From: Michael Werner <werner@mrcoffee.engr.sgi.com>

This patch adds fb support for new multiple agp bridge agpgart api.
Signed-off-by: Mike Werner <werner@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Michael Werner <werner@mrcoffee.engr.sgi.com>

This patch adds drm support for new multiple agp bridge agpgart api.
Signed-off-by: Mike Werner <werner@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Michael Werner <werner@mrcoffee.engr.sgi.com>

This patch allows drivers to allocate memory local to the bridge using
platform specific alloc_page routines.
Signed-off-by: Mike Werner <werner@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Michael Werner <werner@mrcoffee.engr.sgi.com>

This patch gives non-generic platforms a method for using platform specific
agp_find_bridge functions.
Signed-off-by: Mike Werner <werner@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Dave Jones <davej@redhat.com>

From: Michael Werner <werner@mrcoffee.engr.sgi.com>

This patch adds support for initializing and addressing multiple AGP
bridges using the agpgart driver.  In particular, it extends agp_acquire
and agp_allocate_memory so that different bridges can be acquired and
memory allocated within a specific AGP aperature.

From: Brice Goglin <Brice.Goglin@ens-lyon.fr>

It seems that memsetting the whole bridge structure to 0 (instead of juste
the agp_in_use field) fixes Benoit's problem too.  No idea which field was
responsible for this.  New patch attached.

From: Mike Werner <werner@sgi.com>

add bridge assignment missed in agp_allocate_memory

From: Andi Kleen <ak@suse.de>

Here's the correct fix. agp_bridge is defined in drivers/char/agp/agp.h.
It's a bit ugly though.
Signed-off-by: Mike Werner <werner@sgi.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Dave Jones <davej@redhat.com>

The spec says its valid for multiple bits to be set when we read rate,
however we must take care to ensure that we only write 1 bit back.

This removes some bogus warnings that appeared on some boxes.
Signed-off-by: Dave Jones <davej@redhat.com>

We pass this to remap_pfn_range after shifting it right.
Nasty things happen as a result.
Signed-off-by: Dave Jones <davej@redhat.com>

Somehow the conversion to the extra arg for agp_collect_device_status() got dropped.
Signed-off-by: Dave Jones <davej@redhat.com>

into delerium.kernelslacker.org:/mnt/data/src/bk/agpgart

Fix a dm-raid1 deadlock: nested spinlocks with _irq.
Signed-Off-By: Alasdair G Kergon <agk@redhat.com>
From: Tim Burgess <tim.burgess@anu.edu.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

It feels like 2.6.11 is right around the corner.  I would like to disable
ALPS suport for some devices we don't know how to handle properly yet to
cut down on number of complaints that we broke mouse support.

Input: ALPS - do not activate native mode for devices whose data
       we can not handle yet.
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

A patch that I sent in earlier to allow the use of the data address
breakpoint on machines with a hypervisor happened to break things for those
configs, such as for the Maple board, where we don't compile in the
routines for calling the hypervisor.  This patch fixes it.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Reworking the ppc32 mm helps me find interesting bugs in the existing bug,
well, brown paper bag for me, I made this one a while ago.

The routine flush_hash_one_pte() used by ptep_test_and_clear_young()
wasn't properly recaclulating the vaddr from the pte pointer &
page->index.  The result is that we probably never flushed things from
the hash, so that's at least the _second_ bug affecting
ptep_test_and_clear_young(), swap on ppc32 must have been really broken
:(

This fixes it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Signed-off-by: Dave Jones <davej@redhat.com>

From Nigel Cunningham/Karol Kozimor
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Christophe Lucas <c.lucas@ifrance.com>
Signed-off-by: Dave Jones <davej@redhat.com>

Andi Kleen points out that the intel-agp driver actually works on i865's, where
the -mch driver doesn't.  The -mch driver was something of a failed experiment.
Hoping that the code would be much cleaner, I forked the intel-agp driver into
this variant, and removed some legacy bits. The result was a third the size,
but it still was no work of art, worse yet -- it didn't even do what it said on the can.

intel-agp still supports everything that the -mch driver did, so we can just
remove this from the tree and fall back to the old driver.
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

I've seen this happen a few times, and never got to the bottom of it, so hopefully
this will make things a little easier to diagnose what's going on.
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

We need to clear the previously found cap_ptr, or we exit after
finding the first one, even if its not the one plugged into the
bridge we're looking at.
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Stupid inversion meant we passed '0' to userspace, and madness
ensued resulting in very funky visuals.
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

Should fix the 'cant find AGP VGA controller' warnings.
Signed-off-by: Dave Jones <davej@redhat.com>

In this mode, fast writes are just ignored, and treated as regular PCI
writes, but disabling them explicitly doesn't hurt.
Signed-off-by: Dave Jones <davej@redhat.com>

X munges what we hand to it, so with luck, this will lower the possibility
of it handing us something with reserved bits set.
Signed-off-by: Dave Jones <davej@redhat.com>

Signed-off-by: Dave Jones <davej@redhat.com>

If we got far enough to run this code, we *must* have an AMD bridge.
Signed-off-by: Dave Jones <davej@redhat.com>

Some combinations of NVidia GeForce cards and this AGP chipset cause lockups
when operated in AGP x2 mode. Force them to x1 mode for safety.
Signed-off-by: Dave Jones <davej@redhat.com>

B0/B1 steppings of this chipset were horribly broken,
and couldn't do fast writes, or side band addressing.
Looking through the errata on this one reads like a horror story, it's
a miracle it could get AGP x1 working.
Signed-off-by: Dave Jones <davej@redhat.com>

Allow chipset drivers to tell the generic routines not to enable
certain features if they have errata when those features are used. 
Signed-off-by: Dave Jones <davej@redhat.com>

(Basically a distinction between XFree and Xorg)
Signed-off-by: Dave Jones <davej@redhat.com>

The AGP specifications define the following...

      agp2 agp3
000   BAD  BAD
001   x1   x4
010   x2   x8
011   BAD  x8
100   x4   BAD
101   BAD  BAD
110   BAD  BAD
111   BAD  BAD
Signed-off-by: Dave Jones <davej@redhat.com>

The previous checks were against AGP_STAT values, but userspace passes us
an agp command register.