RE: Finished threads remain as zombies on 2.6?

["Tom Evans" <tom@23palmer.net>]

So, has anyone submitted a defect yet? 

-----Original Message-----
From: Tyson Whitehead [mailto:twhitehe@uwo.ca] 
Sent: Monday, August 29, 2005 10:49 AM
To: debian-alpha@lists.debian.org
Cc: Thomas Evans; Thimo Neubauer
Subject: Re: Finished threads remain as zombies on 2.6?

On August 29, 2005 08:33, Thomas Evans wrote:
> Another datapoint - I have a 2nd 164LX, runs Debian testing instead of 
> unstable (libc version 2.3.2.ds1-2) - I built the same kernel 
> (2.6.12.5) that is running on the Debian unstable system.
>
> Threads clean up just fine, so I'm going to hazard a guess that it is 
> not the kernel at this time - porbably something with pthreads in the 
> unstable dist.?

Yup.  Looks like you are right.  I just grabbed the libc6.1 deb for
2.3.2.ds1-22, extracted the contents and dynamically linked the C threadtest
against it (i.e., './LD_LIBRARY_PATH=. ./ld-2.3.2.so ~/tmp/threadtest' in
the extracted lib directory) and the threads clean up just fine.

I checked the libc6.1 change logs for thread and alpha entries between the
two versions and only found the following (the 2.3.5-3 changelog):
  * This version fixes some bugs that are marked as fixed-upstream:
    -Fix broken pthread_cleanup_push on Alpha.  (Closes: #197988)

The proc weirdness issue (not differentiating between threads and
processes), still exists when using the earlier libraries.  I wonder if that
is a pthreads issue or a kernel one?

-T

-- 
 Tyson Whitehead  (-twhitehe@uwo.ca -- WSC-)
 Computer Engineer                          Dept. of Applied Mathematics,
 Graduate Student- Applied Mathematics      University of Western Ontario,
 GnuPG Key ID# 0x8A2AB5D8                   London, Ontario, Canada