Dear Managers,
After sending out the question I did a bit of playing around with the ddd
debugger and to my surprise found the answer. emacs is looking backwards
through /var/adm/wtmp to find out when the last system reboot was. Our
wtmp file was extremely large and very probably corrupt (it often is, as a
perusal of the list archive shows).
Renaming wtmp fixed the problem.
Original message:
>
> Recently emacs has started behaving very strangely on on of our Tru64
> systems. The first time you modify a non-scratch buffer a long hang takes
> place during which emacs uses 40 to 50 seconds of CPU time. After that
> emacs behaves normally, even if you visit new files.
>
> The hang takes place when you change the memory copy of the file, not when
> you save the file. So for example,
> $ emacs myfile
> move cursor to place of interest
> insert a character
> LONG WAIT
> everything normal
>
> We are running Tru64 4.0E patch kit 4 and GNU Emacs 20.4.2 (compiled from
> source).
>
> Does anyone have any ideas on what might cause this problem, or how to go
> about diagnosing it?
>
==============================================================
Bob Vickers R.Vickers_at_cs.rhul.ac.uk
Dept of Computer Science, Royal Holloway, University of London
WWW:
http://www.cs.rhul.ac.uk/home/bobv
Phone: +44 1784 443691
Received on Wed May 16 2001 - 14:10:00 NZST