no 5.5 boot with syslinux 5.01

[Peterm321]

Pausing for 60 seconds
Loading drivers needed to access disk drives. done
Searching for Puppy files ...... puppy_precice_5.5.sfs not found.
Dropping out to initial-ramdisk console ...

I am presuming at this point is that "precice" as spelled above is a typo in the message above not in any of the configuration files.

I once had an issue with unetbootin not changing "pmedia=cd" to "pmedia=usbflash" and by the way if you use a USB HD it may want pmedia=usbhd

Pressing F2 and F3 at boot gives the options and advanced options, you may want to try

puppy pupsfs=sdc1:/puppy_precise_5.5.sfs

at the boot: prompt, where sdc1: is the device suggested in the F3 help. (it may be sda1 or sda2 on your system)

[wiejetzt]

I have 2 mp3 players : One boots fine, the other says farewell to the BIOS after some seconds .

As a workaround i would first try to

Code: Select all

cd /
/init

again
anf if that fails

Code: Select all

mount -t proc none /proc
mount -t sysfs none /sys
cd /
pdev1=sdb1 psubdir=XXY.d ./init

which i use when i boot with the wrong pdev1=sdbX or psubdir=XXX.d parameter.

I tested both methods and both stopped at the same point:

Loading 'de' keyboard layout ... done
Loading drivers needed to access disk drives. done
Searching for Puppy files... done
Loading the 'puppy_precise_5.5.sfs' main file... copying to ram.. done
Setting up the layered filesystem ... done
Performing a 'switch_root' to the layered filesystem...# .

[Karl Godt]

yeah, /sbin/switch looks like this :

#!/bin/sh
#if have dropped to a shell in initramfs, type 'exec switch' to keep going.
sync
umount /proc/bus/usb
umount /sys
umount /proc
#now using cpio archive for initramfs 'initial ramdisk'...
exec switch_root /pup_new /sbin/init

Dunno for now whats happen to /sbin/init inside the precise-5.5.sfs .

It is mounted it seems .

Now look at the output of

Code: Select all

mount
df

commands . Maybe needs a

Code: Select all

cd /initrd/pup_rw
./sbin/init

Will activate my testbox to reproduce it. It's nice weather here Can take some hours ..
echo "Puppy is out for a walk .. "

[wiejetzt]

Pausing for 60 seconds
Loading drivers needed to access disk drives. done
Searching for Puppy files ...... puppy_precice_5.5.sfs not found.
Dropping out to initial-ramdisk console ...

I am presuming at this point is that "precice" as spelled above is a typo in the message above not in any of the configuration files.

I once had an issue with unetbootin not changing "pmedia=cd" to "pmedia=usbflash" and by the way if you use a USB HD it may want pmedia=usbhd

Pressing F2 and F3 at boot gives the options and advanced options, you may want to try

puppy pupsfs=sdc1:/puppy_precise_5.5.sfs

at the boot: prompt, where sdc1: is the device suggested in the F3 help. (it may be sda1 or sda2 on your system)

Hi, Peterm321!
Yes, "precice" is a typing error, I do not know a way to cut and paste at this stage.

I tried in /syslinux/extlinux.conf:

Code: Select all

label precise-pupsfs
  kernel /LINUXes/precise/vmlinuz 
  append initrd=/LINUXes/precise/initrd.gz pupsfs=sda1:/LINUXes/precise/puppy_precise_5.5.sfs pkeys=de

It produced:

Code: Select all

Loading 'de' keyboard layout ...   done
Loading drivers needed to access disk drives.   done
Searching for Puppy files...   done
Loading the 'puppy_precise_5.5.sfs' main file... copying to ram..   done
Setting up the layered filesystem ...   done
Performing a 'switch_root' to the layered filesystem...
[ 9.542677] Kernel panic - not syncing: Attempted to kill init!
[ 9.542752] Pid 1, comm: switch_root Not tainted 3.2.29 #1
[ 9.542783] Call Trace:
[...] [...] ? ...
[...] [...] ? ...
[...] [...] ? ...
[...] [...] ? ...

[wiejetzt]

Now look at the output of

Code: Select all

mount
df

commands . Maybe needs a

Code: Select all

cd /initrd/pup_rw
./sbin/init

Will activate my testbox to reproduce it. It's nice weather here

Code: Select all

#pwd
/
#mount
mount: no /proc/mounts
#df
Filesystem           1K-blocks      Used Available Use% Mounted on
df: /proc/mounts: No such file or directory
#cd /initrd/pup_rw
/bin/sh: cd: can't cd to /initrd/pup_rw
#cd /initrd
/bin/sh: /initrd not found

nice weather - wow

[Karl Godt]

How to mount proc and sys I showed you some posts back .

