numa_enforce_memory_limit tried to be smart and only call lmb_end_of_DRAM
when a memory limit was set via mem= on the command line.  However,
the early boot code will also limit memory added to the lmb system
when iommu=off is specified.  When this happens, the page allocator
is given pages not in the linear mapping and this results in a fatal
data reference to the unmapped page.
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

We used to assume that even numbered threads were the primary
threads, ie those that would be listed and started as a cpu from
open firmware.  Replace a left over is even (% 2) check with a check
for it being a primary thread and update the comments.

Tested with a debug print on pseries, identical code found for cell.
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

64 bit powerpc requires the kexec user space tools avoid overwriting
the static kernel image and translation hash table when choosing
where to put memory image data because it copies the data into place
using the kernels virtual memory system.  Kexec userspace determines
these and other areas blocked by reading properties the kernel adds,
but does not filter these properties when creating the device tree
for the next kernel.

When the second kernel tries to add its values for these properties,
the export via /proc/device-tree is hidden by the pre-existing but
stale values from the flat tree.  Kexec userspace reads the old
property, allocates the new kernel at the old kernel's end, and
gets rejected by the overlap check.

Search and remove these stale properties before adding the new values.
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

There are no users of PPC_MERGE in tree so we can get rid of it.
It was a hold over from the arch/ppc days.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

There are two issues when we enable CONFIG_RELOCATABLE.  The first is due
to the fact that phys_addr_t is now defined in linux/types.h.  The second
is due to the fact that the DMA code changes expose memstart_addr to
prom_init.c
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

Format string bug.  Not exploitable, as this is only writable by root,
but worth fixing all the same.

See 326f6a5cSigned-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

This target is needed to build cuImages with an embedded ramdisk image.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

I noticed, when trying to use, e.g.,
node = find_node_by_prop_value(prev, "booleanprop", "", 0))
to search for all nodes with a certain boolean property, that memcmp()
returns garbage when comparing zero bytes.  It should return zero.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

Adjust amount to reserve based on previous nodes for reserves spanning
multiple nodes. Check if the node active range is empty before attempting
to pass the reserve to bootmem.  In practice the range shouldn't be empty,
but to be sure we check.
Signed-off-by: Jon Tollefson <kniht@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

The issue is the SPU code is not holding the kernel mutex lock while
adding samples to the kernel buffer.

This patch creates per SPU buffers to hold the data.  Data
is added to the buffers from in interrupt context.  The data
is periodically pushed to the kernel buffer via a new Oprofile
function oprofile_put_buff(). The oprofile_put_buff() function
is called via a work queue enabling the funtion to acquire the
mutex lock.

The existing user controls for adjusting the per CPU buffer
size is used to control the size of the per SPU buffers.
Similarly, overflows of the SPU buffers are reported by
incrementing the per CPU buffer stats.  This eliminates the
need to have architecture specific controls for the per SPU
buffers which is not acceptable to the OProfile user tool
maintainer.

The export of the oprofile add_event_entry() is removed as it
is no longer needed given this patch.

Note, this patch has not addressed the issue of indexing arrays
by the spu number.  This still needs to be fixed as the spu
numbering is not guarenteed to be 0 to max_num_spus-1.
Signed-off-by: Carl Love <carll@us.ibm.com>
Signed-off-by: Maynard Johnson <maynardj@us.ibm.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Acked-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

"unsigned int" speed cannot be negative, it's thus pointless
to test if it is.
Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

The pfn of the memory to be removed should be validated prior to
attempting to remove the memory.  In cases where the probe of a
memory section fails during hotplug add, the pfn for the lmb may
not be valid.
Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: Nathan Fontenot <nfont@austin.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

If the vmlinux binary in memory is larger than 4 MiB than it collides
with the initial boot code which is linked at 4 MiB in case of cuBoot.
If the the uncompressed image size (on disk size) is less than 4 MiB
then it would fit. The difference between those two sizes is the bss
section. In cuBoot we have the dtb embedded right after the data
section so it is very likely that the reset of the bss section (in
kernel's start up code) will overwrite the dtb blob. Therefore we
reallocate the dtb. Something similar is allready done to the initrd.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

This patch adds a comment to clarify why atomic_dec_if_positive is being used
to decrement gang's aff_sched_count on SPU context unbind.
Signed-off-by: Andre Detsch <adetsch@br.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

This patch improves redability of the code responsible for trying to find
a node with enough SPUs not committed to other affinity gangs.

