aic7xxx-5.0.15 released

Doug Ledford (dledford@dialnet.net)
Mon, 18 May 1998 19:31:42 -0500


This is a bug-fix release of the aic7xxx 5.0.x driver. This release
concentrates on the various negotiation bugs people have seen in the 5.0.13
and 5.0.14 drivers. Error messages from these drivers could include things
such as invalid SCB during SELTO messages, complete hangs immediately after
attempting to negotiate with certain devices, and other assorted error
messages. This version fixes these problems by committing the horrible sin
of peeking into the data returned by devices in response the the INQUIRY
command. From this data we determine what devices can/can't support
wide/sync negotiation and then issue the negotiation based upon those
results. Personally, I don't like doing it this way, but it's the most
reliable way of knowing what devices we shouldn't or shouldn't negotiate
with.

The other major item is improved support for module loading/unloading.
Thanks to Matthew Jacob for code in this area. This code is, however,
fairly untested as I don't use modules. I would appreciate it if people
could test that this version of the driver properly unloads. The new code
should mean that the driver will no longer leak memory with every module
unload, but could (depending on how the release function is called) cause
bad things to happen. I need some feedback on this from people that do use
modules and have a root filesystem on something other than their aic7xxx
controller.

Minor tweaks include changes to the PCI configuration during driver init
time, which should hopefully help people who have SCSI timeouts that are
caused by the controller being incorrectly intialized WRT termination, and
tweaks to the PCI parity detection code to accomodate machines that are
broken WRT PCI parity interrupts and checking.

As always, my ftp site is ftp.dialnet.net:/pub/linux/aic7xxx. In this site
there are different directories for patches against different kernels. The
5.0.15 patch has been made against the 2.0.33, 2.0.34pre15, and 2.1.102
kernels.

I would very much appreciate feedback in the short term as I would like to
verify that this driver does what I wanted it to do before I take off for
linux-expo.

-- 

Doug Ledford <dledford@dialnet.net> Opinions expressed are my own, but they should be everybody's.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu