firefly-linux-kernel-4.4.55.git
12 years agopowerpc/iommu: Fix iommu pool initialization
Benjamin Herrenschmidt [Fri, 13 Jul 2012 07:45:49 +0000 (17:45 +1000)]
powerpc/iommu: Fix iommu pool initialization

The iommu pool patch has a bug where it would cause a crash when using
only one pool (based on the size of the DMA window).

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agoMerge remote-tracking branch 'kumar/next' into next
Benjamin Herrenschmidt [Fri, 13 Jul 2012 03:38:26 +0000 (13:38 +1000)]
Merge remote-tracking branch 'kumar/next' into next

Freescale updates for 3.6

12 years agopowerpc/eeh: Check handle_eeh_events() return value
Kleber Sacilotto de Souza [Thu, 12 Jul 2012 17:14:36 +0000 (17:14 +0000)]
powerpc/eeh: Check handle_eeh_events() return value

Function eeh_event_handler() dereferences the pointer returned by
handle_eeh_events() without checking, causing a crash if NULL was
returned, which is expected in some situations.

This patch fixes this bug by checking for the value returned by
handle_eeh_events() before dereferencing it.

Signed-off-by: Kleber Sacilotto de Souza <klebers@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org [v3.4+]
12 years agopowerpc/85xx: Add phy nodes in SGMII mode for MPC8536/44/72DS & P2020DS
Jia Hongtao [Thu, 12 Jul 2012 09:36:16 +0000 (17:36 +0800)]
powerpc/85xx: Add phy nodes in SGMII mode for MPC8536/44/72DS & P2020DS

In SGMII riser card different PHY chip are used with different external
IRQ from eTSEC. To support PHY link state auto detect in SGMII mode we
should add another group of PHY nodes for SGMII mode.

For MPC8572DS IRQ6 is used for PHY0~PHY1, IRQ7 is used for PHY2~PHY3.
For MPC8544DS and MPC8536DS IRQ6 is used for PHY0~PHY1.
For P2020DS IRQ5 is used for PHY1~PHY2.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Jia Hongtao <B38951@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/e500: add paravirt QEMU platform
Scott Wood [Wed, 11 Jul 2012 00:26:48 +0000 (19:26 -0500)]
powerpc/e500: add paravirt QEMU platform

This gives the kernel a paravirtualized machine to target, without
requiring both sides to pretend to be targeting a specific board
that likely has little to do with the host in KVM scenarios.  This
avoids the need to add new boards to QEMU just to be able to
run KVM on new CPUs.

As this is the first platform that can run with either e500v2 or
e500mc, CONFIG_PPC_E500MC is now a legitimately user configurable
option, so add a help text.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/mpc85xx_ds: convert to unified PCI init
Scott Wood [Wed, 11 Jul 2012 00:26:49 +0000 (19:26 -0500)]
powerpc/mpc85xx_ds: convert to unified PCI init

Similar to how the primary PCI bridge is identified by looking
for an isa subnode, we determine whether to apply uli exclusions
by looking for a uli subnode.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/fsl-pci: get PCI init out of board files
Scott Wood [Wed, 11 Jul 2012 00:26:47 +0000 (19:26 -0500)]
powerpc/fsl-pci: get PCI init out of board files

As an alternative incremental starting point to Jia Hongtao's patchset,
get the FSL PCI init out of the board files, but do not yet convert to a
platform driver.

Rather than having each board supply a magic register offset for
determining the "primary" bus, we look for which PCI host bridge
contains an ISA node within its subtree.  If there is no ISA node,
normally that would mean there is no primary bus, but until certain
bugs are fixed we arbitrarily designate a primary in this case.

Conversion to a platform driver and related improvements can happen
after this, as the ordering issues are sorted out.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Update corenet64_smp_defconfig
Shengzhou Liu [Wed, 11 Jul 2012 10:40:05 +0000 (18:40 +0800)]
powerpc/85xx: Update corenet64_smp_defconfig

Enable USB, MMC, SATA, LBC, MTD, NAND, SPI, PCIe, EDAC, VFAT, NFS, etc.

Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Update corenet32_smp_defconfig
Shengzhou Liu [Wed, 11 Jul 2012 10:40:04 +0000 (18:40 +0800)]
powerpc/85xx: Update corenet32_smp_defconfig

 - Enable NAND support
 - Enable CONFIG_PCI_MSI and CONFIG_MMC_SDHCI_OF

Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Rename P1021RDB-PC device trees to be consistent
Xu Jiucheng [Tue, 10 Jul 2012 08:39:20 +0000 (16:39 +0800)]
powerpc/85xx: Rename P1021RDB-PC device trees to be consistent

The board is really P1021RDB-PC, so rename from p1021rdb.* to p1021rdb-pc.*

Signed-off-by: Xu Jiucheng <Jiucheng.Xu@freescale.com>
Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/watchdog: move booke watchdog param related code to setup-common.c
Shaohui Xie [Wed, 11 Jul 2012 09:56:08 +0000 (17:56 +0800)]
powerpc/watchdog: move booke watchdog param related code to setup-common.c

Currently, BOOKE watchdog code for checking "wdt" and "wdt_period" is
in setup_32.c, it cannot be used in 64-bit, so move it to a common place
setup-common.c, which will be shared by 32-bit and 64-bit.

Also, replace the simple_strtoul with kstrtol.

Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agosound/aoa: Adapt to new i2c probing scheme
Andreas Schwab [Sat, 9 Jun 2012 13:58:56 +0000 (15:58 +0200)]
sound/aoa: Adapt to new i2c probing scheme

The i2c-powermac driver now creates the i2c devices properly
from the device-tree, including workarounds for broken or
missing device-tree bits, so let's just use the normal probe
methods and get rid of the hand made device creation code.

Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agoi2c/powermac: Improve detection of devices from device-tree
Benjamin Herrenschmidt [Mon, 18 Jun 2012 02:00:50 +0000 (12:00 +1000)]
i2c/powermac: Improve detection of devices from device-tree

This patch adds a number of workarounds for broken Apple device-trees
mostly around sound chips. It handles creating the missing audio codec
devices and works around various issues with missing addresses or
missing compatible properties.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Disable /dev/port interface on systems without an ISA bridge
Haren Myneni [Wed, 11 Jul 2012 05:18:44 +0000 (15:18 +1000)]
powerpc: Disable /dev/port interface on systems without an ISA bridge

Some power systems do not have legacy ISA devices. So, /dev/port is not
a valid interface on these systems. User level tools such as kbdrate is
trying to access the device using this interface which is causing the
system crash.

This patch will fix this issue by not creating this interface on these
powerpc systems.

Signed-off-by: Haren Myneni <haren@us.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agoof: Improve prom_update_property() function
Dong Aisheng [Wed, 11 Jul 2012 05:16:37 +0000 (15:16 +1000)]
of: Improve prom_update_property() function

prom_update_property() currently fails if the property doesn't
actually exist yet which isn't what we want. Change to add-or-update
instead of update-only, then we can remove a lot duplicated lines.

