Signed-off-by: NJason Jin <Jason.jin@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Recently, indirect_pci was changed to test if the bus number requested
is the one hanging straight off the PHB, then it substitutes the bus
number with another one contained in a new "self_busno" field of the
pci_controller structure.

However, this breaks CHRP which didn't initialize this new field, and
which relies on having the right bus number passed to the hardware.

This fixes it by initializing this variable properly for all CHRP bridges
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

The detection of the IBM "Python" PCI host bridge on IBM CHRP
machines such as old RS6000 was broken when we changed
of_device_is_compatible() from strncasecmp to strcasecmp (dropped
the "n" variant) due to the way IBM encodes the chip version.

We fix that by instead doing a match on the model property like
we do for others bridges in that file. It should be good enough
for those machines. If yours is still broken, let me know.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

prom_init was changed to take a new argument, the address
where the kernel is loaded, which is now used to copy the
SMP spin loop down before use.

However, only head_64.S was adapted to pass this new value,
not head_32.S, thus breaking SMP boot on 32-bit SMP CHRP
machines.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

The new merged DMA code will try to access isa_bridge_pcidev when
trying to DMA to/from legacy devices. This is however only defined
on 64-bit. Fixes this for now by adding the variable, even if it
stays NULL. In the long run, we'll make isa-bridge.c common to
32 and 64-bit.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

When the powerpc PCI layer is not configured to re-assign everything,
it currently fails to detect that a PCI to PCI bridge has been left
unassigned by the firmware and tries to allocate resource for the
default window values in the bridge (0...X) (with the notable exception
of a hack we have in there that detects some Apple firmware unassigned
bridge resources).

This results in resource allocation failures, which are generally
fixed up later on but it causes scary warnings in the logs and we
have seen the fixup code fall over in some circumstances (a different
issue to fix as well).

This code improves that by providing a more complete & useful function
to intuit that a bridge was left unassigned by the firmware, and thus
force a full re-allocation by the PCI code without trying to allocate
the existing useless resources first.

The algorithm we use basically considers unassigned a window that
starts at 0 (PCI address) if the corresponding address space enable
bit is not set. In addition, for memory space, it considers such a
resource unassigned also if the host bridge isn't configured to
forward cycles to address 0 (ie, the resource basically overlaps
main memory).

This fixes a range of problems with things like Bare-Metal support
on pSeries machines, or attempt to use partial firmware PCI setup.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

The typesafe version of the powerpc pagetable handling (with
USE_STRICT_MM_TYPECHECKS defined) has bitrotted again.  This patch
makes a bunch of small fixes to get it back to building status.

It's still not enabled by default as gcc still generates worse
code with it for some reason.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

The "phys" argument to machine_init() isn't used and isn't likely to
ever be so let's remove it.
Signed-off-by: NSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

After Becky's work we can almost have different DMA offsets
between on-chip devices and PCI. Almost because there's a
problem with the non-coherent DMA code that basically ignores
the programmed offset to use the global one for everything.
This fixes it.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

This is a much better version of a previous patch to make the parser
tables constant. Rather than changing the typedef, we put the "const" in
all the various places where its required, allowing the __initconst
exception for nfsroot which was the cause of the previous trouble.

This was posted for review some time ago and I believe its been in -mm
since then.
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>
Cc: Alexander Viro <aviro@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The fsl_upm nand driver fails to build because fsl_lbc_lock isn't
exported, the lock is needed by the inlined fsl_upm_run_pattern()
function:

ERROR: "fsl_lbc_lock" [drivers/mtd/nand/fsl_upm.ko] undefined!

Dave Jones purposed to export the lock, but it is better to just uninline
the fsl_upm_run_pattern().

When uninlined we also no longer need the exported fsl_lbc_regs, and
both fsl_lbc_lock and fsl_lbc_regs could be marked static.

While at it, also add some missing includes that we should have included
explicitly.
Reported-by: NDave Jones <davej@redhat.com>
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The StMicro NAND chip (512Mbit, 64MB) is connected to the local bus,
the first local bus' user-programmable machine is configured by the
firmware to work with NAND chips.

