Light-Debian-Core-Live-CD-Wheezy + Porteus-Wheezy

[fredx181]

Hi Toni,

Do you know what icon name and place is needed to show encrypted drive proper?

Doing this shows which icon it searches for:

Code: Select all

killall pup-volume-monitor
pup-volume-monitor &
disown

For crypt I have this:

Code: Select all

Device crypt:
	display_name=crypt
	icon_name=themed_icon|gtk-harddisk	
	type=PupVolume	
	unix_dev=/dev/mapper/crypt
	label=(null)
	fstype=ext4
	uuid=5c149fd2-7212-4b32-8451-cb1718c4175a
	drv_sysname=(null)
	mntpnt=/mnt/live/memory/images/changes-exit
	flags = 0 | PUP_VOLUME_IS_MOUNTABLE

First "icon_name" shows "themed icon", second: gtk-harddisk
In Jwm version there is no icon theme recognized I think.
I didn't test but I think if you copy drive-harddisk in /usr/share/icons/hicolor/48x48/devices to "gtk-harddisk" it should show same icon as harddisk icon.

I'll get back later about the other encryption issue.
The difference between your results and mine could well have to do with the extra squashsfs you loaded but that's just a guess.

Fred

[fredx181]

Hi Toni.
copying /usr/share/icons/hicolor/48x48/devices/drive-harddisk.png to "gtk-harddisk.png" didn't work for me (on Jwm version) but copying to /usr/share/pixmaps/gtk-harddisk.png did show the harddisk icon for crypt on desktop.

I cannot reproduce the problem you have, I think I did almost exactly the same as you did, also with the extra cryptsetup_aespipe.squashfs loaded at boot.
Here's my blkid output (on JWM version):

Code: Select all

root@debian:~# blkid -o list
device     fs_type label    mount point    UUID
-------------------------------------------------------------------------------
/dev/sda1  ntfs             (not mounted)  A260A8A860A8851F
/dev/sda2  ext4             (not mounted)  960e5eb6-6aed-4134-95f7-afd1d43ee9bc
/dev/sda3  ext4    hd-data  /mnt/live/mnt/sda3 da4763af-3d30-45ee-b7d9-77119e9c4251
/dev/loop0 squashfs         /mnt/live/memory/images/01-filesystem.squashfs 
/dev/loop1 squashfs         /mnt/live/memory/images/02-aespipe_cryptsetup.squashfs 
/dev/loop2 crypto_LUKS      (in use)       e097f2c9-3857-4592-abff-41a71995c554
/dev/loop3 squashfs         /mnt/live/memory/images/021-apps-porteus.squashfs 
/dev/sda5  ext4    live-rw-foo (not mounted) 3a76b94a-0e71-4b5f-8505-3564a6cf70f6
/dev/sda6  swap             <swap>         a4ada463-3646-4a74-b508-27ba6126eb52
/dev/mapper/crypt
           ext4             /mnt/live/memory/changes e87d1fce-91d4-4a91-964b-cc36d0b56c46

My boot partition is sda3.
It looks like when the prompt for password appears at boot it's always for "loop2"

Fred

[saintless]

Hi, Fred.

Thank you for the icon tips.

I will make more tests tomorrow about the 500Mb limit problem. I suspect it could be some old hardware or old BIOS issue. I can't mount the encrypted files even manually if the size is less than 500Mb and the file is created with last mk-save encryption GUI.
Mount command asks for file system and what ever I type it errors with wrong file system type. I get the same result on two almost same age computers. I will test on newer hardware.
But default encryption with this command works fine for any size on the same computers:

Code: Select all

cryptsetup luksFormat /dev/loop1

I see lines 220 and 223 in linuxrc /dev/loop2 is specified for encrypted save file. I don't know if this can make problems when more squashfs are loaded on boot but I think it will not be a problem. I will test this also.

Toni

[fredx181]

Hi Toni,

I will make more tests tomorrow about the 500Mb limit problem. I suspect it could be some old hardware or old BIOS issue. I can't mount the encrypted files even manually. Mount commad asks for file system and what ever I type it errors with wrong file system type. I get the same result on two almost same age computers. I will test on newer hardware.

There seems to be a difference if you create encrypted savefile manually.
I compared your manual code in previous post with how it's done in mk-save.
Almost the same except that mk-save searches for a free loop device and uses that and more important maybe: in line 212 for live-boot and line 334 for porteus boot :

Code: Select all

cryptsetup -y -s 256 luksFormat $loop <<< $PPHRASE 2> /dev/null 

Try change these to (leave out the "-s 256"):

Code: Select all

cryptsetup -y luksFormat $loop <<< $PPHRASE 2> /dev/null 

Maybe that is solution.
I just copied this from the porteus make-changes script without knowing what the "-s 256" does.
EDIT; well, on second thought, after looking at the help, I doubt if this makes any difference:

Code: Select all

-s, --key-size=BITS                 The size of the encryption key

In lines 220 and 223 in linuxrc /dev/loop2 is specified for encrypted save file. I don't know if this can make problems when more squashfs are loaded on boot. I will test this also.

