Kernel panic after losing display 
Author Message
 Kernel panic after losing display

I've got a Wish 8.3.1 app that runs on a Solaris 2.6 box and displays
remotely on an HP-UX 10.20 screen. If the user shuts down the HP box
while my app is displayed there (which they shouldn't do, but you *know*
they will!), then the Solaris box eventually generates a kernel panic
and reboots. I haven't had a chance to look at any cores or log files
yet, but I thought I'd see if anyone had any insight into what's going
on.


Sat, 27 Mar 2004 05:06:25 GMT  
 Kernel panic after losing display

Quote:

>I've got a Wish 8.3.1 app that runs on a Solaris 2.6 box and displays
>remotely on an HP-UX 10.20 screen. If the user shuts down the HP box
>while my app is displayed there (which they shouldn't do, but you *know*
>they will!), then the Solaris box eventually generates a kernel panic
>and reboots. I haven't had a chance to look at any cores or log files
>yet, but I thought I'd see if anyone had any insight into what's going
>on.

You may want to add a SIGHUP/SIGINT handler or some such.  TclX can trap signals
for a more gracious shutdown of wish.
--

Tomasoft Engineering, Hayward, CA


Sat, 27 Mar 2004 05:54:28 GMT  
 Kernel panic after losing display
No user space program should be able to generate a kernel
panic unless the kernel is dangerously buggy.

Why on earth are you running Solaris 2.6, anyway?
There is a good chance that you have been rooted and that
you need to do a complete OS reinstall.

Phil

Quote:


>>I've got a Wish 8.3.1 app that runs on a Solaris 2.6 box and displays
>>remotely on an HP-UX 10.20 screen. If the user shuts down the HP box
>>while my app is displayed there (which they shouldn't do, but you *know*
>>they will!), then the Solaris box eventually generates a kernel panic
>>and reboots. I haven't had a chance to look at any cores or log files
>>yet, but I thought I'd see if anyone had any insight into what's going
>>on.

> You may want to add a SIGHUP/SIGINT handler or some such.  TclX can trap signals
> for a more gracious shutdown of wish.



Sat, 27 Mar 2004 06:29:09 GMT  
 Kernel panic after losing display

Quote:

> I've got a Wish 8.3.1 app that runs on a Solaris 2.6 box and displays
> remotely on an HP-UX 10.20 screen. If the user shuts down the HP box
> while my app is displayed there (which they shouldn't do, but you *know*
> they will!), then the Solaris box eventually generates a kernel panic
> and reboots. I haven't had a chance to look at any cores or log files
> yet, but I thought I'd see if anyone had any insight into what's going
> on.

As Phil noted, an app should never generate a kernel panic -
that indicates a bug in the kernel.  I wouldn't be so quick
to think you've been hacked, but I don't think I've ever
heard of Tk doing a kernel panic on Solaris.  The only time
that I can remember Tcl reliably causing a kernel panic was
with some older versions of Linux and *BSD where the posix
threads were buggy and Tcl was compiled with thread support
(this was due to OS core threading bugs).

--
  Jeff Hobbs                     The Tcl Guy
  Senior Developer               http://www.ActiveState.com/
      Tcl Support and Productivity Solutions



Sat, 27 Mar 2004 13:04:35 GMT  
 Kernel panic after losing display

Quote:


> > No user space program should be able to generate a kernel
> > panic unless the kernel is dangerously buggy.

> Agreed.  It sounds like an OS fault (or some *extremely* {*filter*} privileged
> extensions) to me...

> > Why on earth are you running Solaris 2.6, anyway?
> > There is a good chance that you have been rooted and that
> > you need to do a complete OS reinstall.

> Also good advice.  Solaris 8 has been out for ages and is definitely more
> stable than previous versions I've used (not that they were that bad
> either.)

Yes, but there is always the possibility to be too heavy for the
hardware someone owns, where 2.6 may run just fine...

George



Sat, 27 Mar 2004 16:31:23 GMT  
 Kernel panic after losing display
Granted, but are kernel panics "just fine" behaviour?
Since the whole point of running Solaris is for the
legendary (and quite real) stability, if it's panicking
maybe this box should be running Sparc Linux if it is
old and feeble hardware.
My buddy Bob Dinse runs a bunch of old 4 cpu 100MHz
sparc boxen on Sparc Linux quite happily!

