bug-bash
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

2.04 build problem/solution for AIX 4.3.x


From: Nicholas Geovanis
Subject: 2.04 build problem/solution for AIX 4.3.x
Date: Thu, 16 Nov 2000 20:17:09 -0600

Using IBM C and C++ Compiler 3.6.6 on AIX 4.3.2, support/mksignames.c was
core dumping when filling signal_names[] with SIGRTMAX. This turned-out to
be because SIGRTMIN was defined as 888 and SIGRTMAX at 999, both far
larger than NSIG (64). Moping through sys/signal.h, the comments state
that realtime signal constants and functions are defined for
compatibility's sake, but they are not supported (hence the apparently
arbitrary values).

I tried messing with the Posix-compatibility preprocessor defines to keep
the wacky values from being defined, but I couldn't find a combination
that worked for everything. Since mksignames seemed to be relatively
stand-alone, and since I'm no good with autoconf, I thought I'd just
modify the source. The diff is reproduced below (you can safely ignore
the #pragma). All bash test pass and I've seen no other problems in
limited use so far.

The doc states that bash 2.04 works on AIX 4.1.x and AIX 4.2.x and I know
that we built earlier versions here on those platforms, so I'm assuming
that this is a problem introduced with AIX 4.3.x. But all IBM equipment
here is now 4.3.2 or 4.3.3 so I can't test this guess. See ya...

# diff mksignames.orig.c mksignames.c
21a22
> #pragma options source showinc
34a36,38
> #if defined (_AIXVERSION_430) || defined (_AIXVERSION_431)
> char *signal_names[2 * SIGRTMAX];
> #else
35a40
> #endif
39c44,47
< #if defined (SIGRTMAX) || defined (SIGRTMIN)
---
> #if defined (_AIXVERSION_430) || defined (_AIXVERSION_431)
> #  define RTLEN 14
> #  define RTLIM SIGRTMAX
> #elif defined (SIGRTMAX) || defined (SIGRTMIN)
#

* Nick Geovanis
| IT Computing Svcs           Let's have a war,
| Northwestern Univ           We can jack-up the Dow-Jones.
| address@hidden             - Fear
+------------------->




reply via email to

[Prev in Thread] Current Thread [Next in Thread]