From owner-cvs-lib  Tue Mar 25 15:29:19 1997
Return-Path: <owner-cvs-lib>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id PAA20443
          for cvs-lib-outgoing; Tue, 25 Mar 1997 15:29:19 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA20431;
          Tue, 25 Mar 1997 15:29:14 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id KAA16383; Wed, 26 Mar 1997 10:27:23 +1100
Date: Wed, 26 Mar 1997 10:27:23 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199703252327.KAA16383@godzilla.zeta.org.au>
To: fenner@parc.xerox.com, mpp@freefall.freebsd.org
Subject: Re: docs/3047 & cvs commit: src/lib/libc/sys sigaction.2
Cc: cvs-all@freefall.freebsd.org, CVS-committers@freefall.freebsd.org,
        cvs-lib@freefall.freebsd.org, freebsd-gnats-submit@freebsd.org
Sender: owner-cvs-lib@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>silently blocks SIGKILL [and] SIGSTOP".  The real meaning (and an
>alternative reading of the same sentence) is "the system silently
>ignores attempts to block SIGKILL or SIGSTOP"; I just wanted to make
>that sentence less ambiguous.

Actually, it means "the system silently ignores attempts to set the
SIGKILL and SIGSTOP bits in [the kernel's copy of] the signal mask.
[This is different from setting the bits and ignoring them when a
SIGKILL or SIGSTOP arrives, since the application can tell the difference
by reading the kernel's copy of the signal mask.  The copy of the mask
for the first sigaction call is of course unaffected, since it is const.]

>(While writing my test program to make absolutely sure, I noticed that the
>first line should probably read "The mask specified in the sa_mask field"
>or something similar, too...)

Yes.  POSIX says this better.

Bruce