Suggested-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Dong Aisheng <dong.aisheng@linaro.org>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Add "memory" attribute for mfmsr()
Tiejun Chen [Wed, 11 Jul 2012 04:22:46 +0000 (14:22 +1000)]
powerpc: Add "memory" attribute for mfmsr()

Add "memory" attribute in inline assembly language as a compiler
barrier to make sure 4.6.x GCC don't reorder mfmsr().

Signed-off-by: Tiejun Chen <tiejun.chen@windriver.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org
12 years agopowerpc/ftrace: Fix assembly trampoline register usage
roger blofeld [Thu, 21 Jun 2012 05:27:14 +0000 (05:27 +0000)]
powerpc/ftrace: Fix assembly trampoline register usage

Just like the module loader, ftrace needs to be updated to use r12
instead of r11 with newer gcc's.

Signed-off-by: Roger Blofeld <blofeldus@yahoo.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org
12 years agopowerpc/hw_breakpoints: Fix incorrect pointer access
Naveen N. Rao [Fri, 6 Jul 2012 01:30:58 +0000 (01:30 +0000)]
powerpc/hw_breakpoints: Fix incorrect pointer access

If arch_validate_hwbkpt_settings() fails, bp->ctx won't be valid and the
kernel panics. Add a check to fix this.

Reported-by: Edjunior Barbosa Machado <emachado@linux.vnet.ibm.com>
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Put the gpr save/restore functions in their own section
Stephen Rothwell [Thu, 5 Jul 2012 21:09:40 +0000 (21:09 +0000)]
powerpc: Put the gpr save/restore functions in their own section

This allows the linker to know that calls to them do not need to switch
TOC and stop errors like the following when linking large configurations:

powerpc64-linux-ld: drivers/built-in.o: In function `.gpiochip_is_requested':
(.text+0x4): sibling call optimization to `_savegpr0_29' does not allow automatic multiple TOCs; recompile with -mminimal-toc or -fno-optimize-sibling-calls, or make `_savegpr0_29' extern

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Add VDSO version of getcpu
Anton Blanchard [Wed, 4 Jul 2012 20:37:11 +0000 (20:37 +0000)]
powerpc: Add VDSO version of getcpu

We have a request for a fast method of getting CPU and NUMA node IDs
from userspace. This patch implements a getcpu VDSO function,
similar to x86.

Ben suggested we use SPRG3 which is userspace readable. SPRG3 can be
modified by a KVM guest, so we save the SPRG3 value in the paca and
restore it when transitioning from the guest to the host.

I have a glibc patch that implements sched_getcpu on top of this.
Testing on a POWER7:

baseline: 538 cycles
vdso:      30 cycles

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Add a symbol for hypervisor trampolines
Michael Ellerman [Tue, 3 Jul 2012 20:29:41 +0000 (20:29 +0000)]
powerpc: Add a symbol for hypervisor trampolines

Purely for cosmetic purposes, otherwise it can appear that we are in
single_step_pSeries() which is slightly confusing.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/cpuidle: Fixes for pseries_idle hotplug notifier
Deepthi Dharwar [Tue, 3 Jul 2012 20:07:22 +0000 (20:07 +0000)]
powerpc/cpuidle: Fixes for pseries_idle hotplug notifier

Currently the call to pseries_notify_cpuidle_add_cpu(), that takes
action on the cpuidle front when a cpu is added/removed
is being made from smp_xics_setup_cpu().
This caused lockdep issues as
reported https://lkml.org/lkml/2012/5/17/2

On addition of each cpu,
resources were cleared and re-allocated each time, all in critical
section as part of start_secondary() call were interrupts are disabled.
To resolve this issue, the pseries_notify_cpuidle_add_cpu() call is
is being replaced by a hotplug notifier which
would prevent cpuidle resources from being
released and allocated each time cpu is onlined in the critical code path.
It was fixed in https://lkml.org/lkml/2012/5/18/174.

Also it is essential to call cpuidle_enable/disable_device
between  cpuidle_pause_and_lock()  and
cpuidle_resume_and_unlock()  when used externally
to avoid race conditions. Add support for CPU_ONLINE_FROZEN
and CPU_DEAD_FROZEN as part of hotplug notify event for
pseries_idle  and unregister hotplug notifier
while exiting out. The above mentioned issues
are fixed as part of this patch.

Signed-off-by: Deepthi Dharwar <deepthi@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fixup oddity in entry_32.S
Benjamin Herrenschmidt [Sun, 1 Jul 2012 18:47:17 +0000 (18:47 +0000)]
powerpc: Fixup oddity in entry_32.S

When I "fixed" the CONFIG_TRACE_IRQFLAGS case on interrupt entry,
I screwed up a little bit with the test for user space vs. kernel.

The code is fine, there's just some dead code around it. I basically
removed the test and always create the added stack frame whether
coming from user or kernel since in any case we do need to save
a bunch of volatile registers or bad things would happen (we can
take page faults in the kernel for example).

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Use CURRENT_THREAD_INFO instead of open coded assembly
Stuart Yoder [Thu, 5 Jul 2012 04:41:35 +0000 (04:41 +0000)]
powerpc: Use CURRENT_THREAD_INFO instead of open coded assembly

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/85xx: Add ucc uart support for p1025rdb
Zhicheng Fan [Tue, 10 Jul 2012 07:52:10 +0000 (15:52 +0800)]
powerpc/85xx: Add ucc uart support for p1025rdb

Add device tree nodes to enable ucc uart support on P1025RDB.

Signed-off-by: Zhicheng Fan <B32736@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc: select PPC_CLOCK unconditionally for FSL_SOC
Shawn Guo [Fri, 30 Mar 2012 05:38:56 +0000 (13:38 +0800)]
powerpc: select PPC_CLOCK unconditionally for FSL_SOC

Freescale PowerPC SoCs share a number of IP blocks with Freescale
ARM/IMX SoCs, FlexCAN, SSI, FEC, eSDHC, USB, etc.  There are some
effort consolidating those drivers to make them work for both
architectures.

One outstanding difference between two architectures is ARM/IMX will
turn off module clocks during platform initialization for power saving
and expects drivers manage clocks using clk API, while PowerPC
mostly does not do that, and thus does not always build in clk API.

Listing all those driver Kconfig options in "select PPC_CLOCK if" seems
not scalable for long term maintenance, and could easily introduce
Kconfig recursive dependency.  This patch chooses to select PPC_CLOCK
unconditionally for FSL_SOC to always build clk API for PowerPC in.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/qe: set IReady in QE Microcode Upload
Kokoris, Ioannis [Fri, 11 Nov 2011 16:05:11 +0000 (17:05 +0100)]
powerpc/qe: set IReady in QE Microcode Upload

QE Microcode Initialization using qe_upload_microcode() does not work on
P1021 if the IRAM-Ready register is not set after the microcode upload. Add
a definition for the "I-RAM Ready" register and sets it upon microcode
upload completion.

Signed-off-by: Ioannis Kokkoris <ioannis.kokoris@siemens-enterprise.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: MPC8572DS - Update the MSI interrupts into 4-cell format
Jia Hongtao [Tue, 10 Jul 2012 06:08:26 +0000 (14:08 +0800)]
powerpc/85xx: MPC8572DS - Update the MSI interrupts into 4-cell format

With 2-cell format interrupts of MSI PCIe ethernet card can not work.

