Bizarre error message

[jc at trillian.mit.edu (John Chambers)]

Here's a strange error message that I can't track down:

  dl_boot: mmap of /dev/zero failed!

Where it appears is in a little wish tool that I wrote that just runs
"netstat -an" every N seconds, uses regexp to extract the tcp-related
lines, and displays them in a sindow.  It usually  works  just  fine,
giving  me  a  running display of the tcp ports that are in use.  But
occasionally it hangs, can't update its  display  for  many  seconds,
often  with the machine bogging down so that mouse and keyboard input
lags by several minutes.  When it wakes up, I find the above  message
in  the  wish  tools error widget, implying that netstat said this on
its stderr.

I haven't found any clues in any docs as to what this means or how to
fix  whatever problem it implies.  Why netstat (or maybe some library
routine it calls) would want to mmap /dev/zero isn't obvious,  though
I suppose there's also no obvious reason to deny it.

The most obvious reason for mmap to fail would be a memory shortage,
but this happens when the machine is nowhere near saturated.  In the
case a while ago, it happened when top in another window showed:

CPU states:  1.7% user,  1.7% system,  3.2% nice, 96.8% idle
Mem:  257472K av, 214000K used, 43472K free, 36808K shrd, 73980K buff
Swap: 136544K av, 47420K used, 89124K free               55836K cached

This is on a valinux/redhat system that calls itself 2.2.13-0.7.

-
Subcription/unsubscription/info requests: send e-mail with
"subscribe", "unsubscribe", or "info" on the first line of the
message body to discuss-request at blu.org (Subject line is ignored).