On Tue, 7 Feb 2006, Jordan Crouse wrote:
There are additional rumors of strangeness on the DB1550, but those
are unsubstantiated at this point. At least with this patch, it will
compile.
I can confirm strangeness on Au1550, and I've pinned it down to wrong
way of setting the divisor register. The alchemy uart have got a spearate
register for divisor, while the driver tries to set the divisor latch
just like on any 16550 uart.
How would be a good way of solving this? I can imagine a couple of ways:
1. Make divisor latch read/write a bit more abstract, so that a
alternative function can be used for alchemy.
2. Add divisor latch to the alchemy uart register map. But since the
divisor register is located in a single 32-bit register and the driver
will try to access it one byte at a time; byte access would be needed. And
this would add some endian handling...
--
Jon Anders Haugum
