Description of problem: When booting one of our servers which contains a LSIL Megaraid SATA 300-8x controller the megaraid driver loops the following error during bootup: megaraid: 2.20.4.0 (Release Date: Mon Sep 27 22:15:07 EDT 2004) megaraid: probe new device 0x1000:0x0409:0x1000:0x3008: bus 3:slot 14:func 0 megaraid: fw version:[813G] bios version:[H425] megaraid: scanning scsi channel 0 [Phy 0] for non-raid devices megaraid: aborting-1 cmd=12 <c=0 =0 l=0> megaraid abort: 1:0[0:0], fw owner megaraid: resetting the host... megaraid: 1 oustanding commands. Max wait 180 sec megaraid mbox: Wait for 0 commands to complete: 180 megaraid mbox: reset sequence completed sucessfully This gets repated thrice scsi: Device offlined - not reading after error recovery: host 0 channel 0 id 1 lun 0 These messages repeat indefinitly. Version-Release number of selected component (if applicable): kernel-2.6.9-5.EL How reproducible: Always Workaround: Defining acpi=off at the bootprompt fixes the issue for now.
According to Matt Domsch, it is likely a IRQ routing issue. Attaching complete dmesg output and giving a bit more specs about the box: The Mainboard is a Tyan Tomcat i7210 (S5112) with the bios version 105. According to the website at http://www.tyan.com/support/html/b_s5112.html there is a new release which fixes "some compatibility issues with Adaptec SCSI card and ARC-1120 raid card". I'll see if this bios fixes the megaraid issue as well during the next scheduled maintainance window.
Created attachment 115200 [details] kernel bootup log
Still an issue with a recent kernel? If no, is "acpi=noirq" a sufficient workaround? If yes, please attach the dmesg and /proc/interrupts from success & failure cases.
Closing issue. I cannot observe this problem anymore with recent RHEL 4.x.