Phil

Quote:

> Yes, but there is always the possibility to be too heavy for the
> hardware someone owns, where 2.6 may run just fine...



Sat, 27 Mar 2004 23:32:36 GMT  
 Kernel panic after losing display

Quote:

> No user space program should be able to generate a kernel
> panic unless the kernel is dangerously buggy.

Our testers reported a kernel panic & reboot on two different occasions.
When I tried it, it just bombed out with an XIO error, as I would
expect. Something else is going on, but I don't know what, yet.

Quote:
> Why on earth are you running Solaris 2.6, anyway?

No choice in the matter, it's a legacy system and the customer decides
what to run.

Quote:
> There is a good chance that you have been rooted and that
> you need to do a complete OS reinstall.

Not likely... it's a standalone system not connected to any public net.
Also, it's not like most unix systems that are "normally on"; it's
booted specifically for each use, then shutdown afterwards.


Sun, 28 Mar 2004 00:29:31 GMT  
 Kernel panic after losing display
Quote:


> > No user space program should be able to generate a kernel
> > panic unless the kernel is dangerously buggy.

> Our testers reported a kernel panic & reboot on two different occasions.
> When I tried it, it just bombed out with an XIO error, as I would
> expect. Something else is going on, but I don't know what, yet.

Can I suppose that you have build tcl/tk on 2.6 (and not on some other
version) and that you have done testing again on 2.6?
Do you have access to a machine that panic?
If so, perhaps you can examine the log the kernel wrote before it
died. Can you also examine the patches between a a machine working ok
and one that panics? Can you compile tcl/tk under a machine that panic
and see if it continues to panic?

George



Sun, 28 Mar 2004 13:11:38 GMT  
 Kernel panic after losing display

Quote:

> Our testers reported a kernel panic & reboot on two different occasions.
> When I tried it, it just bombed out with an XIO error, as I would
> expect. Something else is going on, but I don't know what, yet.

I banged pretty {*filter*} the actual system this morning and could not
induce a kernel panic by yanking the remote display out from under Tk.
The /var/adm/messages contained only one panic, not two, and it was a
data fault. Tcl/Tk 8.3.1 was compiled on another Solaris 2.6 box, but
not the same one as the target system. At this point I really don't
consider Tcl/Tk the prime suspect any more.

Thanks to the Tcl community for your feedback!



Mon, 29 Mar 2004 01:21:26 GMT  
 Kernel panic after losing display

Quote:


> > Our testers reported a kernel panic & reboot on two different occasions.
> > When I tried it, it just bombed out with an XIO error, as I would
> > expect. Something else is going on, but I don't know what, yet.

> I banged pretty {*filter*} the actual system this morning and could not
> induce a kernel panic by yanking the remote display out from under Tk.
> The /var/adm/messages contained only one panic, not two, and it was a
> data fault. Tcl/Tk 8.3.1 was compiled on another Solaris 2.6 box, but
> not the same one as the target system. At this point I really don't
> consider Tcl/Tk the prime suspect any more.

IIRC, check the patches of the different machines.  In some
patchlevel in the 2.6 era, Sun actually made an *incompatible*
libc change.  If you compiled on one machine, it could cause
a crash on another (unpatched) one, or something similar.  I'm
100% certain of the cause, but not the exact details as it was
a couple years ago.

--
  Jeff Hobbs                     The Tcl Guy
  Senior Developer               http://www.*-*-*.com/
      Tcl Support and Productivity Solutions



Mon, 29 Mar 2004 11:41:54 GMT  
 
 [ 11 post ] 

 Relevant Pages 

1. realtime display of stdout while using Kernel::backtick

2. 3rd post: 16 bits Clarion5 appl. losing display

3. Losing display running 16 bit Clarion4b

4. Lost the System Event Display

5. Lost Lib files - lost source

6. pretty-printing (lose, lose)

7. Y2k: Panic sets in?

8. Report Writer Panic: Please Help!!!!

9. Help...Panic's Off, I fixed it

10. PANIC !!

11. Error 1013 - major puzzle and panic

12. Newbie in a slight panic

 

 
Powered by phpBB® Forum Software