Bug 608
| Summary: | smp machine with 2 scsi cdroms gets scsi aborts | ||
|---|---|---|---|
| Product: | IO/Storage | Reporter: | Mike Wescott (wescott_mike) |
| Component: | SCSI | Assignee: | Mike Anderson (andmike) |
| Status: | REJECTED INVALID | ||
| Severity: | normal | CC: | zwane |
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.5.67 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
|
Description
Mike Wescott
2003-04-20 16:46:03 UTC
I'm trying to collect data for a similar bug, can you tell me wether booting with noapic acts as a workaround? Can you also reproduce with 2.5.68 since that saw a big merge from Justin Gibbs for the aic7xxx driver noapic DOES seem to act as a workaround for 2.5.67 or I got real lucky doing the test. noapic does not work for 2.4.18 2.5.68 fails (didn't try with noapic) my own suspicion is that the problem is somewhere above aic7xxx and scsi probably in sr or cdrom, otherwise ordinary disk activity would trigger the problem. I can reproduce a similar bug with disks, i haven't tried recent 2.4 kernels save 2.4.18-3 as shipped by RedHat. I have a test system which oopses in 2.5.68 due to the aic7xxx driver, my current workaround for that is booting with noapic as it appears to be some sort of race in bottom half handling. I'll supply data as soon as i have an opportunity to collect it. Can't reproduce this on an eight way with two CD-ROMS connected to a 53c700 I suspect it may be a bug in the Adaptec driver with SMP locking and devices that don't do TCQ The dump card state indicates that we are stuck in a DATA-IN phase with all data that the card has already received, delivered to host memory. The card is stuck waiting for a REQ on the bus, either to indicate a new phase or to continue the data transfer. Looks like bad cabling or termination to me. If others are seeing similar problems with other configurations I will need the trace information to be able to comment. Justin, tell me what you need against which kernel and i'll try my best to get it for you. It is simply a matter of attempting a boot on my system. > Looks like bad cabling or termination to me.
No ... more like a buggy CD drive. I replaced the older of the
two drives with a different drive and the problem has disappeared.
|