Looks like it is not a problem as you can see on my blkid output:

Code: Select all

/dev/loop3 squashfs         /mnt/live/memory/images/021-apps-porteus.squashfs 

Fred

[saintless]

Hi, Fred.

Solved but not sure if the news is good.
It has nothing to do with size or compression command. Over 500Mb encrypted save file made with mk-save and mkpfile.sh also started to make problems for me.
Then I found both scripts work wrong because I had testing LUKS formated partition for live-boot-3x and live-boot-2x After removing the LUKS partition all seems fine.

Code: Select all

/dev/sda5: UUID="231c1f89-8465-49de-b7ce-07864594db9b" TYPE="crypto_LUKS" 

mk-save.gtkdlg before removing sda5 LUKS partition:

Code: Select all

root@debian:~# mk-save.gtkdialog
bash: mk-save.gtkdialog: command not found
root@debian:~# mk-save.gtkdlg

CHECKBOX="true"
FILE_BROWSE_SAVEFILENAME=""
FILE_SAVEFILENAME="/live/image/live/changes.dat"
FS="ext2"
LB=""
Size="250"
EXIT="OK"

Can't do passphrase verification on non-tty inputs.
device-mapper: remove ioctl on temporary-cryptsetup-3146 failed: Device or resource busy
device-mapper: remove ioctl on temporary-cryptsetup-3185 failed: Device or resource busy
Formatting 
mkfs.ext2: option requires an argument -- 'L'
Usage: mkfs.ext2 [-c|-l filename] [-b block-size] [-C cluster-size]
        [-i bytes-per-inode] [-I inode-size] [-J journal-options]
        [-G flex-group-size] [-N number-of-inodes]
        [-m reserved-blocks-percentage] [-o creator-os]
        [-g blocks-per-group] [-L volume-label] [-M last-mounted-directory]
        [-O feature[,...]] [-r fs-revision] [-E extended-option[,...]]
        [-t fs-type] [-T usage-type ] [-U UUID] [-jnqvDFKSV] device [blocks-count]
/opt/bin/mk-save.gtkdlg: line 116:  3203 Terminated              yad --on-top --title="Make-Save" --text="  <span size='large' foreground='dark green'><b>*** Formatting $P/$F... ***</b></span>  " --center --undecorated --no-buttons

After removing sda5:

Code: Select all

root@debian:~# mk-save.gtkdlg

CHECKBOX="true"
FILE_BROWSE_SAVEFILENAME=""
FILE_SAVEFILENAME="/live/image/changes.dat"
FS="ext2"
LB="persistence"
Size="250"
EXIT="OK"

Can't do passphrase verification on non-tty inputs.
device-mapper: remove ioctl on temporary-cryptsetup-3024 failed: Device or resource busy
device-mapper: remove ioctl on temporary-cryptsetup-3062 failed: Device or resource busy
Formatting 
mke2fs 1.42.5 (29-Jul-2012)
Filesystem label=persistence
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
Stride=0 blocks, Stripe width=0 blocks
63488 inodes, 253952 blocks
12697 blocks (5.00%) reserved for the super user
First data block=1
Maximum filesystem blocks=67371008
31 block groups
8192 blocks per group, 8192 fragments per group
2048 inodes per group
Superblock backups stored on blocks: 
        8193, 24577, 40961, 57345, 73729, 204801, 221185

Allocating group tables: done                            
Writing inode tables: done                            
Writing superblocks and filesystem accounting information: done 

/opt/bin/mk-save.gtkdlg: line 116:  3080 Terminated              yad --on-top --title="Make-Save" --text="  <span size='large' foreground='dark green'><b>*** Formatting $P/$F... ***</b></span>  " --center --undecorated --no-buttons

mkpfile.sh (mk-save.gtkdlg is link to mkpfile.sh) before removing sda5 LUKS partition:

Code: Select all

root@debian:~# mk-save.gtkdlg
Can't do passphrase verification on non-tty inputs.
device-mapper: remove ioctl on temporary-cryptsetup-3129 failed: Device or resource busy
Device crypt already exists.

After removing sda5 LUKS partition:

Code: Select all

root@debian:~# mk-save.gtkdlg
Can't do passphrase verification on non-tty inputs.
device-mapper: remove ioctl on temporary-cryptsetup-3133 failed: Device or resource busy
device-mapper: remove ioctl on temporary-cryptsetup-3171 failed: Device or resource busy
Formatting /mnt/sda1/live/changes.dat
mke2fs 1.42.5 (29-Jul-2012)
Filesystem label=
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
Stride=0 blocks, Stripe width=0 blocks
50800 inodes, 202752 blocks
10137 blocks (5.00%) reserved for the super user
First data block=1
Maximum filesystem blocks=67371008
25 block groups
8192 blocks per group, 8192 fragments per group
2032 inodes per group
Superblock backups stored on blocks: 
        8193, 24577, 40961, 57345, 73729

Allocating group tables: done                            
Writing inode tables: done                            
Creating journal (4096 blocks): done
Writing superblocks and filesystem accounting information: done 

/opt/bin/mk-save.gtkdlg: line 379:  3190 Terminated              yad --on-top --title="Make-Save" --text="  <span size='large' foreground='dark green'><b>*** Formatting $SAVEBASE... ***</b></span>  " --center --undecorated --no-buttons

Toni

[fredx181]

Hi Toni,

Solved but not sure if the news is good.

Not sure why you are not sure
I'll look at it more tomorrow.
I already found a way to create new encrypted savefile while /dev/mapper/crypt already exists (by make it use "crypt1" or "crypt2").
To avoid error when creating encrypted savefile.
Didn't think it would be that important but it seems it is.

Still can't understand why manual creating worked well for you.

Fred

[saintless]

Hi, Fred.

Still can't understand why manual creating worked well for you.

As I see the problem it is because mk-save and mkpfile.sh produce encrypted save file without formatted ext file system when I have LUKS partition. Manually creating the file and formatting /dev/loop1 produces ext formatted save file that works.

What I don't understand is how I got at least 2 times 500Mb working encrypted save file created with mk-save.gtkdlg.
I'm not sure the problem is completely solved because on the second computer I didn't have separate LUKS partition and created encrypted save file 250Mb with mk-save also didn't work.
I will remaster the main module with cryptsetup and aespipe tomorrow and test all again on both computers.

Toni

[debernardis]

Hello,

I like the idea of a puppified debian and tried it, but network connectivity was a showstopper for me immediately.

No way I could make frisbee setup a static IP connection (I'm using the ethernet lan at work, and I have no problem with that using Frisbee in Slacko).
It kept telling my IP was wrong and resetting to Auto instead that Static on eth0.

SNS only looks for DHCP, no static IP.

Any hint on how should I go on? Thanks

[saintless]

Hi, Debernardis.
Static IP needs boot code additions. Here is information what exactly has to be set in your kernel boot line and example for manually setting up static IP, Netmask, Gateway and DNS. The manually changes can be put in script started from /root/Startup after boot:
http://murga-linux.com/puppy/viewtopic. ... 470#774470

[fredx181]

Hi Toni,
I made small improvements to the make-save scripts:
- It will now create savefile even if /dev/mapper/crypt exists (using crypt1 then or even when crypt1 exists it uses crypt2 etc..).
- Added e2fsck line
Should do it the same as when you create manually.

Fred

[debernardis]

Hi, Debernardis.
Static IP needs boot code additions. Here is information what exactly has to be set in your kernel boot line and example for manually setting up static IP, Netmask, Gateway and DNS. The manually changes can be put in script started from /root/Startup after boot:
http://murga-linux.com/puppy/viewtopic. ... 470#774470

Thank you, this worked, in fact I am posting from that.

[saintless]

I made small improvements to the make-save scripts:

Thank you, Fred!
Tested after creating again sda5 LUKS partition and 200Mb save files works fine for porteus-boot and live-boot-3x.
I will do more tests in the next days. I still need proper testing on the second machine where LUKS partition was missing but I think we have good encrypted save file option for DebianDog now.

Toni

[saintless]

Hi, Fred.
Test on the second computer with mk-save problem shows encrypted save file made with new mk-save scripts works fine.
I think the problem on this computer (without separate LUKS partition) was from creating more than one save file with mk-save.gtkdlg in one session. The first one is dev/mapper/crypt and formating works. The second one fails because /dev/mapper/crypt is already created and LuksClose is not executed for crypt before creating second save.

Toni

[mcewanw]

Hi Toni and Fred,

I've been busy coding/testing the past couple of weeks.

New 0.9.5 version of pAVrecord released. Substantial under-the-hood changes and bug-fixes. Release announcement and download (dotpet for Puppy and deb package for DebianDog) via here:

http://www.murga-linux.com/puppy/viewto ... 582#656582

Cheers, William

[saintless]

a couple days ago I acheived remedial save function with DebianPup equal to the solution saintless posted for booting puppy with DD (DebianDog) initrd. This assumes booting to ram from sda3 chang appropriately for your setup or add searh for savefile function.

