1. Changed around the DRIVER_LOCK for 2.0.33. This improves reliability
and removes some problems I could force with the kernel spinning
in an infinite loop.
2. Changes to the abort/reset code. These changes should squash the bug
related to No active SCB for reconnecting target, issuing BUS
DEVICE RESET messages that could get printed out after a bus
recovery operation was performed.
3. Changes to abort/reset code. These changes should squash the bug
related to commands queued exceeds queue depth messages.
4. Changes to abort/reset code. These changes should squash the bugs
that could cause us to never recover from a reset operation
depending on the device negotiation parameters.
5. Changes to the negotiation code in regards to tag commands. This should
help devices that die if both WIDE and Sync negotiation is enabled.
It also helps the reset code.
6. Fixed the same bug in multiple places where we incorrectly set the
SCSISEQ register to 0 instead of the current value &'ed with
~ENSELO.
7. Sparc updates (thanks to Dave Miller for testing these for me and for
providing the initial work which I simply incorporated). This
driver should work "out of the box" on Sparc machines with a PCI
bus now. Additionally, it may work out of the box on PPC machines
as well.
8. Changes to message code to get rid of the "panic: AWAITING_MSG for scb
that doesn't have a waiting message" problem (closely related to
abort/reset fixes).
