The attached patch converts the "Intel Triton IDE support" into a
"Generic PCI bus-mastering DMA support". Actually these are quite
small changes. I just had to change the way that PCI IDE interfaces
are probed a little. Hopefully I haven't broken anything. At least
it now works with my SiS 5571 chipset... :)
I don't know about these changes. I do however agree that DMA support
attempts in the generic IDE code should be protected by a single
define.
I have a ns87415 driver for sparc64 which is pending merge into the
tree, that driver is entirely seperate and does support DMA. I took
the easy way out in my changes and just made all checks for
CONFIG_IDE_TRITON which protected IDE dma stuff in the generic layer
into tests for CONFIG_IDE_TRITON or CONFIG_IDE_NS87415.
Later,
David "Sparc" Miller
davem@caip.rutgers.edu