Have now fired it up :

Code: Select all

cd /pup_new
chroot .
/sbin/init

did not work for me, busybox init gave a message new to me, but not about need to be nr 1 of all processes.

Rebooted by Cntrol+Alt+Del and that time I just

Code: Select all

exec switch

and viola, posting from it now.

There are probably more possibilities, but it seems that "exec" is needed to leave the initial ramdisk .

The kernel line to simulate that rdsh ramdiskshell dropout :

Code: Select all

cat /proc/cmdline
initrd=initrd.gz pmedia=usbflash pfix=ram debug panic=60 pkeys=de BOOT_IMAGE=vmlinuz pdev1=sdNONEXISTANT psubdir=NONEXISTANT

default puppy
display boot.msg
prompt 1
timeout 150
F1 boot.msg
F2 help.msg
label puppy
kernel vmlinuz
append initrd=initrd.gz pmedia=usbflash pfix=ram debug panic=60 pkeys=de

(# extlinux --version
extlinux 3.73 Copyright 1994-2009 H. Peter Anvin)

[Peterm321]

Loading the 'puppy_precise_5.5.sfs' main file... copying to ram.. done
Setting up the layered filesystem ... done
Performing a 'switch_root' to the layered filesystem...
[ 9.542677] Kernel panic - not syncing: Attempted to kill init!
[ 9.542752] Pid 1, comm: switch_root Not tainted 3.2.29 #1
[ 9.542783] Call Trace:
....

(?) From my past experience from a while back so my memory is faded, is that the boot process is still not happy with where exactly it considers the boot partition should be. It is now loading the sfs file OK, it has copied it to ram and it is possible that it is mounting the extracted filesystem to an irregular mountpoint, that is to say it is creating a mountpoint based on the pmedia setting and trying to mount the extracted filesystem from the precise sfs file to that mountpoint? I'm guessing of course.

The vmlinuz file is really a cpio xz compressed archive which can be extracted and looked at to see how the boot process works, but myself I wasnt keen on the idea unless there was no other way.

What I can remember is that when I fixed the config to pmedia=usbflash the kernel panics were history.

By the way I know little about this /syslinux/extlinux.conf file. Maybe it represents another boot installation utility?

When I used undetbootin (on Windows) and bootflash(Native Linux), setting up the USB drive as bootable involved copying the sysinux bin file but also the files

isolinux.cfg
and
syslinux.cfg

I think unetbootin is using the syslinux.cfg file but I amended both files to contain pmedia=usbflash and no problems since. There is no extlinux.conf file in sight.

A tip which might be useful is that you dont have to re-run unetbootin / bootflash etc if you want to change puppy linux version that you might wish to test (that is without relying on a savefile or using a frugal or full install). For example when I changed from Slacko 5.4 to Raring 3.8 I deleted the files vmlinuz, initrd.gz and puppy_slacko...sfs from the USB drive and extracted the equivalents from the Raring ISO and copied them over. This helps avoid having to re-edit the config files when changing versions.

[wiejetzt]

How to mount proc and sys I showed you some posts back .

Have now fired it up :

Code: Select all

cd /pup_new
chroot .
/sbin/init

did not work for me, busybox init gave a message new to me, but not about need to be nr 1 of all processes.

Rebooted by Cntrol+Alt+Del and that time I just

Code: Select all

exec switch

and viola, posting from it now.

There are probably more possibilities, but it seems that "exec" is needed to leave the initial ramdisk .

The kernel line to simulate that rdsh ramdiskshell dropout :

Code: Select all

cat /proc/cmdline
initrd=initrd.gz pmedia=usbflash pfix=ram debug panic=60 pkeys=de BOOT_IMAGE=vmlinuz pdev1=sdNONEXISTANT psubdir=NONEXISTANT

default puppy
display boot.msg
prompt 1
timeout 150
F1 boot.msg
F2 help.msg
label puppy
kernel vmlinuz
append initrd=initrd.gz pmedia=usbflash pfix=ram debug panic=60 pkeys=de

(# extlinux --version
extlinux 3.73 Copyright 1994-2009 H. Peter Anvin)

#mount -t proc none /proc
#mount -t sysfs none /sys
#cd /pup_new
#chroot .
chroot: can't execute '/bin/sh': No such file or directory
#ls -la /bin/sh
lrwxrwxrwx 1 0 0 7 Mar 10 10:13 /bin/sh -> busybox

Shortening extlinux.conf to "default vmlinuz initrd=initrd.gz pkeys=de" does not help, neither do pmedia, pdev1, pfix, debug, panic.