An additional check is also added, to avoid taking into account gangs that
have no SPU affinity.
Signed-off-by: Andre Detsch <adetsch@br.ibm.com>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

No need to zero the entire buffer, just the head and tail indices.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

With most file readers (eg cat, dd), reading a context's regs file will
result in two reads: the first to read the data, and the second to
return EOF. Because each read performs a spu_acquire_saved, we end up
descheduling and re-scheduling the context twice.

This change does a simple check to see if we'd return EOF before
calling spu_acquire_saved(), saving the extra schedule operation.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Currently, read() on the sputrace log will block until the read buffer
is full. This makes it difficult to retrieve the end of the buffer, as
the user will need to read with the right-sized buffer.

In a similar method as 91553a1b5e0df006a3573a88d98ee7cd48a3818a, this
change makes the switch_log return if there has already been data
read.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Currently, we use ctx->mapping_lock and ctx->switch_log->lock for the
context switch log. The mapping lock only prevents concurrent open()s,
so we require the switch_lock->lock for reads.

Since writes to the switch log buffer occur on context switches, we're
better off synchronising with the state_mutex, which is held during a
switch. Since we're serialised througout the buffer reads and writes,
we can use the state mutex to protect open and release too, and
can now kfree() the log buffer on release. This allows us to perform
the switch log notify without taking any extra locks.

Because the buffer is only present while the file is open, we can use
it to prevent multiple simultaneous openers.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Currently, read() on the sputrace buffer will only return data when
the user buffer is exhausted. This may mean that we never see the
end of the event log, unless we read() with exactly the right-sized
buffer.

This change makes sputrace_read not block if we have data ready to
return.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Currently, sputrace will start logging to the event buffer before the
log buffer has been open()ed. This results in a heap of "lost samples"
warnings if the sputrace file hasn't yet been opened.

Since the buffer is reset on open() anyway, there's no need to enable
logging when no-one has opened the log.

Because open clears the log, make it return EBUSY for mutliple open
calls.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>

Support for new features needed by the PPC 405EZ boards
introduced some errors in the MAL and EMAC feature handling.
This broke 'allmodconfig' builds as CONFIG_PPC_DCR_NATIVE is
not set for those.

This patch fixes these errors by wrapping the code in the
appropriate #ifdefs.
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

This patch adds support for the GPIO functions of PPC40x and PPC44x
SOCs.
Signed-off-by: Steve Falco <sfalco@harris.com>
Acked-by: Stefan Roese <sr@denx.de>
Acked-by: Sean MacLennan <smaclennan@pikatech.com>
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

Adds support for a HCU4 PPC405GPr based board from Netstal Maschinen AG.
Signed-off-by: Niklaus Giger <niklaus.giger@member.fsf.org>
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

Add simple defconfig for the AMCC PowerPC 405EZ Acadia evaluation board
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

This adds a cuboot wrapper for the AMCC PowerPC 405EZ Acadia board.  The
clocking code is derived from U-Boot, originally written by Stefan Roese.
Signed-off-by: Josh Boyer <jwboyer@linux.ibm.com>

This adds a common board file for almost all of the "simple" PowerPC 40x
boards that exist today.  This is intended to be a single place to add
support for boards that do not differ in platform support from most of the
evaluation boards that are used as reference platforms.  Boards that have
specific requirements or custom hardware setup should still have their own
board.c file.

