The xics code does update the default server information when the boot
cpu is removed.  This patch recognizes when the boot cpu is being
removed and updates the appropriate information based on the new 'boot
cpu'.

Failure to update this information can causes us to leave irqs pinned
to cpus that are being removed, especially when removing the boot cpu.
The cpu is removed from the kernel, but cpu dlpar remove operations
fail since we cannot return the cpu to the hypervisor.
Signed-off-by: NNathan Fonteno <nfont@austin.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

It appears that xics.c has its own of_get_cpu_node().  Remove this and
use the common one from prom.c.
Signed-off-by: NNathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This splits off the kexec path bits of the xics_teardown_cpu() routine
into its own xics_kexec_teardown_cpu() routine.  With the previous
combined routine the CPPR for a cpu that is being removed may have its
CPPR reset in the plpar_eoi() call (which explicitly sets the CPPR to
a non-zero value).  Splitting of the kexec bits of the code prevents
this from happening in the cpu remove path.

Once again, this does not cause the cpu remove from the kernel to
fail, but it does cause cpu dlpar operations to not be able to return
the cpu to the hypervisor.
Signed-off-by: NNathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The affinity mask in the virq descriptor needs to be set before we
reset the affinity for the virq.  Without doing this the call to get
the new irq server fails and we end up leaving the virq pinned to the
cpu we are removing.

This does not fail the cpu remove from the kernel, but it does prevent
cpu dlpar remove operations from returning the cpu to the hypervisor.
Signed-off-by: NNathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Currently, the kernel uses CONFIG_DEVICE_TREE to wrap a kernel image
with a fdt blob which means for any given configuration only one dts
file can be selected and so support for only one board can be built

This moves the selection of the default .dts file out of the kernel
config and into the bootwrapper makefile.  The makefile chooses which
images to build based on the kernel config and the dts source file
name is taken directly from the image name.  For example "cuImage.ebony"
will use "ebony.dts" as the device tree source file.

In addition, this patch allows a specific image to be requested from the
command line by adding "cuImage.%" and "treeImage.%" targets to the list
of valid built targets in arch/powerpc/Makefile.  This allows the default
dts selection to be overridden.

Another advantage to this change is it allows a single defconfig to be
supplied for all boards using the same chip family and only differing in
the device tree.

Important note: This patch adds two new zImage targets; zImage.dtb.% and
zImage.dtb.initrd.% for zImages with embedded dtb files.  Currently
there are 5 platforms which require this: ps3, ep405, mpc885ads, ep88xc,
adder875-redboot and ep8248e.  This patch *changes the zImage filenames*
for those platforms.  ie. 'zImage.ps3' is now 'zImage.dtb.ps3'.

This new zImage.dtb targets were added so that the .dts file could be
part of the dependancies list for building them.
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Here's a dumb simple implementation of fake NUMA nodes for PowerPC.
Fake NUMA nodes can be specified using the following command line
option

numa=fake=<node range>

node range is of the format <range1>,<range2>,...<rangeN>

Each of the rangeX parameters is passed using memparse().  I find the
patch useful for fake NUMA emulation on my simple PowerPC machine.
I've tested it on a numa box with the following arguments

numa=fake=512M
numa=fake=512M,768M
numa=fake=256M,512M mem=512M
numa=fake=1G mem=768M
numa=fake=
without any numa= argument

The other side-effect introduced by this patch is that; in the case
where we don't have NUMA information, we now set a node online after
adding each LMB.  This node could very well be node 0, but in the case
that we enable fake NUMA nodes, when we cross node boundaries, we need
to set the new node online.
Signed-off-by: NBalbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Looks like "[POWERPC] kdump shutdown hook support" broke builds when
CONFIG_DEBUGGER=n and CONFIG_KEXEC=y, such as in g5_defconfig:

arch/powerpc/kernel/crash.c: In function 'default_machine_crash_shutdown':
arch/powerpc/kernel/crash.c:388: error: '__debugger_fault_handler' undeclared (first use in this function)
arch/powerpc/kernel/crash.c:388: error: (Each undeclared identifier is reported only once
arch/powerpc/kernel/crash.c:388: error: for each function it appears in.)

Move the debugger hooks to under CONFIG_DEBUGGER || CONFIG_KEXEC, since
that's when the crash code is enabled.

(I should have caught this with my build-script pre-merge, my bad. :( )
Signed-off-by: NOlof Johansson <olof@lixom.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Add 512x support using the psc_ops framework established
with the previous patch.

All 512x PSCs share the same interrupt so add
IRQF_SHARED to irq flags.
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

PSC devices are different between the mpc5200 and the mpc5121
this patch localizes the differences in preparation for adding mpc5121
support to the psc uart driver.
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Minimal /dts-v1/ device tree for mpc5121 ads.

port-number property in uart nodes
will go away after the driver learns to use aliases
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

512x is very similar to 83xx and most
of this is patterned after code from 83xx.

New platform:
    changed:
	arch/powerpc/Kconfig
	arch/powerpc/platforms/Kconfig
	arch/powerpc/platforms/Kconfig.cputype
	arch/powerpc/platforms/Makefile
    new:
	arch/powerpc/platforms/512x/*
	include/asm-powerpc/mpc512x.h
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NGrant Likely <grant.likely@secretlab.ca>

Merge branch 'for-2.6.25' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerpc into for-2.6.25

Of_get_parent and of_find_compatible_node do a of_node_get, and thus a
corresponding of_code_put is needed in both the error case and the normal
return case.

The problem was found using the following semantic match.
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@@
type T,T1,T2;
identifier E;
statement S;
expression x1,x2,x3;
int ret;
@@

  T E;
  ...
* E = \(of_get_parent\|of_find_compatible_node\)(...);
  if (E == NULL) S
  ... when != of_node_put(...,(T1)E,...)
      when != if (E != NULL) { ... of_node_put(...,(T1)E,...); ...}
      when != x1 = (T1)E
      when != E = x3;
      when any
  if (...) {
    ... when != of_node_put(...,(T2)E,...)
        when != if (E != NULL) { ... of_node_put(...,(T2)E,...); ...}
        when != x2 = (T2)E
(
*   return;
|
*   return ret;
)
  }
// </smpl>
Signed-off-by: NJulia Lawall <julia@diku.dk>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The functions of_find_compatible_node and of_find_node_by_type both
call of_node_get on their result.  So any error handling code
thereafter should call of_node_put(np).  This is taken care of in the
case where there is a goto out, but not when there is a direct return.

The function irq_alloc_host puts np into the returned structure, which is
stored in the global variable mpc8xx_pic_host, so the reference count
should be set for the lifetime of that variable.  The current solution ups
the reference count again in the argument to irq_alloc_host so that it can
be decremented on the way out.  This seems a bit unnecessary, and also
doesn't work in the case where irq_alloc_host fails, because then the
reference count only goes does by one, whereas it should go down by two.  A
better solution is to not increment the reference count in the argument to
irq_alloc_host and only decrement it on the way out in an error case.

The problem was found using the following semantic match.
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@@
type T,T1,T2;
identifier E;
statement S;
expression x1,x2,x3;
int ret;
@@

  T E;
  ...
* E = \(of_get_parent\|of_find_compatible_node\)(...);
  if (E == NULL) S
  ... when != of_node_put(...,(T1)E,...)
      when != if (E != NULL) { ... of_node_put(...,(T1)E,...); ...}
      when != x1 = (T1)E
      when != E = x3;
      when any
  if (...) {
    ... when != of_node_put(...,(T2)E,...)
        when != if (E != NULL) { ... of_node_put(...,(T2)E,...); ...}
        when != x2 = (T2)E
(
*   return;
|
*   return ret;
)
  }
// </smpl>
Signed-off-by: NJulia Lawall <julia@diku.dk>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Of_get_parent and of_find_compatible_node do an of_node_get, and thus a
corresponding of_code_put is needed in the error case.

