You are not logged in.
initially I installed [arch linux and boot loader on ext. usb-HD following
http://www.danynativel.com/blog/2013/02 … ypted-ssd/ but using UUIDs only (instead of sdXn)
(I later found out that this guide was outdated; with some battles I nevertheless completed the install (and the post-install tweaking it proposes). use official arch linux install guide instead. )
ext. HD part. usage end of UUID# size GUID
-------------------------------------------------------------------
sda1 BIOS boot ...f3be 2M EF02 (BIOS)
sda2 /boot ...38db 100M 8300 (linux fs)
sda3 swap ...6941 8G 8200 (swap)
sda4 vgroup ...2072 1.9T 8E00 (LVM)
# grub-install /dev/sda
# nano /etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="cryptdevice=/dev/disk/by-uuid/42cab5e0-887b-4bef-b898-d7533ae32072:vgroup"
# grub-mkconfig -o /boot/grub/grub.cfg
the system worked great!
until I did
#pacman -Syy
#pacman -Syu
which downloaded a couple of hundred MB and afaik also upgraded the kernel.
now the ext. HD no longer boots, waiting 10 s for an UUID of the archbang system disk and after another try gives up.
The boot-repair CD didn't fix it; report after boot-repair executed from CD-boot:
http://paste.ubuntu.com/7765072/
then I mounted the ext. HD under Ubuntu 12.04. the ext. HD became sdc:
/dev/sdc2: UUID="8b5830cc-db47-411c-b2ae-ed11fe2e38db" TYPE="ext2"
/dev/sdc4: UUID="42cab5e0-887b-4bef-b898-d7533ae32072" TYPE="crypto_LUKS"
/dev/mapper/udisks-luks-uuid-42cab5e0-887b-4bef-b898-d7533ae32072-uid1000: UUID="TQJMlF-pZ2d-t94e-mNjE-nEBh-atiH-7avVdK" TYPE="LVM2_member"
/dev/mapper/vgroup-root: UUID="9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f" TYPE="ext4"
/dev/mapper/vgroup-home: UUID="6add2f29-38f5-4ec8-96cf-c680c5c3fa11" TYPE="ext4"
and did (admit w/o much thinking)
# sudo grub-install --root-directory=/mnt /dev/sdc
finished w/o error.
plugged the ext. HD into a machine w/o any int. HD; so it became sda.
still doesn't boot, obviously looking for UUID from ubuntu system disk.
result obtained from boot-repair CD analysis after grub-install: http://paste.ubuntu.com/7765242/
then tried booting archlinux live-CD and did
# cryptsetup luksOpen /dev/sda4 enc-lvm
>gave passphrase
# mount /dev/mapper/vgroup-root /mnt
# mount /dev/sda2 /mnt/boot
# mount /dev/mapper/vgroup-home /mnt/home
# arch-chroot /mnt
# grub-install /dev/sda
# grub-mkconfig -o /boot/grub/grub.cfg
#exit
then unmounted, rebooted with this result:
[Edit n:] now put pack the 320GB int. HD (the ext. HD becoming sdb) and did
# grub-mkconfig -o /boot/grub/grub.cfg
again, but booting ext. HD still doesn't ask for passphrase. here is the latest http://paste.ubuntu.com/7771346/ (ignore all the sda stuff.)
it shows an error in " == StdErr Messages: ==" section for sdb:
cat: write error: Broken pipe
File descriptor 8 (/proc/2845/mounts) leaked on lvscan invocation. Parent PID 16067: bash
No volume groups found
don't know what this means.
[edit n+1:] boot screen again, note, there is not a single sdb device in #blkid eventhough I booted from sdb:
/etc/mkinitcpio.conf
HOOKS="base udev block autodetect modconf keyboard keymap encrypt lvm2 filesystems"
[edit n+2:]
$ sudo blkid
/dev/sda[...]
/dev/sdb2: UUID="8b5830cc-db47-411c-b2ae-ed11fe2e38db" TYPE="ext2"
/dev/sdb4: UUID="42cab5e0-887b-4bef-b898-d7533ae32072" TYPE="crypto_LUKS"
/dev/mapper/udisks-luks-uuid-42cab5e0-887b-4bef-b898-d7533ae32072-uid1000: UUID="TQJMlF-pZ2d-t94e-mNjE-nEBh-atiH-7avVdK" TYPE="LVM2_member"
/dev/mapper/vgroup-root: UUID="9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f" TYPE="ext4"
/dev/mapper/vgroup-home: UUID="6add2f29-38f5-4ec8-96cf-c680c5c3fa11" TYPE="ext4"
$ sudo lvmdiskscan
...
/dev/dm-1 [ 1.81 TiB] LVM physical volume
...
