You are not logged in.
Hi all!
I've just got a new SanDisk Extreme SSD 120GB (SDSSDX-120G-G25), but during booting of Arch I'm getting a flood of errors like this:
[ 13.343829] ata2: EH complete
[ 13.344870] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0
[ 13.345218] ata2.00: BMDMA stat 0x25
[ 13.345489] ata2.00: failed command: READ DMA
[ 13.345895] ata2.00: cmd c8/00:08:a7:4b:f9/00:00:00:00:00/ed tag 0 dma 4096 in
[ 13.345896] res 51/10:01:01:00:00/00:00:00:00:00/00 Emask 0x81 (invalid argument)
[ 13.346927] ata2.00: status: { DRDY ERR }
[ 13.347432] ata2.00: error: { IDNF }
[ 13.383816] ata2.00: configured for UDMA/133
I found a few places where it's mentioned that the problem might be connected to AHCI and NCQ. I've got an old motherboard Gigabyte 945PL-S3G, which supports only SATA II. In BIOS there are a few options for the SATA mode:
- Disabled: Disabled SATA controller.
- Auto: Auto arrange by BIOS.
- Combined Mode: PATA and SATA are combined. Max. of 2 IDE drives in each channel.
- Enhanced Mode: Enable both SATA and PATA. Max. of 6 IDE drives are supported.
- Non-combined.
However, there is no AHCI option, so I'm not sure if the motherboard supports AHCI (couldn't find anything about on Gigabyte's web site). Currently, the SATA mode is set to Auto.
I found a few threads mentioning that the problem may disappear if add libata.force=noncq to Grub's menu.lst. I've tried that, but nothing's changed.
I've tried setting MODULES="ahci" in /etc/mkinitcpio.conf running mkinitcpio -p linux, as suggested here: https://wiki.archlinux.org/index.php/AHCI But again, nothing's changed. dmesg|grep -i ahci returns nothing, which looks like saying that AHCI is not enabled.
I guess the reason for all these problems might be that my motherboard doesn't support AHCI and NCQ. But how to make sure that it is so? If AHCI is supported, how to enable it? If NCQ is not supported, how to disable it to remove the errors I listed in the beginning?
I would really appreciate any help. Thanks!
Last edited by beloglazov (2012-04-23 22:47:18)
Offline
I've just tried this test: http://andyduffell.com/techblog/?p=852 and got zeros, which means TRIM is working fine (I'm not sure if it's somehow related to the problem or not).
Offline
I've just done one more test: first without libata.force=noncq in menu.lst, and then with it just to check if NCQ gets disabled.
1. Without NCQ
dmesg|grep -i ncq
[ 0.000000] Kernel command line: root=/dev/sdb1 ro #libata.force=noncq
[ 0.823563] ata1.00: 1953523055 sectors, multi 16: LBA48 NCQ (depth 0/32)
[ 0.838897] ata2.00: 234441648 sectors, multi 16: LBA48 NCQ (depth 0/32)
2. With NCQ
dmesg|grep -i ncq
[ 0.000000] Kernel command line: root=/dev/sdb1 ro libata.force=noncq
[ 0.830121] ata1.00: FORCE: horkage modified (noncq)
[ 0.831176] ata1.00: 1953523055 sectors, multi 16: LBA48 NCQ (not used)
[ 0.846808] ata2.00: FORCE: horkage modified (noncq)
[ 0.848934] ata2.00: 234441648 sectors, multi 16: LBA48 NCQ (not used)
So looks like NCQ does get disabled, but the errors I listed in the first message still appear, which means that they are not caused by NCQ. Here is my menu.lst:
timeout 5
default 0
color light-blue/black light-cyan/blue
# (0) Arch Linux
title Arch Linux
root (hd0,0)
kernel /boot/vmlinuz-linux root=/dev/sdb1 ro
initrd /boot/initramfs-linux.img
# (1) Arch Linux
title Arch Linux Fallback
root (hd0,0)
kernel /boot/vmlinuz-linux root=/dev/sdb1 ro
initrd /boot/initramfs-linux-fallback.img
Please let me know if you have any ideas. Thanks!
Offline
badblocks -w revealed some bad blocks. I'm going to run it again to confirm.
Offline
It is confirmed, on the second pass badblocks -w has found 1079 bad blocks (1055/0/24 errors). I'm returning the disk and going for Intel 520 120GB.
Offline