Emacs19 crashes OpenWindows? Network traffic monitor?

Emacs19 crashes OpenWindows? Network traffic monitor?

Post by Mark Hastin » Fri, 29 Apr 1994 10:41:37



Sorry if this is a common question, but every time I've tried running
emacs-19 on my Solaris 2.3 machine, it has crashed my OpenWindows Xserver.
It also happened on other machines here, but none have any patches applied.
The core file shows that it's in some font code in the server.  Does anyone
know if there is a patch?  For now, I just run 'emacs -nw' in an xterm,
but I want to play with the fonts and color...

Also, on SunOS 4.x, I used to run /usr/bin/traffic, which was a suntools
program that visually displayed network traffic (more digestable than snoop).
Are there any PD workalike programs?

Thanks for any tips,
        Mark Hastings

 
 
 

Emacs19 crashes OpenWindows? Network traffic monitor?

Post by Casper H.S. D » Fri, 29 Apr 1994 17:17:26



>Sorry if this is a common question, but every time I've tried running
>emacs-19 on my Solaris 2.3 machine, it has crashed my OpenWindows Xserver.
>It also happened on other machines here, but none have any patches applied.
>The core file shows that it's in some font code in the server.  Does anyone
>know if there is a patch?  For now, I just run 'emacs -nw' in an xterm,
>but I want to play with the fonts and color...

This is a known bug in OW 3.3 Xsun.  Get patch 101362-06 or later (-11 = cur)

Quote:>Also, on SunOS 4.x, I used to run /usr/bin/traffic, which was a suntools
>program that visually displayed network traffic (more digestable than snoop).
>Are there any PD workalike programs?

There's xtraffic, but it requires rpc.etherd which isn't part of Solaris 2.x.

Casper

 
 
 

1. Memory leak/kernel crash under heavy network traffic load

We are running a 2.4.2-based kernel on an IBM 405GP processor. Under
ethernet-to-ethernet bridging, with an extremely heavy continuous
traffic load (25k 128-byte packets/sec into each port), the system
runs out of memory and crashes.

Our working theory is that the ethernet ports are generating hardware
interrupts so frequently that the softirq function net_tx_action() in
net/core/dev.c never gets a chance to run.  Hence, completed transmit
skbs on the completion_queue are never freed.

If I replace dev_kfree_skb_irq() with dev_kfree_skb() in the ethernet
driver transmit complete ISR, the system does not run out of memory,
since the skb is freed immediately rather than being queued for later
disposal by
net_tx_action().

Does this seem like a plausible explanation and also, is it kosher to
call dev_kfree_skb() from an ISR? Note we don't expect the kernel to
keep up with such a heavy traffic load, but only to behave gracefully
(i.e. not crash).

Thanks for any information.

Joe Perri
Viadux, Inc. (formerly RC Networks, Inc.)
9890 Towne Centre Dr. #200
San Diego, CA 92122

2. finger command, different details

3. Crash during network traffic

4. Sound Galaxy Nova 16

5. Machine crashes with heavy network traffic

6. Chipset, Mouse and Video Card types?

7. Problem: Crash into the debugger at havvy network-traffic...

8. OCR Program

9. Crash during network traffic

10. Real-Time Monitoring & Accounting of Network traffic

11. network traffic monitoring without kstat

12. tools to monitor network traffic from the server side

13. how to monitor current network traffic on my machine