/dev/sdb1 [ 1.00 MiB]
/dev/sdb2 [ 97.00 MiB]
/dev/sdb3 [ 7.90 GiB]
/dev/sdb4 [ 1.81 TiB]
0 disks
28 partitions
0 LVM physical volume whole disks
1 LVM physical volume
$ sudo pvscan
PV /dev/dm-1 VG vgroup lvm2 [1.81 TiB / 39.01 GiB free]
Total: 1 [1.81 TiB] / in use: 1 [1.81 TiB] / in no VG: 0 [0 ]
$ sudo lvscan
ACTIVE '/dev/vgroup/root' [16.00 GiB] inherit
ACTIVE '/dev/vgroup/home' [1.76 TiB] inherit
$ sudo pvs
PV VG Fmt Attr PSize PFree
/dev/dm-1 vgroup lvm2 a- 1.81t 39.01g
$ sudo lvdisplay
--- Logical volume ---
LV Name /dev/vgroup/root
VG Name vgroup
LV UUID JRiGGp-OnoE-WDrZ-h4wL-EFQI-qthw-emtfPE
LV Write Access read/write
LV Status available
# open 0
LV Size 16.00 GiB
Current LE 4096
Segments 1
Allocation inherit
Read ahead sectors auto
- currently set to 256
Block device 252:2
--- Logical volume ---
LV Name /dev/vgroup/home
VG Name vgroup
LV UUID Df4gcB-J72X-4vb8-DRBj-Tybv-9GDu-KkI4wa
LV Write Access read/write
LV Status available
# open 0
LV Size 1.76 TiB
Current LE 460800
Segments 1
Allocation inherit
Read ahead sectors auto
- currently set to 256
Block device 252:3
==> Building image from preset: /etc/mkinitcpio.d/linux.preset: 'default'
-> -k /boot/vmlinuz-linux -c /etc/mkinitcpio.conf -g /boot/initramfs-linux.img
==> Starting build: 3.15.3-1-ARCH
-> Running build hook: [base]
-> Running build hook: [udev]
-> Running build hook: [block]
-> Running build hook: [autodetect]
-> Running build hook: [modconf]
-> Running build hook: [keyboard]
-> Running build hook: [keymap]
-> Running build hook: [encrypt]
-> Running build hook: [lvm2]
-> Running build hook: [filesystems]
==> Generating module dependencies
==> Creating gzip initcpio image: /boot/initramfs-linux.img
==> Image generation successful
==> Building image from preset: /etc/mkinitcpio.d/linux.preset: 'fallback'
-> -k /boot/vmlinuz-linux -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-fallback.img -S autodetect
==> Starting build: 3.15.3-1-ARCH
-> Running build hook: [base]
-> Running build hook: [udev]
-> Running build hook: [block]
-> Running build hook: [modconf]
-> Running build hook: [keyboard]
-> Running build hook: [keymap]
-> Running build hook: [encrypt]
-> Running build hook: [lvm2]
-> Running build hook: [filesystems]
==> Generating module dependencies
==> Creating gzip initcpio image: /boot/initramfs-linux-fallback.img
==> Image generation successful
adding
GRUB_CMDLINE_LINUX="lvmwait=/dev/mapper/vgroup-root rootdelay=20"
into /etc/default/grub also didn't help
/boot/grub/grub.cfg :
#
# DO NOT EDIT THIS FILE
#
# It is automatically generated by grub-mkconfig using templates
# from /etc/grub.d and settings from /etc/default/grub
#
### BEGIN /etc/grub.d/00_header ###
insmod part_gpt
insmod part_msdos
if [ -s $prefix/grubenv ]; then
load_env
fi
if [ "${next_entry}" ] ; then
set default="${next_entry}"
set next_entry=
save_env next_entry
set boot_once=true
else
set default="0"
fi
if [ x"${feature_menuentry_id}" = xy ]; then
menuentry_id_option="--id"
else
menuentry_id_option=""
fi
export menuentry_id_option
if [ "${prev_saved_entry}" ]; then
set saved_entry="${prev_saved_entry}"
save_env saved_entry
set prev_saved_entry=
save_env prev_saved_entry
set boot_once=true
fi
function savedefault {
if [ -z "${boot_once}" ]; then
saved_entry="${chosen}"
save_env saved_entry
fi
}
function load_video {
if [ x$feature_all_video_module = xy ]; then
insmod all_video
else
insmod efi_gop
insmod efi_uga
insmod ieee1275_fb
insmod vbe
insmod vga
insmod video_bochs
insmod video_cirrus
fi
}
if loadfont unicode ; then
set gfxmode=auto
load_video
insmod gfxterm
set locale_dir=$prefix/locale
set lang=en_US
insmod gettext
fi
terminal_input console
terminal_output gfxterm
if [ x$feature_timeout_style = xy ] ; then
set timeout_style=menu
set timeout=5
# Fallback normal timeout code in case the timeout_style feature is
# unavailable.
else
set timeout=5
fi
### END /etc/grub.d/00_header ###
### BEGIN /etc/grub.d/10_linux ###
menuentry 'Arch Linux' --class arch --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-simple-9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f' {
load_video
set gfxpayload=keep
insmod gzio
insmod part_gpt
insmod ext2
set root='hd1,gpt2'
if [ x$feature_platform_search_hint = xy ]; then
search --no-floppy --fs-uuid --set=root --hint-bios=hd1,gpt2 --hint-efi=hd1,gpt2 --hint-baremetal=ahci1,gpt2 8b5830cc-db47-411c-b2ae-ed11fe2e38db
else
search --no-floppy --fs-uuid --set=root 8b5830cc-db47-411c-b2ae-ed11fe2e38db
fi
echo 'Loading Linux linux ...'
