- any device teardown must happen between agp_remove_bridge and
  agp_put_bridge, before agp_remove_bridge users can still call into
  the code
- it's releasing a reference to the wrong device

Original patch from: Christoph Hellwig
Further munging from Roland Dreier and myself.
Signed-off-by: Dave Jones <davej@redhat.com>

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

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

agp=off disabled the agpgart driver, but not the chipset specific drivers,
which still did their init routines. Chaos ensued.
Signed-off-by: Dave Jones <davej@redhat.com>

From: Peer.Chen@uli.com.tw
Signed-off-by: Dave Jones <davej@redhat.com>

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

virt_to_page(phys_to_virt(masked_addr)) still works.  But the 460GX
sticks bits in the middle, so the free blows up.

I've tested the agp_allocate_memory() change on Intel 460GX.  I don't have
hardware to test the Intel MCH change, but it looks to me like the same
problem.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Dave Jones <davej@redhat.com>

Most AGP chipset drivers announce the bridge found, and i460gx used
to, but it seems to have gotten dropped somewhere along the way.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Dave Jones <davej@redhat.com>

It's not entirely clear whether this is needed, or it was working around
the missing PCI Posting workaround.  Until we know any better, make
the generic routine do what the per chipset routines do.

Spotted-by: Alan Cox
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>

Also remove a pointless wrapper.
Signed-off-by: Dave Jones <davej@redhat.com>

Calls to change_page_attr() need an explicit call to
global_flush_tlb() afterwards.  The AGP code didn't
do this in a number of cases.  This patch makes
map_page_into_agp/unmap_page_from_agp do the calls
themselves, which takes care of most of the problem.

The Intel AGP driver also has some slightly different calls to what
map_page_into_agp() does, as it changes 4 contiguous pages.
Introduce explicit flushes afterwards there too.

Thanks to Alan Cox for pointing this out.
Signed-off-by: Dave Jones <davej@redhat.com>

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

on_each_cpu does the right thing in the UP case, so we can
kill those ugly ifdefs.
Signed-off-by: Dave Jones <davej@redhat.com>

As pci_find_device is going away soon I have converted this file to use
pci_get_device instead. for_each_pci_dev is just a macro wrapper around
pci_get_device.  I have compile tested it. If anyone has this hardware
and could test it that would be great.
Signed-off-by: Hanna Linder <hannal@us.ibm.com>
Signed-off-by: Dave Jones <davej@redhat.com>

As pci_find_device is going away soon I have converted this file to use
pci_get_device instead. I have compile tested it. If anyone has this hardware
and could test it that would be great.
Signed-off-by: Hanna Linder <hannal@us.ibm.com>
Signed-off-by: Dave Jones <davej@redhat.com>

As pci_find_device is going away soon I have converted this file to use
pci_get_device instead. I have compile tested it. If anyone has this hardware
and could test it that would be great.
Signed-off-by: Hanna Linder <hannal@us.ibm.com>
Signed-off-by: Dave Jones <davej@redhat.com>

As pci_find_device is going away soon I have converted this file
to use pci_get_device instead. I have compile tested it.
If anyone has this hardware and could test it that would be great.
Signed-off-by: Hanna Linder <hannal@us.ibm.com>
Signed-off-by: Dave Jones <davej@redhat.com>

As pci_find_device is going away soon I have converted this file
to use pci_get_device instead. I have compile tested it.
If anyone has this hardware and could test it that would be great.
Signed-off-by: Hanna Linder <hannal@us.ibm.com>
Signed-off-by: Dave Jones <davej@redhat.com>

Unfortunately the restoring of SPE state was causing data corruption
since we were restoring based on the size of the altivec context and not
the SPE context.  Also, fixed setting of last_task_used_spe on
start_thread, flush_thread, and exit_thread. 
Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

It's "const volatile __iomem *". The "const" was missing.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

