Re: procps-1.2.2

Albert D. Cahalan (
Thu, 18 Dec 1997 02:27:50 -0500 (EST)

Brian Feeny writes:

> I upgraded to procps-1.2.2 as per the Changes file for the
> latest kernels, and noticed this:

First, the Changes file should have procps-1.2.5 instead.
Versions below that have a /tmp security hole.

> luna:~$ ps -aux
> warning: `-' deprecated; use `ps aux', not `ps -aux'
> Why did they do away with the "-"? Isn't it "standard" to use a - for
> arguments to a program? So now we are suppose to do like "ps aux"?

Make sure "man" can find the new man page (not some old preformatted one)
and read it. For those that haven't upgraded to procps-1.2.5 yet:

The standard interpretation of "ps -aux" is:

-a select all processes with a tty, but omit session leaders
-u select by effective user ID (list some EUID nuumbers...)
-x illegal

The standard equivalent to "ps aux" would be "ps -ef".
Most modern unix systems will interpret both.

Older SysV systems only accepted "ps -ef". The "ps aux" is
from BSD. Older BSD systems also accepted "ps -aux", but that
conflicts with POSIX and the Unix98 standard.

The development version of ps accepts most AIX and Digital Unix
syntax. It also accepts as much BSD syntax as possible, taken
directly from several *BSD man pages. Also, 16 gnu long options.

ObKernel: :-)

The kernel does not seem to supply info for these fields:

CP/CPU short-term cpu usage factor (for scheduling)
CURSIG current signal
SESS session pointer
JOBC job control count; count of processes qualifying PGID for job control
SL sleep time (in seconds; 127 = infinity)
RE core residency time (in seconds; 127 = infinity)

Have I missed something? These are needed for full BSD behavior.