Earlier cleanup efforts of the ppc64 IOMMU code have mostly been targeted
at simplifying the allocation schemes and modularising things for the
various platforms.  The IOMMU init functions are still a mess.  This is an
attempt to clean them up and make them somewhat easier to follow.

The new rules are:

1. iommu_init_early_<arch> is called before any PCI/VIO init is done
2. The pcibios fixup routines will call the iommu_{bus,dev}_setup functions
   appropriately as devices are added.

TCE space allocation has changed somewhat:

* On LPARs, nothing is really different. ibm,dma-window properties are still
  used to determine table sizes.
* On pSeries SMP-mode (non-LPAR), the full TCE space per PHB is split up
  in 256MB chunks, each handed out to one child bus/slot as needed. This
  makes current max 7 child buses per PHB, something we're currently below
  on all machine models I'm aware of.
* Exception to the above: Pre-POWER4 machines with Python PHBs have a full
  GB of DMA space allocated at the PHB level, since there are no EADS-level
  tables on such systems.
* PowerMac and Maple still work like before: all buses/slots share one table.
* VIO works like before, ibm,my-dma-window is used like before.
* iSeries has not been touched much at all, besides the changed unit of
  the it_size variable in struct iommu_table.

Other things changed:
* Powermac and maple PCI/IOMMU inits have been changed a bit to conform to
  the new init structure
* pci_dma_direct.c has been renamed pci_direct_iommu.c to match
  pci_iommu.c (see separate patch)
* Likewise, a couple of the pci direct init functions have been renamed.
Signed-off-by: Olof Johansson <olof@austin.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch renames pci_dma_direct.c to pci_direct_iommu.c to comply to the
naming convention of the other iommu files.

This is part of the iommu cleanup, but broken out as a separate patch since
for mainline, a BK rename is more appropriate.  Still, we need a patch to
apply for non-BK-based trees (-mm)
Signed-off-by: Olof Johansson <olof@austin.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch is from Jake Moilanen <moilanen@austin.ibm.com>.

Log machine check errors to error log and NVRAM.
Signed-off-by: Dave Altobelli <dalto@austin.ibm.com>
Signed-off-by: Jake Moilanen <moilanen@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch is from Jake Moilanen <moilanen@austin.ibm.com>.

Changed the naming conventions for head.S to more closely follow the Linux
naming conventions.
Signed-off-by: Jake Moilanen <moilanen@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch is from Jake Moilanen <moilanen@austin.ibm.com>.

Clean-up of traps.c.  Moved the machine dependent calls to a ppc_md call,
and moved the pSeries specific code to ras.c.

I also changed the naming convention to more closely follow the Linux
standards.
Signed-off-by: Jake Moilanen <moilanen@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch is from Nathan Fontenot <nfont@austin.ibm.com> originally.

The PPC64 EEH code needs a small update to start using the
ibm,read-slot-reset-state2 rtas call if available.  The currently used
ibm,read-slot-reset-state call will be going away on future machines.

This patch attempts to use the newer rtas call if available and falls back
the older version otherwise.  This will maintain EEH slot checking
capabilities on all future and current firmware levels.
Signed-off-by: Nathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Most special purpose registers on the ppc64 have both the SPR number, and
the various fields within the register defined in asm-ppc64/processor.h.
So far that's not true for the performance counter control registers, MMCR0
and MMCRA.  They have the SPR numbers defined, but the internal fields are
defined in the oprofile code and (just a few) in traps.c where they're
actually used.

This patch moves all the MMCR0 and MMCRA definitions, plus the MSR
performance monitor bit, MSR_PMM, into processor.h.
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>

OOM kills have been observed with 70% of the pages in lowmem being in the
writeback state.  If we count those pages in sc->nr_scanned, the VM should
throttle and wait for IO completion, instead of OOM killing.

(akpm: this is how the code was designed to work - we broke it six months
ago).
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch is required to fix compile errors for m32r.

This was originally given by the following patch:
  [PATCH] move irq_enter and irq_exit to common code
  http://www.ussg.iu.edu/hypermail/linux/kernel/0411.1/1738.html

I think it was maybe accidentally dropped only for the m32r arch due to a
patching conflict with the other patches or something like that.
Signed-off-by: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

I'm being at least sometimes deferred to for hugetlb maintenance.
I also originally wrote the fs methods, and generally get stuck
working on it on a regular basis. So here is a MAINTAINERS entry
reflecting that.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

oprofile m32r arch updates, including some API changes.
Signed-off-by: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Kconfig symbols are always uppercase.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Acked-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

clear_page_range()/4-level page table updates.

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

Patch from Dave Jiang

Signed-off-by: Dave Jiang

This is the IQ80332 platform port that's based off the IOP33x CPU. The IQ80332 is an PCI-express CRB based off the IOP332 processor. Otherwise functionalites are fairly similar to IQ80331.
Signed-off-by: Russell King

Patch from Dave Jiang

Signed-off-by: Dave Jiang

Major cleanup of the 321 PCI defines to make them more coherent. Unified some groups that were per platform to common proc specific. Removed some magic numbers.
Signed-off-by: Russell King

Patch from Dave Jiang

Signed-off-by: Dave Jiang

Substitute 2 asm instr w/ one recommend by RMK.
Signed-off-by: Russell King

Patch from Deepak Saxena

The IXP46x chips from Intel include the same I2C controller as found
on the IOP3xx.  This patch adds a hook to the platform-level init code
to add that device to the device tree if running on a 46x CPU. A
separate patch has been sent upstream via Greg KH for the changes to the
I2C driver to support both IOP and IXP systems.

