Commit Graph

2538 Commits

Author SHA1 Message Date
Paul Mackerras
09eee63471 powerpc: Move U3 IOMMU driver to arch/powerpc/sysdev
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-28 08:29:36 +10:00
Paul Mackerras
bfab1019b7 powerpc: Fix new-world powermac detection
My G5 was being reported as an OldWorld in /proc/cpuinfo, which is
obviously not right... :)

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 22:45:33 +10:00
Paul Mackerras
6316222ea0 powerpc: Introduce toreal/fromreal assembly macros
On 32-bit platforms, these convert from kernel virtual addresses
to real (physical addresses), like tophys/tovirt but they use
the same register for the source and destination.  On 64-bit
platforms, they do nothing because the hardware ignores the top
two bits of the address in real mode.

These new macros are used in fpu.S now.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 22:44:39 +10:00
Paul Mackerras
80579e1f4a powerpc: 32-bit CHRP SMP fixes
Untested, but "should" work...  at least this way it compiles.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 22:42:04 +10:00
Paul Mackerras
eef69e3cfb powerpc: remove duplicate screen_info from setup_32.c
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 22:21:37 +10:00
Kumar Gala
4b24b01c6c [PATCH] powerpc: Moved dcr support to arch/powerpc
Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 20:51:22 +10:00
Kumar Gala
cffb09ce6b [PATCH] powerpc: Fix warning related to do_dabr
do_dabr() is not relevant on 40x or Book-E processors so dont build it

Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 20:51:19 +10:00
David Gibson
e37bc5df8e [PATCH] powerpc: Purge bootinfo.h
With ARCH=powerpc we assume the presence of a device tree, so we don't
require any support for the old bi_recs method of passing boot
parameters.  Likewise, we've never needed it for ppc64, but we still
had an include/asm-ppc64/bootinfo.h from which nothing was used.  This
patch removes that file, and all references to it in arch/ppc64 and
arch/powerpc.  A related, unused variable 'boot_mem_size' is also
removed from setup_32.c.  The bootinfo stuff remains in ARCH=ppc for
the time being.

Built and booted on Power5 (ARCH=ppc64 and ARCH=powerpc), built for
32-bit powermac (ARCH=powerpc and ARCH=ppc).

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 20:50:46 +10:00
David Gibson
25c8a78b1e [PATCH] powerpc: Fix handling of fpscr on 64-bit
The recent merge of fpu.S broken the handling of fpscr for
ARCH=powerpc and CONFIG_PPC64=y.  FP registers could be corrupted,
leading to strange random application crashes.

The confusion arises, because the thread_struct has (and requires) a
64-bit area to save the fpscr, because we use load/store double
instructions to get it in to/out of the FPU.  However, only the low
32-bits are actually used, so we want to treat it as a 32-bit quantity
when manipulating its bits to avoid extra load/stores on 32-bit.  This
patch replaces the current definition with a structure of two 32-bit
quantities (pad and val), to clarify things as much as is possible.
The 'val' field is used when manipulating bits, the structure itself
is used when obtaining the address for loading/unloading the value
from the FPU.

While we're at it, consolidate the 4 (!) almost identical versions of
cvt_fd() and cvt_df() (arch/ppc/kernel/misc.S,
arch/ppc64/kernel/misc.S, arch/powerpc/kernel/misc_32.S,
arch/powerpc/kernel/misc_64.S) into a single version in fpu.S.  The
new version takes a pointer to thread_struct and applies the correct
offset itself, rather than a pointer to the fpscr field itself, again
to avoid confusion as to which is the correct field to use.

Finally, this patch makes ARCH=ppc64 also use the consolidated fpu.S
code, which it previously did not.

Built for G5 (ARCH=ppc64 and ARCH=powerpc), 32-bit powermac (ARCH=ppc
and ARCH=powerpc) and Walnut (ARCH=ppc, CONFIG_MATH_EMULATION=y).
Booted on G5 (ARCH=powerpc) and things which previously fell over no
longer do.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 20:48:50 +10:00
Paul Mackerras
fda262b897 [PATCH] ppc64: remove arch/ppc64/kernel/setup.c
and use setup_64.c from the merged tree instead.  The only difference
between them was the code to set up the syscall maps.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 20:20:42 +10:00
Paul Mackerras
cb4ab974ae powerpc: Remove common stuff from setup_64.c
This should have been in commit 03501dab03
but got missed by accident.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-27 11:08:31 +10:00
Paul Mackerras
b6a4ce526a powerpc: 32-bit powermac needs the mpc106 code
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 22:54:17 +10:00
Paul Mackerras
bd561c79dc powerpc: Fix incorrect timer register addresses in mpic.c
We were computing the wrong address for the MPIC timer registers,
so when we went to initialize them we would have been hitting some
unrelated ioremap... oops.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 21:55:33 +10:00
Paul Mackerras
fa39dc437a powerpc32: Limit memory to lowmem if !CONFIG_HIGHMEM.
This trims off the extra unusable memory from the lmb structure,
so we don't try to use it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 21:54:21 +10:00
Paul Mackerras
c49888203d powerpc: Fixes to get the Longtrail CHRP a bit further
Talk about buggy firmware...  the OF on the Longtrail returns 0
from the claim client service rather than -1 when the claim fails.
It also has no device_type on the /memory node and blows up if
the output buffer for package-to-path is too big.

