[solved] Puppy ok on CD, DOA on HD

[Alex Johnson]

Be gentle. I'm new.

I tried Puppy 4.00 on a LiveCD and I really liked it. It was pretty good and it got my network up with no fuss. It looked okay and best of all it was very small and fast. I installed it on my HDD as a full ext3 partition and it seemed to run great but I couldn't get sound working and a few other nuisances so I tried out a couple other distros. I replaced my Puppy partition with Suse and used that for a while, but I really liked Puppy better so I went back and installed Puppy to a second partition. Now the problems start.

From day 1 Puppy had no trouble with my hardware. It detected my network adapter (RealTek RTL-8169) and my onboard audio (intel-hda). I just couldn't hear anything with all the volumes set to maximum. When I reinstalled and boot from HDD now Puppy can't detect either piece of hardware and when I try to load the drivers manually it prevents me saying that hardware isn't connected. It also won't mount my FAT32 "shared" partition and says there is an error in the disk. But I can boot from the CD just fine and when I do it always detects by network and audio. And when booted from CD I can mount my FAT32 partition easily and read and write files on it.

I've reformatted the Puppy ext3 partition and reinstalled 4 times. I always have the same symptoms now. CD boot is functional but HD boot is unable to use any of my hardware. Yet when I booted from HD on my very first installation it was just fine.

Please someone help me! Puppy 4.00 is my favorite distro I've played with so far but I can't live with booting off a CD all the time!

Hardware details below....
Network: Module r8169, Bus pci, Description RealTek RTL-8169 Gigabit Ethernet
Audio: hda-intel 00:1b.0 Class 0403: 8086:293e

This is just a regular old Gigabyte GA-P35-S3G with a GeForce 7300GT video, Core 2 Quad, and 2GB RAM (yeah, I don't have to have a small linux like Puppy, but I really like it). There is a 120GB Seagate IDE drive partitioned into 4 parts (Puppy ext3, Suse ext3, "shared" FAT32, and unused FAT32).

[muggins]

Hi Alex,

how have you installed puppy to hard disk, frugal or full install? If the latter, then it doesn't access the zdrv_400.sfs file, which contains a lot of the drivers.

If you do have a full install, just boot your pupCD in RAM, (by entering puppy pfix=ram in first few secs when you boot the CD),
then mount the partition where you've installed puppy to. Assuming it's partition /dev/hda2, then do:

Code: Select all

cp -r /lib/* /mnt/hda2/lib

(If not hda2, change to whatever puppy's partition number is)

[Alex Johnson]

muggins, I did a full install (I thought it was clear by saying I devoted a full ext3 partition to it) but all the drivers are available. The problem is that Puppy says I have no hardware to detect. I try the connect wizard and it says no network hardware is detected on the machine. I tell it to load the r8169 driver and it refuses saying the hardware that driver is for is not present in the system. I also use the Alsa autodetect and it says no hardware of any kind is present, although Alsa finds it fine from CD. But just to be sure I followed your instructions and rebooted from HD. It still couldn't find any of my hardware and though the filesystem of my FAT32 partition was corrupt and wouldn't mount it.

[hillside]

After you have installed to the hard drive, if you boot the cd, does the cd still work properly? If so, I would consider doing a frugal install. Frugal operates almost exactly like running from cd. Puppy was not designed to operate from a full install. It can often work that way and there are quite a number of people who use it that way, but you may be unlucky enough that your system doesn't support full install well.

Even though frugal seems crazy if you're not used to the idea, it works really, really well. If your system has adequate memory, I would say a minimum of about 256 megs, frugal is almost bulletproof and always purrs like a kitten.

[alienjeff]

Puppy was not designed to operate from a full install.

That is misinformation.

It can often work that way and there are quite a number of people who use it that way, but you may be unlucky enough that your system doesn't support full install well.

Please provide an example of a system that "doesn't support full install well."

[Mr.J.S.]

Do 1 thing - FORmat Your ext3 partition into ext 2 partition and also make a 2 GB linux swap partition. After that, make a frugal install

I do not have any basis of saying this But I can say that Hit and Trial method may work. It might help you also (do back up your files before)

[alienjeff]

...and also make a 2 GB linux swap partition.

Alex Johnson has 2G of RAM. Suggesting a 2G swap partition is abject overkill.

Mr.J.S.

Joined: Today
Posts: 11

PostPosted: Today, at 11:18 am

"I do not have any basis of saying this..."

I'm biting my tongue.

[Alex Johnson]

Please, I'm interested in making Puppy my full-time OS. Frugal seems like what one would use to either 1) to share a computer with another OS that isn't friendly (we all know who that would be) or 2) to be frugal where resources are scarce. I don't see a reason to do this because 1) I will not have any form of Windows on this box and 2) my box is obviously well beyond the requirements of Puppy. If in setting up Puppy to work from an image they somehow broke native operation, I'd consider it, but I don't believe they broke anything...full install worked perfectly for me for a week before I wiped it and tried Suse next. I wanted to go back to Puppy but it wouldn't work again. I can't think of an explanation for that behavior...works initially, then never again.

