Moved the low hanging fruit that was either identical or close
to it between ppc32 & ppc64 for PCI into pci-common.c
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

In the places we can move to using pci_bus_to_host, this allows us
to make pci_bus_to_host static and remove its export.
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Make the ppc32 pcibios_alloc_controller take a device node to match
the ppc64 prototypes and have it set arch_data.
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Make the pci_controller struct use global_number for the PHB domain number
instead of index to match what ppc64 does and reuse its pci_domain_nr code.

Introduced a pci-common.c to handle shared code between ppc32 & ppc64.
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

There are no in kernel users of any off these functions and some of
them were not even EXPORT_SYMBOL:

- pci_bus_io_base()
- pci_bus_io_base_phys()
- pci_bus_mem_base_phys()
- pci_resource_to_bus()
- phys_to_bus()
- pci_phys_to_bus()
- pci_bus_to_phys()
- pci_init_resource()
- resource_fixup()
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

In pcibios_fixup_bus(), bridges that are subordinate
to transparent bridges were still relocating their
IORESOURCE_IO and IO_RESOURCE_MEM start and end values.

Fix this by preventing the transparent bridge from
relocating the start and end values, thus allowing the
subordinate non-transparent bridge full molestation rights.
Signed-off-by: NYork Sun <yorksun@freescale.com>
Signed-off-by: NAndy Fleming <afleming@freescale.com>
Signed-off-by: NJon Loeliger <jdl@freescale.com>
Signed-off-by: NKumar Gala <galak@kernel.crashing.org>

Commit 3d5134ee left debugging turned on
in arch/powerpc/kernel/pci_64.c.  This turns it off again.
Signed-off-by: NMichael Ellerman <michael@ellerman.id.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Currently iSeries will recalibrate the cputime_factors in the first
settimeofday() call.

It seems the reason for doing this is to ensure a resaonable time delta after
time_init().  On current kernels (with udev), this call is made 40-60 seconds
into the boot process, by moving it to a late initcall it is called
approximately 5 seconds after time_init() is called.  This is sufficient to
recalibrate the timebase.
Signed-off-by: NTony Breeds <tony@bakeyournoodle.com>
CC: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Remove references to pSeries and OpenFirmware in the __secondary_hold
usage comment.  __secondary_hold is a generic routine and can be used
by other platforms.
Signed-off-by: NGeoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Add a printout of the params value to early_init_devtree.
This value is handy to have for comparison when debugging the
bootwrapper code.
Signed-off-by: NGeoff Levand <geoffrey.levand@am.sony.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

On pSeries the firmware features are not setup until ppc_md.init_early,
so we can't do the firmware feature sections fixups till after this.

Currently firmware feature sections is only used on iSeries which inits
the firmware features much earlier.  This is a bug in waiting on
pSeries.
Signed-off-by: NMichael Neuling <mikey@neuling.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

For POWERPC, stolen time accounts for cycles lost to the hypervisor or
PURR cycles attributed to the other SMT thread.  Hence, when a PURR is
available, we should still calculate stolen time, irrespective of being
virtualised.
Signed-off-by: NMichael Neuling <mikey@neuling.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This patch adds irq_create_direct_mapping().  This routine is
an alternative to irq_create_mapping(), for irq controllers that
can use linux virq numbers directly as hardware numbers.
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>

A future patch will need the logic at the end of irq_create_mapping()
which setups a virq and installs it in the irq_map. So split it out
into a new function irq_setup_virq().
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>

Uninline virq_to_hw and export it so modules can use it.  The alternative
would be to export the irq_map array instead, but it's an infrequently
called function, and keeping the array unexported seems considerably
cleaner.
Signed-off-by: NOlof Johansson <olof@lixom.net>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

cpu_purr_data is a per-cpu array used to account for stolen time on
partitioned systems.  It used to be the case that cpus accessed each
others' cpu_purr_data, so each entry was protected by a spinlock.

However, the code was reworked ("Simplify stolen time calculation")
with the result that each cpu accesses its own cpu_purr_data and not
those of other cpus.  This means we can get rid of the spinlock as
long as we're careful to disable interrupts when accessing
cpu_purr_data in process context.
Signed-off-by: NNathan Lynch <ntl@pobox.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

With this, when kexec-ing, we copy the code and start the slaves on
their journey to the next kernel's spin loop as soon as we copy the
kexec image into place.