BTW: The aforementioned Dumping lines appear, when I omit pkeys=de:
Dumping last lines of /tmp/bootinit.log ...
ls: /lib/keymaps//L*.gz: No such file or directory
Dumping last lines of kernel log ...

BTW2: When precise 5.5 successfully boots with syslinux 4.05, the logging lines are displayed with the usual font.
But when failing with syslinux 5.01, a font larger and thinner ist used.

BTW3: quirky-5.4.91, Fedora-17-i686-Live-LXDE, archbang-03.03.2013-i686, boot with syslinux 5.01.
I already mentioned this for KNOPPIX_V7.0.5CD-2012-12-21-DE, Porteus-XFCE-v2.0-i486, SalineOS-1.6-i386,
xubuntu-12.04.1-desktop-i386, pmagic_2013_02_28, systemrescuecd-x86-3.4.1:

Code: Select all

default vesamenu.c32
timeout 80
menu background #00000000
menu label Welcome
menu color border 0 #00eeeeee #00000000
menu color sel 7 #ffffffff #33eeeeee
menu color label 0 #ffeeeeee #00000000
menu color tabmsg 0 #ffeeeeee #00000000
menu color unsel 0 #ffeeeeee #00000000
menu color hotsel 0 #ff000000 #ffffffff
menu color hotkey 7 #ffffffff #ff000000
menu color timeout_msg 0 #ffffffff #00000000
menu color timeout 0 #ffffffff #00000000
menu color cmdline 0 #ffffffff #00000000

label archbang
  kernel /LINUXes/archbang/vmlinuz
  append initrd=/LINUXes/archbang/archiso.img archisobasedir=LINUXes/archbang archisolabel=test 
label fedora
  kernel /LiveOS/vmlinuz0
  append initrd=/LiveOS/initrd0.img root=LABEL=test rootfstype=auto rw liveimg
label quirky
  kernel /LINUXes/quirky/kernel.qky 
  append psubdir=/LINUXes/quirky pkeys=de 

label precise
  MENU DEFAULT
  kernel /LINUXes/precise/vmlinuz 
  append initrd=/LINUXes/precise/initrd.gz psubdir=/LINUXes/precise pkeys=de 
label slacko
  kernel /LINUXes/slacko/vmlinuz 
  append initrd=/LINUXes/slacko/initrd.gz psubdir=/LINUXes/slacko pkeys=de 

label knoppix
  kernel /KNOPPIX/linux
  append initrd=/KNOPPIX/minirt.gz noimage noddc nocomposite vga=normal nodrm nomodeset no3d tz=localtime lang=de  
label porteus
  kernel /LINUXes/porteus/vmlinuz 
  append initrd=/LINUXes/porteus/initrd.xz login=root kmap=de autoexec=xf-compositing~off from=/LINUXes/porteus/Porteus-XFCE-v2.0-i486.iso  
label salineos
  kernel /LINUXes/salineos/vmlinuz
  append initrd=/LINUXes/salineos/initrd.img live-media-path=/LINUXes/salineos/ boot=live config noprompt 
label xubuntu
  kernel /LINUXes/xubuntu/vmlinuz
  append initrd=/LINUXes/xubuntu/initrd.lz boot=casper iso-scan/filename=/LINUXes/xubuntu/xubuntu-12.04.1-desktop-i386.iso console-setup/ask_detect=false console-setup/layoutcode=de noprompt
label pmagic
  kernel /LINUXes/pmagic/bzImage 
  append initrd=/LINUXes/pmagic/initrd.img iso_location=/LINUXes/pmagic edd=off load_ramdisk=1 prompt_ramdisk=0 rw vga=normal loglevel=9 max_loop=256 vmalloc=384MiB keymap=de consoleboot 
label sysrescd
  kernel /LINUXes/sysrescd/rescue32
  append initrd=/LINUXes/sysrescd/initram.igz setkmap=de isoloop=/LINUXes/sysrescd/systemrescuecd-x86-3.4.1.iso

label wary
  kernel /LINUXes/wary/vmlinuz 
  append initrd=/LINUXes/wary/initrd.gz psubdir=/LINUXes/wary pkeys=de 
label racy
  kernel /LINUXes/racy/vmlinuz 
  append initrd=/LINUXes/racy/initrd.gz psubdir=/LINUXes/racy pkeys=de 

