That alleged hyperthreading-related bug

[Mark Geisert <mark at maxrnd dot com>]

Hi, I don't know how additional narrowing-down info about this can be
posted without sounding like a redundant annoying "me too" but here goes.

I can easily reproduce it on a Dell Workstation PWS360 running WinXP SP1.
The CPU is a 2.8Ghz Pentium 4, 800MHz FSB.  1GB RAM installed but no more
than 350MB or so is used when I reproduce the bug.  It's an i875P mobo.
I cannot reproduce it on a 2-CPU Win2K machine, echoing other reports.

I use the 10-line shell script posted quite a while ago that repeatedly
evaluates $(pwd) and checks for zero length result.  On the theory that
anything provoking a problem will provoke it faster with more copies, I
invoke the script with
	./ht& ./ht& ./ht& ./ht&
to create four copies, then recall that line 20 or 30 times to cause LOTS
of processes to be created.

What I see within a minute or so is (numbers differ between messages):
4 [exiting thread] bash 2956 cygthread::stub: erroneous thread activation
2 [exiting thread] bash 4264 cygthread::stub: erroneous thread activation
3 [exiting thread] bash 5496 cygthread::stub: erroneous thread activation

The pid mentioned in the messages always refers to a process started
within the last 10 or so processes while recalling that command above.
Once in a while I also see "Test failed.. Path empty." but can't tie that
to a specific pid without changing the script (easy but haven't done it).

The processes can all be killed in one fell swoop with
	jobs -p | xargs kill
Hope this helps without annoying,

..mark

P.S.: Here's the script I use...

#! /bin/bash
mypath=$(pwd)
while [[ ! -z $mypath ]]
do
  mypath=$(pwd)
  if [[ -z $mypath ]]
  then
    echo "Test failed.. Path is empty."
  fi
done


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/