2.6.27 SATA drives not accessible at boot time, 2.6.24 working
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Linux |
Expired
|
Unknown
|
|||
linux (Ubuntu) |
Won't Fix
|
Medium
|
Unassigned | ||
Bug Description
System info:
Hardware:
Mainboard MSI 965P Neo2, Intel 965 chipset, ICH8 southbridge, chipset 82801H, One SATA Samsung Harddisk 300GB, One SATA optical drive, both connected to the chipset's SATA channels 3 and 4 (behaviour is unchanged if connected to channels 1 and 2).
SATA is running in IDE mode, AHCI mode is not supported by the hardware.
Software:
Affects all kernel versions from version 2.6.26 to 2.6.28-13-generic and (untested) maybe later versions also
Doesn't affect kernel version 2.6.24-19-generic, Ubuntu 8.04 or ubuntu 8.10, vanilla 2.6.25, or earlier versions
Symptoms:
Booting ubuntu 8.10 with the included kernel 2.6.27-7-generic from either the live CD or as a system on disk upgraded via apt-get dist-upgrade stops at the point where the root partition is to be mounted. The BusyBox initramfs shell prompt appears after a timeout period, and when listing the "/dev" directory I can see that it doesn't contain any disk related device nodes, even the "dev/disk" directory is missing.
The output of dmesg is showing that all four SATA channels are correctly recognised and initialised, but they all show a message
"SATA link down", just as if no disk was connected to them. There are no further SATA/ATA related error messages in the dmesg output.
Reproducibility:
Booting the same system with a still-installed 2.6.24-19 from Ubuntu 8.04 works without any problems. The dmesg output is mostly the same as with 2.6.27, the only notable difference is that instead of the message "ata3: SATA link down" and "ata4: SATA link down" it now show the correct initialisation messages for my drives in the same place in the dmesg output.
Reproducibility UPDATE:
Issue persists for Ubuntu 9.04 with kernel 2.6.28-13-generic, either precompiled version or locally compiled.
Reproducibility UPDATE #2:
Ussue also persists for Ubuntu 9.10 beta with kernel 2.6.31-14
Reproducibility UPDATE #3:
Issue persists for Ubuntu 10.04 with kernel 2.6.32-22-server
Manually trying to load the sd_mod module from the initramfs prompt didn't bring the missing device nodes into existence either.
Connecting a USB flash memory storage device and mounting it in the BusyBox initramfs shell works, however (in which case the necessary USB-disk-related device node is created automatically under /dev), and this way I was able to capture the dmesg output and attach it below.
The attached files are:
1: The output of lspci to further specify my hardware configuration
2: The output of dmesg from the initramfs shell of kernel 2.6.27
3: As a comparison the output of dmesg from the working kernel 2.6.24 setup
Changed in linux: | |
assignee: | nobody → ubuntu-kernel-team |
importance: | Undecided → Medium |
status: | Incomplete → Triaged |
Changed in linux: | |
status: | Unknown → In Progress |
tags: | removed: needs-upstream-testing |
description: | updated |
description: | updated |
tags: | added: 2.6.28 2.6.28-13-generic 9.04 |
Changed in linux: | |
status: | In Progress → Expired |
Here is the dmesg output captured from the initramfs shell. Note: the sda disk correctly initialized and mounted in this listing is only the USB flash memory storage device which i plugged in to capture the dmesg output... The internal drives are still unusable.