The first board ported to this is the AMCC PowerPC 405EZ Acadia board.
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

This adds the AMCC PowerPC 405EZ chip to the cputable
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

Add the base DTS for the AMCC PowerPC 405EZ Acadia evalution board.
In addition to some of the normal PPC 40x peripherals, the Acadia
board has:
- 64 MiB PSRAM
- NOR and NAND flash
- Two USB 1.1 host ports
- Two CAN 2.0 ports
- ADC and DAC connectors
- LCD display

This adds the basic platform support to build from.
Signed-off-by: Josh Boyer <jwboyer@linux.vnet.ibm.com>

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  MIPS: Kill unused <asm/debug.h> inclusions
  MIPS: IP32: Add platform device for CMOS RTC; remove dead code
  RTC: M48T35: new RTC driver
  MIPS: IP27: Switch over to RTC class driver
  MIPS: DS1286: New RTC driver
  MIPS: IP22/28: Switch over to RTC class driver
  MIPS: PCI: Scan busses when they are registered
  MIPS: WGT634U: Add reset button support
  MIPS: BCM47xx: Use the new SSB GPIO API
  MIPS: BCM47xx: Remove references to BCM947XX
  MIPS: WGT634U: Add machine detection message
  MIPS: Align .data.cacheline_aligned based on CONFIG_MIPS_L1_CACHE_SHIFT
  MIPS: show_cpuinfo prints the type of the calling CPU
  MIPS: Fix wrong branch target in new spin_lock code.
  MIPS: Have a heart for a lonely, lost header file ...

proc_clear_tty() gets called with interrupts off (while holding the task list
lock) from sys_setid. This means that it needs the _irqsave version of the
locking primitives.
Reported-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Tested-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

The metronome driver produces warnings when built on x86-64 as it assumes that
size_t is an int. Use %Zd instead.
Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Read & write functions now call a generic transfer function, so identical
code in both routines could be eliminated. The result is easier to maintain
and smaller in source and binary code. Also, fix some checkpatch warnings.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

MPC5200 needs to have pipelining disabled for ATA to work. MPC5200B does not.
So, for the latter, don't touch the original setting from the bootloader.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

Current device trees do not have the device_type = soc property set
anymore.  Fix up the cuImage bootwrapper fragment to still find the IMMR
nodes.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>

When we skip unrecognized options in xfs_fs_remount we should just break
out of the switch and not return because otherwise we may skip clearing
the xfs-internal read-only flag.  This will only show up on some
operations like touch because most read-only checks are done by the VFS
which thinks this filesystem is r/w.  Eventually we should replace the
XFS read-only flag with a helper that always checks the VFS flag to make
sure they can never get out of sync.

Bug reported and fix verified by Marcel Beister on #xfs.
Bug fix verified by updated xfstests/189.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Eric Sandeen <sandeen@sandeen.net>
Signed-off-by: Timothy Shimmin <tes@sgi.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

* 'build_fix' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2:
  ocfs2: fix build error

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (158 commits)
  powerpc: Fix CHRP PCI config access for indirect_pci
  powerpc/chrp: Fix detection of Python PCI host bridge on IBM CHRPs
  powerpc: Fix 32-bit SMP boot on CHRP
  powerpc: Fix link errors on 32-bit machines using legacy DMA
  powerpc/pci: Improve detection of unassigned bridge resources
  hvc_console: Fix free_irq in spinlocked section
  powerpc: Get USE_STRICT_MM_TYPECHECKS working again
  powerpc: Reflect the used arguments in machine_init() prototype
  powerpc: Fix DMA offset for non-coherent DMA
  powerpc: fix fsl_upm nand driver modular build
  powerpc/83xx: add NAND support for the MPC8360E-RDK boards
  powerpc: FPGA support for GE Fanuc SBC610
  i2c: MPC8349E-mITX Power Management and GPIO expander driver
  powerpc: reserve two DMA channels for audio in MPC8610 HPCD device tree
  powerpc: document the "fsl,ssi-dma-channel" compatible property
  powerpc: disable CHRP and PMAC support in various defconfigs
  OF: add fsl,mcu-mpc8349emitx to the exception list
  powerpc/83xx: add DS1374 RTC support for the MPC837xE-MDS boards
  powerpc: remove support for bootmem-allocated memory for the DIU driver
  powerpc: remove non-dependent load fsl_booke PTE_64BIT
  ...