When adding or deleting from the source list make sure to find matches
by comparing against the new source address, not the group address.
Also, check each addr in the list rather than just the first one.
And, finally, only delete from list when there's a match rather than
vice-versa.  Drop the effort to keep list sorted, since it's not done
on full-state api and can create an sl_addr entry that the delta api
won't be able to delete. Without these fixes sl_count can be corrupted
which can allow for kernel memory corruption.
Signed-off-by: Chris Wright <chrisw@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

into intel.com:/data/home/aegl/BK/linux-ia64-release-2.6.10

- undo the pfm_context_free() call in pfm_handle_work().
  The logic is different. In the case of a controlling
  task exiting while monitored in pfm_handle_work. The
  context is freed in pfm_close() and not pfm_handle_work().
signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

Andrew took the patch that created this, and Linus applied
it to his BK tree.  Remove my copy so that we can merge.
Signed-off-by: Tony Luck <tony.luck@intel.com>

into nuts.davemloft.net:/disk1/BK/net-2.6

- fix pfm_force_terminate() to really cleanup the state.
  There was a bug affecting sessions which use
  PFM_FL_OVFL_BLOCK flag. It could lead to an assertion
  failure in pfm_load_regs() in the case where the
  owner of the session died unexpectedly. The fix is to
  do the full unload operation.

- simplifies the debug output in pfm_flush_pmds()

- simplifies the debug output in pfm_overflow_handler()

- fix a minor typos in a comment in pfm_flush().
signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

Fix ZBOOT_ROM options help texts to describe what's going on there
bettwe.

Make ZBOOT_ROM depend on ZBOOT_ROM_TEXT and ZBOOT_ROM_BSS being
different values.  This allows all*config-created configurations
to work where ZBOOT_ROM_TEXT and ZBOOT_ROM_BSS were both zero.

emard<at>softhome!net discovered an embarrasing bug in dvb_net.c where
the NET_REMOVE_IF ioctl will cause an Oops when called with an invalid
interface number. 
Signed-off-by: Johannes Stezenbach <js@linuxtv.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The hisax driver looks at dev_avm->irq before calling pci_enable_device(),
which means it requests the wrong IRQ.  This patch fixes it.

Thanks to Thorsten Doil for reporting the problem and testing the fix.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Avoid documenting something which isn't there.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Add some configuration information to Documentation/rocket.txt that was
included with the external driver package.
Signed-off-by: James Nelson <james4765@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

I recently learnt that limiting RAM with by using only "mem=xxxM" is no
good on machines equipped with PCI.  In my case (vanilla 2.6.9) the cardbus
bridge on my laptop got mapped to the unused RAM area which resulted in
wierd errors due to the collision.

The right solution is to use "mem=" together with "memmap=" to mark the
unused RAM area reserved.

Henceforth I force the kernel to use 2016MiB by passing "mem=2016M
memmap=32M#2016M" instead of just "mem=2016M".
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

- av7110: fixed av7110_before_after_tune()/av7110_fe_lock_fix(): firmware >=
  261d: wait for empty message queue, firmware <= 261c: wait 50ms

- av7110: add __user and __iomem annotations, remove some unnecessary cast
  (patch by C.Y.M)

- av7110: __av7110_send_fw_cmd(): added some sanity checks suggested by
  Werner Fin

- av7110: added support for full-featured DVB-C cards: 13c2:0000 Siemens
  DVB-C (full-length card) VES1820/Philips CD1516 and 13c2:0003 Haupauge DVB-C
  2.1 VES1820/ALPS TDBE2

- av7110: follow saa7146 changes, remove superflous casts, and other misc. 
  minor cleanups

- av7110: Fixed race condition between driver and av7110 while accessing the
  COMMAND register in DPRAM.  See
  http://www.linuxtv.org/mailinglists/vdr/2004/01-2004/msg00331.html

- budget: various cleanups by Adrian bunk 
Signed-off-by: Michael Hunold <hunold@linuxtv.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>