Mount script
#!/bin/sh
mkdir -p /media/sda3 && mount /dev/sda3 /media/sda3
mount /media/sda3/live/debian-save.2fs /initrd/pup_ro5
cp -r -p /initrd/pup_ro5/. /initrd/pup_rw/

restart X to see changes or add to init

save script
#!/bin/sh
cp -r -p /initrd/pup_rw/. /initrd/pup_ro5/

But work continues to implant full puppy options

Edit: the official fatdog64 option is to specify a kernel argument pupsave=ram:/path/to/savefile #mounts save file on /initrd/pup_rw ram layer
or pupsave=save:/path/to/savefile #mounts savefile on /initrd/pup_save - which doesn't exist outside of real FatDog64 initrd.
neither worked for me, maybe not enough testing.

Hi, Stemsee.
I have much better option for you (and for anyone who like to work on adapting puppy scripts to work on debian).
Download this archive with your 3.15-pae kernel:
http://smokey01.com/saintless/Fredx181/ ... pae.tar.gz

Content and changes:
1. initrd.gz (better compression and now support save changes in folder with latest init script from woof-ce - it was not with the original initrd.gz) Also added search for ydrv_puppy.sfs and adrv_puppy.sfs inside /live on boot. If modules with the same names present they will be loaded on boot.

2. kernel-modules.sfs - better compression only (untouched).

3. vmlinuz (untouched)

4. debdogsave - directory that will save changes on boot inside if /live is on sda1 and ext3 formatted. If you use different drive and filesystem change the lines pointing to sda1,ext3 in /live/debdogsave/etc/rc.d/PUPSTATE (I'm sure you are better than me in editing puppy configuration files).
debdogsave has proper links to /live/image and /live/cow for PUPMODE=12 so RemasterDog and RemasterCow will work proper.
Works also with save file debdogsave.2fs inside /live or anywhere if you copy the content of debdogsave inside and make the changes for path and name debdogsave.2fs.

5. The name of main module changed to 01-filesystem.squashfs.sfs (just add .sfs at the end for easy testing). All ubuntu prefix in distrospec changed to debian.

Use this boot code for sda1:

Code: Select all

title DebianDog-Puppy kernel 3.15 testing
root=(hd0,0)
kernel /live/vmlinuz pmedia=atahd psubdir=live pfix=fsck
initrd /live/initrd.gz 

At least further testing is easy now since loading 2 more sfs modules on boot is possible and save file and save folder work with manually setting up the partition in /live/debdogsave/etc/rc.d/PUPSTATE
And the most important for me - Debian structure in the main module is untouched.

EDIT: New archive for testing here. Previuous one removed:
http://www.murga-linux.com/puppy/viewto ... 468#788468

[stemsee]

I knew that my lame attempt would provoke you into providing the most formidable solution!!!

Win win!

I am going to test now. But before I do it appears that updating the kernel means changing only vmlinux and kernel-modules.sfs. So I will test with my 3.16-rc1.

Also initrd.gz can load 01-filesystem.squashfs without name change, just change puppy.sfs to 01-filesystem.squashfs in initrd.gz

[fredx181]

Hi, Fred.
Test on the second computer with mk-save problem shows encrypted save file made with new mk-save scripts works fine.
I think the problem on this computer (without separate LUKS partition) was from creating more than one save file with mk-save.gtkdlg in one session. The first one is dev/mapper/crypt and formating works. The second one fails because /dev/mapper/crypt is already created and LuksClose is not executed for crypt before creating second save.

Toni

Yes, I had the feeling already that it was because of existing /dev/mapper/crypt.
Nice that it works now!

Fred

[stemsee]

There is something wrong with that initrd.gz ... kernel panic no matter what combination I make. Are you sure you uploaded the correct one? I even copied the init to my normal initrd and it booted 01-filesystem.squashfs ... maybe the repack? Did you test this actual combo on your laptop?

[mcewanw]

Toni and Fred,

Sorry, I couldn't resist adding something extra to pAVrecord, so have re-uploaded the latest version, which is now version 0.9.6. I'll leave that for user testing now. Check here for details:

http://www.murga-linux.com/puppy/viewto ... 706#785706

Cheers, William

[saintless]

There is something wrong with that initrd.gz ... kernel panic no matter what combination I make. Are you sure you uploaded the correct one? I even copied the init to my normal initrd and it booted 01-filesystem.squashfs ... maybe the repack? Did you test this actual combo on your laptop?

Yes, I did and tested again now. See the picture:


Replacing init is not enough. You need also to change distrospecs.
If you are using different kernel make sure it supports xz decompression. initrd.gz is xz compressed now.
One more thing to try: change pfix=fsck to pfix=nocopy
Copy2Ram takes much time on my machine and maybe this creates problem on some hardware. I think puppy is setup by default copy2ram if more than 256Mb Ram is available.

You can try also the nopae module. It works on one more desktop PC for me. initrd.gz is the same but packed separately (this why the md5sum is different).
http://smokey01.com/saintless/Fredx181/ ... pae.tar.gz
Try initrd.gz from nopae archive to test but both work for me.