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Jia Hongtao <B38951@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: MPC8572DS - Fix eTSEC is not available on core1 of AMP boot
Jia Hongtao [Tue, 10 Jul 2012 06:08:25 +0000 (14:08 +0800)]
powerpc/85xx: MPC8572DS - Fix eTSEC is not available on core1 of AMP boot

The issue log on core1 is:
root@mpc8572ds:~# ifconfig eth0 10.192.208.244
net eth0: could not attach to PHY
SIOCSIFFLAGS: No such device

To attach PHY node mdio@24520 should not be disabled in dts of core1.
Because all PHYs are controlled through this node as follows:

mdio@24520 {
phy0: ethernet-phy@0 {
interrupts = <10 1 0 0>;
reg = <0x0>;
};
phy1: ethernet-phy@1 {
interrupts = <10 1 0 0>;
reg = <0x1>;
};
phy2: ethernet-phy@2 {
interrupts = <10 1 0 0>;
reg = <0x2>;
};
phy3: ethernet-phy@3 {
interrupts = <10 1 0 0>;
reg = <0x3>;
};

tbi0: tbi-phy@11 {
reg = <0x11>;
device_type = "tbi-phy";
};
};

Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Jia Hongtao <B38951@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/watchdog: replace CONFIG_FSL_BOOKE with CONFIG_PPC_FSL_BOOK3E
Shaohui Xie [Fri, 11 May 2012 05:33:40 +0000 (13:33 +0800)]
powerpc/watchdog: replace CONFIG_FSL_BOOKE with CONFIG_PPC_FSL_BOOK3E

CONFIG_FSL_BOOKE is only defined in 32-bit, CONFIG_PPC_FSL_BOOK3E is
defined in both 32-bit and 64-bit, so use CONFIG_PPC_FSL_BOOK3E to make
driver work in 32-bit & 64-bit.

Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/mpic: Use the MPIC_LARGE_VECTORS flag for FSL MPIC.
Varun Sethi [Mon, 9 Jul 2012 08:46:35 +0000 (14:16 +0530)]
powerpc/mpic: Use the MPIC_LARGE_VECTORS flag for FSL MPIC.

We should use the MPIC_LARG_VECTORS flag while intializing the MPIC.
This prevents us from eating in to hardware vector number space (MSIs)
while setting up internal sources.

Signed-off-by: Varun Sethi <Varun.Sethi@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/e500: make load_up_spe a normal fuction
Liu Yu [Thu, 1 Mar 2012 01:20:19 +0000 (09:20 +0800)]
powerpc/e500: make load_up_spe a normal fuction

So that we can call it when improving SPE switch like book3e did for fp
switch.

Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Olivia Yin <hong-hua.yin@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Add BSC9131 RDB Support
Prabhakar Kushwaha [Thu, 22 Mar 2012 04:54:15 +0000 (10:24 +0530)]
powerpc/85xx: Add BSC9131 RDB Support

BSC9131RDB is a Freescale reference design board for BSC9131 SoC. The
BSC9131 is integrated SoC that targets Femto base station market. It
combines Power Architecture e500v2 and DSP StarCore SC3850 core
technologies with MAPLE-B2F baseband acceleration processing elements.

The BSC9131 SoC includes the following function and features:
    . Power Architecture subsystem including a e500 processor with 256-Kbyte
    shared L2 cache
    . StarCore SC3850 DSP subsystem with a 512-Kbyte private L2 cache
    . The Multi Accelerator Platform Engine for Femto BaseStation Baseband
      Processing (MAPLE-B2F)
    . A multi-standard baseband algorithm accelerator for Channel
      Decoding/Encoding, Fourier Transforms, UMTS chip rate processing, LTE
      UP/DL Channel processing, and CRC algorithms
    . Consists of accelerators for Convolution, Filtering, Turbo Encoding,
      Turbo Decoding, Viterbi decoding, Chiprate processing, and Matrix
      Inversion operations
    . DDR3/3L memory interface with 32-bit data width without ECC and 16-bit
      with ECC, up to 400-MHz clock/800 MHz data rate
    . Dedicated security engine featuring trusted boot
    . DMA controller
    . OCNDMA with four bidirectional channels
    . Interfaces
    . Two triple-speed Gigabit Ethernet controllers featuring network
      acceleration including IEEE 1588. v2 hardware support and
      virtualization (eTSEC)
    . eTSEC 1 supports RGMII/RMII
    . eTSEC 2 supports RGMII
    . High-speed USB 2.0 host and device controller with ULPI interface
    . Enhanced secure digital (SD/MMC) host controller (eSDHC)
    . Antenna interface controller (AIC), supporting three industry standard
      JESD207/three custom ADI RF interfaces (two dual port and one single
      port) and three MAXIM's MaxPHY serial interfaces
    . ADI lanes support both full duplex FDD support and half duplex TDD
      support
    . Universal Subscriber Identity Module (USIM) interface that facilitates
      communication to SIM cards or Eurochip pre-paid phone cards
    . TDM with one TDM port
    . Two DUART, four eSPI, and two I2C controllers
    . Integrated Flash memory controller (IFC)
    . TDM with 256 channels
    . GPIO
    . Sixteen 32-bit timers

The DSP portion of the SoC consists of DSP core (SC3850) and various
accelerators pertaining to DSP operations.

 BSC9131RDB Overview
 ----------------------
    BSC9131 SoC
    1Gbyte DDR3 (on board DDR)
    128Mbyte 2K page size NAND Flash
    256 Kbit M24256 I2C EEPROM
    128 Mbit SPI Flash memory
    USB-ULPI
    eTSEC1: Connected to RGMII PHY
    eTSEC2: Connected to RGMII PHY
    DUART interface: supports one UARTs up to 115200 bps for console display

 Linux runs on e500v2 core and access some DSP peripherals like AIC

Signed-off-by: Ramneek Mehresh <ramneek.mehresh@freescale.com>
Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com>
Signed-off-by: Akhil Goyal <Akhil.Goyal@freescale.com>
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Rajan Srivastava <rajan.srivastava@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agoRevert "powerpc/p3060qds: Add support for P3060QDS board"
Timur Tabi [Thu, 5 Jul 2012 22:07:53 +0000 (17:07 -0500)]
Revert "powerpc/p3060qds: Add support for P3060QDS board"

This reverts commit 96cc017c5b7ec095ef047d3c1952b6b6bbf98943.

The P3060 was cancelled before it went into production, so there's no point
in supporting it.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: use the BRx registers to enable indirect mode on the P1022DS
Timur Tabi [Thu, 5 Jul 2012 15:08:28 +0000 (10:08 -0500)]
powerpc/85xx: use the BRx registers to enable indirect mode on the P1022DS

In order to enable the DIU video controller on the P1022DS, the FPGA needs
to be switched to "indirect mode", where the localbus is disabled and
the FPGA is accessed via writes to localbus chip select signals CS0 and CS1.

