Re: [FYI 1/2] {test-protocols} tap/awk: handle exit statuses > 256 ( seen on few korn shells)

[Stefano Lattarini]

Hi Eric, thanks for the links.

On Wednesday 28 September 2011, Eric Blake wrote:
> On 09/28/2011 08:34 AM, Stefano Lattarini wrote:
> > Some Korn shells, when a child process die due to signal number
> > n, can leave in $? an exit status of 256+n instead of the more
> > standard 128+n.  Apparently, both behaviours are allowed by
> > POSIX,
> 
> Indeed.  Here's some further reading:
> 
> http://austingroupbugs.net/view.php?id=51
>
Very useful: I've added a reference to it in the comments of
tap-driver.sh; see attached patch.

> http://austingroupbugs.net/view.php?id=347
>
This does not seem relevant for the issue under discussion to me.  Maybe
wrong link?  Or am I missing something?

> [SNIP]
>
> > so be prepared to handle them both.
> 
> Yes, portability does demand we handle both.  Also, would you mind 
> preparing an autoconf patch to mention this?
> 
Patch coming up soonish (hopefully this evening, otherwise in a couple
of days).

Regards,
  Stefano

From 5014b1c294a4b8982ab3c7c78e2272fdcdff468d Mon Sep 17 00:00:00 2001
Message-Id: <address@hidden>
From: Stefano Lattarini <address@hidden>
Date: Wed, 28 Sep 2011 20:48:13 +0200
Subject: [PATCH] tap/awk: improve comments about Korn shell signal handling 
issues

* lib/tap-driver.sh: Link an Austin Group discussion about how
shells are allowed, and even encouraged, to set the special
variable `$?' to values greater than 256 to report termination
of a child by a signal.  Improve and extend comments about our
workarounds for unusual korn shell signals' propagation.

Thanks to Eric Blake for the pointers.
---
 ChangeLog         |   10 ++++++++++
 lib/tap-driver.sh |   18 +++++++++++-------
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index c28f9e4..7a78db6 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,15 @@
 2011-09-28  Stefano Lattarini  <address@hidden>
 
+       tap/awk: improve comments about Korn shell signal handling issues
+       * lib/tap-driver.sh: Link an Austin Group discussion about how
+       shells are allowed, and even encouraged, to set the special
+       variable `$?' to values greater than 256 to report termination
+       of a child by a signal.  Improve and extend comments about our
+       workarounds for unusual korn shell signals' propagation.
+       Thanks to Eric Blake for the pointers.
+
+2011-09-28  Stefano Lattarini  <address@hidden>
+
        maintcheck: fix usage of `cd' instead of `$(am__cd)'
        * lib/am/inst-vars.am (am__uninstall_files_from_dir): Use
        `$(am__cd)', not plain `cd'.
diff --git a/lib/tap-driver.sh b/lib/tap-driver.sh
index e9f1037..c911991 100755
--- a/lib/tap-driver.sh
+++ b/lib/tap-driver.sh
@@ -117,14 +117,17 @@ fi
 
 {
   (
-    # Ignore common signals (in this subshell only!) to avoid potential
+    # Ignore common signals (in this subshell only!), to avoid potential
     # problems with Korn shells.  Some Korn shells are known to propagate
     # to themselves signals that have killed a child process they were
-    # waiting for (this is done at least for SIGINT -- and usually only
-    # for it in truth); this would cause a premature exit in this subshell,
-    # so that the awk script would never seen the exit status it expects
-    # on its last input line (and which is displayed below by the last
-    # `echo $?' command), and would thus die reporting an internal error.
+    # waiting for; this is done at least for SIGINT (and usually only for
+    # it, in truth).  Without the `trap' below, such a behaviour could
+    # cause a premature exit in the current subshell, e.g., in case the
+    # test command it runs gets terminated by a SIGINT.  Thus, the awk
+    # script we are piping into would never seen the exit status it
+    # expects on its last input line (which is displayed below by the
+    # last `echo $?' statement), and would thus die reporting an internal
+    # error.
     # For more information, see the Autoconf manual and the threads:
     # <http://lists.gnu.org/archive/html/bug-autoconf/2011-09/msg00004.html>
     # 
<http://mail.opensolaris.org/pipermail/ksh93-integration-discuss/2009-February/004121.html>
@@ -458,7 +461,8 @@ function get_test_exit_message(status)
     # shells, when a child process die due to signal number n, can leave
     # in $? an exit status of 256+n instead of the more standard 128+n.
     # Apparently, both behaviours are allowed by POSIX (2008), so be
-    # prepared to handle them both.
+    # prepared to handle them both.  See also Austing Group report ID
+    # 0000051 <http://www.austingroupbugs.net/view.php?id=51>
     exit_details = sprintf(" (terminated by signal %d?)", status - 256)
   else
     # Never seen in practice.
-- 
1.7.2.3