linux /vmlinuz-linux root=UUID=9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f rw lvmwait=/dev/mapper/vgroup-root rootdelay=20 cryptdevice=/dev/disk/by-uuid/42cab5e0-887b-4bef-b898-d7533ae32072:vgroup
echo 'Loading initial ramdisk ...'
initrd /initramfs-linux.img
}
submenu 'Advanced options for Arch Linux' $menuentry_id_option 'gnulinux-advanced-9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f' {
menuentry 'Arch Linux, with Linux linux' --class arch --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-linux-advanced-9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f' {
load_video
set gfxpayload=keep
insmod gzio
insmod part_gpt
insmod ext2
set root='hd1,gpt2'
if [ x$feature_platform_search_hint = xy ]; then
search --no-floppy --fs-uuid --set=root --hint-bios=hd1,gpt2 --hint-efi=hd1,gpt2 --hint-baremetal=ahci1,gpt2 8b5830cc-db47-411c-b2ae-ed11fe2e38db
else
search --no-floppy --fs-uuid --set=root 8b5830cc-db47-411c-b2ae-ed11fe2e38db
fi
echo 'Loading Linux linux ...'
linux /vmlinuz-linux root=UUID=9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f rw lvmwait=/dev/mapper/vgroup-root rootdelay=20 cryptdevice=/dev/disk/by-uuid/42cab5e0-887b-4bef-b898-d7533ae32072:vgroup
echo 'Loading initial ramdisk ...'
initrd /initramfs-linux.img
}
menuentry 'Arch Linux, with Linux linux (fallback initramfs)' --class arch --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-linux-fallback-9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f' {
load_video
set gfxpayload=keep
insmod gzio
insmod part_gpt
insmod ext2
set root='hd1,gpt2'
if [ x$feature_platform_search_hint = xy ]; then
search --no-floppy --fs-uuid --set=root --hint-bios=hd1,gpt2 --hint-efi=hd1,gpt2 --hint-baremetal=ahci1,gpt2 8b5830cc-db47-411c-b2ae-ed11fe2e38db
else
search --no-floppy --fs-uuid --set=root 8b5830cc-db47-411c-b2ae-ed11fe2e38db
fi
echo 'Loading Linux linux ...'
linux /vmlinuz-linux root=UUID=9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f rw lvmwait=/dev/mapper/vgroup-root rootdelay=20 cryptdevice=/dev/disk/by-uuid/42cab5e0-887b-4bef-b898-d7533ae32072:vgroup
echo 'Loading initial ramdisk ...'
initrd /initramfs-linux-fallback.img
}
}
### END /etc/grub.d/10_linux ###
### BEGIN /etc/grub.d/20_linux_xen ###
### END /etc/grub.d/20_linux_xen ###
### BEGIN /etc/grub.d/30_os-prober ###
### END /etc/grub.d/30_os-prober ###
### BEGIN /etc/grub.d/40_custom ###
# This file provides an easy way to add custom menu entries. Simply type the
# menu entries you want to add after this comment. Be careful not to change
# the 'exec tail' line above.
### END /etc/grub.d/40_custom ###
### BEGIN /etc/grub.d/41_custom ###
if [ -f ${config_directory}/custom.cfg ]; then
source ${config_directory}/custom.cfg
elif [ -z "${config_directory}" -a -f $prefix/custom.cfg ]; then
source $prefix/custom.cfg;
fi
### END /etc/grub.d/41_custom ###
### BEGIN /etc/grub.d/60_memtest86+ ###
### END /etc/grub.d/60_memtest86+ ###
appreciate your help. thanks!
***
[edit n+3:] maybe I found the solution: http://mzanfardino.wordpress.com/2012/0 … rch-linux/
(reinstall udev and mkinitcpio, recreate the initial ramdisk environment)
Had no time yet to try it; will report back soon.
