ftp://ftp.dialnet.net/pub/linux/aic7xxx
It applies against 2.0.33 cleanly. It does not work under 2.1.x, but I'm
sure there will be an announcement shortly of a forward port :)
This driver includes only two real fixes. First, we migrated back to the
interrupt handler based completion method due to poor swapping
performance under the bottom half completion method. Second, it includes
a horrible code hack and option to get the driver to work on buggy Tyan
Thunder II motherboards with aic7895 controller chipsets (the BIOS
incorrectly sets the IRQs on channel A and channel B as two different
IRQs when they are actually sharing the same IRQ, if this has been fixed
in later Tyan BIOSes I'm not sure, but it exists in at least some of the
Tyan motherboard BIOS versions, see drivers/scsi/README.aic7xxx).
Special note:
As far as the 2.0.x kernel series is concerned, this patch has now
officially migrated from active development status to maintenance status.
The patch is reliable enough that I'm using it on our production servers,
and to my knowledge there are no more outstanding TODO items (at least
that are feasible given the tools at my disposal). This patch is, in all
actuality, now more reliable than the Nov05 patch I had previously. I
say this because I found a bug in the Nov05 patch that could cause bad
thigns under certain conditions, and that (among others) has already been
fixed in the 5.0.5 patch. For instance, that particular bug is likely to
be the culprit of the few reports I have of the RedHat-5.0 boot disk not
working on a few people's aic7xxx controller. Any linux distributors
that have opted to use my patch in their distribution (either on boot
disks or in the running kernel after installation) I would recommend
either upgrading to the 5.0.5 or contacting me to get a patch against the
Nov05 driver to fix the single most important bug in the Nov05 code which
is a memory scribble bug that can result in the sporadic reboots, or
messages like "swap_free: bad swap entry 0x.." and other odd things. If
people are interested, then I can give the full details of what this bug
is as well as why it doesn't effect the majority of people using my Nov05
patch.
As far as the 2.1.x kernel series is concerned, this patch is still under
development. There are changes I intend to make that won't go into the
2.0.x patch versions that relate specifically to the new IO/APIC
interrupt handlers, new SCSI mid level code, spin locks, SMP issues, etc.
I have not yet started on this particular project, but that's next on my
list of items for the aic7xxx driver.
--------------------------------------
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