The problem was found using the following semantic match.
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@@
type T,T1,T2;
identifier E;
statement S;
expression x1,x2,x3;
int ret;
@@

  T E;
  ...
* E = \(of_get_parent\|of_find_compatible_node\)(...);
  if (E == NULL) S
  ... when != of_node_put(...,(T1)E,...)
      when != if (E != NULL) { ... of_node_put(...,(T1)E,...); ...}
      when != x1 = (T1)E
      when != E = x3;
      when any
  if (...) {
    ... when != of_node_put(...,(T2)E,...)
        when != if (E != NULL) { ... of_node_put(...,(T2)E,...); ...}
        when != x2 = (T2)E
(
*   return;
|
*   return ret;
)
  }
// </smpl>
Signed-off-by: NJulia Lawall <julia@diku.dk>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The 8572 is a dual core processor, not reason not to describe both
cores in the device tree.
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

One is intoduced by me (of_node_put() absence) and another was
present already (not checking for NULL).

Found by Stephen Rothwell.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Enable math emulation and ucc_geth and some PHYs mpc83xx boards use.
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

calling platform_device_register after platform_device_alloc causes
this:

kobject (c3841a70): tried to init an initialized object, something is seriously wrong.
Call Trace:
[c381fe20] [c0007bb8] show_stack+0x3c/0x194 (unreliable)
[c381fe50] [c01322a8] kobject_init+0xb8/0xbc
[c381fe60] [c01591cc] device_initialize+0x30/0x9c
[c381fe80] [c015ee34] platform_device_register+0x1c/0x34
[c381fea0] [c02f1fe0] of_fsl_spi_probe+0x21c/0x22c
[c381ff30] [c02f2044] fsl_spi_init+0x54/0x160
[c381ff60] [c02f3924] __machine_initcall_mpc832x_rdb_mpc832x_spi_init+0x120/0x138
[c381ff70] [c02e61b4] kernel_init+0x98/0x284
[c381fff0] [c000f740] kernel_thread+0x44/0x60

fixed by calling platform_device_add (second half of
platform_device_register) instead.
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

arch/powerpc/platforms/83xx/mpc832x_rdb.c: In function ‘mpc832x_rdb_setup_arch’:
arch/powerpc/platforms/83xx/mpc832x_rdb.c:104: warning: ‘np’ is used uninitialized in this function
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The e300 c3 and c4 variants support hardware performance monitor counters
which are identical to those found in the e500.
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Some of the more recent e300 cores have the same performance monitor
implementation as the e500.  e300 isn't book-e, so the name isn't
really appropriate.  In preparation for e300 support, rename a bunch
of fsl_booke things to say fsl_emb (Freescale Embedded Performance Monitors).
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Cleaned up IRQ layout and removed unsused ISU allocations.
Fixed RTC address typo from /dts-v1/ conversion.
Incorporated list suggestions to use an "iomega," vendor prefix,
and to use a node reference rather than a hard path.

Signed-off-by: Jon Loeliger <jdl@@jdl.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

cpm_uart_core has a dependency on fsl,cpm-brg/clock-frequency, this
means that a .dts that uses the cpm uart driver needs to supply a
clock-frequency entry for get_brgfreq to return a meaningful number.
Signed-off-by: NBryan O'Donoghue <bodonoghue@codehermit.ie>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The 8313 rdb has a ds1339 at address 0x68.
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Currently, update_mmu_cache will crash if given a no-access PTE.
There's no need to synchronize dcache/icache unless it's an exec
mapping -- however, due to the existence of older glibc versions that
execute out of a read-but-no-exec page, readability is tested instead.

This assumes no exec-only mappings; if such mappings become supported,
they will need to go through the kmap_atomic() version of
dcache/icache synchronization.

This fixes a bug reported by some users where the kernel would crash
while dumping core on a threaded program.
Signed-off-by: NScott Wood <scottwood@freescale.com>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The patch to legacy_serial.c (1a7507c7,
Reduce code duplication in legacy_serial, add UART parent types) changed
the semantics for opb ports from type = "opb" || compatible = "ibm,opb"
to type = "opb" && compatible = "ibm,opb".

The result is serial ports on our QS21s (Cell blades) don't get found,
and for some reason the machine doesn't boot at all - possibly it's
panicking due to lack of a console?

The fix is to add two entries to the of_device_id table, one that looks
for type = "opb" and the other compatible = "ibm,opb".
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This ensures that the syscall and the (fast) vdso versions of
clock_getres() will return the same resolution.
Signed-off-by: NTony Breeds <tony@bakeyournoodle.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

WARNING: drivers/cdrom/viocd.o(.text+0x504): Section mismatch in reference from the function .viocd_probe() to the function .init.text:.find_capability()
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

WARNING: vmlinux.o(.text+0x3017c): Section mismatch in reference from the function .vio_create_viodasd() to the function .devinit.text:.vio_register_device_node()
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

I got this warning from gcc:
arch/powerpc/platforms/cell/axon_msi.c:118: warning: 'tmp' may be used uninitialized in this function

Which turns out to be a false positive, but pointed out that it was
possible for the error path in find_msi_translator() to do an extra
of_node_put on a node.  This fixes it by localising the ref counting
a bit.  As a side effect, the warning goes away.
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

There's a brown-paper-bag bug in axon_msi, we pass the address of our
FIFO directly to the hardware, without DMA mapping it.  This leads to
DMA exceptions if you enable MSI & the IOMMU.

The fix is to correctly DMA map the fifo, dma_alloc_coherent() does
what we want - and we need to track the virt & phys addresses.
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Now that we create of_platform devices earlier on cell, we can make the
axon_msi driver an of_platform driver.  This makes the code cleaner in
several ways, and most importantly means we have a struct device.
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Although of_platform_device's can have a shutdown routine, at the moment
the bus code doesn't actually call it. So add the required glue to
hook the shutdown routine.
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Currently cell publishes OF devices at device_initcall() time, which
means the earliest a driver can bind to a device is also device_initcall()
time.  We have a driver we want to register before other devices, so
publish the devices at subsys_initcall() time.

This should not cause any behaviour change for existing drivers, as they
are still bound at device_initcall() time.
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Iterating through a device node's parents is simple enough, but dealing
with the refcounts properly is a little ugly, and replicating that logic
is asking for someone to get it wrong or forget it all together, eg:

while (dn != NULL) {
	/* loop body */
	tmp = of_get_parent(dn);
	of_node_put(dn);
	dn = tmp;
}

So add of_get_next_parent(), inspired by of_get_next_child().  The
contract is that it returns the parent and drops the reference on the
current node, this makes the loop look like:

while (dn != NULL) {
	/* loop body */
	dn = of_get_next_parent(dn);
}
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

asm/commproc.h was renamed to asm/cpm1.h
sysdev/commproc.h was renamed to platforms/8xx/mpc8xx.h
m8xx_pic_init was renamed to mpc8xx_pics_init
Signed-off-by: NScott Wood <scottwood@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

According to the 8349EA ref man, the second PCI PHB IRQ is 67.  Thanks to Peter
Van Ackeren for finding this.
Signed-off-by: NKim Phillips <kim.phillips@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Reference count for the "neighbor" spu context was not
being correctly decremented after usage.
So, contexts used as reference during SPU affinity setup
were not being deallocated, leading to a memory leak.
Signed-off-by: NAndre Detsch <adetsch@br.ibm.com>
Signed-off-by: NJeremy Kerr <jk@ozlabs.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>