The kernel doesn't know exactly which slaves are spinning in
kexec_wait.  This allows us to pass more than max-cpus to the
next kernel.  But it also means that we might leave some behind.

Moving the code here means they have the time it takes us to
clear the hash table to wake up and move on.  Moving the code
any earlier would reuqire walking the image description to
search for the code, which could span multiple pages.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Currently the powerpc kernel has a 64-bit only feature,
COHERENT_ICACHE used for those CPUS which maintain icache/dcache
coherency in hardware (POWER5, essentially).  It also has a feature,
SPLIT_ID_CACHE, which is used on CPUs which have separate i and
d-caches, which is to say everything except 601 and Freescale E200.

In nearly all the places we check the SPLIT_ID_CACHE, what we actually
care about is whether the i and d-caches are coherent (which they will
be, trivially, if they're the same cache).

This tries to clarify the situation a little.  The COHERENT_ICACHE
feature becomes availble on 32-bit and is set for all CPUs where i and
d-cache are effectively coherent, whether this is due to special logic
(POWER5) or because they're unified.  We check this, instead of
SPLIT_ID_CACHE nearly everywhere.

The SPLIT_ID_CACHE feature itself is replaced by a UNIFIED_ID_CACHE
feature with reversed sense, set only on 601 and Freescale E200.  In
the two places (one Freescale BookE specific) where we really care
whether it's a unified cache, not whether they're coherent, we check
this feature.  The CPUs with unified cache are so few, we could
consider replacing this feature bit with explicit checks against the
PVR.

This will make unifying the 32-bit and 64-bit cache flush code a
little more straightforward.
Signed-off-by: NDavid Gibson <dwg@au1.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

APUS (the Amiga Power-Up System) is not supported under arch/powerpc
and it's unlikely it ever will be.  Therefore, this patch removes the
fragments of APUS support code from arch/powerpc which have been
copied from arch/ppc.

A few APUS references are left in asm-powerpc in .h files which are
still used from arch/ppc.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

A Power6 can give up CPU cycles on a dedicated CPU (as opposed to a
shared CPU) to other shared processors if the administrator asks for it
(via the HMC).

This enables that to work properly on P6.

This just involves setting a bit in the CAS structure as well as the
VPA.  To donate cycles, a CPU has to have all SMT threads idle and
have the donate bit set in the VPA.  Then call H_CEDE.

The reason why shared processors just aren't used is because dedicated
CPUs are guaranteed an actual processor, yet the system is still able to
increase the capacity of the shared CPU pool.

Also rename the VPA's cpuctls_task_attrs field to a more accurate name.
Signed-off-by: NJake Moilanen <moilanen@austin.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This patch moves things around a little bit in the new common signal.c
and signal.h files to remove the last #ifdef in the middle of the
common do_signal().
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

set_dabr() and thread.dabr exist on 32 bits as well nowadays (they
actually may do something even, depending on what CPU you have).

So this removes the ifdef.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The code for creating signal frames was still duplicated and split
in strange ways between 32 and 64 bits, including the SA_ONSTACK
handling being in do_signal on 32 bits but inside handle_rt_signal
on 64 bits etc...

This moves the 64 bits get_sigframe() to the generic signal.c,
cleans it a bit, moves the access_ok() call done by all callers to
it as well, and adapts/cleanups the 3 different signal handling cases
to use that common function.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The powerpc signal code still had some obsolete freezer bits that
have long been removed from x86 (it's now done in generic code).
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

do_signal has exactly the same behaviour on 32bit and 64bit and 32bit
compat on 64bit for handling 32bit signals.  Consolidate all these
into one common function in signal.c.  The only odd left over is
the try_to_free in the 32bit version that no other architecture has
in mainline (only in i386 for some odd SuSE release).  We should
probably get rid of it in a separate patch.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

restore_sigmask is exactly the same on 32 and 64bit, so move it to
common code.  Also move _BLOCKABLE to signal.h to avoid defining it
multiple times.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

sys_sigaltstack is the same on 32bit and 64 and we can consolidate it
to signal.c.  The only difference is that the 32bit code uses ints
for the unused register paramaters and 64bit unsigned long.  I've
changed it to unsigned long because it's the same width on 32bit.

