Description of problem: Our systems hang for 5-10 seconds on boot, while trying to load the aic79xx module. Version-Release number of selected component (if applicable): RH Academic Desktop 3 R4 How reproducible: Every boot on our Sun W2100z workstations. Steps to Reproduce: 1. Start boot 2. Watch messages until it gets to "Loading aic79xx module." message. 3. Wait. Actual results: 5-10 second wait. Expected results: Minimal wait at best. Additional info:
The default selection timeout on SCSI is .25 sec. Times 16 targets, gives 4 seconds, so it is going to take a while, especially on an unpopulated SCSI bus. 5-6 seconds seems normal. Please add this to modules.conf: options aic79xx aic79xx=verbose,seltime:3 Re-make the initrd, and reboot. This reduces the selection timeout to 32ms. This should be safe for modern SCSI devices. See it this makes a difference, and post the results of /var/log/messages. If you can, point out where the long the delay occurs in the log. Tom
This actually increased the load time to ~30 seconds. There were no messages to /var/log/messages; this is all being done before it's actually mounted any file systems. *checks* It wasn't much different without that line. (~27 seconds.)
So, is anything being done on this?
When you get to "Loading aic79xx module." the system will log a bunch of messages related to module/adapter initialization, and SCSI bus scanning. These will be in dmesg and /var/log/messages after the system finishes booting. Please add this to modules.conf: options aic79xx aic79xx=verbose Re-make the initrd, and reboot. Then post /var/log/messages showing the boot messages. Also check the log for any other errors on this adapter during normanl use.
Created attachment 110441 [details] /var/log/messages describing boot process As requested. I don't see much useful information on there, though. There are no other error messages regarding the aic79xx.
This is a stab, but would you try booting with "noapic" on the kernel command line? See it this makes a difference.
Adding noapic to the kernel boot parameters worsened the problem. The computers did do anything after the Loading aic79xx module for at least 5+ minutes at which time they were given a hard reboot. In addition we got the following error message for probably every IRQ. ERROR: Unable to locate IOAPIC for IRQ ## \n
Going back through old bug reports.... I overlooked this because it was still in "needinfo". Is this still an issue for you? As I said in comment #1, it is not unusual for the loading of a SCSI driver to take 4-5 seconds per SCSI bus. The time is required to scan the bus. I see this on a system here: # time modprobe aic79xx real 0m26.795s user 0m0.010s sys 0m0.290s The delay occurs _after_ the following is printed on the console: PCI: Enabling device 00:03.0 (0156 -> 0157) PCI: Enabling device 00:03.1 (0156 -> 0157) <Adaptec 39320 Ultra320 SCSI adapter> aic7902: Ultra320 Wide Channel A, SCSI Id=7, PCI 33 or 66Mhz, 512 SCBs <Adaptec 39320 Ultra320 SCSI adapter> aic7902: Ultra320 Wide Channel B, SCSI Id=7, PCI 33 or 66Mhz, 512 SCBs while the busses are scanned. Does this match when you see the delay? There is not much that can be done other than to try some module parameters to limit the number of target IDs or the timeout during the scan. Let me know if you need assistance with this. Otherwise I will close this BZ.
This bug is filed against RHEL 3, which is in maintenance phase. During the maintenance phase, only security errata and select mission critical bug fixes will be released for enterprise products. Since this bug does not meet that criteria, it is now being closed. For more information of the RHEL errata support policy, please visit: http://www.redhat.com/security/updates/errata/ If you feel this bug is indeed mission critical, please contact your support representative. You may be asked to provide detailed information on how this bug is affecting you.