QE GPIO pin is used to poll the NAND's Ready-Not-Busy signal.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Support for the SBC610 VPX Single Board Computer from GE Fanuc (PowerPC
MPC8641D).

This patch adds support for the registers held in the devices main FPGA,
exposing extra information about the revision of the board through cpuinfo.
Signed-off-by: NMartyn Welch <martyn.welch@gefanuc.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The Freescale Elo DMA driver binds to all DMA channels in the device tree that
are compatible with "fsl,eloplus-dma-channel".  This conflicts with the sound
drivers for the MPC8610 HPCD.  On this board, the SSI uses two DMA channels and
therefore those channels are not available for general purpose use.  We
change the compatible properties for these channels "fsl,ssi-dma-channel".
This works because the sound drivers don't actually check the compatible
property when it grabs channels.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Because CHRP and PMAC are by default enabled, several non-CHRP and non-PMAC
PowerPC defconfigs will have these Kconfig options set erroneously.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The RTC is sitting on the I2C1 bus at address 0x68. RTC interrupt signal
is connected to the IPIC's EXT3 interrupt line.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Early versions of the Freescale DIU framebuffer driver depended on a bootmem
allocation of memory for the video buffer.  The need for this feature was
removed in commit 6b51d51a, so now we can remove the platform-specific code
that allocated that memory.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

b38fd42f added false dependencys
to order the load of upper and lower halfs of the pte, but only
adjusted whitespace instead of deleting the old load in the iside
handler, letting the hardware see the non-dependent load.

This patch removes the extra load.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Uses mpc83xx_add_bridge in fsl_pci.c

Adds second register tuple to pci node register property
as done for 83xx device trees in a previous patch.
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Acked-by: NGrant Likely <grant.likely@secretlab.ca>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Modify mpc83xx_add_bridge to get config space register base address from
the device tree instead of immr + hardcoded offset.

83xx pci nodes have this change:
    register properties now contain two address length tuples:
	First is the pci bridge register base, this has always been there.
	Second is the config base, this is new.

This is documented in dts-bindings/fsl/83xx-512x-pci.txt

The changes accomplish these things:
    mpc83xx_add_bridge no longer needs to call get_immrbase
    it uses hard coded addresses if the second register value is missing
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The class of the MPC5121 pci host bridge is PCI_CLASS_BRIDGE_OTHER
while other freescale host bridges have class set to
PCI_CLASS_PROCESSOR_POWERPC.

This patch makes fixup_hide_host_resource_fsl match
PCI_CLASS_BRIDGE_OTHER in addition to PCI_CLASS_PROCESSOR_POWERPC.
Signed-off-by: NJohn Rigby <jrigby@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

In the standalone setup the board's CPLD disables the PCI internal
arbiter, thus any access to the PCI bus will hang the board.

The common way to disable particular devices in the device tree is to
put the "status" property with any value other than "ok" or "okay"
into the device node we want to disable.

So, when there is no PCI arbiter on the bus the u-boot adds status =
"broken (no arbiter)" property into the PCI controller's node, and so
marks the PCI controller as unavailable.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Specifying user-selectable option in the qe_lib/Kconfig was a bad idea
because the qe_lib/Kconfig is included into the top level Kconfig, and
thus the QE_GPIO option appears at the top level menu.

This patch effectively moves the QE_GPIO option under the platform menu
instead.
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Modify the Kconfig so that Freescale QUICC Engine (QE) support is a selectable
option, thereby allowing users to compile kernels without any QE support.

The drawback is that QE support is now disabled by default on platforms that
have a QE, and so a defconfig is needed to enable QE and QE devices (like
UCC GETH).  Fortunately, all the current relevant defconfigs do that already.
Signed-off-by: NTimur Tabi <timur@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Support for the SBC610 VPX Single Board Computer from GE Fanuc (PowerPC MPC8641D).