[edit n+4:] edit n+3 didn't help, eventhough
#pacman -Syu downloaded and installed another 180MB (incl. another kernel).
doing more searches I found two posts describing the same problem of device not detected; in both cases remained unsolved:
https://bbs.archlinux.org/viewtopic.php?id=47288
https://bbs.archlinux.org/viewtopic.php?id=80355
reading archwiki's mkinitcpio, I then modified HOOKS using base, systemd, ... sd-encrypt, sd-lvm2, ... and got a more detailed error-log but the cause not finding the ext. usb device sdbX is unchanged.
now it doesn't even look first for
/dev/sdb4: UUID="42cab5e0-887b-4bef-b898-d7533ae32072" TYPE="crypto_LUKS"
but tries to find (dev/mapper/vgroup-root) UUID="9dbd2e7d-8f4e-4fc4-8e46-64b29155e00f" which is inside the /dev/sdb4
best,
0.
p.s.: could this problem be related to http://ubuntuforums.org/showthread.php?t=2066130 ?
Last edited by 0liver (2014-07-25 12:00:26)
Offline
Don't mix and match old guides or ones for other distros, it only confuses - particularly for the init system.
Specify the UUID for cryptdevice in the same format as you do for root, i.e. "cryptdevice=UUID=...".
Offline
thanks.
set
GRUB_CMDLINE_LINUX_DEFAULT="cryptdevice=UUID=42cab5e0-887b-4bef-b898-d7533ae32072:vgroup"
did run
# grub-mkconfig -o /boot/grub/grub.cfg
rebooted, see the syntax-change in grub when viewing the commands with 'e', but nothing changed. sdb still not present. doesn't ask for key.
Last edited by 0liver (2014-07-19 15:27:49)
Offline
What do you mean with "sdb still not present"? Did you go to the busybox emergency shell and /dev/sdb does not turn up?
If it does not, it did turn up booting a CD iso in your tries before or not? If it does turn up, but you mean the boot still times out waiting for vgroup-root (which i suspect) then try to unlock the LUKS device sdb manually from the emergency shell for starters
cryptsetup open /dev/sdb4 lvm
vgscan
vgchange -ay
and let it try boot again with "exit". Does that work?
Offline
> What do you mean with "sdb still not present"? Did you go to the busybox emergency shell and /dev/sdb does not turn up?
right! that's it.
> If it does not, it did turn up booting a CD iso in your tries before or not?
it turns up when I boot latest arch iso live CD and I can open it with
#cryptSetup luksOpen /dev/sbd4 enc-lvm
I can reboot with Grub from sdb:
http://www.directupload.net/file/d/3688 … mh_jpg.htm
http://www.directupload.net/file/d/3688 … yf_jpg.htm
http://www.directupload.net/file/d/3688 … 7f_jpg.htm
but then sdb is gone (and I see only sda which is not at all involved):
http://www.directupload.net/file/d/3688 … mt_jpg.htm
can't do #cryptSetup luksOpen /dev/sbd4 enc-lvm
device doesn't exist!
Last edited by 0liver (2014-07-19 19:34:05)
Offline
I just re-read your initial post. You tried so many things, it is not clear what the current setup is (apart from the grub config you show in the linked images above in post 5).
Obviously it makes one wonder where the grub install is, if the drive you installed it to is not shown in the busybox. Did you install that to your first disk (sda)? That's the only idea I would have for it. Is the Windows on it still booting? The Bios is set to legacy (not EFI)? Did you check from the arch live cd which partition is bootable?
Offline
> You tried so many things, it is not clear what the current setup is (apart from the grub config you show in the linked images above in post 5).
I undid the unsuccessful test with HOOKS using systemd, sd-encrypt, sd-lvm2
> Obviously it makes one wonder where the grub install is, if the drive you installed it to is not shown in the busybox.
Grub finds sdbx o/w it couldn't start on usb-disk. (I've learned that grub detetects devices differently. once boot starts, the device list is regenerated. this is the moment when sdbx is no longer found. the hook 'usb' is deprecated now, so won't help I guess.)
> Did you install that to your first disk (sda)?
no, grub2 is installed it on usb-disk (sdb).
> That's the only idea I would have for it. Is the Windows on it still booting?
yes.
> The Bios is set to legacy (not EFI)?
it is a thinkpad T60 (2007); can't find any bios-EFI option so guess it doesn't yet have EFI.
it booted with internal HD present (as well as with int. HD removed) before from usb-HD.
just removed int. HD (as well as live-CD) and booted with usb-disk only. get the same grub screen and then timeout as already shown http://www.directupload.net/file/d/3688 … mt_jpg.htm
except that #blkid list is now empty!
> Did you check from the arch live cd which partition is bootable?
sda2 has flag Boot (cfdisk shows).
#parted /dev/sdb print # since cfdisk doesn't support gpt:
Model: Seagate Backup+ RD (scsi)
Disk /dev/sdb: 2000GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Disk Flags:
Number Start End Size File system Name Flags
1 1049kB 2098kB 1049kB BIOS boot partition bios_grub
2 3146kB 105MB 102MB ext2 Linux filesystem
3 106MB 8590MB 8484MB Linux swap
4 8591MB 2000GB 1992GB Linux LVM lvm
I wouldn't keep researching this issue if I were convinced that after re-install of arch linux on usb-disk and after an update of the kernel the same boot problem wouldn't reappear.
the broken boot was already observed with older arch linux versions as described here: http://mzanfardino.wordpress.com/2012/0 … rch-linux/ (and already linked in my first post; sadly what helped him doesn't help me, as also already pointed out above.)
