• Welcome to OS2World OLD-STATIC-BACKUP Forum.
 

News:

This is an old OS2World backup forum for reference only. IT IS READ ONLY!!!

If you need help with OS/2 - eComStation visit http://www.os2world.com/forum

Main Menu

JFS + Kernel bug

Started by pasha, 2009.10.08, 10:45:01

Previous topic - Next topic

pasha

Hi

  I found bug (thanks for yoda's  focusing). If you create >16..19 divers letters in your system - next don't show volume label in JFS. I deep to this problem. Result. If you have 16..19 drivers letter in system, you can lost data in all your JFS volume.

  As you can detect this. Create via net use 19 drives letter, say label x:, where x: your JFS drive. You must see lost or damage label. Next, I very recommended immediately reboot or be ready to buy Jrescuer.

  This is bug of os2krnl. In os4 patch for os2krnl is present too now.


warpspace

Would it not make sense to pass this serious bug to IBM, perhaps Mensys should seek a solution?

pasha

Quote from: warpspace on 2009.10.08, 12:47:10
Would it not make sense to pass this serious bug to IBM, perhaps Mensys should seek a solution?
I don't know. I must say about this bug, because I am author of Jresucuer ;)

warpcafe

Hi,

Quote from: warpspace on 2009.10.08, 12:47:10
Would it not make sense to pass this serious bug to IBM, perhaps Mensys should seek a solution?

yes, along with the other things that have accumulated since the 90s.
Seriously, if someone passes that to IBM, they would refer you to the website stating the end-of-life and end-of-support announcement for OS/2.
Leaving aside the questions about whether the current JFS IFS driver we have is officially IBM software, there is the small but important fact that we are talking about a bug in the kernel. Of OS/2. Occurring when the JFS volume is => 16th drive letter... how many business customers of IBM are affected by this... what do you think? So there is no business case for them to fix it. Even if there was, who the hell would be doing this at IBM? I'm pretty sure that the last people who are able to do this are contractors that have already left years ago.
Let's face it: This will not change unless someone gets a soldering iron out and fiddles a workaround into the ridiculously undocumented nuts and bolts of OS/2 that "we" have source-code level access to.

Cheers,
Thomas
"It is not worth an intelligent man's time to be in the majority.
By definition, there are already enough people to do that"
- G.H. Hardy

Pete

Hi

I doubt if IBM would have any interest in fixing anything OS/2 related - they have been trying to bury OS/2 and pretend it does not exist since 1996.

It is not clear whether this bug is in both the iBM JFS and the eCS BootableJFS builds; maybe Pasha can enlighten us further?

I would also guess that a good proportion of drive letters in a system with over 19 drive letters would be network drives. Probably a good case for installing NetDrive to get around the problem - or does that fail also in this case?

Regards

Pete

pasha

#5
Quote from: Pete on 2009.10.08, 16:34:36
Hi

I doubt if IBM would have any interest in fixing anything OS/2 related - they have been trying to bury OS/2 and pretend it does not exist since 1996.

It is not clear whether this bug is in both the iBM JFS and the eCS BootableJFS builds; maybe Pasha can enlighten us further?

I would also guess that a good proportion of drive letters in a system with over 19 drive letters would be network drives. Probably a good case for installing NetDrive to get around the problem - or does that fail also in this case?

Regards

Pete

It is bug in kernels. I can say why, but not ready say where.
Yes, I can drop 1-3 day and fix current jfs.ifs.  But problem be in kernel, I don't know now what is rewrite with ifs struct.

In any case this problem be fixed at near 2 month. Via mensys JFS or via OS/4 patch. I already repeat - problem is in kernel. And I don't know now where. I detect problem only.


mobybrick

Pasha: Can you confirm: OS/4 kernel does NOT have this problem, i.e. it is fixed in it?

pasha

Quote from: mobybrick on 2009.10.08, 19:19:35
Pasha: Can you confirm: OS/4 kernel does NOT have this problem, i.e. it is fixed in it?

OS/4 has this problem too.

RobertM

Does this:

"If you have 16..19 drivers letter in system..."

mean between 16-19 drive letters (ie: that many used), or a letter that the 16th-19th (ie: skipped letters)?

Best,
Rob


|
|
Kirk's 5 Year Mission Continues at:
Star Trek New Voyages
|
|


mobybrick

Back in the old Warp 3 days I opened an APAR with IBM whereby selective install bombed out if there were more than 15 or so drive letters. This was fixed in Warp 3, but I remember at the time IBM telling me proudly (whilst the fix for Warp 3 was being developed) that this issue didn't exist in Warp 4.

Whilst this does not sound like exactly the same bug, the process years ago whereby Warp 4 was tested (and OK) for 15 drive letters, makes me believe that there has been a regression somewhere. But of course, Warp 4 was pre LVM days, and the problem could be there.

Moby.