Debugging a system hang.

[Naoki]

Hi Dave,

I couldn't get any more of the trace because that was all I could fit on
the post it note when copying it down before kicking the box in the
guts :)

Just tried again with : 2.6.12-1.1447_FC4smp #1 SMP Fri Aug 26 21:03:12
EDT 2005 x86_64 x86_64 x86_64 GNU/Linux  & glibc-2.3.5-10.3

Started the rsync,  and then saw a blip of activity after the file list
was built :

 1  0      0 625444  14764  47896    0    0     0     3 1015    18 50  0
50  0
 0  1      0 556360  21636  84184    0    0  1518    13 1359   700 18  5
50 27
 0  1      0 134984  52840 274760    0    0  5975   680 2792  3755  1 23
49 27
 0  0      0  24640  75672 320568    0    0   546  6373 2378  3252  2 21
40 37
 0  0      0  24688  75672 320568    0    0     0     0 1122   155  0  0
100  0

Then nothing, a long list of 100% idle..  I CTRL-C'ed it and ran 'dmesg'
which has still not come back with anything and the box is now down to
the point of being unable to SSH into it.  Ohh, exciting.

There is no trace on the console and when I type in 'root' and press the
return key I get a newline created.. And then a whole lot of nothing.  I
can get chars on the screen but CTRL-ALT-DEL fail to provide joy.



On Mon, 2005-09-05 at 22:41 -0400, Dave Jones wrote:
> On Tue, Sep 06, 2005 at 11:38:45AM +0900, Naoki wrote:
> 
>  > The closest thing to a trace I've managed to see is :
>  > 
>  > <3>Debug : Sleeping function called from invalid context at
>  > include/linux/rwsem.h:43
>  > in_atomic:(0) irqs_disabled:(1)
> 
> Below this line should be a trace of functions that were called
> just before this happened. Without that, there isn't enough
> info to go on.
> 
> You should also try the latest errata kernel.
> 
> 		Dave
>