We run OSF v3.2 in Lazy-Memory-Management mode on an Alpha AXP 3000/400, with
the DEC supplied X-server, Xdec, and Emacs --with-x-toolkit. In a previous
SUMMARY we reported that setting DISPLAY to local:0, in order to use the shared
memory transport, caused a mismatch between Xdec and Emacs (both versions 19.28
and 19.29), resulting in severe capture of CPU cycles by the two processes.
The problem went away when we reset DISPLAY to :0.
However, under Emacs-19.29 there is still a problem of runaway CPU usage by
Emacs processes originating in servers other than the workstation itself, e.g.,
one started by a user who rlogin's from another machine. Xdec is not involved
in this case, just the Emacs process. Moreover, the process remains a runaway
orphan even after the parent rlogin is terminated.
This problem does *not* occur under Emacs-19.28, and we recommend reverting to
that version until the problem is resolved.
After my previous SUMMARY, Richard Stallman <rms_at_gnu.ai.mit.edu> suggested to
step through XTread_socket and see what kinds of events it reads. I am afraid
that I have neither the time nor the knowhow to locate the problem in the Emacs
software, but the suggestion is open to anyone interested.
Cheers -- Amos Yahil
Professor of Astrophysics
===============================================================================
Office: State University of New York
Astronomy Program, E.S.S. Bldg.
Stony Brook, NY 11794-2100, USA
Vox: 516-632-8224
Fax: 516-632-8240
E-mail: Amos.Yahil_at_sunysb.edu (always auto-forwarded)
Public keyID: C1B21A21
===============================================================================
Received on Wed Jul 26 1995 - 15:26:46 NZST