linux-stable.git/drivers/bcma/driver_chipcommon.c, branch linux-3.8.y

Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus

2012-12-14T22:27:45+00:00

Pull MIPS updates from Ralf Baechle:
 "The MIPS bits for 3.8.  This also includes a bunch fixes that were
  sitting in the linux-mips.org git tree for a long time.  This pull
  request contains updates to several OCTEON drivers and the board
  support code for BCM47XX, BCM63XX, XLP, XLR, XLS, lantiq, Loongson1B,
  updates to the SSB bus support, MIPS kexec code and adds support for
  kdump.

  When pulling this, there are two expected merge conflicts in
  include/linux/bcma/bcma_driver_chipcommon.h which are trivial to
  resolve, just remove the conflict markers and keep both alternatives."

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (90 commits)
  MIPS: PMC-Sierra Yosemite: Remove support.
  VIDEO: Newport Fix console crashes
  MIPS: wrppmc: Fix build of PCI code.
  MIPS: IP22/IP28: Fix build of EISA code.
  MIPS: RB532: Fix build of prom code.
  MIPS: PowerTV: Fix build.
  MIPS: IP27: Correct fucked grammar in ops-bridge.c
  MIPS: Highmem: Fix build error if CONFIG_DEBUG_HIGHMEM is disabled
  MIPS: Fix potencial corruption
  MIPS: Fix for warning from FPU emulation code
  MIPS: Handle COP3 Unusable exception as COP1X for FP emulation
  MIPS: Fix poweroff failure when HOTPLUG_CPU configured.
  MIPS: MT: Fix build with CONFIG_UIDGID_STRICT_TYPE_CHECKS=y
  MIPS: Remove unused smvp.h
  MIPS/EDAC: Improve OCTEON EDAC support.
  MIPS: OCTEON: Add definitions for OCTEON memory contoller registers.
  MIPS: OCTEON: Add OCTEON family definitions to octeon-model.h
  ata: pata_octeon_cf: Use correct byte order for DMA in when built little-endian.
  MIPS/OCTEON/ata: Convert pata_octeon_cf.c to use device tree.
  MIPS: Remove usage of CEVT_R4K_LIB config option.
  ...

bcma: unify naming schema for clock functions

2012-12-10T20:47:30+00:00

Signed-off-by: John W. Linville

bcma: register watchdog driver

2012-12-06T19:58:57+00:00

Register the watchdog driver to the system if this is a SoC. Using the
watchdog on a non SoC device, like a PCIe card, will make the PCIe
card die when the timeout expired, but starting it again is not
supported by bcma.

Signed-off-by: Hauke Mehrtens 
Signed-off-by: John W. Linville

bcma: add methods for watchdog driver

2012-12-06T19:58:56+00:00

The watchdog driver wants to set the watchdog timeout in ms and not in
ticks, which is depending on the SoC type and the clock.
Calculate the number of ticks per millisecond and provide two functions
for the watchdog driver. Also return the ticks or millisecond the timer
was set to in case the provided value was bigger than the max allowed
value.

Signed-off-by: Hauke Mehrtens 
Signed-off-by: John W. Linville

bcma: set the pmu watchdog if available

2012-12-06T19:58:56+00:00

Mostly all bcma based devices have a PMU and the PMU watchdog should be
used and not the old one in chip common. This patch also calculates the
maximal number the watchdog could be set to.

Signed-off-by: Hauke Mehrtens 
Signed-off-by: John W. Linville

bcma: add bcma_chipco_alp_clock

2012-12-06T19:58:56+00:00

For devices without a PMU the alp clock is always 20000000.

Signed-off-by: Hauke Mehrtens 
Signed-off-by: John W. Linville

bcma: add comment to bcma_chipco_gpio_control

2012-11-21T20:55:51+00:00

Add description to the function.

Signed-off-by: Hauke Mehrtens 
Patchwork: http://patchwork.linux-mips.org/patch/4588
Acked-by: Florian Fainelli

bcma: add bcma_chipco_gpio_pull{up,down}

2012-11-21T20:55:51+00:00

Add functions to access the GPIO registers for pullup and pulldown.
These are needed for handling gpio registration.

Signed-off-by: Hauke Mehrtens 
Patchwork: http://patchwork.linux-mips.org/patch/4586
Acked-by: Florian Fainelli

bcma: add locking around GPIO register accesses

2012-11-21T20:55:51+00:00

The GPIOs are access through some registers in the chip common core.
We need locking around these GPIO accesses, all GPIOs are accessed
through the same registers and parallel writes will cause problems.

Signed-off-by: Hauke Mehrtens 
Patchwork: http://patchwork.linux-mips.org/patch/4585
Acked-by: Florian Fainelli

bcma: just do the necessary things in early register on SoCs

2012-10-19T19:52:18+00:00

Some parts of the initialization for chip common and the pcie core are
accessing the sprom struct, but it is not initialized at that stage.
Just do the necessary thing in the early register on SoCs and not the
complete initialization to read out the nvram from the flash chip.
After it is possible to read out the nvram, the sprom should be parsed
from it and the full initialization of the cores should be run.

Signed-off-by: Hauke Mehrtens 
Signed-off-by: John W. Linville