Messages in this thread | | | Subject | IDE DMA/irq timeouts | From | Nick Burrett <> | Date | 23 Sep 1999 15:10:53 +0100 |
| |
Hi,
Since upgrading our servers to kernel version 2.2.10 and 2.2.12 I have noticed increasing instability with the hard discs.
The servers I use (all single processor configurations) present the following errors:
hdb: irq timeout: status=0xd0 { Busy } hda: DMA disabled ide0: reset: success hdb: irq timeout: status=0xd0 { Busy } ide0: reset: success hdb: irq timeout: status=0xd0 { Busy } ide0: reset: success
and
hdc: timeout waiting for DMA hdc: irq timeout: status=0xd0 { Busy } hdc: DMA disabled ide1: reset: success
and
hda: dma_intr: status=0x71 { DriveReady DeviceFault SeekComplete Error } hda: dma_intr: error=0x10 { SectorIdNotFound }, LBAsect=3487119, sector=1452 hda: DMA disabled hdb: DMA disabled ide0: reset: success
I originally suspected that this was a problem with the hard discs as the BIOS has never detected FUJITSU MPC3102AT hard discs correctly (the BIOS recognises them as 648Mb hard discs, where they are actually 9765Mb). But our recent experiences with two brand new (i.e. totally unused) MPC3102AT hard discs suggested that the fault may lie elsewhere.
We installed the MPC3102AT into a computer, partitioned and formatted the disk. Running e2fsck on the disk displayed many disk errors and send several inodes to lost and found. The same errors occurred on the second MPC3102AT hard disc. We placed the hard discs into another computer (a PII 333), where the BIOS correctly recognised them this time, and exactly the same problem occurred. The kernel used in both cases was 2.2.12.
The IRQ/DMA timeouts are problematic as they are seem to be related to some major disk corruption. The kernel doesn't report the disk errors until too late, and we end up losing a about 90Mb of data to lost+found.
The hard discs we use are: FUJITSU MPC3102ATE, Maxtor 91080D5 Maxtor 91010D6 IBM-DTTA-371440
The servers are: Gigabyte 6A-6BXD Intel 440BX AGP chipset (motherboard) Pentium II 400Mhz 512Mb RAM Intel EtherexpressPro 10/100
The hdparm settings we use on all hard discs are:
/dev/hdc: multcount = 16 (on) I/O support = 1 (32-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) nowerr = 0 (off) readonly = 0 (off) readahead = 8 (on) geometry = 19590/16/63, sectors = 19746720, start = 0
I've heard in the past that overclocking the CPU can cause the sort of problem that I am seeing. However the CPUs are not overclocked on any of the servers.
Does anybody have any thoughts on this ?
Regards,
Nick Burrett.
- 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/
| |