Dave,
> I believe 2609 describes a regression in 1.0.13.  It is very easy to 
> recreate when jfs.o is built as a module.  A fix for 2609 is in drop 
> 1.0.14 which we're working on today.

Currently i have removed jfs from my system because i had to do some 
work. I will try it 1.0.14 on a separate disk next weekend. In all my 
previous tries i had compiled jfs as a module that gets loaded via 
initrd (my root was also jfs)
 
> 
> > 
> > I would like to have some hints how to isolate the problem. Is there some kind 
> > of kernel debugger that can be activated (i.e. via a special key via keyboard) 
> > so that i can see where the kernel hangs. I think my mainboard does not have a
> > nmi button. For me it looks like some kind of deadlock in jfs.
> 
> 
> I find kgdb very useful.  It requires two machines to be connected via a 
> null-modem serial cable.  Detailed instructions can be found at 
> http://kgdb.sourceforge.net/

I will try it also

> I'm guessing this has to do with fsck.jfs using an ioctl, BLKSSZGET. 
> This is new to fsck.jfs, but it seems the md driver doesn't like it. 
> You can probably ignore it for now and we'll figure out what fsck.jfs 
> should be using.
Ok, i will ignore that, no problem at all
> We've had lots of problems with the mailing list.
Yes, the smtp server in the usf was not respnding ;-)
 
Thanks a lot for your info i will try it again on a separate disk. I 
already have converted all used filesystems to reiserfs because i had 
to much trouble with jfs. I will try a compete jfs system later.

btw, its better to use my office address:
[EMAIL PROTECTED]

------------------------------------------------------
Armin Diehl
[EMAIL PROTECTED]

_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion

Reply via email to