Last edited by 0liver (2014-07-20 14:07:26)
Offline
Thanks for the update/answers and explanations of your tries. A couple are not clear to me (I adapt to your quoting style for now):
>> no, grub2 is installed it on usb-disk (sdb).
conflicts with
>> sda2 has flag Boot (cfdisk shows).
And then you continue with gparted output from /dev/sdb
I guess the sda2 was a typo there. Reading up on GPT here, it seems that a "bootable" flag is incompatible with the GPT specification per se, but some BIOS' want it, others don't (no idea about a sturdy T60. Your sdb disk technically does not need gpt but, yes, it used to work until that update).
Anyhow, I do not have a further idea currently. Being pragmatic what I would do is go to the Arch iso, chroot into that sdb and re-install the bootloader. (Why not give syslinux a try, for example.) Maybe someone else has another idea.
edit: I happen to browse over this bug. Follow the launchpad link and read the error description (I know, it is not your hardware, but the sympton of the device disappearing is oddly similar). According to that you can try another simple thing: eliminate the rootdelay option you added. What's the kernel on the iso you use?
Last edited by Strike0 (2014-07-21 21:28:50)
Offline
>> no, grub2 is installed it on usb-disk (sdb).
> conflicts with
>> sda2 has flag Boot (cfdisk shows).
sda is internal disk. (when an int. disk is present)
it has its own grub installed.
it boots independent from usb-disk.
and I want usb-disk to boot independent from whatever (or none) int. disk.
anyway, to make it as simple as possible, as said above, I removed int. disk (sda).
note, that in arch iso the usb-disk then became sda. (this was the reason to use UUIDs only)
> And then you continue with gparted output from /dev/sdb
/dev/sdb is usb-disk - before I removed int. disk.
(I used parted and not gparted, but I guess the g-prefix stands for gnome or gui which doesn't run in arch linux iso shell).
> I guess the sda2 was a typo there.
it isn't; I included it since you talked about boot flags and suspected it booted from int. disk.
> Reading up on GPT here, it seems that a "bootable" flag is incompatible with the GPT specification per se, but some BIOS' want it, others don't (no idea about a sturdy T60. Your sdb disk technically does not need gpt but, yes, it used to work until that update).
I didn't change anything on boot flags since install. the install worked fine. boot is broken since the kernel upgrade. (it did run mkinitcpio as part of #pacman -Syu)
> I would do is go to the Arch iso, chroot into that sdb and re-install the bootloader.
did this several times before.
> (Why not give syslinux a try, for example.) Maybe someone else has another idea.
tried this in advanced options in grub menu. exactly same effect. doesn't see usb-disk.
> edit: I happen to browse over this bug. Follow the launchpad link and read the error description (I know, it is not your hardware, but the sympton of the device disappearing is oddly similar).
the page you linked is very technical. can't really understand it. there are other posts which describe usb-dev had disappeared. i linked to them above as follows:
"device not detected; in both cases remained unsolved: ... https://bbs.archlinux.org/viewtopic.php?id=80355"
inner quote from this post:
>>> THE Problem: Even though, as explained above, GRUB initially
detects Flash as (hd0,x) (or sda), the kernel seemingly detects the local
HDD as sda, AND FAILS TO DETECT THE USB STICK AT ALL.
This became to be obviously true when i listed the contents of /dev,
after being dropped to the recovery shell
(or entering it manually with the `break' option in the kernel line).
There was sda1, sda2 and sda5 (the partitions of the local HDD)
and no sdb at all.
he apparently 'solved' this as follows
I copied over a kernel file named `kernel-2.6.25.29-114'
and an initrd file named `initramfs-2.6.25.20-114' over to
Arch's /boot from the /boot of the Linux on my HDD, and
using those two, together (and only together), can boot Arch.
Does anyone have an idea why my vmlinuz26 and kernel26.img
should not work? I'm clueless on this.
I don't have such files on my archlinux iso. On usb-disk (now called sda2 or /boot once mounted) I have:
grub (directory with grub.cfg)
initramfs-linux[-fallback].img
vmlinuz-linux
(however I could swap the int. disk from my 3rd notbok (T61) which has recent archbang install to get its corresponding /boot files.
Edit: copied archbang's initramfs-linux[-fallback].img and vmlinuz-linux and replaced those on usb-disk but archbang install doesn't use luks/lvm and thus doesn't have the encrypt and lvm2 hooks and on boot of usb-disk (the only hook being udev) it's waiting directly for root device UUID=...e00f
more importantly, in boot rescue shell, #blkid sees the archbang's sdaX but no sdbX.
