Problems with NOP on an IBM Thinkpad 240x laptop

[TheBlackSheep]

I bought one of these laptops (thinkpad 240x) off ebay as a cheap alternative to a netbook (it's a PIII 500Mhz 128Mb model and has a 10.4" screen@800x600 and a cheap PCMCIA Wifi card) for use on holiday/traveling etc.

I stuck in a 20Gb drive and loaded puppy 4.1 although recently changed this to NOP413 (frugal installs on a relatively small DOS partition but with a larger ext3 partition for files outside the system) which looks and works generally very well on here.

Couple of things I'd like to make better if I can;

When booting the system appears to hang on the "Booting the kernel" bit for over a minute - I haven't seen this on other systems, in fact most systems (even equivalent performance ones) are generally fully loaded to screen within 45 seconds. I'm going to try a different hard-disk (a 30Gb one) to see if that improves the situation but I have a feeling it's actually to do with some hardware analysis by the kernel process at this point - would it make sense to use the retro version of puppy to try and improve this here?

the shutdown works fine but for some reason a restart doesn't - it seems to go right through the shutdown process but when restarting after the BIOS screen it just hangs. I updated the BIOS to the latest when I originally got it so I presume that's ok. This scenario was present on Dingo as well as the current version of NOP. I've got "acpi=force" set as this seemed to work better than "off" or "on" but this problem remains.

Any help here most grateful.

Chris

[gary]

I had a similar problem recently that ended up being the HD itself. It turns out the drive had sectors that were going bad, but weren't quite fully bad yet. Remember, that disk data is ultimately analog at the magnetic domain level. This means that individual bit strength may vary somewhat from bit to another. Everytime a sector is read (usually 512 bytes), a CRC is also read which guarantees its data integrity. If the CRC notes an error, the sector is re-read. This continues until a max retry count is exceeded (assigned by the driver or a higher up layer), or the data gets read intact. Keep in mind that single bits, or even sectors are incredibly small surface areas, and because of that, rarely go bad individually. Instead multiple weak sectors, if any, would be the norm. All of this means that reads from certain areas of a disk may take longer, even though no error gets reported.

When you installed your 20gig drive you didn't say whether it was new, used, eBay unknown, etc.? One easy way to verify all sectors is to use the badblocks command as follows:

badblocks -sv /dev/mydevice

Where mydevice is your hard drive designator (sda1, hda2, etc.). This will read every disk sector (nondestructively) complete with CRC checking and report any sectors either outright bad (hard errors) or requiring retries (soft errors). Just typing badblocks in the console will give you a summary of its options, or Man badblocks for a more complete explaination of its use.

[TheBlackSheep]

Hi Gary

thanks for the info - I think you were partially right with the hardware issue although I think the drive itself might be ok - I replaced it with the 30gb drive but had a torrid time trying to get it to work. The design of the small interconnector between the IDE drive and the m/b creates a very poor fit (the drive is mounted on the plastic case cover and slots in when you replace the drive) - when I held the drive(s) in place manually with my fingers the boot time was reduced to ~1 minute with no hanging on the "Booting the kernel" element. When I replaced the drive correctly on the cover it either wouldn't boot at all or took the two minutes or so again. I've since reverted to a bit of folded paper (not for the first time!) to try and hold the drive in the correct position and it seems to be fine now.

Now just the restart issue!

Note: Have tried changing the thread subject as this is not an NOP specific issue but a puppy general one (Dingo et al.)

Chris

[gary]

Well, here's some wisdom from an old (relatively anyway - LOL) retired test enggie. Never confuse a symptom with the actual underlying problem. In this case (no pun), it sounds like there is a connectivity problem in the drive connector. It may manifest itself more in Puppy for whatever reason (more error retries, etc.?), but the fundamental problem is still connectivity. If you can boot at all, I would run the badblocks test I mentioned earlier to establish if any read errors are being exhibited. Since you have tried 2 different drives with similar results, the problem seems to lie with the LT's connector. This could be a broken solder connection or an intermittent short. Shimming the drive with paper merely addresses the symptom, not the problem.

BTW, I measured my own startup time from pushing power button to full desktop screen at about 50 seconds. My "booting to kernal" event lasts about 10 seconds. This is on a 2Ghz AMD3200+ with 512meg ram, and a frugal HD install with Pup 4.12.