The good news is that in the mean time my computer is still usable in Suse and from the Puppy CD. But I want to get the Puppy installed natively. Playing with CDs is only good so long and Suse works but I still prefer Puppy.

[Sit Heel Speak]

Let's take it from the beginning. Questions:

1. When you attempt to boot your full install, do you in fact come to the Puppy desktop?

2. Boot from the live-CD, open an rxvt terminal window. Issue the command

uname -a

and paste here the output (I want to know if he's using the sd* 2.6.25 or the hd* 2.6.21.7 version of Puppy 4.00).

3. Is the Seagate connected to the mainboard's own onboard IDE controller, to the onboard SATA controller, or to an add-in controller card?

4. What channels are the CD and the Seagate on, master or slave, are you certain they are jumpered correctly?

5. When you partitioned the first (ext3 Puppy) partition, did you click "Manage flags" and tick "boot"?

6. After you install Puppy to the first partition and Grub to the root superblock, let us see what /boot/grub/menu.lst looks like. I just need to see down through the section you are booting Puppy from.

HTH, SHS

[Alex Johnson]

Let's take it from the beginning.

Alright, that's what I expected.

1. When you attempt to boot your full install, do you in fact come to the Puppy desktop?
Yes.

2. Boot from the live-CD, open an rxvt terminal window. Issue the command

uname -a

and paste here the output (I want to know if he's using the sd* 2.6.25 or the hd* 2.6.21.7 version of Puppy 4.00).
When booting from HD: Linux puppypc 2.6.25.5-1.1-default #1 SMP
When booting from CD: Linux puppypc 2.6.21.7 #1
I don't understand how they are different since the install came from the CD. However, the answer may be in the /boot/grub/menu.lst since I see Suse is using 2.6.25.5-1.1

3. Is the Seagate connected to the mainboard's own onboard IDE controller, to the onboard SATA controller, or to an add-in controller card?
mainboard's PATA cable as master via cable select, slave channel not connected

4. What channels are the CD and the Seagate on, master or slave, are you certain they are jumpered correctly?
CD in SATA-0 as /dev/sr0, Seagate in PATA-master as /dev/sda

5. When you partitioned the first (ext3 Puppy) partition, did you click "Manage flags" and tick "boot"?
No. I have a longer explanation below.
During the initial partitioning I didn't select "boot" for any partition, assuming the distro would set that during install. Puppy was the first install and it went on /dev/sda4. When I later installed Suse to /dev/hda4, YaST marked the partition as "boot". I later tried to install Puppy to /dev/sda3 to keep both installs alive until I had the sound problem with Puppy worked out (recall I couldn't hear anything on Puppy even though it recognized my intel-hda audio). When I reinstalled to /dev/sda3 "boot" was ticked for /dev/sda4.

6. After you install Puppy to the first partition and Grub to the root superblock, let us see what /boot/grub/menu.lst looks like. I just need to see down through the section you are booting Puppy from.
Copied below.
# Modified by YaST2. Last modification on Mon Jun 30 16:07:49 CDT 2008
default 0
timeout 8
##YaST - generic_mbr
gfxmenu (hd0,3)/boot/message
##YaST - activate

###Don't change this comment - YaST2 identifier: Original name: linux###
title openSUSE 11.0
root (hd0,3)
kernel /boot/vmlinuz-2.6.25.5-1.1-default root=/dev/disk/by-id/scsi-SATA_ST3120814A_5LS06RTV-part4 splash=silent showopts vga=0x31a
initrd /boot/initrd-2.6.25.5-1.1-default

###Don't change this comment - YaST2 identifier: Original name: failsafe###
title Failsafe -- openSUSE 11.0
root (hd0,3)
kernel /boot/vmlinuz-2.6.25.5-1.1-default root=/dev/disk/by-id/scsi-SATA_ST3120814A_5LS06RTV-part4 showopts ide=nodma apm=off acpi=off noresume nosmp noapic maxcpus=0 edd=off x11failsafe vga=0x31a
initrd /boot/initrd-2.6.25.5-1.1-default

title Puppy Linux 4.00
root (hd0,3)
kernel /boot/vmlinuz root=/dev/sda3
# kernel /boot/vmlinuz root=/dev/sda3 splash=silent showopts vga=0x31a
# pmedia=idehd
initrd /boot/initrd

(Note: I've changed the Puppy section several times to try combinations of the options tacked on by Suse and Puppy.)

[hillside]

Puppy was not designed to operate from a full install.

That is misinformation.

I stand corrected. The last thing I want to do is spread misinformation.

Alex,
It looks like someone stepped up to help. Good luck.

[HairyWill]

Frugal seems like what one would use to either 1) to share a computer with another OS that isn't friendly (we all know who that would be) or 2) to be frugal where resources are scarce.

It is a shame it is called frugal as that implies second class.
Providing you have sufficient ram (256MB ish for puppy 4.00) a frugal install is not inferior to a full install. They have different strengths and weaknesses. There are long term puppy users in both camps. My personal preference is for frugal.

If you are short on RAM than it is generally accepted that a full install is preferable.