)
> According to that you can try another simple thing: eliminate the rootdelay option you added.
I had done this already before I ran latest test shown in image link "http://www.directupload.net/file/d/3688 … yf_jpg.htm"
you see, it no longer has wait 20s option. default is 10s.
> What's the kernel on the iso you use?
3.14.1.1-ARCH
downloaded as archlinux-2014.05.01-dual.iso
and just in case it matters, the complete system upgrade on usb-disk installed linux 3.15.5-1 as shown above in "http://www.directupload.net/file/d/3686 … 6x_jpg.htm"
what puzzles me: once failed boot and in emergency shell, I see in hooks-directory only 4 empty files:
encrypt keymap lvm2 udev
eventhough I have in mkinitcpio.conf (which of course matches the HOOKS in emergency shell's file buildconfig):
HOOKS="base udev autodetect block modconf keyboard keymap encrypt lvm2 filesystems"
is this expected behaviour?
(by the way, I also tried swapping autodetect and block without success.)
thanks for your continued help!
Last edited by 0liver (2014-07-22 13:58:11)
Offline
Please stick to one setup, the way you mix output and comments from different setups with every paragraph is not helpful for anyone trying to follow you.
To the question: Yes, that is expected behaviour. See the right hand column in the table: https://wiki.archlinux.org/index.php/Mk … time_hooks
Regarding the thread you link: kernel 2.6.25.29-114 is too old to help us find your bug I would say.
You write you "tried" syslinux from grub? https://wiki.archlinux.org/index.php/Syslinux
One question I cannot see answered from your problematic update to 3.15.5-1 is whether the initramfs creation was successful or not.
Have you tried to chroot into the usb-drive from the Arch iso and re-install linux and grub (or syslinux instead)?
Offline
> Please stick to one setup, the way you mix output and comments from different setups with every paragraph is not helpful for anyone trying to follow you.
sorry, I tried to keep it simple, but couldn't avoid the mix-up. will try harder.
> Regarding the thread you link: kernel 2.6.25.29-114 is too old to help us find your bug I would say.
sure, ok, I wanted to show that I'm not the only one missing devices at boot.
> You write you "tried" syslinux from grub? https://wiki.archlinux.org/index.php/Syslinux
sorry again, I tried 'arch linux with linux linux' from grub advanced menu.
> One question I cannot see answered from your problematic update to 3.15.5-1 is whether the initramfs creation was successful or not.
yes, you see the log at about middle of my first post.
> Have you tried to chroot into the usb-drive from the Arch iso and re-install linux and grub (or syslinux instead)?
yes. I boot arch iso CD and then, - assuming sda is usb-disk with int. disk removed -
# loadkeys de
# cryptsetup luksOpen /dev/sda4 enc-lvm
...passphrase...
# mount /dev/mapper/vgroup-root /mnt
# mount /dev/sda2 /mnt/boot
# mount /dev/mapper/vgroup-home /mnt/home
# arch-chroot /mnt
# mkinitcpio -p linux
# grub-install /dev/sda
# grub-mkconfig -o /boot/grub/grub.cfg [gives a lot of warnings (as with very first install which worked) "...is leaking..." but finds boot images]
#exit
# umount /mnt/boot
# umount /mnt/home
# umount /mnt
# reboot
and choosing usb-disk as boot device at startup.
Last edited by 0liver (2014-07-22 20:15:00)
Offline
Browsing over bugs, maybe this recent report is one that affects you: https://bugs.archlinux.org/task/41273
Offline
I patched /usr/lib/initcpio/install/block according https://bugs.archlinux.org/task/41273
rebooted, chrooted and
#mkinitcpio -p linux
now results in
==> Building image from preset: /etc/mkinitcpio.d/linux.preset: 'default'
-> -k /boot/vmlinuz-linux -c /etc/mkinitcpio.conf -g /boot/initramfs-linux.img
==> ERROR '/lib/modules/3.15.3-1-ARCH' is not a valid kernel module directory
(... same for initramfs-linux-fallback.img)
I only have /lib/modules/3.15.5-1-ARCH/
so i did:
#pacman -Syy
#pacman -Syu
results in installing amongst others of linux 3.15.5-1 in chrooted environment:
http://www.directupload.net/file/d/3692 … 7r_jpg.htm
apart from an error in the proposed patch /usr/lib/initcpio/install/block
"line 21: +: command not found:"
http://www.directupload.net/file/d/3692 … wx_jpg.htm
line 21 in block is this line:
+ add_checked_modules 'drivers/usb/storage/uas.ko.gz'
#if add_checked_modules -f '(_cs|sl811_hcd|isp116x_hcd)' '/usb/host'; then
# blockdevs+=(['usb_storage?']=1 ['sd_mod?']=1 ['sr_mod?']=1)
# add_checked_modules 'drivers/usb/storage/ums-*'
#fi
if add_checked_modules -f '(_cs|sl811_hcd|isp116x_hcd)' '/usb/host'; then
blockdevs+=(['usb_storage?']=1 ['sd_mod?']=1 ['sr_mod?']=1)
add_checked_modules 'drivers/usb/storage/ums-*'
+ add_checked_modules 'drivers/usb/storage/uas.ko.gz'
fi
# firewire
means, I conclude, uas.ko.gz was not added.
