Kudos to Doug Ledford, who with an assist from the Adaptec
folks' have come up with the solution to my long headache!
Original problem:
The following machine had a new disk added:
Dual PPro 200 in SuperMicro P6DNE, 192MB EDO RAM
Adaptec AHA-2940 [BIOS ver 1.16] SCSI controller, with
ID#0 Seagate Barracuda (ST15150N) 4GB Fast-SCSI-2 HDD
ID#1 Seagate Hawk (ST32430N) 2GB Fast-SCSI-2 HDD
ID#3 Plextor UltraPlex 14x/32x Ultra-SCSI CD-ROM
ID#4 Ricoh MP6201S 6x/2x/2x CD-Rewritable
NEW -> ID#5 Seagate Hawk (ST15230WC) 4GB HDD SCA w/ 80-to-50
ID#6 Fujitsu M2512A 230MB Magneto-Optical Drive
Matrox Millenium 8MB PCI, D-Link DE-220 NE2000 compat NIC,
Logitech TrackMan Marble+, SB16 Sound Blaster clone non-PnP.
Result:
[A] The new drive cannot be recognised.
[B] When the CD-ROMs are removed, the drive is recognised fine,
but on bootup the kernel says `Using Wide (16-bit) Transfers'
and in general messes everything up. The new HDD is read by
the kernel as being of 32MB size.
[C] Swapping in a Seagate Elite (ST410800W) 8.4GB Fast-SCSI-2-Wide
with a 68-to-50 converter results in the HDD being read as a
drive of the right size, but problem occur with the partition
tables being off-balance somehow.
Resolution:
1. Changed Adaptec AHA-2940 SCSI controller to AHA-3985 3-channel SCSI
RAID card, which can be used by Linux as three separate AHA-2940's.
Part of the problem had been with signal strength and cabling. I'd
used Adaptec's special cables with pass-through terminators for the
SCA drive, which cannot be terminated and whose 80-to-50 converter
had not been equipped with a terminator.
** Now all drives can be seen by the system just fine. I guess with
SCSI you really need good cabling and multiple channels to hold more
than 4 devices, as I found out the hard way.
2. Made new kernel with Doug Ledford's new 5.1.0-pre10 patch [I had
problems-- lots-- compiling 2.1.121, and resorted to using the
version 2.1.115 patch for pre7, then pigpiled on the pre9-10
patches. [I actually used 2.1.117 as 2.1.115 does not boot...]
** Now the drive is recognised properly by the kernel and is now
functioning just fine, at least for the last several hours, in
which I compiled kernel about twenty times. I will be adding
more drives to experiment with Linux's soft RAID presently ...
** The Elite 9 using a Wide(68)-to-Narrow(50) converter also seems
to work fine, but I am not going to be using it in this box ...
I would like to thank Doug for his help [this is not the
first time that he'd helped me out with AIC7XXX problems]
and call for a round of cheers for him and for Adaptec's
seeing the light! [Hip, hip, hurray ....]
Bo-Yin Yang
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/faq.html