[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: defined(IRIX)
From: |
Paul Eggert |
Subject: |
Re: defined(IRIX) |
Date: |
Thu, 26 Sep 2002 10:03:51 -0700 (PDT) |
> From: Bill Moseley <address@hidden>
> Date: Thu, 26 Sep 2002 06:43:13 -0700 (PDT)
>
> We have a memory manager that aligns reads on sizeof(long), but that
> causes sigbus on some platforms when sizeof(long) == 4 but we need 8 byte
> alignment.
For native builds, there's a simple way to attack this problem. Run
a little test program with AC_TRY_RUN and see whether it dumps core.
Perhaps that is what the Postgres configure does.
For cross-builds, currently we don't have a good hook for running
programs remotely. The simplest workaround is to use an alignment of
8; that should work on all platforms that I know of. You can use the
output of 'config.guess' and guess from that, if you really care about
wasting a few bytes in cross-builds.
- defined(IRIX), Bill Moseley, 2002/09/26
- Re: defined(IRIX), Philip Willoughby, 2002/09/26
- Re: defined(IRIX), Bill Moseley, 2002/09/26
- Re: defined(IRIX),
Paul Eggert <=
- Re: defined(IRIX), Peter Eisentraut, 2002/09/26
- Re: defined(IRIX), Akim Demaille, 2002/09/27
- Re: defined(IRIX), Thomas Dickey, 2002/09/27
- Re: defined(IRIX), Philip Willoughby, 2002/09/27
- Re: defined(IRIX), Bill Moseley, 2002/09/27
- Re: defined(IRIX), Peter Eisentraut, 2002/09/28
- Re: defined(IRIX), Thomas Dickey, 2002/09/28
- Re: defined(IRIX), Peter Eisentraut, 2002/09/29