A number of MPC8641D based route interrupts for on-board interrupts through
a FPGA based interrupt controller, which is chained with the
MPC8641D's mpic. This patch provides a basic driver to allow basic routing
of interrupts to the mpic.
Signed-off-by: NMartyn Welch <martyn.welch@gefanuc.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Add interrupt info to the MPC8536DS .dts for the RTC
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

The comment in the code was asking "Do we have to do this?", and according
to x86 and s390 the answer is no, the scheduler will do it before calling
the arch hook.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

A single full sync (mb()) is requrired to order the mmio to the qirr reg
with the set or clear of the message word.  However, test_and_clear_bit
has the effect of smp_mb() and we are not doing any other io from here,
so we don't need a mb per bit processed.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

Several printks were broken at word boundaries for line length.   Some
even referred to old function names.   Using __func__ and changing the
text slightly for the format allows these printk formats to fit on one
line.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

It is physically per-cpu, and we want the irq layer to treat it that way.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

EOI normally has the side effect of returning the cpu to the base
priority to recieve the next interrupt.  This is actually controlled
by the top byte of the xirr register.   When we are exiting the
kernel in kexec we must eoi the ipi for the next kernel because we
never return from the handler, but we want to leave interrupt
delivery blocked until the next kernel takes action.

Since the hardware ipi vector is fixed, its easiest to just do the
eoi explicitly.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

This factors out processors joining and unjoining the Global Interrupt
Queue into a separate function.

There is a bit of math to calculate the arguments to rtas to join
or leave the global interrupt queue, and a warning on failure
afterwards.  Make a helper for the 3 callers.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

We only need to check the ibm,interrupt-server#-size property once, not
once per global server and thread.

We can use !CONFIG_SMP cpu masks and hard_smp_processor_id() to avoid an ifdef.
Put the node when breaking out of the loop on lpar systems.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

Trim unneeded includes from xics.c.  We don't use signals or gfp
flags, we use only OF functions and don't need prom, and the 8259
is now handled by our caller.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

The xirr is 32 bits in hardware, but the hypervisor requries the upper
bits of the register to be clear on the hcall.  By changing the type
from signed to unsigned int we can drop masking it back to 32 bits.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

Now that xics_update_irq_servers is called only from init and hotplug
code, it becomes possible to clean up the ordering of functions in the
file, grouping them but the interfaces they implement.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

xics supports only one ipi per cpu, and expects software to use some
queue to know why the interrupt was sent.  In Linux, we use a an array
of bitmaps indexed by cpu to identify the message.  Currently the bits
are set in smp.c and decoded in xics.c, with the data structure in a
header file.   Consolidate the code in xics.c similar to mpic and other
interrupt controllers.

Also, while making the the array static, the message word doesn't need
to be volatile as set_bit and test_clear_bit take care of it for us, and
put it under ifdef smp.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

Previously the FDT header field boot_cpuid_phys wasn't actually used
on ppc32.  Instead the physical boot cpuid was assumed to be 0 for
!CONFIG_SMP.
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

There is an old workaround in the sysdev/Makefile for dealing
with arch/ppc vs. arch/powerpc compiles.  This is no longer
needed as arch/ppc is dead.
Signed-off-by: NJosh Boyer <jwboyer@linux.vnet.ibm.com>
Acked-by: NKumar Gala <galak@kernel.crashing.org>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>

Currently, every time we determine which irq server to use, we check if
default_server, which is the id of the bootcpu, is still online.  But
default_server is a hardware cpu, not the logical cpu id needed to index
cpu_online_map.

Since the default server can only go offline during a cpu hotplug event,
explicitly check the default server and choose the new one when we move
irqs away from the cpu being offlined.

This has the added benefit of only needing the boot_cpuid to be updated
and not relying on the cpu being marked offline during migrate_irqs_away.

Also, since xics_update_irq_servers only reads device tree information, we
can call it before xics_init_host in xics_init_IRQ and then default_server
will always be valid when we can reach get_irq_server via the host ops.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>