To obtain the address of CS0 and CS1, the platform driver uses an "indirect
pixis mode" device tree node.  This node assumes that the localbus 'ranges'
property is sorted in chip-select order.  That is, reg value 0 maps to
CS0, reg value 1 maps to CS1, etc.  This is how the 'ranges' property is
supposed to be arranged.

Unfortunately, the 'ranges' property is often mis-arranged, and not just on
the P1022DS.  Linux normally does not care, since it does not program the
localbus.  But the indirect-mode code on the P1022DS does care.

The "proper" fix is to have U-Boot fix the 'ranges' property, but this would
be too cumbersome.  The names and 'reg' properties of all the localbus
devices would also need to be updated, and determining which localbus device
maps to which chip select is board-specific.

Instead, we determine the CS0/CS1 base addresses the same way that U-boot
does -- by reading the BRx registers directly and mapping them to physical
addresses.  This code is simpler and more reliable, and it does not require
a U-boot or device tree change.

Since the indirect pixis device tree node is no longer needed, the node is
deleted from the DTS.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/p2041rdb: add NAND node in device tree
Shaohui Xie [Fri, 29 Jun 2012 09:41:11 +0000 (17:41 +0800)]
powerpc/p2041rdb: add NAND node in device tree

NAND on p2041 uses CS1 as chip select.

Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc: remove Wind River SBC8560 support
Paul Gortmaker [Tue, 26 Jun 2012 19:54:27 +0000 (15:54 -0400)]
powerpc: remove Wind River SBC8560 support

This reference board dates back to 2004, and is largely a legacy
EOL product.  The MPC8560 is a pre e500v2 CPU.  The SBC8548 is
a more modern, better e500v2 target for people to use as a
reference board with today's kernels, should they require one.

Removing support for it will also allow us to remove some
sbc8560 specific quirk handling in 8250 UART code, and some
MTD mapping support.

Cc: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Add P1024rdb dts support
Tang Yuantian [Thu, 24 May 2012 09:08:27 +0000 (17:08 +0800)]
powerpc/85xx: Add P1024rdb dts support

Signed-off-by: Jin Qing <b24347@freescale.com>
Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Tang Yuantian <Yuantian.Tang@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Add P1024rdb board support
Tang Yuantian [Thu, 24 May 2012 09:08:28 +0000 (17:08 +0800)]
powerpc/85xx: Add P1024rdb board support

The p1024rdb has the similar feature as the p1020rdb. Therefore, p1024rdb use
the same platform file as the p1/p2 rdb board.
Overview of P2020RDB platform
- DDR3 1G
- NOR flash 16M
- 3 Ethernet interfaces
- NAND Flash 32M
- SPI EEPROM 16M
- SD/MMC
- 2 USB ports
- 4 TDM ports

Signed-off-by: Jin Qing <b24347@freescale.com>
Signed-off-by: Li Yang <leoli@freescale.com>
Signed-off-by: Tang Yuantian <Yuantian.Tang@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/p1010rdb: add EEPROMs to device tree
Gustavo Zacarias [Wed, 23 May 2012 14:35:18 +0000 (11:35 -0300)]
powerpc/p1010rdb: add EEPROMs to device tree

Add EEPROM to the P1010RDB device tree.
The 24c01 acts as a memory SPD so it shouldn't be overwritten without
care.
The 24c256 is a general purpose memory.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agoRevert "powerpc/85xx: p2020rdb - move the NAND address."
Sebastian Andrzej Siewior [Wed, 9 May 2012 19:53:07 +0000 (21:53 +0200)]
Revert "powerpc/85xx: p2020rdb - move the NAND address."

This reverts commit 0c00f65653389a408dfbbee7578e671664eea26a.
The initial commit was my fault. There are two boards out there:
P2020RDB and P2020RDB-PC. I wasn't aware of that and assumed that I have
a RDB board in front of me while I the RDB-PC. This patch makes it work
for the RDB-PC variant and breaks it for the RDB. Now there is a device
tree file available for the RDB-PC which was not there earlier. So with
this revert, everything gets back to normal :)

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/82xx: add SPI support for mgcoge
Holger Brunck [Tue, 8 May 2012 13:57:22 +0000 (15:57 +0200)]
powerpc/82xx: add SPI support for mgcoge

Add spi support for mgcoge into the platform code and the dts
file. Additionaly SPIDEV is switched on in the defconfig and the
updates for the newer kernel version are committed. The SPI
interface is used to drive the Maxim DS3106 clock chip.

Signed-off-by: Holger Brunck <holger.brunck@keymile.com>
cc: Heiko Schocher <hs@denx.de>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/83xx: update defconfig for kmeter1
Holger Brunck [Tue, 8 May 2012 13:57:21 +0000 (15:57 +0200)]
powerpc/83xx: update defconfig for kmeter1

Switch on UBIFS, HOTPLUG and TIPC and update the config to
the latest kernel version.

Signed-off-by: Holger Brunck <holger.brunck@keymile.com>
cc: Heiko Schocher <hs@denx.de>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/83xx: fix RGMII AC values workaround for km83xx
Christian Herzig [Tue, 8 May 2012 13:57:20 +0000 (15:57 +0200)]
powerpc/83xx: fix RGMII AC values workaround for km83xx

Fix RGMII workaround code in km83xx.c for MPC8360E and MPC8358E that it
correctly identifes all affected SoC chip models and applies the
workarounds appropriate for 2.0 and 2.1 revisions as per Freescale
MPC8360ECE Errata document Rev.5(9/2011) item QE_ENET10.

Signed-off-by: Christian Herzig <christian.herzig@keymile.com>
Signed-off-by: Holger Brunck <holger.brunck@keymile.com>
cc: Heiko Schocher <hs@denx.de>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/83xx: use for_each_node_by_name for km83xx.c
Holger Brunck [Tue, 8 May 2012 13:57:19 +0000 (15:57 +0200)]
powerpc/83xx: use for_each_node_by_name for km83xx.c

Signed-off-by: Holger Brunck <holger.brunck@keymile.com>
cc: Heiko Schocher <hs@denx.de>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/p1022ds: Add RTC support
Jerry Huang [Tue, 17 Apr 2012 01:42:35 +0000 (09:42 +0800)]
powerpc/p1022ds: Add RTC support

Add the RTC support into the p1022ds device tree

Signed-off-by: Jerry Huang <Chang-Ming.Huang@freescale.com>
Acked-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/85xx: Enable MTD/NOR/NAND options by default in defconfig
Shengzhou Liu [Fri, 27 Apr 2012 03:01:00 +0000 (11:01 +0800)]
powerpc/85xx: Enable MTD/NOR/NAND options by default in defconfig

Enable MTD/NOR/NAND options by default in mpc85xx_defconfig and
mpc85xx_smp_defconfig to support NOR, NAND flash.

Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/fsl: Distribute interrupts on all CPUs by default
Kim Phillips [Fri, 13 Apr 2012 22:26:15 +0000 (17:26 -0500)]
powerpc/fsl: Distribute interrupts on all CPUs by default

At least for crypto/IPSec, doing so provides users with a better
performance experience out of the box.

Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc/fsl/pci: Fix when quirk_fsl_pcie_header is freed up
Matias Garcia [Mon, 19 Sep 2011 15:35:11 +0000 (11:35 -0400)]
powerpc/fsl/pci: Fix when quirk_fsl_pcie_header is freed up

Change quirk_fsl_pcie_header from __init to __devinit to ensure if we
have a runtime access (like via an FPGA being loaded after boot on the
PCIe link) that we dont access randomly freed memory.

Signed-off-by: Matias Garcia <mgarcia@rossvideo.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
12 years agopowerpc: Turn on BPF_JIT in ppc64_defconfig
Michael Ellerman [Thu, 21 Jun 2012 17:52:29 +0000 (17:52 +0000)]
powerpc: Turn on BPF_JIT in ppc64_defconfig

Matt added BPF_JIT support in commit 0ca87f05, but currently none of our
defconfigs build it. Turn that sucker on.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: IOMMU fault injection
Anton Blanchard [Sun, 24 Jun 2012 18:26:17 +0000 (18:26 +0000)]
powerpc: IOMMU fault injection

Add the ability to inject IOMMU faults. We enable this per device
via a fail_iommu sysfs property, similar to fault injection on other
subsystems.

An example:

...
0003:01:00.1 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) (rev 02)

To inject one error to this device:

echo 1 > /sys/bus/pci/devices/0003:01:00.1/fail_iommu
echo 1 > /sys/kernel/debug/fail_iommu/probability
echo 1 > /sys/kernel/debug/fail_iommu/times

As feared, the first failure injected on the be3 results in an
unrecoverable error, taking down both functions of the card
permanently:

be2net 0003:01:00.1: Unrecoverable error in the card

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Call dma_debug_add_bus for PCI and VIO buses
Anton Blanchard [Sun, 24 Jun 2012 18:25:28 +0000 (18:25 +0000)]
powerpc: Call dma_debug_add_bus for PCI and VIO buses

The DMA API debug code has hooks to verify all DMA entries have been
freed at time of hot unplug. We need to call dma_debug_add_bus for
this to work.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/vio: Separate vio bus probe and device probe
Anton Blanchard [Sun, 24 Jun 2012 18:24:44 +0000 (18:24 +0000)]
powerpc/vio: Separate vio bus probe and device probe

Similar to PCI, separate the bus probe from device probe. This allows
us to attach bus notifiers for DMA debug and IOMMU fault injection
before devices have been probed.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/vio: Remove dma not supported warnings
Anton Blanchard [Sun, 24 Jun 2012 18:23:53 +0000 (18:23 +0000)]
powerpc/vio: Remove dma not supported warnings

During boot we see a number of these warnings:

vio 30000000: Warning: IOMMU dma not supported: mask 0xffffffffffffffff, table unavailable

The reason for this is that we set IOMMU properties for all VIO
devices even if they are not DMA capable.

Only set DMA ops, table and mask for devices with a DMA window.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/pseries: Fix whitespace in eeh
Michael Neuling [Tue, 26 Jun 2012 18:12:01 +0000 (18:12 +0000)]
powerpc/pseries: Fix whitespace in eeh

Signed-off-by: Michael Neuling <mikey@neuling.org>
Acked-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/perf: Use perf_instruction_pointer in callchains
Anton Blanchard [Tue, 26 Jun 2012 01:04:20 +0000 (01:04 +0000)]
powerpc/perf: Use perf_instruction_pointer in callchains

We use SIAR or regs->nip for the instruction pointer depending on
the PMU configuration, but we always use regs->nip in the callchain.

Use perf_instruction_pointer so the backtrace is consistent.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/perf: Always use pt_regs for userspace samples
Anton Blanchard [Tue, 26 Jun 2012 01:02:15 +0000 (01:02 +0000)]
powerpc/perf: Always use pt_regs for userspace samples

At the moment we always use the SIAR if the PMU supports continuous
sampling. Unfortunately the SIAR and the PMU exception are not
synchronised for non marked events so we can end up with callchains
that dont make sense.

The following patch checks the HV and PR bits for samples coming from
userspace and always uses pt_regs for them. Userspace will never have
interrupts off so there is no real advantage to using the SIAR for
non marked events in userspace.

I had experimented with a patch that did a similar thing for kernel
samples but we lost a significant amount of information. I was
unable to profile any of our early exception code for example.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/perf: Move code to select SIAR or pt_regs into perf_read_regs
Anton Blanchard [Tue, 26 Jun 2012 01:01:36 +0000 (01:01 +0000)]
powerpc/perf: Move code to select SIAR or pt_regs into perf_read_regs

The logic to choose whether to use the SIAR or get the information
out of pt_regs is going to get more complicated, so do it once in
perf_read_regs.

We overload regs->result which is gross but we are already doing it
with regs->dsisr.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/perf: Create mmcra_sihv/mmcra_sipv helpers
Anton Blanchard [Tue, 26 Jun 2012 01:00:13 +0000 (01:00 +0000)]
powerpc/perf: Create mmcra_sihv/mmcra_sipv helpers

We want to access the MMCRA_SIHV and MMCRA_SIPR bits elsewhere so
create mmcra_sihv and mmcra_sipr which hide the differences between
the old and new layout of the bits.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Enforce usage of RA 0-R31 where possible
Michael Neuling [Mon, 25 Jun 2012 13:33:25 +0000 (13:33 +0000)]
powerpc: Enforce usage of RA 0-R31 where possible

Some macros use RA where when RA=R0 the values is 0, so make this
the enforced mnemonic in the macro.

Idea suggested by Andreas Schwab.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Add defines for RA 0-R31
Michael Neuling [Mon, 25 Jun 2012 13:33:24 +0000 (13:33 +0000)]
powerpc: Add defines for RA 0-R31

R0 is special since it'll be 0.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Enforce usage of R0-R31 where possible
Michael Neuling [Mon, 25 Jun 2012 13:33:23 +0000 (13:33 +0000)]
powerpc: Enforce usage of R0-R31 where possible

Enforce the use of R0-R31 in macros where possible now we have all the
fixes in.

R0-R31 macros are removed here so that can't be used anymore.  They
should not be defined anywhere.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Introduce new __REG_R macros
Michael Neuling [Mon, 25 Jun 2012 13:33:22 +0000 (13:33 +0000)]
powerpc: Introduce new __REG_R macros

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Start using ___PPC_RA/B/S/T where necessary
Michael Neuling [Mon, 25 Jun 2012 13:33:21 +0000 (13:33 +0000)]
powerpc: Start using ___PPC_RA/B/S/T where necessary

Now have ___PPC_RA/B/S/T we can use it in some places.  These are
places where we can't use the existing defines which will soon enforce
R0-R31 usage.

