I have the following system:
- Pentium II 333, 512 MB RAM on a Gigabyte GA686 double processor board
- RedHat 5.2 based + all 2.2 Kernel upgrades
- Linux 2.2.3ac4 Kernel, all drivers as module
- 3COM 3C905B (Cyclone) NIC @ 100 MBit running off the ubiquitous Cisco
2924XL switch
- two Adaptec 2940UW scsi adaptors:
- adaptor 0:
50 internal chain:
- TEAC CDROM CD-532S, ID 3, async narrow
- IBM DCAS-34440 4 GB drive, ID 0, sync 20 MBit/sec
This drive is terminated
68 internal chain:
- IBM 9 GB UW SCSI Disk, ID 1, sync 40 MBit/sec
(DDRS-39130W)
- via connector in an external box:
- IBM 4 GB UW SCSI Disk, ID 5, sync 40 MBit/sec
(DDRS-34560D)
- IBM 4 GB UW SCSI Disk, ID 6, sync 40 MBit/sec
(DDRS-34560D)
- Box has an terminator connected.
68 external chain _NOT_ connected
terminator (on DCAS disk)
|
IBM DCAS 4 GB, ID 0
|
TEAC CDROM-Drive, ID 3
|
Adaptec 2940 UW, Termination: Low off, High on
|
IBM 9GB UW Disk, ID 1
|
external connector, 68 pins
|
IBM 4 GB UW Disk, ID 5
|
IBM 4 GB UW Disk, ID 6
|
terminator 68 pins
This looks correct to me.
- adaptor 1:
68 internal chain:
- IBM 9 GB UW SCSI Disk, ID 1, sync 40 MBit/Sec
(DDRS-39130W)
terminated
68 external chain:
- external box:
- IBM 9 GB UW SCSI Disk, ID 3, sync 40 MBit/Sec
(DDRS-39130W)
- Box has an terminator connected.
terminator (on 9 GB Disk disk)
|
IBM 9GB UW Disk, ID 1
|
Adaptec 2940 UW, Termination: Low off, High off
|
external connector, 68 pins
|
IBM 9GB UW Disk, ID 3
|
terminator 68 pins
This chain looks equally correct to me.
The whole system runs fine (> 50 days of uptime) under 2.0.36pre15,
2.0.36 and now 2.0.37ac8. I run with
aic7xxx="verbose.tag_info:{{16.16.16.16.16.16.16}.{16.16.16.16.16.16.16}.{16.16.16.16.16.16.16}}"
and all relevant drives report a tag queue of 16. This is with aic7xxx
driver 5.1.7 from the 2.0.37ac8 kernel.
When I decided to upgrade the kernel to 2.2.3-ac4, the whole machine
broke down. The kernel never even completed scanning of the drive
chains. The first disk (DCAS) was reported as "synchrous at 20
MBit/sec", then a timeout occured, the scb was aborted and the loop
repeated. Sometimes, the second drive (9 GB UW on the internal chain)
was found at 40 MBit/sec but it never went as far as detecting the
TEAC CDROM. The new driver identifies itself as "5.1.12".
I'm at a sort of loss here. The machine runs fine under 2.0.3x but it
seems to be much more sensitive to SCSI trouble (timeouts, parity
errors and all this stuff is reported in tons while booting) under
2.2.3ac4
I was able to boot the machine into single user mode when activating
the Low/High termination on the first Adaptec controller and just
connecting the DCAS drive off the 50 pin chain. Even then I had lots
of trouble with aborts, resets and retries while booting or running
bonnie on a local partition of the drive. Normal operation is
impossible.
So I'm back to 2.0.3x. I do suspect a driver problem, I simply cannot
believe, that my SCSI chains work fine with 2.0.3x and are incorrect
with 2.2.x.
(And oh yes, the machine does _heavy_ SCSI activity under normal
circumstances. Its a full feed newsserver / reader).
I even rebuilt the machine with a third Adaptec 2940 to move the DCAS
and the CDROM onto a different chain to get all 68pin wide chains. No
go. Runs fine under 2.0.3x but not under 2.2.3ac4
Kind regards
Henning
-- Dipl.-Inf. (Univ.) Henning P. Schmiedehausen -- hps@tanstaafl.de TANSTAAFL! Consulting - Unix, Internet, SecurityHutweide 15 Fon.: 09131 / 50654-0 "There ain't no such D-91054 Buckenhof Fax.: 09131 / 50654-20 thing as a free Linux"
- 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/