ps (maybe ksh?) bug (SysVr2)

Curtis Jackson rcj at burl.UUCP
Fri Aug 2 04:04:10 AEST 1985


Before I describe this problem, know ye that I run a recent (02/21/85)
version of David Korn's ksh.

We have been having a problem where 'ps' stops in the middle of printing
out its process entries.  It always stops after a full entry has been
output (makes sense, since terminal output is line-buffered); the shell
(ksh) then merrily eats (as in discards) all input until the user hits
a return, at which point the ps finishes printing and the shell prompt
is given as usual.

Some clues:

a) There are only four of us that I know of who are heavy users of 'ps',
and three of us have seen the problem.  The fourth is the only one
who doesn't use the '-f' option; he always does either 'ps' alone or
'ps -al'.  The rest of us use '-f' alone or in concert with '-e' or '-a'.

b) The last process given by ps before it halts is (seemingly) random.

c) The problem seems to occur MUCH more frequently when the user is
running shl (shell layers); this is what tends to make me suspect the
shell rather than 'ps' itself.

d) Most of the time (maybe always; I can't duplicate it at will to find
out), the COMMAND name of the process printed \after/ the user has hit
a return is null (i.e., nothing in the COMMAND field).  That process
seems to be as random as the process before ps halts; except that I
can't remember a time when this process first printed after the return
is entered was NOT owned by init (ppid=1).

Thanks so much for any (non-null) pointers to an answer to this mystery,
-- 

The MAD Programmer -- 919-228-3313 (Cornet 291)
alias: Curtis Jackson	...![ ihnp4 ulysses cbosgd mgnetp ]!burl!rcj
			...![ ihnp4 cbosgd akgua masscomp ]!clyde!rcj



More information about the Comp.unix.wizards mailing list