On 10/09/2013 05:47 AM, Amit Daniel Kachhap wrote:

hi,

The patches in the series are work in progress and as discussed with Al Stone
and Hanjun include changes in the GTDT acpi table to add missing elements for ARM
platforms. Elements added are Timer frequency, interrupt count(currently 8 supported),
timer ID(This will be used to invoke the correct controller).

Still more TODO's are identified.
* To add timer clock names.
* if this __init invoking mechanism is finalised then other similar devices can
  use this like clock, GIC etc.
* GTDT can ultimately be made CTDT(Common timer descriptor table) to other classes of
  timers like external exynos MCT.

Changes in V2:
* Added new fields in GTDT acpi table.
* Infrastructure to invoke the correct timer controller.

Thanks in advance.
Amit


Amit Daniel Kachhap (6):
   irqdomain: Add a new API irq_create_default_mapping
   ACPI: ARM: Update acpi_register_gsi to register with the core IRQ
     subsystem
   irqchip: gic: Fix to make dt node check optional in case of ACPI
   acpi: gtdt: Update GTDT tables to add more fields
   clocksource:acpi: Add macro CLOCKSOURCE_ACPI_DECLARE
   clocksource: arch_timer: acpi: Use GTDT table to gather data

  drivers/acpi/plat/arm/boot.c         |   27 +++++++-
  drivers/clocksource/arm_arch_timer.c |  111 +++++++++++++++++++++++++++++-----
  drivers/clocksource/clksrc-of.c      |   32 ++++++++++
  drivers/irqchip/irq-gic.c            |    2 +
  include/acpi/actbl3.h                |   22 +++++++
  include/asm-generic/vmlinux.lds.h    |    6 ++
  include/linux/clocksource.h          |   13 ++++
  include/linux/irqdomain.h            |    2 +
  kernel/irq/irqdomain.c               |   35 +++++++++++
  9 files changed, 232 insertions(+), 18 deletions(-)

This is a *really* interesting approach to the problem.  I think
I like it :).  I've read through the patches a couple of times but
I'm still sorting out my thoughts so I'm going to think on them a
little longer.  My initial reactions are pretty positive, though.

One thing that occurs to me is that we should probably fix iasl to
match the table change, and that perhaps we should consider
recommending a change to the ACPI spec to match the table change (both
relatively straightforward things to do, really).

Hanjun: any comments from your side?

--
ciao,
al
-----------------------------------
Al Stone
Software Engineer
Linaro Enterprise Group
al.stone@linaro.org
-----------------------------------