Classic Pup 2.14X -- Updated 2 series

[MinHundHettePerro]

XMMS:
Have the same problem as davesurrey just posted above . Haven't looked for the libs yet, though, probably have them lying around somewhere .

MUT:
hda1 - hda15 are present in /dev but, only hdb1 -hdb9. So the five extra partitions of OlddogNewtricks's will be shown, while hdb10 and hdb11 of davesurrey's will not. Myself, I was missing sdc9, only sdc1 - sdc8 in /dev - so I made a /dev/sdc9 and voilà: sdc9 .

cheers/
MHHP

[BarryK]

Well just finished compiling gcc again, lol this time it went to 2074MB working directory, lol and the actual gcc was 209MB extracted or 64MB pet, I think I had too many languages, like java etc, I'll try again this time with less.
What an experience, lol, wish I knew what Barry usually puts in and takes out.
ttuuxxx

I think that I might have documented some of it, back in late 2007, in my daily news or the forum.
Also, the T2 build system is a good place to look. They have configure options that you should use, plus some patches.

[davesurrey]

OlddogNewtricks wrote

I created 5 temporary ext2 partitions with gparted & as you can see they are all seen. although pmount hasn't put them in order. No problems with gparted either.

Hi and thanks for helping by doing this test.
I noted that you only have one disk and so all your partitions were on hda.
Mine were on hdb.
So I sacrificed my Windows Seven RC1 partition just now on my hda disk and made partitions up to hda12.
Just as you reported they all show in MUT and Pmount so I conclude this is something to do with the second disk.

So I tried to resize and reformat hdb11 using Gaprted on 214X4 but it wouldn't do that. I used Gparted on Ubuntu 9.04 and that worked so I ended up with a newly formatted hdb11 and an hdb12 partitions. Again 214X4 MUT wouldn't see hdb10, 11 or 12.

Next I tried reformatting hdb10,11 and 12 using ttuuxxx's 420 v2 Deeper Thought just in case there was an inode problem with Ubuntu. Deeper Thought was happy to reformat hdb11 and 12. But again 214X4 failed to see hdb10,11 or 12.

214X4 seemed happy to reformat any partition on disk hda but not after hdb9 on disk hdb. The disk hda is 76GiB. Adding all the space up to hdb9 I get 128GiB and to hdb10 I get 148 GiB.

Therefore I believe this is a fundamental problem that affects 214X (and probably other 2 series??) with disks above the magic 137GB (48bit LBA) limit. I guess that would be an fdisk issue?

I then added "pfix=debug" to the grub stanza for 214X4 and saw that it reported

mounting /dev/hdb9 on /mnt/hdb9 failed

and the same for hdb10,11 and 12.

I guess hdb9 just squeezes in as its boot files will be within the 137GB limit so it can boot okay. But Gparted shows superblock errors for hdb10, 11 and 12. As I would now expect.

Puppy 4 series doesn't have this limitation nor does any other linux distro I have so it's definitely not a problem with my BIOS or MOBO.

Puppy 4 series uses gparted 0.3.3 but 214X seems to say it also uses 0.3.3 which is strange as they clearly work differently. For example in 4 series you have to select whether to view hda or hdb at the start of Gparted but in 214X you can toggle between disks using the tab at the top right side.

Anyway I believe GParted is associated with fdisk (??) so I looked at fdisk.

In Deeper Thought doing in terminal

Code: Select all

fdisk /dev/sdb 

and then

Code: Select all

p

for print partition table I get the full 12 partitions of sdb printed out.

Doing this in 214X works okay for /dev/hda but not for /dev/hdb. No table is printed.

Sorry for very long post but I hope this will convince folk that there is an issue here and what it is associated with. So, is it possible to "update" fdisk (or would it be libparted??) so it can recognise over the 137 GB barrier?

I may as well go on.

Minor niggles include:

1. As OlddogNewtricks noted the partition icons on the desktop are not contiguous. ie they go
hda1, hda10, hda11,hda12,hda2.hda3......
same for hdb not surprisingly.

2. What is a surprise is that it gives an icon for hda4 (even though this is the extended partition) but doesn't do it for hdb4. Very weird.