The macros being changed here are being used in inline asm, which
can't convert to enforce the R0-R31 usage.

bpf_jit uses a mix of both generated and non-generated with the same
code, so just convert all these to use the ___PPC_R versions which
won't enforce R usage later.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Introduce new ___PPC_RA/B/S/T macros
Michael Neuling [Mon, 25 Jun 2012 13:33:20 +0000 (13:33 +0000)]
powerpc: Introduce new ___PPC_RA/B/S/T macros

These are currently the same as __PPC_RA/B/S/T but we'll wrap them
soon.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fix VSX macros so register names aren't wrapped
Michael Neuling [Mon, 25 Jun 2012 13:33:19 +0000 (13:33 +0000)]
powerpc: Fix VSX macros so register names aren't wrapped

We need to do this so we can enforce the name of a and b in called
macros PPC_RA/B later.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fixes for instructions not using correct register naming
Michael Neuling [Mon, 25 Jun 2012 13:33:18 +0000 (13:33 +0000)]
powerpc: Fixes for instructions not using correct register naming

These macros are using integers where they could be using logical
names since they take registers.

We are going to enforce this soon, so fix these up now.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Change LOAD_REG_ADDR to use real register names
Michael Neuling [Mon, 25 Jun 2012 13:33:17 +0000 (13:33 +0000)]
powerpc: Change LOAD_REG_ADDR to use real register names

LOAD_REG_ADDR define is just a wrapper around real instructions so we
can just use real register names here (ie. lower case).

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Change mtcrf to use real register names
Michael Neuling [Mon, 25 Jun 2012 13:33:16 +0000 (13:33 +0000)]
powerpc: Change mtcrf to use real register names

mtocrf define is just a wrapper around the real instructions so we can
just use real register names here (ie. lower case).

Also remove braces in macro so this is possible.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Move and fix MTMSR_EERI definition
Benjamin Herrenschmidt [Wed, 4 Jul 2012 04:49:12 +0000 (14:49 +1000)]
powerpc: Move and fix MTMSR_EERI definition

Move this duplicated definition to ppc_asm.h and remove the
braces which prevent the use of %rN register names

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Merge VCPU_GPR
Michael Neuling [Mon, 25 Jun 2012 13:33:15 +0000 (13:33 +0000)]
powerpc: Merge VCPU_GPR

Merge the defines of VCPU_GPR from different places.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Merge STK_REG/PARAM/FRAMESIZE
Michael Neuling [Mon, 25 Jun 2012 13:33:14 +0000 (13:33 +0000)]
powerpc: Merge STK_REG/PARAM/FRAMESIZE

Merge the defines of STACKFRAMESIZE, STK_REG, STK_PARAM from different
places.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/pasemi: Move lbz/stbciz to ppc-opcode.h
Michael Neuling [Mon, 25 Jun 2012 13:33:13 +0000 (13:33 +0000)]
powerpc/pasemi: Move lbz/stbciz to ppc-opcode.h

move lbz/stbciz to ppc-opcode.h.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Convert to %r for all GPR usage
Michael Neuling [Mon, 25 Jun 2012 13:33:12 +0000 (13:33 +0000)]
powerpc: Convert to %r for all GPR usage

Now all the fixes are in place, let's rock-n-roll!

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fix usage of register macros getting ready for %r0 change
Michael Neuling [Mon, 25 Jun 2012 13:33:10 +0000 (13:33 +0000)]
powerpc: Fix usage of register macros getting ready for %r0 change

Anything that uses a constructed instruction (ie. from ppc-opcode.h),
need to use the new R0 macro, as %r0 is not going to work.

Also convert usages of macros where we are just determining an offset
(usually for a load/store), like:
std r14,STK_REG(r14)(r1)
Can't use STK_REG(r14) as %r14 doesn't work in the STK_REG macro since
it's just calculating an offset.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Modify macro ready for %r0 register change
Michael Neuling [Mon, 25 Jun 2012 13:33:09 +0000 (13:33 +0000)]
powerpc: Modify macro ready for %r0 register change

The assembler doesn't take %r0 register arguments in braces, so remove them.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Add defines for R0-R31
Michael Neuling [Mon, 25 Jun 2012 13:33:08 +0000 (13:33 +0000)]
powerpc: Add defines for R0-R31

We are going to use these later and convert r0 to %r0 etc.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agoMerge branch 'merge' into next
Benjamin Herrenschmidt [Tue, 10 Jul 2012 09:16:43 +0000 (19:16 +1000)]
Merge branch 'merge' into next

We want to bring in the latest IRQ fixes

12 years agotty/hvc_opal: Fix debug function name
Benjamin Herrenschmidt [Wed, 4 Jul 2012 05:51:10 +0000 (15:51 +1000)]
tty/hvc_opal: Fix debug function name

udbg_init_debug_opal() should be udbg_init_debug_opal_raw() as
the caller in arch/powerpc/kernel/udbg.c expects

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc/numa: Avoid stupid uninitialized warning from gcc
Benjamin Herrenschmidt [Thu, 5 Jul 2012 16:30:33 +0000 (16:30 +0000)]
powerpc/numa: Avoid stupid uninitialized warning from gcc

Newer gcc are being a bit blind here (it's pretty obvious we don't
reach the code path using the array if we haven't initialized the
pointer) but none of that is performance critical so let's just
silence it.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
12 years agopowerpc: Fix build of some debug irq code
Benjamin Herrenschmidt [Tue, 10 Jul 2012 08:37:56 +0000 (18:37 +1000)]
powerpc: Fix build of some debug irq code

There was a typo, checking for CONFIG_TRACE_IRQFLAG instead of
CONFIG_TRACE_IRQFLAGS causing some useful debug code to not be
built

This in turns causes a build error on BookE 64-bit due to incorrect
semicolons at the end of a couple of macros, so let's fix that too

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org [v3.4]
12 years agopowerpc: More fixes for lazy IRQ vs. idle
Benjamin Herrenschmidt [Tue, 10 Jul 2012 08:36:40 +0000 (18:36 +1000)]
powerpc: More fixes for lazy IRQ vs. idle

Looks like we still have issues with pSeries and Cell idle code
vs. the lazy irq state. In fact, the reset fixes that went upstream
are exposing the problem more by causing BUG_ON() to trigger (which
this patch turns into a WARN_ON instead).

We need to be careful when using a variant of low power state that
has the side effect of turning interrupts back on, to properly set
all the SW & lazy state to look as if everything is enabled before
we enter the low power state with MSR:EE off as we will return with
MSR:EE on. If not, we have a discrepancy of state which can cause
things to go very wrong later on.

This patch moves the logic into a helper and uses it from the
pseries and cell idle code. The power4/970 idle code already got
things right (in assembly even !) so I'm not touching it. The power7
"bare metal" idle code is subtly different and correct. Remains PA6T
and some hypervisor based Cell platforms which have questionable
code in there, but they are mostly dead platforms so I'll fix them
when I manage to get final answers from the respective maintainers
about how the low power state actually works on them.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org [v3.4]
12 years agoMerge tag 'regulator-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Linus Torvalds [Mon, 9 Jul 2012 20:43:02 +0000 (13:43 -0700)]
Merge tag 'regulator-3.5' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "A smallish fix for a lock dependency issue which affects a bunch of
  Qualcomm boards that do unusually complicated things with their
  regulators, the API is unlikely to be called by any other system."