This also fixes a bug with calling alloc_up with align == 0, where
we did _ALIGN_UP(alloc_bottom, 0) which will end up as 0.

Lastly, we now check the return value (in r3) from calling the
prom, and return -1 from call_prom if we get a negative value back.
That is supposed to indicate that the requested client service
doesn't exist.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 21:52:53 +10:00
Paul Mackerras
bbd0abda9c powerpc: Merge 32-bit CHRP support.
SMP still needs more work but UP gets as far as starting userspace
at least.  This uses the 64-bit-style code for spinning up the cpus.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 21:45:56 +10:00
Paul Mackerras
303d72a000 powerpc: Don't limit pmac_get_rtc_time to return only positive values
If the machine's clock is set to a bogus value, this check resulted
in userland waiting effectively forever for the RTC value to change,
so remove the check.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 17:22:33 +10:00
Paul Mackerras
6d0124fc06 powerpc: Fix interrupt-tree parsing
The interrupt-tree parsing code wasn't offsetting interrupt numbers
by 16 on 32-bit platforms with an i8259 interrupt controller, and
it was confused about the encoding of interrupt sense and level
(which is different for i8259 and openpic interrupt controllers,
just to make things interesting).

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 17:19:06 +10:00
Paul Mackerras
830825d6c3 powerpc: Pull out MPC106 (grackle) initialization code into its own file
This is so that the 32-bit CHRP code can use it.  The MPC106
initialization code is now in arch/powerpc/sysdev/grackle.c and
is controlled by CONFIG_PPC_MPC106.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 17:16:38 +10:00
Paul Mackerras
03501dab03 powerpc: Pull common bits of setup_{32,64}.c into setup-common.c
This creates a new arch/powerpc/kernel/setup-common.c with various
bits that setup_32.c and setup_64.c had in common - functions like
machine_shutdown/restart/power_off, show_cpuinfo, set_preferred_console
etc.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 17:11:18 +10:00
Paul Mackerras
033ef338b6 powerpc: Merge rtas.c into arch/powerpc/kernel
This splits arch/ppc64/kernel/rtas.c into arch/powerpc/kernel/rtas.c,
which contains generic RTAS functions useful on any CHRP platform,
and arch/powerpc/platforms/pseries/rtas-fw.[ch], which contain
some pSeries-specific firmware flashing bits.  The parts of rtas.c
that are to do with pSeries-specific error logging are protected
by a new CONFIG_RTAS_ERROR_LOGGING symbol.  The inclusion of rtas.o
is controlled by the CONFIG_PPC_RTAS symbol, and the relevant
platforms select that.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 17:05:24 +10:00
Paul Mackerras
f9bd170a87 powerpc: Merge i8259.c into arch/powerpc/sysdev
This changes the parameters for i8259_init so that it takes two
parameters: a physical address for generating an interrupt
acknowledge cycle, and an interrupt number offset.  i8259_init
now sets the irq_desc[] for its interrupts; all the callers
were doing this, and that code is gone now.  This also defines
a CONFIG_PPC_I8259 symbol to select i8259.o for inclusion, and
makes the platforms that need it select that symbol.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 16:47:42 +10:00
Paul Mackerras
25635c71e4 ppc: Use the indirect_pci.c from arch/powerpc/sysdev
This defines a CONFIG_INDIRECT_PCI symbol to control whether it
gets used or not, and fixes the Kconfig to select that symbol for
platforms that need it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 16:36:55 +10:00
Michael Ellerman
d9953105ce [PATCH] powerpc: Move ras.c into arch/powerpc/platforms/pseries
ras.o is only built for CONFIG_PPC_PSERIES, so move it into
arch/powerpc/platforms/pseries. Update Makefiles to suit.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 16:00:29 +10:00
Kumar Gala
6d7f58b04d [PATCH] powerpc: Some minor cleanups to setup_32.c
* Removed of_show_percpuinfo and just report CPU frequency in generic
  show_cpuinfo code.
* Killed OCP and PPC_SYS related code which doesn't belong in the
  merge tree

Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 15:55:53 +10:00
Kumar Gala
0fd6f71794 [PATCH] powerpc: Add support for Book-E timer config to generic_calibrate_decr
We need to initialize some control SPRS for timers on Book-E before
we start taking decrementer interrupts.

Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 15:55:46 +10:00
Becky Bruce
ab590ccb42 [PATCH] ppc: Fix m82xx_pci build
A recent patch updated the name of pci_assign_all_busses to
pci_assign_all_buses.  This instance of its use wasn't corrected
by the original patch to use the new name.

Builds cleanly on ads8272.

Signed-off-by: Becky Bruce <becky.bruce@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 15:55:06 +10:00
Kumar Gala
60dda2565b [PATCH] powerpc: some prom.c cleanups
On !CONFIG_PPC_MULTIPLATFORM _machine is defined as 0.  This is ok, but
we can't assign a value to _machine then.

We may not have CONFIG_PCI available, so only build in support for
find_parent_pci_resource(), request_OF_resource(), release_OF_resource()
if PCI is enabled.  This is probably not the long term fix but works out
for now.

Make reg_property64 contain 64-bit elements on a 32-bit machine.

Mark the deprecated prom.c functions as __deprecated.

Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 15:52:56 +10:00
Kumar Gala
d2e6151257 [PATCH] powerpc: Make sure we have an RTC before trying to adjust it
Its valid for ppc_md.set_rtc_time to be NULL.  We need to check
that its non-NULL before trying to update the RTC.

Signed-off-by: Kumar K. Gala <kumar.gala@freescale.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-26 15:52:37 +10:00
Stephen Rothwell
340601043d powerpc: Add a shutdown member to vio_driver
Add a shutdown member to struct vio_driver. We also need vio_bus_shutdown()
which converts from struct device to struct vio_dev and knows how to extract
the struct vio_driver.

Original patch adjusted for different location of vio.c.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
2005-10-24 17:40:23 +10:00
Stephen Rothwell
7c7eb28481 powerpc: iseries: Fix a bogus comment
A comment in lpevents.c refers to code that's actually in HvCallEvent.h.
The code in HvCallEvent.h is pretty obvious, so just remove the comment
altogether.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
2005-10-24 15:42:24 +10:00
Stephen Rothwell
6fdf5392ca powerpc: don't duplicate name between vio_driver and device_driver
Just set the name field directly in the device_driver structure
contained in the vio_driver struct.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
2005-10-24 15:42:12 +10:00
Stephen Rothwell
580460194e powerpc: Merge arch/ppc64/kernel/vio.c into arch/powerpc/kernel/vio.c
Merge arch/ppc64/kernel/vio.c into arch/powerpc/kernel/vio.c, update
the Makefiles to make it work, and make ARCH=ppc64 still work.

Michael's version put vio.c in arch/powerpc/sysedv but after consolting
Paulus, this one puts it in arch/powerpc/kernel.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
2005-10-24 15:41:44 +10:00
Paul Mackerras
85e568ea58 powerpc: Make coff boot wrapper load the kernel at 8M
Previously it claimed 7MB starting at the 9M point and loaded the
kernel there.  That meant that prom_init put the flattened device
tree above 16M.  On the 601 that caused the early device tree scan
to fail, since only 16MB are mapped with BATs on the 601.  Moving
this down to 8MB allows prom_init to put the flattened device tree
between 15M and 16M, so it works on the 601.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-23 17:26:32 +10:00
Paul Mackerras
a575b80717 powerpc: Run on old powermacs.
Old powermacs have a number of differences from current machines:
- there is no interrupt tree in the device tree, just interrupt
  or AAPL,interrupt properties
- the chosen node in the device tree is called /chosen@0
- the OF claim method doesn't map the memory, so we have to do
  an explicit map call as well
- there is no /chosen/cpu property on SMP machines
- the NVRAM isn't structured as a set of partitions.

This adapts the merged powermac support code to cope with these
issues.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-23 17:23:21 +10:00
Paul Mackerras
96c4450760 powerpc: Fix time code for 601 processors
The 601 doesn't have the timebase register; instead it has an RTCL
register that counts nanoseconds and wraps at 1000000000, and an
RTCU register that counts seconds.  This makes the necessary changes
for the merged time code to use the RTCL/U registers when the kernel
is running on a 601.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-23 17:14:56 +10:00
Paul Mackerras
985990137e Merge changes from linux-2.6 by hand 2005-10-22 16:51:34 +10:00
Paul Mackerras
8342894475 ppc64: Use arch/powerpc/platforms/powermac for powermac build.
This switches the ARCH=ppc64 build to use arch/powerpc/platforms/powermac
instead of arch/ppc64/kernel/pmac*, and deletes the latter set of files.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 16:06:27 +10:00
Paul Mackerras
35499c0195 powerpc: Merge in 64-bit powermac support.
This brings in a lot of changes from arch/ppc64/kernel/pmac_*.c to
arch/powerpc/platforms/powermac/*.c and makes various minor tweaks
elsewhere.  On the powermac we now initialize ppc_md by copying
the whole pmac_md structure into it, which required some changes in
the ordering of initializations of individual fields of it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 16:02:39 +10:00
Paul Mackerras
ef1a12817a ppc64: Simplify secondary CPU startup on powermacs
... for consistency with ppc32, and because this way is neater.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 15:55:58 +10:00
Paul Mackerras
529294d59f ppc64: Rearrange btext initialization for consistency with ppc32
Moved init_boot_display from arch/ppc64/kernel/pmac_setup.c to
arch/ppc64/kernel/btext.c and declared it in asm-ppc64/btext.h.
Call it from init_early rather than pmac_init_early.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 15:36:52 +10:00
Paul Mackerras
f7abbc190b ppc64: Add a `primary' argument to pci_process_bridge_OF_ranges
... for consistency with ppc32 and to make the powermac merge easier.
Also make it use just a single resource in the host bridge for multiple
consecutive elements of the ranges property.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 15:03:21 +10:00
Paul Mackerras
a5b518ed31 ppc64/powerpc: Fix time initialization on SMP systems
This moves smp_space_timers from arch/ppc64/kernel/smp.c to
arch/powerpc/kernel/time.c and makes it initialize last_jiffy[]
instead of paca[].next_jiffy_update_tb, since last_jiffy[] is
now what the time code uses.  It also declares smp_space_timers
in include/asm-powerpc/time.h and gets rid of an ifdef in
div128_by_32.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 14:55:23 +10:00
Paul Mackerras
e2b5530698 ppc64: Fix delivery of RT signals to 32-bit processes.
An error in merging led to 32-bit processes getting the wrong link
register value on entry to RT signal handlers, and the wrong stack
chain as well.  This fixes it.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 14:46:33 +10:00
Paul Mackerras
5c8c56ebdf powerpc: Move agp_special_page export to where it is defined
... instead of exporting it in arch/*/kernel/ppc_ksyms.c.

Signed-off-by: Paul Mackerras <paulus@samba.org>
2005-10-22 14:42:51 +10:00
Chris Wright
63172cb3d5 [PATCH] typo fix in last cpufreq powernow patch
Not sure how it slipped by, but here's a trivial typo fix for powernow.

Signed-off-by: Chris Wright <chrisw@osdl.org>
[ It's "nurter" backwards.. Maybe we have a hillbilly The Shining fan? ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 17:08:30 -07:00
Dave Jones
0213df7431 [PATCH] cpufreq: fix pending powernow timer stuck condition
AMD recently discovered that on some hardware, there is a race condition
possible when a C-state change request goes onto the bus at the same
time as a P-state change request.

Both requests happen, but the southbridge hardware only acknowledges the
C-state change.  The PowerNow! driver is then stuck in a loop, waiting
for the P-state change acknowledgement.  The driver eventually times
out, but can no longer perform P-state changes.

It turns out the solution is to resend the P-state change, which the
southbridge will acknowledge normally.

Thanks to Johannes Winkelmann for reporting this and testing the fix.

Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com>
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 14:28:58 -07:00
David Gibson
3078fcc1d1 [PATCH] ppc64: Fix typo bug in iSeries hash code
This fixes a stupid typo bug in the iSeries hash table code.

When we place a hash PTE in the secondary bucket, instead of setting the
SECONDARY flag bit, as we should, we (redundantly) set the VALID flag.

This was introduced with the patch abolishing bitfields from the hash
table code.  Mea culpa, oops.  It hasn't been noticed until now because
in practice we don't hit the secondary bucket terribly often.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 12:24:41 -07:00
Linus Torvalds
2c86c83bf4 Merge master.kernel.org:/home/rmk/linux-2.6-arm 2005-10-21 12:23:07 -07:00
Benjamin Herrenschmidt
5d96551541 [PATCH] ppc64: Fix pages marked dirty abusively
While working on 64K pages, I found this little buglet in our
update_mmu_cache() implementation.

The code calls __hash_page() passing it an "access" parameter (the type
of access that triggers the hash) containing the bits _PAGE_RW and
_PAGE_USER of the linux PTE.  The latter is useless in this case and the
former is wrong.  In fact, if we have a writeable PTE and we pass
_PAGE_RW to hash_page(), it will set _PAGE_DIRTY (since we track dirty
that way, by hash faulting !dirty) which is not what we want.

In fact, the correct fix is to always pass 0. That means that only
read-only or already dirty read write PTEs will be preloaded. The
(hopefully rare) case of a non dirty read write PTE can't be preloaded
this way, it will have to fault in hash_page on the actual access.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-21 12:17:43 -07:00