3. Also Pmount order goes
hda10,hda11,hda12,hda1,hda2......

Again sorry for the long and boring post.
But hope it helps
Dave

[davesurrey]

MinHundHettePerro
Saw your post only after I had written my dead-sea-scroll above.
Well spotted about the missing files in/dev.

I just tried to copy them so I have a set of hdb10 to hdb16.

Obviously my methiod of just copy and paste isn't correct as for example if I copy hdb2 and rename it hdb12 then whatever was on hdb2 is now also hdb12, after a reboot.

Also still get the "failed to mount" messages on boot-up.

Could you please let me know what exactly you did to get this working.

Many thanks.
Dave

[tasmod]

Hi all,

I have the problem over partitions with Gparted with a second disk.

Working on a large primary disk was no problem but when I tried a secondary disk it really ended up screwing it up.
It was recognised OK at first scan. I then tried to delete the two partitions it had (Old windows) this seemed to complete Ok.

However, when I tried to create three new partitions, all reported set up OK but has a small warning triangle at the side. They all showed up but would not mount, it gave error mounting.

I tried all formats with no joy. Finally i removed the drive to another machine and used disk doctor tools to create the partitions again. They were there originally but reported errors. I suspect a faulty write to the partition table.

So it seems to be an issue with a second drive that causes problems. Spooky if it's important, mine wasn't thankfully.

----

Overall everything is running fine. I'm hoping to settle it all in, then stow it upstairs as a torrent seeder during the night.

[MinHundHettePerro]

.............Could you please let me know what exactly you did to get this working.

Many thanks.
Dave

Delete the /dev/hdb10-hdb16 that you somehow copied.
Then, if you want to make a new block device hdb10 (if your highest hdb is hdb9)

Code: Select all

ls -l /dev/hdb9

let N be the number in the 5th column of the output (that's the major type), then

Code: Select all

ls -l /dev|grep ' N,'

find the highest number of the 6th column, let M be this number + 1 (that's the minor type). Now

Code: Select all

mknod /dev/hdb10 b N M

Repeat for hdb11 ...
It is also possible to copy the block-devices from another distro that has the missing block devs. Be sure to use the -dpR switch then. I nicked mine from pup2_218.sfs. But, maybe this could screw up the minor type number.
Now, if the partitions are healthy and formatted etc., I guess you're ready to mount them with pmount, mut or hotpup. Using the mount command on the command-line, I suppose, would require a mkdir /mnt/hdb10 first.

Hope that helps/
MHHP

[davesurrey]

MHHT

Hi well that sorted it all out. Sort of.

MUT, Pmount and fdisk now ident all partitions, GParted doesn't give errors. But ....it still gives "mounting /dev/hdb9 on /mnt/hdb9 failed " and for hdb10,11 and 12 on booting !?

I still feel it's more than a coincidence it happens around the 137GB limit.

Any idea what causes it? Is this common to all 2-series?

Cheers
Dave

[Aitch]

davesurrey

As far as I know, the 137Gb limit stuff is only to do with bios, i.e. the motherboard, or whether you have an 80conductor ide cable, since linux doesn't worry about scanning harddrives for size

Some confusion exists because of windoze problems, going back to win98/2k days, when ata100/133 were introduced on the faster/larger 7200rpm harddrives
Win98 had to have drive translation, DDA, to point the os to a bios jump routine to locate the 48bit lba data needed to access above 137Gb, because of bios & OS limitations [fat16, too]
Win2K needed a registry tweak, but neither would see the higher ata100/133 [udma] without the 80conductor cable
Older bx440 chipsets could be problematic, and a fix for any motherboard/chipset problems, short of changing mobos, is to install a later pci ide/sata controller card with a later chipset onboard

see

http://en.wikipedia.org/wiki/AT_Attachm ... _interface

http://en.wikipedia.org/wiki/SATA

That said, I don't think your problem is to do with this '137Gb limit', though maybe to do with your bios settings, which should have plug'n'pray turned off, and it may be worthwhile exploring the possibility of a bios update, and check your harddrive cabling & master/slave jumpering

HTH

Aitch

[davesurrey]

Hello Aitch,
Thanks for getting involved in this issue and your comments.

I am no expert but I do believe that exceeding the 137GB limit is not just a question of having the right BIOS but also one needs an OS that can support LBA. I don't know where Linux stands in this and you may well be right that it's a non issue (assumingly because its been built into the kernel for a while.)

My point is though that the very same setup (BIOS, MOBO, HDD etc) works happily over the 137GB limit for Ubuntu, fedora, debian, Arch Crunchbang, Slitaz, Tiny Core and more to the point all versions of Puppy that I have on it (including puppy 4.0, 412, 421, 413 ..) except version 214X and now I've just found TeenPup also (which is based on 214).

So I can't see it would be BIOS, PnP, Cable or any other way H/W related except something in (or not in ) the OS.

If I am wrong then I will happily stand corrected and be happy to learn but I can't see logically any other explanation at this time.

Cheers
Dave

[OlddogNewtricks]

I'm no help with a 2nd hard drive, all I have is an old laptop drive I can connect using a USB to IDE connector cable. However this is not recognised by any pup before 2.17. Either a module is needed or this old kernel won't support this setup.
Could that be a clue limited support in this older kernel.

[ttuuxxx]

I'm no help with a 2nd hard drive, all I have is an old laptop drive I can connect using a USB to IDE connector cable. However this is not recognised by any pup before 2.17. Either a module is needed or this old kernel won't support this setup.
Could that be a clue limited support in this older kernel.

Theres one way that you can find out
Barry rebuild a 2 series called 2.18 like 2 weeks back, one version has the old kernel like whats in 2.14.1R/X and one thats newer.
just download the older one and try it, then download the newer one and try it, If not thats ok. lol

older kernel
http://puppylinux.ca/puppyfiles/test/pu ... 6.18.1.iso

newer kernel
http://puppylinux.ca/puppyfiles/test/pu ... p2-218.iso

ttuuxxx

[OlddogNewtricks]

Hi ttuuxx,
No joy there. A boot up message says 'IDM_VALIDATE_PARTITION_TABLE DISK READ FAILED.
Gparted sees it & wants to create a 'disklabel' but fails.

MinHundHettePerro seems to know whats what. However I wondered if you could simoly copy one of the hdb's in /dev to another folder rename hdb10 & pop it back.

[ttuuxxx]

Hi ttuuxx,
No joy there. A boot up message says 'IDM_VALIDATE_PARTITION_TABLE DISK READ FAILED.
Gparted sees it & wants to create a 'disklabel' but fails.

MinHundHettePerro seems to know whats what. However I wondered if you could simoly copy one of the hdb's in /dev to another folder rename hdb10 & pop it back.

Ok so your talking about 2.18 the retro kernel right, and if so then its a kernel things and basically I can't fix it, lol
unless you guys figure out something, who knows, works fine for me, well I don't have that many partitions.
ttuuxxx

[OlddogNewtricks]

This is where the 2.6.21.5 kernel & modules in 2.17 would solve a lot of problems.

[raffy]

Agree that the newer kernel in Puppy 2.17 solves some hardware issues. A good example also is the chip of the dual-core processors: Puppy 2.17 handles it well, but earlier 2 series get confused (for example, it sees an extra/small IDE drive in the system which must be in the chip).

It will be good to try the newer kernel, but only after kernel 2.6.18.1 of the 2 series has been tested and packaged as the updated 2.14R.

Plus, one can instead use Puppy 4 series to cope with the inadequacies of kernel 2.6.18.1 in the Puppy 2 series.

As to Puppy2 seeing a large drive (say 160 GB), I've seen it done by Puppy 2.13 (in a bios that can't actually see the full 160 GB drive). I had 5 partitions in that drive and MUT saw them all.

[ttuuxxx]

Agree that the newer kernel in Puppy 2.17 solves some hardware issues. A good example also is the chip of the dual-core processors: Puppy 2.17 handles it well, but earlier 2 series get confused (for example, it sees an extra/small IDE drive in the system which must be in the chip).

It will be good to try the newer kernel, but only after kernel 2.6.18.1 of the 2 series has been tested and packaged as the updated 2.14R.

Plus, one can instead use Puppy 4 series to cope with the inadequacies of kernel 2.6.18.1 in the Puppy 2 series.

As to Puppy2 seeing a large drive (say 160 GB), I've seen it done by Puppy 2.13 (in a bios that can't actually see the full 160 GB drive). I had 5 partitions in that drive and MUT saw them all.

Well I'm using 2 sata2 hard drives 750GB & 500GB and 2.14.1X sees them just fine I think the issue is number of partitions, like over 9, but it looks like it works for some, and not for others.
ttuuxxx

[sikpuppy]

The ext2/3 File system can handle Tb of info, let along a few hundred Gb.

The sfs that puppy frugal uses may go pearshaped after the first few Gb, but not a full install.

I would be more worried about the condition of the filesystem. Puppy doesn't do a disk check on a semi-regular basis like some other distos. I would imagine that it's more a function of an unclean file system that hasn't been checked.

To do a basic file check using the fsck tool, boot into a live session of puppy 4.x using the "puppy pfix=ram" option, open the GParted tool in the System menu and choose the drive you want to check, then right clicking when the screen comes up.

Choose "Check" from the right click menu and then hit apply after reading the warning

From the command line you can also use the ef2sprogs tool, e2fsck.

Run it using e2fsck -f . Then it will spring up a "Do you want to fix? Y/N" type query in the terminal if there is any issues. Type e2fsck --help to see all the command line switches, including ones to just repair the disc without bothering you.

And don't do any of this on a filesystem that is mounted (or live). Always unmount any discs (it should probably warn you anyway).

[davesurrey]

ttuxxx wrote

Well I'm using 2 sata2 hard drives 750GB & 500GB and 2.14.1X sees them just fine I think the issue is number of partitions, like over 9, but it looks like it works for some,

Yes I think that's correct. A quick look into /dev shows nodes up to 15 or 16 for sda and hda ie SATA an PATA disks. But for sdb and hdb they only go to 8 and 9 respectively. Hence my original problem as I have more than 9 partitions on the second disk of this test machine.

Interestingly I looked into puppy421 and see sdb, sdc and sdd go up to 15. I guess in the knowledge that SATA is now much more common.

OlddogNewtricks wrote:

However I wondered if you could simoly copy one of the hdb's in /dev to another folder rename hdb10 & pop it back.

I tried doing this sort of copy, rename, paste but as I mentioned above it also copies the partition "data" (I guess its a symlink issue?) such that if I use hdb1 to make hdb10 then whatever distro was in hdb1 carries over to hdb10. No good if I have anything in hdb10 already, which I do.

MHHT's post was what fixed it.
Using the cli to add new nodes using

Code: Select all

mknod /dev/hdb10  b  X  Y

I found that you can calc X and Y by the following:

X is 3 for hda and hdb
X is 22 for hdc and hdd

Y is calculated by Y= (Z + P)
where
Z is 0 for hda and hdc
Z is 64 for hdb and hdd.

P is the partition number.

So for example to make hdd5
use mknod /dev/hdd5 b 22 69

A lot easier than it seems.

The only remaining issue is that for reasons I don't understand I still get :

mounting /dev/hdb9 on /mnt/hdb9 failed "

and also for hdb10,11 and 12.
Oh well.

One last thing. I noticed that if I add a partition, add the icon to the desktop, and then delete the partition, the icon remains. Any ideas how to get rid of it in 214X?

Many thanks to all who helped and showed interest. Sorry to bore the rest of you

Cheers
Dave

[davesurrey]

@sikpuppy
I think our posts must have crossed.

Thanks for the good advice but I had already checked that I didn't have any corruption of the disk. I did feel it unlikely as all the other distros on it worked fine.

As I wrote above I think it's now resolved apart from the error message.

Thanks again
Dave

[MinHundHettePerro]

Added a few partitions to my external 250 GB USB drive, sdc1 - sdc14, i.e. 13 partitions to display in MUT/pmount/hotpup.
Hotpup displays ok, MUT displays ok, provided that I change to higher screen resolution (my standard 800x600 can only show down to a cut in half sdc9).
Pmount does not show all partitions; maximum 13 hd partitions and 12 sd parts .

Fix: edit /usr/sbin/pmount
for sd (USB/SATA), go to lines 505 and 523 and add needed number of parts, insert corresponding lines at line 508 (also re-number where needed in lines 508-519).

for hd, the lines are 370 - 392.

It's much easier than it sounds!

@davesurrey
Does /mnt/hdb9 exist? Also, with some disks I've had to explicitly give the fs-type with the mount command.

cheers/
MHHP