[wiejetzt]

Hi, Peterm321!
I experienced, that neither puppylinux nor knoppix nor ubuntu can be relied on 100% to boot under all circumstances.
(Some Fujitsu-Siemens monitors are too special, for example.)
Thus I have to have several distros at hand to be able to work at different machines (recovery).
And I cannot rely on grub4dos reinstallation 100%, because puppylinux boot is not failsafe.
Syslinux can be installed distro-independent.

Maybe we have to wait for a syslinux > 5.01.
But obviously puppy 5.5 relies on the bootloader, because there is no problem with syslinux 4.05.
And as far as I can see, just puppy 5.5 does not work with syslinux 5.01.

[Karl Godt]

ls: /lib/keymaps//L*.gz: No such file or directory

Somehow I start to suspect that the .iso download is slightly to short or that the initrd.gz wasn't copied right onto the usb or onto some badblocks, thus corrupting the initrd.gz .

chroot: can't execute '/bin/sh': No such file or directory

Current woof2 has BusyBox v1.21.0 for initrd.gz , while it had been for a long time v1.4.2 like i had and some may have v1.17 and v1.19 . Busybox changes a lot . Will test all these when I have time .

[wiejetzt]

ls: /lib/keymaps//L*.gz: No such file or directory

Somehow I start to suspect that the .iso download is slightly to short or that the initrd.gz wasn't copied right onto the usb or onto some badblocks, thus corrupting the initrd.gz .

Hi, Karl!
I tested on a 16GB-Stick, then a 4GB-SD-Card. After that I downloaded the iso again and did a md5sum check. After copying files out of an iso, I sync before closing the iso. Then I tested on /dev/sda1 internal harddisk. *sigh*

[Karl Godt]

wattnu, dat is jo underlich ..

Second possibility :

pdev=sdb3 psubdir=M529 ./init
mount /proc
mount /sys

Code: Select all

cd /pup_new
chroot .
/etc/rc.d/rc.sysinit
login root

<ramram|backback>

and login ran /etc/profile which launched X .

# cat /proc/cmdline
video=640x480 pdev1=sdNONE psubdir=NOTEXISTENT
# extlinux --version
extlinux 3.73 Copyright 1994-2009 H. Peter Anvin
# . /etc/DISTRO_SPECS
# echo $DISTRO_NAME $DISTRO_VERSION
macpup 529
# uname -r
3.2.18
# ps|head -n4
PID USER TIME COMMAND
1 root 0:01 /bin/sh
2 root 0:00 [kthreadd]
3 root 0:00 [ksoftirqd/0]

initrd.gz : BusyBox v1.20.0.git
.sfs : BusyBox v1.19.3

*

Now had difficulties to load the correct ip drivers to be able to post here .
Had "Couldn't set tty to PPP discipline: Invalid argument" by pppd

sh-3.00# modprobe -v ppp_async
insmod /lib/modules/3.2.18/kernel/lib/crc-ccitt.ko
insmod /lib/modules/3.2.18/kernel/drivers/net/ppp/ppp_async.ko
sh-3.00# modprobe -v ppp_deflate
insmod /lib/modules/3.2.18/initrd/kernel/lib/zlib_deflate/zlib_deflate.ko
insmod /lib/modules/3.2.18/kernel/drivers/net/ppp/ppp_deflate.ko
sh-3.00# modprobe -v ppp_synctty
insmod /lib/modules/3.2.18/kernel/drivers/net/ppp/ppp_synctty.ko

But the ppp_synctty driver might have done it .

Its fun to sail in a nutshell Would be more fun with a full static busybox

[wiejetzt]

Now I took a USB stick, gparted new partition table, new ext2, boot flag set.
Download of slacko-5.5-4G.iso, md5sum OK. Same old story:
slacko boots with syslinux 4.05 on 2nd USB stick, but does not with syslinux 5.01.
Same result of "chroot ."

But one slight difference: "Dropping out to initial-ramdisk console ..."
the 1st time is in larger and thinner font as mentioned,
but 2nd time after "pdev=sdb1 psubdir=/LINUXes/slacko ./init" the font is as usual.

