > Or convince the kernel developers that patching the dsdt is the best
> > solution and that the dsdt-in-initrd-patch needs to be merged into the
> > kernel. Maybe then it's getting enabled in Fedora. 
> 
> No idea how to do that.

Just use a different mailinglist (e.g. acpi-devel or lkml) and request
the inclusion of that patch. But I doubt that you'll be successful. But
maybe the right people will notice the problem behind your request and
will try to solve it somehow.

CU
thl