While at it, we fix a minor issue with the code where we were sending 
the wrong ptr to platform_add_devices(). I am not sure how it ever 
worked...

Signed-off-by: Deepak Saxena
Signed-off-by: Russell King

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

Since we no longer pass a struct request to the MMC host drivers,
it seems pointless including linux/blkdev.h.  Instead, include
linux/highmem.h and asm/scatterlist.h as appropriate.
Signed-off-by: Russell King <rmk@arm.linux.org.uk>

The regular executable loader path doesn't need the locking,
because it's the only user of its VM. But the same is not true
at library load time. So get the mmap semaphore.

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

On x86-64, the attached patch is required to fix

> kernel/sys.c: In function `sys_setsid':
> kernel/sys.c:1078: error: `tty_sem' undeclared (first use in this function)
> kernel/sys.c:1078: error: (Each undeclared identifier is reported only once
> kernel/sys.c:1078: error: for each function it appears in.)

kernel/sys.c needs the tty_sem declaration from linux/tty.h.

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

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

* add drive->media != ide_disk check to ide_abort()
* kill ide_cdrom_abort() and idedisk_abort()
* split __ide_abort() out of ide_abort()
* call driver->abort() inside ide_abort()
* convert the only user of driver->abort()
* fix default_abort() and idescsi_atapi_abort()
* make idescsi_atapi_abort() static
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

* split __ide_error() out of ide_error()
* call driver->error() inside ide_error()
* convert all users of driver->error()
* fix default_cleanup() and idescsi_atapi_error()
* make idescsi_atapi_error() static
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

* move idedisk_error() and ide_cdrom_error() to ide-io.c
* rename idedisk_error() to ide_ata_error()
* rename ide_cdrom_error() to ide_atapi_error()
* use ide_{ata,atapi}_error() in ide_error()
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

* move idedisk_dump_status() and ide_dump_status() to ide-lib.c
* rename idedisk_dump_status() to ide_dump_ata_status()
* use ide_dump_{ata,atapi}_status() in ide_dump_status()
* use ide_dump_status() in ide-cd.c, ide-disk.c and ide-scsi.c
* make ide_dump_opcode() and ide_dump_atapi_status)() static
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

* add it to ide-lib.c and cleanup ide{disk}_dump_status()
* as a bonus it fixes unknown opcode reporting in ide_dump_status()
* use ide_dump_opcode() in ide_dump_atapi_status()
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

also kill default_sense() in ide.c
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

From: Jim Paris <jim@jtan.com>

Bartlomiej: I have an ICH3-M controller on my laptop.  The BIOS is
leaving the prog-if as 0x8E (primary = legacy, secondary = native).
When the PCI interrupt is routed (either in the IDE driver's
pci_enable_device, or earlier if pci=routeirq is used), unhandled
interrupts cause IRQ 9 to be disabled, breaking most of my other
hardware.  This seems to be caused by having the nonexistant secondary
interface set to native mode.  According to the datasheet I checked,
having different modes for primary/secondary is not an allowed
combination anyway, so the following PCI quirk checks for this case
and forces both interfaces to legacy if true.

It may make sense to make this more generic (this problem may affect
other PCI IDs as well), or it may be better solved in the IDE driver,
at least when pci=routeirq is not used.  But the following patch does
work well for me.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

From: Prarit Bhargava <prarit@sgi.com>

Please see discussion related to this patch here,

http://marc.theaimsgroup.com/?l=linux-ide&m=110485007824374&w=2Acked-by: Jens Axboe <axboe@suse.de>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

Ben did not disagree to hide the 3 additional lines printed when booting
my ibook with the 'quiet' cmdline option.
I think these debug printk have no real value for normal operation.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

Reported to work OK by Daniel Robitaille <robitaille@gmail.com>.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

On Mon, 03 Jan 2005 21:44:33 +0000, Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
>
> Different PCI functions are but nothing requires that the PCI function
> that is the IDE controller is only the IDE controller. In some cases
> other logic lives in the "spare" BAR register areas of the device.
>
> One example where the weird design makes it obvious is the CS5520. Here
> the 5520 bridge has the IDE in one BAR and all sorts of other logic
> (including the xBUS virtual ISA environment) in the same PCI function.
> On that chip a pci_disable_device on the IDE pci_dev turns off mundane
> things like the timer chips keyboard and mouse 8).
>
> Other vendors do equally evil things and providing the chip reports IDE
> class and has the IDE BARs set up nobody else is any the wiser and
> presumably gate count goes down.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>

Oh, bugger...

The recent mmzone.h change triggered some very, very ugly things in
next_zone(), is_highmem(), is_normal() and zone_idx() because it changed
the size of "struct zone" subtly.

Pointer subtraction is no fun when sizeof of object gets weird and poor
gcc blows its brains out trying to optimize that...  This slowed down an
ARM cross-build by a factor of 2.5. 

Turn the pointer subtraction tests into pointer addition tests instead
where possible, making them trivial to deal with (and gets better code,
in addition to faster compile).

zone_idx() still does a subtraction.