I nevertheless did (and captured end of result of)
# grub-mkconfig -o /boot/grub/grub.cfg
http://www.directupload.net/file/d/3692 … 89_jpg.htm
but on reboot still no usb-device. timeout as before.
what's the syntax error in https://bugs.archlinux.org/task/41273 please?
edit 2: ok, I see, the "+" is part of the patch-syntax. will correct and try again.
***
edit 1: it must be the error described there: just tried to boot the usb-disk again with an usb-stick plugged addionally; after timeout for the usb-disk, #blkid lists the usb-stick as /dev/sda1 !
no other device, except /dev/sr0 if a CD is in the tray.
***
just deleted the '+' in the hooks script:
...
add_checked_modules 'drivers/usb/storage/uas.ko.gz'
and redid
#mkinitcpio - p linux
and now it boots into passphrase prompt. thanks again for your help and finding the right recent bug I'd run into!
[edit 3] p.s.: there was some trouble at boot:
http://www.directupload.net/file/d/3692 … 88_jpg.htm
/dev/mapper/enc-swap isn't ok;
wrongly looks on /dev/sdb3 instead of /dev/sda3 (not related to the UAS-bug or rather -'feature-request;' see below),
but home dir works eventhough it shows "Dependency failed for Cryptography Setup Home":
http://www.directupload.net/file/d/3692 … lo_jpg.htm
(instead of screenshot I'd prefer upload of this log's text, but where is it stored?)
***
but everything else looks good. even startx (gnome) works.
Last edited by 0liver (2014-07-23 15:27:33)
Offline
and now it boots into passphrase prompt. thanks again for your help and finding the right recent bug I'd run into!
A million times no. You don't use USB-attached SCSI, so this "bug" (notice it's a feature request) is irrelevant to you.
Last edited by falconindy (2014-07-23 13:17:46)
Offline
the bug description says
> Without this [uas] module some newer usb devices are not recognized.
and I did nothing else than applying the block-script-patch and now it works.
do you see another reason why it boots now that it finally detects all usb-devices? (though I understand that uas - usb attached scsi means something different)
Last edited by 0liver (2014-07-23 13:22:19)
Offline
no. yes. the s in sdx imho historically means scsi. right?
is it possible that your bug fix fixes more than only the uas issue?
it makes sense to me; my usb-disk (parted says: Model: Seagate Backup+ RD (scsi), Disk /dev/sdb: 2000GB) is a "newer usb device."
Offline
is it possible that your bug fix fixes more than only the uas issue?
Not unless the 'uas' module supports more than just UAS. Despite the confusing name, I suspect that it only supports UAS.
it makes sense to me; my usb-disk (parted says: Model: Seagate Backup+ RD (scsi), Disk /dev/sdb: 2000GB) is a "newer usb device."
So, one might actually make a better assertion about this by looking at dmesg, or the output of 'lsmod' from a successfully booted system (i.e. a live cd).
Offline
yes, it uses UAS.
didn't know the seagate 2TB slim is an UAS device; couldn't find a detailed data sheet yet.
dmesg:
…
[ 0.806704] usb 5-2: new high-speed USB device number 2 using ehci-pci
[ 0.933403] usbcore: registered new interface driver usb-storage
[ 0.934141] scsi6 : uas
[ 0.934238] usbcore: registered new interface driver uas
[ 0.935300] scsi 6:0:0:0: Direct-Access Seagate Backup+ RD A905 PQ: 0 ANSI: 6
…
lsmod:
...
cdrom 35191 1 sr_mod
ata_generic 3434 0
pata_acpi 3579 0
uas 17010 2
usb_storage 50327 1 uas
atkbd 17006 0
libps2 4571 2 atkbd,psmouse
ahci 24299 0
libahci 21772 1 ahci
ata_piix 25496 1
libata 174089 5 ahci,pata_acpi,libahci,ata_generic,ata_piix
scsi_mod 138333 5 uas,usb_storage,libata,sd_mod,sr_mod
ehci_pci 4152 0
uhci_hcd 34795 0
ehci_hcd 64619 1 ehci_pci
usbcore 188381 5 uas,uhci_hcd,usb_storage,ehci_hcd,ehci_pci
usb_common 1712 1 usbcore
executed on now again booting seagate 2TB usb-disk on thinkpad T60 (has USB 2.0 only;
a note about USB Attached SCSI Protocol (UASP): "Although UAS was added in the USB 3.0 standard, it can also be used at USB 2.0 speeds, assuming compatible hardware," see https://en.wikipedia.org/wiki/USB_Attached_SCSI (page version: 21:26, 6 June 2014, retrieved 17:20, 23 July 2014)
Last edited by 0liver (2014-07-25 11:51:21)
Offline
Awesome you got it solved! Thanks for the mention
I reckon it is up to you whether the logs are linked permanently or not.
