| View previous topic :: View next topic |
| Author |
Message |
mutehim
n00b
Joined: 31 May 2004
Posts: 26
|
 Posted: Mon Jun 17, 2013 11:28 am Post subject: [SOLVED] pm-hibernate - slow resume caused by cfg80211 |
 |
|
Hi there,
after years of abstinence i want back to gentoo. I am pretty impressed about portage working more flawlessly then i remembered. Anyway, i have an encrypted drive setup using cryptsetup with luks.
Therefore i have my own initram to mount my root ang get the keyfiles for all other devices including the swap.
Now, i want to hibernate. I figured out, that my swap (/dev/mapper/luksSwap) has the internal number 254:1.
When i enter (echo 254:1 > /sys/power/resume) on the busybox after mounting all my encrypted stuff everything works fine, but to slow.
So my questions are:
- What is the proper way to figure out the drive numer (254:1) - i pretty much guessed by looking in the folder /sys/dev/block?
- What happens if i resume from swap without having hibernated? Is there a check to perform before doing resume?
- Resuming takes twice as long as hibernating. Is there a way to debug this?
If i get this done right, i will write a short howto since using encrypted swap for hibernation is a not well covered topic without using genkernel.
Thanks in advance.
My init-file looks like:
| Code: |
#!/bin/busybox sh
TARGET=/dev/sda6
NAME=luksRoot
MAP=/dev/mapper/$NAME
ROOT=/mnt/root
INIT=/sbin/init
SWAP=/dev/sda5
SWAPMAP=luksSwap
SWAPKEY=/root/luks/swap
SWAPKEY=$ROOT/$SWAPKEY
mount -t proc proc /proc
mount -t sysfs sysfs /sys
mount -t devtmpfs none /dev
/sbin/cryptsetup luksOpen $TARGET $NAME
if [ ! -d $ROOT ]; then mkdir $ROOT; fi
mount $MAP $ROOT
/sbin/cryptsetup luksOpen -d $SWAPKEY $SWAP $SWAPMAP
#debug
/bin/busybox sh
echo 254:1 /sys/power/resume
umount /dev
umount /proc
umount /sys
exec switch_root $ROOT $INIT
|
Sorry for my english.
M
Last edited by mutehim on Wed Jun 19, 2013 10:27 am; edited 1 time in total |
|
| Back to top |
|
 |
Hu
Moderator
Joined: 06 Mar 2007
Posts: 21593
|
| Posted: Mon Jun 17, 2013 10:23 pm Post subject: |
|
|
| If I am interpreting your initramfs correctly, you have designed it to cause filesystem corruption on resume. When you boot to resume, you must not mount any filesystems which were mounted at hibernation time. |
|
| Back to top |
|
|
mutehim
n00b
Joined: 31 May 2004
Posts: 26
|
| Posted: Tue Jun 18, 2013 7:48 am Post subject: |
|
|
| Quote: | | If I am interpreting your initramfs correctly, you have designed it to cause filesystem corruption on resume |
not particularly  - honestly i have not thought about that, stupid ...
Does a readonly mount cause corruptions aswell?
I guess it is not possible to store the keyfiles on the root-system if i want to resume the system without datacorrution. Therefore i changed my init to mount a encrypted container directly stored in the initram, which holds the two keys for my root and swap device.
Now it look like:
| Code: |
#!/bin/busybox sh
TARGET=/dev/sda9
NAME=luksRoot
MAP=/dev/mapper/$NAME
ROOT=/mnt/root
INIT=/sbin/init
KEYFILE=/keycont
SWAP=/dev/sda5
SWAPMAP=luksSwap
SWAPKEY=/mnt/keys/swap
ROOTKEY=/mnt/keys/root
mount -t proc proc /proc
mount -t sysfs sysfs /sys
mount -t devtmpfs none /dev
echo "Setting up loopback device"
losetup /dev/loop0 /keycont
echo "Open container"
/sbin/cryptsetup luksOpen /dev/loop0 luksKeys
mount -o ro /dev/mapper/luksKeys /mnt/keys
echo "Opening Swap"
/sbin/cryptsetup luksOpen -d $SWAPKEY $SWAP $SWAPMAP
echo "copy root keyfile"
# Here, i copy the key because i want to unmount the container, but i dont want to open the device yet.
cp $ROOTKEY /tmp/root
ROOTKEY=/tmp/root
# is this a security risk?
echo "Unmount key container"
umount /mnt/keys
cryptsetup luksClose luksKeys
losetup -d /dev/loop0
echo "Trying to resume"
echo 254:1 > /sys/power/resume
echo "Boot from root"
/sbin/cryptsetup luksOpen -d $ROOTKEY $TARGET $NAME
mount $MAP $ROOT
umount /dev
umount /proc
umount /sys
exec switch_root $ROOT $INIT
|
Hopefully this is more likely designed to boot my system than to corrupt it.
Anyway. It boots fine and resume works aswell but, again, very slow (2-3 times the hibernation time).
Thanks!
Edit:
I can answer most of my (stupid) questions by myself now:
Q&A:
What is the proper way to figure out the drive numer (254:1) - i pretty much guessed by looking in the folder /sys/dev/block?
| Code: | ls -l /dev/mapper/luksRoot
brw------- 1 root root 254, 1 Jun 18 21:33 /dev/mapper/luksSwap
|
There we are => (254:1)
What happens if i resume from swap without having hibernated? Is there a check to perform before doing resume?
Seems like the kernel does this check and there is no need to treat this, right?
Resuming takes twice as long as hibernating. Is there a way to debug this?
Yes, use dmesg. In my case, time was wasted by cfg80211 which is used by the Networkmanager.
In my case adding
| Code: | echo "RestartServices NetworkManager NetworkManagerDispatcher"
>> /etc/hibernate/common.conf |
solved the problem.
Besides no confirmation for b) and the question wheater or not it is save to readonly mount a used filesystem i am done. Thanks |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Kernel & Hardware