* tag 'regulator-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: Fix recursive mutex lockdep warning

12 years agogspca_sn9c20x: Fix NULL pointer dereference
Hans de Goede [Sun, 8 Jul 2012 17:41:14 +0000 (19:41 +0200)]
gspca_sn9c20x: Fix NULL pointer dereference

Don't call v4l2_ctrl_g_ctrl on ctrls which the model cam in question
does not have.

Reported-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
[ Taken directly, since Mauro is on vacation ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'virtio-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty...
Linus Torvalds [Mon, 9 Jul 2012 20:39:56 +0000 (13:39 -0700)]
Merge tag 'virtio-for-linus' of git://git./linux/kernel/git/rusty/linux-2.6-for-linus

Pull minor virtio-balloon fix from Rusty Russell:
 "Theoretical fix, which greatly simplifies upcoming balloon patches
  which will go in via some vm tree."

* tag 'virtio-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  virtio-balloon: fix add/get API use

12 years agoMerge tag 'rpmsg-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad...
Linus Torvalds [Mon, 9 Jul 2012 20:24:00 +0000 (13:24 -0700)]
Merge tag 'rpmsg-3.5-fixes' of git://git./linux/kernel/git/ohad/rpmsg

Pull rpmsg fixes from Ohad Ben-Cohen:
 "Fixing two (somewhat rare) endpoint-related race issues, both of which
  were reported by Fernando Guzman Lugo."

* tag 'rpmsg-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/rpmsg:
  rpmsg: make sure inflight messages don't invoke just-removed callbacks
  rpmsg: avoid premature deallocation of endpoints

12 years agoMerge tag 'remoteproc-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 9 Jul 2012 20:23:21 +0000 (13:23 -0700)]
Merge tag 'remoteproc-3.5-fixes' of git://git./linux/kernel/git/ohad/remoteproc

Pull remoteproc fixes from Ohad Ben-Cohen:
 "Two build-related remoteproc fixes for 3.5."

* tag 'remoteproc-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
  remoteproc: fix missing CONFIG_FW_LOADER configurations
  remoteproc/omap: fix randconfig unmet direct dependencies

12 years agoMerge tag 'hwspinlock-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 9 Jul 2012 20:22:40 +0000 (13:22 -0700)]
Merge tag 'hwspinlock-3.5-fixes' of git://git./linux/kernel/git/ohad/hwspinlock

Pull hwspinlock fix from Ohad Ben-Cohen:
 "A single hwspinlock core fix for multiple hwspinlock devices
  scenarios, from Shinya Kuribayashi."

* tag 'hwspinlock-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/hwspinlock:
  hwspinlock/core: use global ID to register hwspinlocks on multiple devices

12 years agoMerge tag 'iommu-fixes-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 9 Jul 2012 17:57:21 +0000 (10:57 -0700)]
Merge tag 'iommu-fixes-v3.5-rc5' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU fixes from Joerg Roedel:
 "The patches fix several issues in the AMD IOMMU driver, the NVidia
  SMMU driver, and the DMA debug code.

  The most important fix for the AMD IOMMU solves a problem with SR-IOV
  devices where virtual functions did not work with IOMMU enabled.  The
  NVidia SMMU patch fixes a possible sleep while spin-lock situation
  (queued the small fix for v3.5, a better but more intrusive fix is
  coming for v3.6).  The DMA debug patches fix a possible data
  corruption issue due to bool vs u32 usage."

* tag 'iommu-fixes-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/amd: fix type bug in flush code
  dma-debug: debugfs_create_bool() takes a u32 pointer
  iommu/tegra: smmu: Fix unsleepable memory allocation
  iommu/amd: Initialize dma_ops for hotplug and sriov devices
  iommu/amd: Fix missing iommu_shutdown initialization in passthrough mode

12 years agovirtio-balloon: fix add/get API use
Michael S. Tsirkin [Mon, 2 Jul 2012 07:33:08 +0000 (10:33 +0300)]
virtio-balloon: fix add/get API use

Since ee7cd8981e15bcb365fc762afe3fc47b8242f630 'virtio: expose added
descriptors immediately.', in virtio balloon virtqueue_get_buf might
now run concurrently with virtqueue_kick.  I audited both and this
seems safe in practice but this is not guaranteed by the API.
Additionally, a spurious interrupt might in theory make
virtqueue_get_buf run in parallel with virtqueue_add_buf, which is
racy.

While we might try to protect against spurious callbacks it's
easier to fix the driver: balloon seems to be the only one
(mis)using the API like this, so let's just fix balloon.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (removed unused var)
12 years agoMerge branch 'for-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj...
Linus Torvalds [Sun, 8 Jul 2012 16:09:27 +0000 (09:09 -0700)]
Merge branch 'for-3.5-fixes' of git://git./linux/kernel/git/tj/cgroup

Pull cgroup fixes from Tejun Heo:
 "The previous cgroup pull request contained a patch to fix a race
  condition during cgroup hierarchy umount.  Unfortunately, while the
  patch reduced the race window such that the test case I and Sasha were
  using didn't trigger it anymore, it wasn't complete - Shyju and Li
  could reliably trigger the race condition using a different test case.

  The problem wasn't the gap between dentry deletion and release which
  the previous patch tried to fix.  The window was between the last
  dput() of a root's child and the resulting dput() of the root.  For
  cgroup dentries, the deletion and release always happen synchronously.
  As this releases the s_active ref, the refcnt of the root dentry,
  which doesn't hold s_active, stays above zero without the
  corresponding s_active.  If umount was in progress, the last
  deactivate_super() proceeds to destory the superblock and triggers
  BUG() on the non-zero root dentry refcnt after shrinking.

  This issue surfaced because cgroup dentries are now allowed to linger
  after rmdir(2) since 3.5-rc1.  Before, rmdir synchronously drained the
  dentry refcnt and the s_active acquired by rmdir from vfs layer
  protected the whole thing.  After 3.5-rc1, cgroup may internally hold
  and put dentry refs after rmdir finishes and the delayed dput()
  doesn't have surrounding s_active ref exposing this issue.

  This pull request contains two patches - one reverting the previous
  incorrect fix and the other adding the surrounding s_active ref around
  the delayed dput().

  This is quite late in the release cycle but the change is on the safer
  side and fixes the test cases reliably, so I don't think it's too
  crazy."

* 'for-3.5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: fix cgroup hierarchy umount race
  Revert "cgroup: superblock can't be released with active dentries"

12 years agoLinux 3.5-rc6
Linus Torvalds [Sun, 8 Jul 2012 00:23:56 +0000 (17:23 -0700)]
Linux 3.5-rc6

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sun, 8 Jul 2012 00:21:59 +0000 (17:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/linux-security

Pull security docs update from James Morris.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  security: Minor improvements to no_new_privs documentation

12 years agovfs: make O_PATH file descriptors usable for 'fchdir()'
Linus Torvalds [Sat, 7 Jul 2012 17:17:00 +0000 (10:17 -0700)]
vfs: make O_PATH file descriptors usable for 'fchdir()'

We already use them for openat() and friends, but fchdir() also wants to
be able to use O_PATH file descriptors.  This should make it comparable
to the O_SEARCH of Solaris.  In particular, O_PATH allows you to access
(not-quite-open) a directory you don't have read persmission to, only
execute permission.