(I also wonder who came up with this awkward calling convention.. :))
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This patch moves the code in signal_32.c and signal_64.c for handling
syscall restart into a common signal.c file and converge around a single
implementation that is based on the 32 bits one, using trap, ccr
and r3 rather than the special "result" field for deciding what to do.

The "result" field is now pretty much deprecated. We still set it for
the sake of whatever might rely on it in userland but we no longer use
it's content.

This, along with a previous patch that enables ptracers to write to
"trap" and "orig_r3" should allow gdb to properly handle syscall
restarting.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This patch removes the #ifdef CONFIG_PPC64 around setting the DABR.

The actual setting of the SPR inside of the set_dabr() function is dependent
on CONFIG_PPC64 || CONFIG_6xx but you can always provide a ppc_md hook to
override that.  We should improve support for different HW breakpoints
facilities but this is a first step.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Allow ptrace to set dabr in the thread structure for both 32 and 64 bits,
though only 64 bits actually uses that field, it's actually defined in both.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

One of the gratuitous difference between 32 and 64-bit ptrace is
whether you can whack the MSR:FE0 and FE1 bits from ptrace.  This
patch forbids it unconditionally.  In addition, the 64-bit kernels
used to return the exception mode in the MSR on reads, but 32-bit
kernels didn't.  This patch makes it return those bits on both.

Finally, since ptrace-ppc32.h and ptrace-ppc64.h are mostly empty now, and
since the previous patch made ptrace32.c no longer need the MSR_DEBUGCHANGE
definition, we just remove those 2 files and move back the remaining bits
to ptrace.c (they were short lived heh ?).
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This patch allows a ptracer to write to the "trap" and "orig_r3" words
of the pt_regs.

This, along with a subsequent patch to the signal restart code, should
enable gdb to properly handle syscall restarting after executing a separate
function (at least when there's no restart block).

This patch also removes ptrace32.c code toying directly with the registers
and makes it use the ptrace_get/put_reg() accessors for everything so that
the logic for checking what is permitted is in only one place.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

CHECK_FULL_REGS() exist on both 32 and 64 bits, so there's no need
to make it conditional on CONFIG_PPC32.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This folds back the ptrace-common.h bits back into ptrace.c and removes
that file. The FSL SPE bits from ptrace-ppc32.h are folded back in as
well.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The powerpc ptrace interface is dodgy at best. We have defined our
"own" versions of GETREGS/SETREGS/GETFPREGS/SETFPREGS that strangely
take arguments in reverse order from other archs (in addition to having
different request numbers) and have subtle issue, like not accessing
all of the registers in their respective categories.

This patch moves the implementation of those to a separate function
in order to facilitate their deprecation in the future, and provides
new ptrace requests that mirror the x86 and sparc ones and use the
same numbers:

   PTRACE_GETREGS    : returns an entire pt_regs (the whole thing,
                       not only the 32 GPRs, though that doesn't
                       include the FPRs etc... There's a compat version
                       for 32 bits that returns a 32 bits compatible
                       pt_regs (44 uints)

   PTRACE_SETREGS    : sets an entire pt_regs (the whole thing,
                       not only the 32 GPRs, though that doesn't
                       include the FPRs etc... Some registers cannot be
                       written to and will just be dropped, this is the
                       same as with POKEUSR, that is anything above MQ
                       on 32 bits and CCR on 64 bits. There is a compat
                       version as well.

   PTRACE_GETFPREGS  : returns all the FP registers -including- the FPSCR
                       that is 33 doubles (regardless of 32/64 bits)

   PTRACE_SETFPREGS  : sets all the FP registers -including- the FPSCR
                       that is 33 doubles (regardless of 32/64 bits)

And two that only exist on 64 bits kernels:

   PTRACE_GETREGS64  : Same as PTRACE_GETREGS, except there is no compat
                       function, a 32 bits process will obtain the full 64
                       bits registers

   PTRACE_SETREGS64  : Same as PTRACE_SETREGS, except there is no compat
                       function, a 32 bits process will set the full 64
                       bits registers

The two later ones makes things easier to have a 32 bits debugger on a
64 bits program (or on a 32 bits program that uses the full 64 bits of
the GPRs, which is possible though has issues that will be fixed in a
later patch).