This page on the wiki discusses the difference.
http://puppylinux.org/wiki/how-tos/gene ... ullinstall

[rcrsn51]

Your menu.lst entry for Puppy is scrambled. If Puppy is actually installed in sda3, the entry should be

title Puppy
root (hd0,2)
kernel /boot/vmlinuz root=/dev/sda3 ro vga=normal

Note the use of (hd0,2). Also, a full install does not need the initrd line. It is only required by a frugal install.

Someone can correct me if I'm wrong, but I believe that in v4.0, a PATA drive is still identified as hda, not sda.

[Alex Johnson]

Your menu.lst entry for Puppy is scrambled. If Puppy is actually installed in sda3, the entry should be

title Puppy
root (hd0,2)
kernel /boot/vmlinuz root=/dev/sda3 ro vga=normal

Note the use of (hd0,2). Also, a full install does not need the initrd line. It is only required by a frugal install.

Someone can correct me if I'm wrong, but I believe that in v4.0, a PATA drive is still identified as hda, not sda.

Along with many other things, I tried that before posting here. When using (hd0,2) it fails during grub with File not found and returns me to the grub menu to select a new boot entry. I think this is because the first installation on sda4 made that the primary boot partition, but now I'm installing Puppy in sda3. Since grub is in sda4, root needs to be (hd0,3).

As for hda vs sda, since I've been using 4.00 (Dingo) via the CD and that calls my partitions /dev/sda1-4, I'm quite confident sda is the correct nomenclature in 4.00.

[rcrsn51]

The "root" command sets the drive path for the GRUB commands that follow it.

Boot off the Live CD and mount sda3. Can you find the file /boot/vmlinuz?

While you are in the /boot folder of sda3, make a dummy file named "marker". Just go to a console and type the command: touch marker

Verify that you now have the file /mnt/sda3/boot/marker.

Reboot off your hard drive. When the GRUB menu comes up, press the 'c' key to get the GRUB command prompt. Type the command: find /boot/marker

GRUB will search for the file that you just created. Where does it find it?

[Sit Heel Speak]

Hi Alex,

A trick you might be stumbling over: the two versions of Puppy-4.00 final, do not designate your disks the same way. Not our fault; Messrs. Torvalds, Molnar, Morton et al changed the way the kernel calls disks, beginning at about kernel 2.6.18.9. That kernel and newer (including all Linuxes which use a 2.6.25 kernel, such as your Suse) will usually call that disk sda. Kernel 2.6.18.8 and older will usually call it hda. However, Barry seems to have stayed with the old convention in the k2.6.21.7 Puppy 4 (your live-CD); see here for what the assignments look like on my own install of that Puppy (the Quantum Atlas'es labelled sd* are true SCSI, the top two are ATA).

Therefore, I would try the following section to start Puppy:

Code: Select all

title Puppy 4.00 on first disk, third partition
root (hd0,2)
kernel /boot/vmlinuz real_root=/dev/hda3 ro vga=normal

or, better yet, for "belt-and-suspender safety" use

Code: Select all

title Puppy 4.00 on first disk, third partition
root (hd0,2)
kernel /boot/vmlinuz real_root=/dev/hda3 ro vga=normal irqpoll noacpi

and see if that does it fer ya.

[Alex Johnson]

Boot off the Live CD and mount sda3. Can you find the file /boot/vmlinuz?

Yes

Reboot off your hard drive. When the GRUB menu comes up, press the 'c' key to get the GRUB command prompt. Type the command: find /boot/marker

GRUB will search for the file that you just created. Where does it find it?

Failed I was unable to do this. I created the file but the slick version of grub I had did not have a command prompt option. Whatever I typed was appended to the selected boot option.

Therefore, I would try the following section to start Puppy:

Failed I attempted this but grub died quickly with these messages:
VFS: Cannot open root device "<NULL>" or unknown-block(8,9)
Please append a correct "root=" boot option

[rcrsn51]

The "find" command is your best friend for solving GRUB problems. If you look at the top of your menu.lst, you will see an entry for "gfxmenu". Comment out this line by putting a # in front.

Then reboot and see if you can get to the GRUB command prompt.

[Sit Heel Speak]

(disregard, I need to think about this)

[Sit Heel Speak]

I suggest you try the other version of Puppy 4.00, the build with the 2.6.25 kernel. It's at http://distro.ibiblio.org/pub/linux/dis ... inux/test/.

I notice, looking through the kernel compile-time config files, that CONFIG_PATA_JMICRON is not set in the 2.6.21.7 Puppy kernel (i.e. the JMicron PATA controller is too new for the 2.6.21.7 kernel to have had support for it), but it is set=y in the 2.6.25 Puppy kernel. According to this, that mainboard carries a JMicron 368 IDE chip, which I suspect is your PATA controller.

Earlier successes might have owed to your putting Puppy on a lower area of the disk, at least that's my initial guess. Speedstep and assorted audio device drivers were compiled as modules, not into the kernel directly, in 2.6.21.7, and this may at least partially explain your other earlier woes.

The 2.6.25 kernel is quite new, quite rough-edged, so I suggest you try the 4.1-alpha-3 as well, and be prepared to report a lot of glitches to help Barry and the rest of us straighten them out...

HTH, SHS