Noticed during development of multithread support for ksh93.

Reported-by: ольга крыжановская <olga.kryzhanovska@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: stable@kernel.org # O_PATH introduced in 3.0+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agocgroup: fix cgroup hierarchy umount race
Tejun Heo [Sat, 7 Jul 2012 23:08:18 +0000 (16:08 -0700)]
cgroup: fix cgroup hierarchy umount race

48ddbe1946 "cgroup: make css->refcnt clearing on cgroup removal
optional" allowed a css to linger after the associated cgroup is
removed.  As a css holds a reference on the cgroup's dentry, it means
that cgroup dentries may linger for a while.

Destroying a superblock which has dentries with positive refcnts is a
critical bug and triggers BUG() in vfs code.  As each cgroup dentry
holds an s_active reference, any lingering cgroup has both its dentry
and the superblock pinned and thus preventing premature release of
superblock.

Unfortunately, after 48ddbe1946, there's a small window while
releasing a cgroup which is directly under the root of the hierarchy.
When a cgroup directory is released, vfs layer first deletes the
corresponding dentry and then invokes dput() on the parent, which may
recurse further, so when a cgroup directly below root cgroup is
released, the cgroup is first destroyed - which releases the s_active
it was holding - and then the dentry for the root cgroup is dput().

This creates a window where the root dentry's refcnt isn't zero but
superblock's s_active is.  If umount happens before or during this
window, vfs will see the root dentry with non-zero refcnt and trigger
BUG().

Before 48ddbe1946, this problem didn't exist because the last dentry
reference was guaranteed to be put synchronously from rmdir(2)
invocation which holds s_active around the whole process.

Fix it by holding an extra superblock->s_active reference across
dput() from css release, which is the dput() path added by 48ddbe1946
and the only one which doesn't hold an extra s_active ref across the
final cgroup dput().

Signed-off-by: Tejun Heo <tj@kernel.org>
LKML-Reference: <4FEEA5CB.8070809@huawei.com>
Reported-by: shyju pv <shyju.pv@huawei.com>
Tested-by: shyju pv <shyju.pv@huawei.com>
Cc: Sasha Levin <levinsasha928@gmail.com>
Acked-by: Li Zefan <lizefan@huawei.com>
12 years agoRevert "cgroup: superblock can't be released with active dentries"
Tejun Heo [Sat, 7 Jul 2012 22:55:47 +0000 (15:55 -0700)]
Revert "cgroup: superblock can't be released with active dentries"

This reverts commit fa980ca87d15bb8a1317853f257a505990f3ffde.  The
commit was an attempt to fix a race condition where a cgroup hierarchy
may be unmounted with positive dentry reference on root cgroup.  While
the commit made the race condition slightly more difficult to trigger,
the race was still there and could be reliably triggered using a
different test case.

Revert the incorrect fix.  The next commit will describe the race and
fix it correctly.

Signed-off-by: Tejun Heo <tj@kernel.org>
LKML-Reference: <4FEEA5CB.8070809@huawei.com>
Reported-by: shyju pv <shyju.pv@huawei.com>
Cc: Sasha Levin <levinsasha928@gmail.com>
Acked-by: Li Zefan <lizefan@huawei.com>
12 years agohwspinlock/core: use global ID to register hwspinlocks on multiple devices
Shinya Kuribayashi [Sat, 7 Jul 2012 10:37:42 +0000 (13:37 +0300)]
hwspinlock/core: use global ID to register hwspinlocks on multiple devices

Commit 300bab9770 (hwspinlock/core: register a bank of hwspinlocks in a
single API call, 2011-09-06) introduced 'hwspin_lock_register_single()'
to register numerous (a bank of) hwspinlock instances in a single API,
'hwspin_lock_register()'.

At which time, 'hwspin_lock_register()' accidentally passes 'local IDs'
to 'hwspin_lock_register_single()', despite that ..._single() requires
'global IDs' to register hwspinlocks.

We have to convert into global IDs by supplying the missing 'base_id'.

Cc: stable <stable@vger.kernel.org>
Signed-off-by: Shinya Kuribayashi <shinya.kuribayashi.px@renesas.com>
[ohad: fix error path of hwspin_lock_register, too]
Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
12 years agoMerge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Linus Torvalds [Sat, 7 Jul 2012 18:20:59 +0000 (11:20 -0700)]
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm

Pull ARM fixes from Russell King:
 "Last merge window, we had some updates from Al cleaning up the signal
  restart handling.  These have caused some problems on ARM, and while
  Al has some fixes, we have some concerns with Al's patches but we've
  been unsuccesful with discussing this.

  We have got to the point where we need to do something, and we've
  decided that the best solution is to revert the appropriate commits
  until Al is able to reply to us.

  Also included here are four patches to fix warnings that I've noticed
  in my build system, and one fix for kprobes test code."

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
  ARM: fix warning caused by wrongly typed arm_dma_limit
  ARM: fix warnings about atomic64_read
  ARM: 7440/1: kprobes: only test 'sub pc, pc, #1b-2b+8-2' on ARMv6
  ARM: 7441/1: perf: return -EOPNOTSUPP if requested mode exclusion is unavailable
  ARM: 7443/1: Revert "new way of handling ERESTART_RESTARTBLOCK"
  ARM: 7442/1: Revert "remove unused restart trampoline"
  ARM: fix set_domain() macro
  ARM: fix mach-versatile/pci.c warning

12 years agosecurity: Minor improvements to no_new_privs documentation
Andy Lutomirski [Thu, 5 Jul 2012 18:23:24 +0000 (11:23 -0700)]
security: Minor improvements to no_new_privs documentation

The documentation didn't actually mention how to enable no_new_privs.
This also adds a note about possible interactions between
no_new_privs and LSMs (i.e. why teaching systemd to set no_new_privs
is not necessarily a good idea), and it references the new docs
from include/linux/prctl.h.

Suggested-by: Rob Landley <rob@landley.net>
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: James Morris <james.l.morris@oracle.com>
12 years agoMerge tag 'ecryptfs-3.5-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 6 Jul 2012 22:32:18 +0000 (15:32 -0700)]
Merge tag 'ecryptfs-3.5-rc6-fixes' of git://git./linux/kernel/git/tyhicks/ecryptfs

Pull eCryptfs fixes from Tyler Hicks:
 "Fixes an incorrect access mode check when preparing to open a file in
  the lower filesystem.  This isn't an urgent fix, but it is simple and
  the check was obviously incorrect.

  Also fixes a couple important bugs in the eCryptfs miscdev interface.
  These changes are low risk due to the small number of users that use
  the miscdev interface.  I was able to keep the changes minimal and I
  have some cleaner, more complete changes queued up for the next merge
  window that will build on these patches."

* tag 'ecryptfs-3.5-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
  eCryptfs: Gracefully refuse miscdev file ops on inherited/passed files
  eCryptfs: Fix lockdep warning in miscdev operations
  eCryptfs: Properly check for O_RDONLY flag before doing privileged open