I've been seeing a weird error on my server console lately. We're using Dell PowerEdge 2950 and 2950 III running Oracle's Enterprise Linux 4U5, source code from RHEL AS 4u5, with no patches applied. Straight "base distribution" (I know, I know, bad idea but bare with me). I reported it to Dell's PE mailing list, but didn't get any conclusive answers. So I'm listing it here, for reference.
I've found various reports of similar errors on different hardware but all seem to revolve around Red Hat's 4 release (OEL, RHEL, CentOS, Fedora) with various updates. See [1] [2]
Some sources say it's a DMA-related issue. Others suggest to turn off ACPI using kernel boot parameters acpi=off and another says it's simply the HAL daemon probing your devices. You can configure the hald to stop reporting probes of the CD-ROM. Haven't tried that yet.
I'm gonna try patching 4U5 to the el4_u5_patch channel and see if that resolves it, perhaps update it to the el4_latest instead. Otherwise, I'm gonna see if simply silencing hald is gonna help.
Update: Dell has an answer on this. The BIOS enables DMA but the Linux kernel used in EL4 disables it. Use hdparm -d0 /dev/hda to disable the use of DMA for the hda device (the cd-rom). Use -d1 to enable it.
Update2: The Dell Linux mailing list suggested more fixes on the issue. Others have similar problems. They all include the CD/DVD. You may try to tell /etc/lvm/lvm.conf to ignore removable media. You can disable the virtual CD in the BIOS and one mentioned to removed CD-related lines from /etc/fstab.
Sep 1 15:36:26 kernel: hda: irq timeout: status=0xd0 { Busy }
Sep 1 15:36:26 kernel: hda: irq timeout: error=0x00
Sep 1 15:36:26 kernel: hda: ATAPI reset complete
I've found various reports of similar errors on different hardware but all seem to revolve around Red Hat's 4 release (OEL, RHEL, CentOS, Fedora) with various updates. See [1] [2]
Some sources say it's a DMA-related issue. Others suggest to turn off ACPI using kernel boot parameters acpi=off and another says it's simply the HAL daemon probing your devices. You can configure the hald to stop reporting probes of the CD-ROM. Haven't tried that yet.
I'm gonna try patching 4U5 to the el4_u5_patch channel and see if that resolves it, perhaps update it to the el4_latest instead. Otherwise, I'm gonna see if simply silencing hald is gonna help.
Update: Dell has an answer on this. The BIOS enables DMA but the Linux kernel used in EL4 disables it. Use hdparm -d0 /dev/hda to disable the use of DMA for the hda device (the cd-rom). Use -d1 to enable it.
Update2: The Dell Linux mailing list suggested more fixes on the issue. Others have similar problems. They all include the CD/DVD. You may try to tell /etc/lvm/lvm.conf to ignore removable media. You can disable the virtual CD in the BIOS and one mentioned to removed CD-related lines from /etc/fstab.
Comments
What was your final solution? Do you now enable DMA for the cd-rom in a startup script now and does that work?
Is your BIOS/firmware up to date?