Finally, while at it, the patch removes a whole bunch of code duplication
between ptrace32.c and ptrace.c, in large part by having the former call
into the later for all requests that don't need any special "compat"
treatment.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The powerpc ptrace code has some weirdness, like a ptrace-common.h file that
is actually ppc64 only and some of the 32 bits code ifdef'ed inside ptrace.c.

There are also separate implementations for things like get/set_vrregs for
32 and 64 bits which is totally unnecessary.

This patch cleans that up a bit by having a ptrace-common.h which contains
really common code (and makes a lot more code common), and ptrace-ppc32.h and
ptrace-ppc64.h files that contain the few remaining different bits.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

The handling of PPC_PTRACE_GETFPREGS is broken on 32 bits kernel,
it will only return half of the registers. Since that call didn't
initially exist for 32 bits kernel (added recently), rather than
fixing it, let's just remove it.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

This rewrites pretty much from scratch the handling of MMIO and PIO
space allocations on powerpc64.  The main goals are:

 - Get rid of imalloc and use more common code where possible
 - Simplify the current mess so that PIO space is allocated and
   mapped in a single place for PCI bridges
 - Handle allocation constraints of PIO for all bridges including
   hot plugged ones within the 2GB space reserved for IO ports,
   so that devices on hotplugged busses will now work with drivers
   that assume IO ports fit in an int.
 - Cleanup and separate tracking of the ISA space in the reserved
   low 64K of IO space. No ISA -> Nothing mapped there.

I booted a cell blade with IDE on PIO and MMIO and a dual G5 so
far, that's it :-)

With this patch, all allocations are done using the code in
mm/vmalloc.c, though we use the low level __get_vm_area with
explicit start/stop constraints in order to manage separate
areas for vmalloc/vmap, ioremap, and PCI IOs.

This greatly simplifies a lot of things, as you can see in the
diffstat of that patch :-)

A new pair of functions pcibios_map/unmap_io_space() now replace
all of the previous code that used to manipulate PCI IOs space.
The allocation is done at mapping time, which is now called from
scan_phb's, just before the devices are probed (instead of after,
which is by itself a bug fix). The only other caller is the PCI
hotplug code for hot adding PCI-PCI bridges (slots).

imalloc is gone, as is the "sub-allocation" thing, but I do beleive
that hotplug should still work in the sense that the space allocation
is always done by the PHB, but if you unmap a child bus of this PHB
(which seems to be possible), then the code should properly tear
down all the HPTE mappings for that area of the PHB allocated IO space.

I now always reserve the first 64K of IO space for the bridge with
the ISA bus on it. I have moved the code for tracking ISA in a separate
file which should also make it smarter if we ever are capable of
hot unplugging or re-plugging an ISA bridge.

This should have a side effect on platforms like powermac where VGA IOs
will no longer work. This is done on purpose though as they would have
worked semi-randomly before. The idea at this point is to isolate drivers
that might need to access those and fix them by providing a proper
function to obtain an offset to the legacy IOs of a given bus.
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Maybe the type should have been char[] instead of __u8[]
in the first place, but this will do.
Signed-off-by: NSegher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: NPaul Mackerras <paulus@samba.org>

Previously, registering this early console would just result
in dropping early buffered printk output until a udbg_putc
was registered.

However, commit 69331af7
clears the CON_PRINTBUFFER flag on the main console when a
CON_BOOT (early) console has been registered, resulting in
the buffered messages never being displayed to the user.

This fixes the problem by making sure we don't register udbg_console
on platforms that don't implement udbg_putc.
Signed-off-by: NMilton Miller <miltonm@bga.com>
Acked-by: NMark A. Greer <mgreer@mvista.com>
Signed-off-by: NPaul Mackerras <paulus@samba.org>