Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines
Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.
REPEAT: Do NOT report bugs for outdated packages!
https://wiki.archlinux.org/title/Bug_reporting_guidelines
Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.
REPEAT: Do NOT report bugs for outdated packages!
FS#7459 - Different names of (virtual) hard drives, install CD boot vs. hda boot
Attached to Project:
Arch Linux
Opened by Dmitry Golubovsky (golubovsky) - Monday, 18 June 2007, 14:55 GMT
Last edited by Tobias Powalowski (tpowa) - Tuesday, 19 June 2007, 13:36 GMT
Opened by Dmitry Golubovsky (golubovsky) - Monday, 18 June 2007, 14:55 GMT
Last edited by Tobias Powalowski (tpowa) - Tuesday, 19 June 2007, 13:36 GMT
|
DetailsDescription:
When booting from the installation CD (LinuxTag 2007 FTP ISO), QEMU/KVM's virtual PIIX3 IDE controller is used as a scsi-over-ata device, and disks are sdaX. When booting the unstalled system from virtual disk, the controller is used as a pure IDE device, and disks are hdaX. Comparison of dmesg's in both cases shows that libata is not loaded (and neither is scsi-over-ata layer) when booting from hda. Additional info: See this forum message for details: http://bbs.archlinux.org/viewtopic.php?id=34225 Not tested on real PIIX controllers. Steps to reproduce: Install ArchLinux from CD. Analyze dmesg output. Boot the installed system. Analyze dmesg output. PS This may be an issue with QEMU/KVM emulation, when booting from a "hard disk" puts the virtual IDE controller in the state when it cannot be correctly identified as requiring loading libata, contrary to the boot from "cdrom". Otherwise there is some difference in the boot scripts between those on the CD and installed. |
This task depends upon
Please test new kernel from http://archlinux.org/~tpowa/rc-kernel/
if not its your own stuff to get things correct.
FS#7445, but they are not relevant). I did not do anything specific to enable hwdetect: my goal was to try to install Arch Linux on a virtual machine in the most straightforward way, and see what happens.A side question: if hwdetect is enabled in CD boot (likely it is) then why isn't it enabled by default in what is installed?
Earlier I installed x86_64 version of Duke, disks show up as sda in both CD and hda boots.
So I'm just trying to narrow the problem down - it may be one of:
- my own omission that running quickinst I skipped some steps
- KVM/QEMU issue how it emulates block devices
- kernel issue when different device configuration is compiled on CD and in kernel26
- installed boot scripts which run hardware detection differently from CD ones