After "mkdir /mnt/sdb1", "mount -t ext2 /dev/sdb1 /mnt/sdb1"
and "pdev=sdb1 psubdir=/LINUXes/slacko ./init" the sfs file was successfully found.
At "Performing a 'switch_root'" it stopped.

[Karl Godt]

Have now Dl

http://ftp.nluug.nl/ftp/pub/os/Linux/di ... 5.5-4G.iso[/b]

which has
# extlinux -v
extlinux 3.73 Copyright 1994-2009 H. Peter Anvin

AND

http://ftp.nluug.nl/ftp/pub/os/Linux/di ... se-5.5.iso[/b]

which has
# /mnt/+mnt++root+precise-5.5.iso+puppy_precise_5.5.sfs/sbin/extlinux -v
extlinux 3.73 Copyright 1994-2009 H. Peter Anvin

I really don't understand why you use some other quirky linux distro's extlinux to confuse everyone here on the forum .
You should post a BUG REPORT on their's forum then !

[Peterm321]

Now I took a USB stick, gparted new partition table, new ext2, boot flag set.
Download of slacko-5.5-4G.iso, md5sum OK. Same old story:

Hi wiejetzt

Re: the above (and Karls post Today, at 08:24)

I once tried the 4G version of Slacko on my Netbook and I recall the hardware not liking it a great deal. I switched to Raring (good non PAE Kernel - prefer non PAE) it works well with unetbootin. You may want to consider raring and/or another version of Slacko and/or getting rid extlinux, requesting help in an extlinux based thread etc etc.

[rcrsn51]

What wiejetzt has failed to mention in all of this discussion is where he is getting syslinux 5.01 and why he is using it to boot Puppy. As Karl stated above, all recent woof-built Puppies have syslinux 3.73, which works fine.

As a test, I compiled syslinux 5.01 in Precise. Sure enough, it had exactly the problems that wiejetzt described. The same was true with 5.00. However, 4.06 worked.

So it appears that syslinux 5.x introduced something that is incompatible with how Puppy builds its kernel/initrd.gz.

[Karl Godt]

I can only suspect that extlinux v5 is passing the boot parameters corrupted furhther .

Code: Select all

echo "pmedia='$pmedia' PMEDIA='$PMEDIA' psubdir='$psubdir' PSUBDIR='$PSUBDIR' pdev1='$pdev1' PDEV1='$PDEV1'"

In the initrd shell would probably reveal things .

[Karl Godt]

sh-3.00# ls -1 /root/my-documents/syslinux-3.86/extlinux/

ext2_fs.h
extlinux
main.c
Makefile

sh-3.00# ls -1 /root/my-documents/syslinux-4.06/extlinux/

btrfs.h
extlinux
fat.h
main.c
Makefile
mountinfo.c
mountinfo.h
ntfs.h

sh-3.00# ls -1 /root/my-documents/syslinux-5.01/extlinux/

btrfs.h
extlinux
fat.h
main.c
Makefile
misc.h
mountinfo.c
mountinfo.h
ntfs.h
xfs_fs.h
xfs.h
xfs_sb.h
xfs_types.h

[wiejetzt]

What wiejetzt has failed to mention in all of this discussion is where he is getting syslinux 5.01 and why he is using it to boot Puppy. As Karl stated above, all recent woof-built Puppies have syslinux 3.73, which works fine.

As a test, I compiled syslinux 5.01 in Precise. Sure enough, it had exactly the problems that wiejetzt described. The same was true with 5.00. However, 4.06 worked.

So it appears that syslinux 5.x introduced something that is incompatible with how Puppy builds its kernel/initrd.gz.

Got syslinux 5.01 from syslinux.org, included are binaries, enabling to install immediately.
That syslinux 5.x introduced something is one conclusion, another conclusion is that other distros have no problem with syslinux 5.01, only puppy 5.5.

I am sorry that caused so much trouble. I thought it is helpful to point out bugs. Obviously I have to be VERY careful, to report something that might be a bug, but may be not. Delete my account.

[01micko]

I am sorry that caused so much trouble. I thought it is helpful to point out bugs. Obviously I have to be VERY careful, to report something that might be a bug, but may be not. Delete my account.

I haven't got lot of time lately but this thread is on my radar.

From the start I understood that your problem was with syslinux >= 5.0x and any recent puppy linux.

It is a bug and should be dealt with. Don't pack up your bat and ball and go home just yet!