[PATCH 0/4] async_tx: fix dependency handling and related cleanups

[Dan Williams <dan.j.williams@intel.com>]

Injecting channel-switch-interrupts has been broken for a while now.  It
has not been a problem in practice because the only in-tree driver that
relied on this functionality was the iop3xx version of iop-adma, and it
had a bug-masking local workaround.  Three side benefits arise from this
fix:

1/ dma_async_tx_descriptor sheds two list_heads
2/ Locking is made sane in that dma drivers no longer need to directly
   touch dma_async_tx_descriptor.lock
3/ dma_device.device_dependency_added is no longer needed

Testing shows that iop-adma now gets by without the 'watchdog'
workaround.

---

Dan Williams (4):
      iop-adma: remove the workaround for missed interrupts on iop3xx
      async_tx: kill ->device_dependency_added
      async_tx: fix multiple dependency submission
      async_tx: checkpatch says s/__FUNCTION__/__func__/g


 crypto/async_tx/async_memcpy.c         |    6 -
 crypto/async_tx/async_memset.c         |    6 -
 crypto/async_tx/async_tx.c             |  203 ++++++++++++++++++++++++++------
 crypto/async_tx/async_xor.c            |   12 +-
 drivers/dma/dmaengine.c                |    3 
 drivers/dma/ioat_dma.c                 |   12 --
 drivers/dma/iop-adma.c                 |   21 +--
 include/asm-arm/arch-iop13xx/adma.h    |    5 -
 include/asm-arm/hardware/iop3xx-adma.h |    8 -
 include/asm-arm/hardware/iop_adma.h    |    2 
 include/linux/dmaengine.h              |   11 --
 11 files changed, 185 insertions(+), 104 deletions(-)

-- 
Dan