This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: the same gcc binary emits warnings on one machine but not on another
- From: Florian Weimer <fw at deneb dot enyo dot de>
- To: john smith <wempwer at gmail dot com>
- Cc: gcc-help at gcc dot gnu dot org
- Date: Thu, 20 Oct 2016 23:15:10 +0200
- Subject: Re: the same gcc binary emits warnings on one machine but not on another
- Authentication-results: sourceware.org; auth=none
- References: <CAKmQUfZnzT9AMG5wE7fnePA4mzUmSN-C6aMNs6RFN1+LZDrz2w@mail.gmail.com>
* john smith:
> We use the same cross-compiler on two Linux machines: machine A and
> machine B. It produces some warnings on machine A but not on machine
> B when used to compile the same file and with the same set of options
> (specifically we use -Wall and -Wextra). We are absolutely sure that
> it's the same compiler - it's in the same version and has the same
> md5sum on both machines. However, these 2 Linux machines differ - A
> is x32 Ubuntu, B is x64 bit Slackware.
It's unlikely that the two compiler versions are the same. In the
past, Debian has backported some new upstream warnings to an older
version, and Slackware probably won't have these backports.
In addition, some warnings are issued from the optimizers, and they
can be very target-specific in practice. I saw this with -Woverflow
warnings in particular because fold uses different heuristics on IBM
targets compared to the rest.