One other method to paste for another time, see: https://bbs.archlinux.org/viewtopic.php?id=57855
I also see the report already has been fixed by falconindy with commit. If you now prepend [ Solved] and add "USB" to the "UAS" to your thread title, other users will find a working resolution easier.
Offline
@Strike0: Thanks modified title and posted only the relevant parts of dmesg & lsmod inline.
It still puzzle me why this usb/uas-disk booted before the system upgrade.
p.s.: Seagate offers a "Firmware update for Seagate Backup Plus and Seagate Slim USB 3.0 in MacOS:" http://knowledge.seagate.com/articles/e … uage=en_US *
because maybe Mac has trouble detecting this device, which I conclude from a post in an Ubuntu/Mac related forum post.**
*) only found the FW-update-tool for MacOS, but maybe related to other OS**.
**) http://forum.parallels.com/showthread.p … t=usb+scsi
Their problem occured with a Seagate Backup+ SL
It runs Ubuntu in a VM (under Parallels Desktop 9 on a MBP running OS X 10.7); when plugged via an USB-2 hub it was detected (eventhough read-only)
Offline
Yes, puzzling. Maybe it has to do with an USB module being used instead, because the UAS module itself was declared broken before 3.15.rc2: https://git.kernel.org/cgit/linux/kerne … f92a532cd0
Offline
OT: I also fixed the swap encryption setup now and I'd like to add a reminder about 'persistent device naming' https://wiki.archlinux.org/index.php/Pe … ice_naming :
/dev/disk/by-PARTUUID/…uuid… [or /dev/disk/by-PARTLABEL/…]
must be used in (#makeswap, #swapon and in) /etc/crypttab, which in my install only contains one line
enc-swap /dev/disk/by-PARTUUID/…swap-partition-uuid... /dev/urandom cipher=aes-cbc-essiv:sha256,size=128,hash=sha512,swap
partuuid's can be obtained with
# ls -l /dev/disk/by-partuuid/
or
# lsblk -f|nopaste
see http://pastie.org/private/fs8sdpcdnn3szmoe5yibbw
(Thanks again, Strike0 for the pastie-hint)
/dev/disk/by-UUID/...uuid… doesn't work.
/dev/sdXN can be used but is very dangerous (data loss!) because its name can change at boot. see warning in https://wiki.archlinux.org/index.php/Dm … sk_support
my swap-entry in /etc/fstab (which was created during install with
#genfstab -U -p /mnt >> /mnt/etc/fstab
to get UUIDs instead of disk labels) is:
/dev/mapper/enc-swap swap swap defaults 0 0
luckily Grub (on usb-disk) doesn't care if the boot command uses hd0,gpt2 or hd1,gpt2; it boots the usb-disk if there is an internal disk or not.
Here is the boot duration with the SGT Backup Plus slim usb-3.0-disk on usb-2.0 port of a Thinkpad T60 (2 GHz T7200 core2duo CPU):
10s from BIOS boot-device select to disk's passphrase prompt.
19s after passphrase to MATE-login manager** (didn't enable autologin).
20s until MATE** desktop displayed.
Now, never satisfied , because a "Dual Port USB 3.0 HUB to Express Card Adapter - NEC Chipset Renesas uPD720202" (~ 10 euro) works great with usb-3.0 disks* when used as non-boot-disk, it'd be good to find a boot manager which could boot a usb-3.0 disk via an usb-3-pcmcia adapter. (I tried without succes the http://www.plop.at boot manager which can boot usb-2.0-disks attached to pcmcia usb-2.0 adapter cards in usb-1.x-only computers.)
*) e.g., effective 63 MB/s transfer rate when copying data from one (non-encrypted) usb-3 disk to another (luks/lvm). Tested on an Dell E5520, i5-2410M, 2.30GHz; link with more details on request.
**) http://wiki.mate-desktop.org/download - fine and fast stuff!
Last edited by 0liver (2014-07-25 20:02:56)
Offline
when I tried to boot on a desktop via USB3 port, it didn't ask for volume passphrase and boot failed with root device not found time-out. after rebooting on USB2, modifying
MODULES= ""
to
MODULES= "xhci_hcd"
in /etc/mkinitcpio.conf
and
#mkcpio -p linux
it booted via usb3. — thank you for your help, mihi42, your "Schuss ins Blaue" worked fine
Edit 1: MODULES= "uhci_hcd xhci_hcd"
with xhci_hcd, uhci_hcd is also needed, e.g. on older mainboards like Thinkpad T60/T61, to make the keyboard work for typing the volume passphrase.
Last edited by 0liver (2014-11-16 21:58:54)
Offline