GNU bug report logs - #51591
webkitgtk fails to build on i686-linux; possibly a clang issue

Previous Next

Full log

View this message in rfc822 format

From: Mark H Weaver <mhw <at> netris.org>
To: Maxime Devos <maximedevos <at> telenet.be>, Liliana Marie Prikler <liliana.prikler <at> ist.tugraz.at>, 51591 <at> debbugs.gnu.org
Subject: bug#51591: webkitgtk fails to build on i686-linux; possibly a clang issue
Date: Sat, 06 Nov 2021 02:09:23 -0400

Hi Maxime,

Maxime Devos <maximedevos <at> telenet.be> writes:

> Mark H Weaver schreef op vr 05-11-2021 om 15:42 [-0400]:
>> +             ;; XXX Until we switch back to using GCC,
>> +             ;; work around <https://bugs.gnu.org/51591>.
>> +             ,@(if (string=? "i686-linux" (%current-system))
>> +                   '((substitute*
>> "Source/WTF/wtf/CheckedArithmetic.h"
>> +                       (("#define USE_MUL_OVERFLOW 1")
>> +                        "#define USE_MUL_OVERFLOW 0")))
>> +                   '())
>
> For cross-compilation, this should use (%current-target-system)
> instead. Also, I'd presume the same issue exists for the Hurd.
> I'd suggest: (target-x86-32?). That procedure is not
> yet in master (only core-updates{-frozen{,-batched-changes}}),
> so it would need to be cherry picked.

You're certainly right that the simple test above is not a good example
of how best to check if we're building for i686-linux.

For that matter, a more proper fix would be the one I proposed in my
earlier email: to apply a patch to Source/WTF/wtf/CheckedArithmetic.h
that changes "CPU(ARM)" to "(CPU(ARM) || CPU(XXX))", where XXX is the
appropriate symbol for 32-bit x86.  Such a patch could even be submitted
upstream.  That's really the kind of approach I'd like to encourage.

In this case, I had very little available time to work on it, but wanted
to get the problem fixed on i686-linux ASAP.  Given this, and the fact
that this code is a temporary workaround until the core-updates-frozen
merge, and also my assumption that we are quite far from being able to
usefully cross-compile WebKitGTK and its dependencies, I don't feel
*too* badly about cutting corners here.

Having said all of this, I certainly agree that the conditional above
should be changed to follow our best practices.  Would you like to
propose a patch?

Thanks very much for your vigilance.

     Regards,
       Mark

-- 
Disinformation flourishes because many people care deeply about injustice
but very few check the facts.  Ask me about <https://stallmansupport.org>.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.