[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: test failure on FreeBSD 12.0
|
From: |
Bruno Haible |
|
Subject: |
Re: test failure on FreeBSD 12.0 |
|
Date: |
Sun, 03 May 2020 11:58:01 +0200 |
|
User-agent: |
KMail/5.1.3 (Linux/4.4.0-177-generic; KDE/5.18.0; x86_64; ; ) |
Hi Paul,
> > The attached patch (for the 'branch-1.4' branch) fixes the issue.
>
> I think the actual bug lies elsewhere: m4 is written as if int32_t arithmetic
> wraps around silently, but that's not what C does. If I'm right, the bug is
> not
> in the FreeBSD C compiler, it's in m4.
OK, let's clean up the incorrect assumptions about signed integer arithmetic
first.
Here's a patch that does this, in the caller of the 'ntoa' function. I don't see
incorrect assumptions in the code of 'ntoa' itself.
Unfortunately, this patch does not fix the test failure on FreeBSD 12.0. That
is,
after this patch, my previous patch is still needed.
Bruno
0001-Don-t-assume-that-signed-integer-arithmetic-wraps-ar.patch
Description: Text Data
0002-Work-around-a-compiler-optimization-bug-on-FreeBSD-1.patch
Description: Text Data