From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:48121) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ebnpL-0004H7-K7 for qemu-devel@nongnu.org; Wed, 17 Jan 2018 08:32:28 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ebnpH-0004Ca-H1 for qemu-devel@nongnu.org; Wed, 17 Jan 2018 08:32:27 -0500 Received: from mx1.redhat.com ([209.132.183.28]:47216) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1ebnpH-0004Br-AA for qemu-devel@nongnu.org; Wed, 17 Jan 2018 08:32:23 -0500 Date: Wed, 17 Jan 2018 13:32:01 +0000 From: "Daniel P. Berrange" Message-ID: <20180117133201.GN19227@redhat.com> Reply-To: "Daniel P. Berrange" References: <20180117131821.18700-1-f4bug@amsat.org> <20180117131821.18700-2-f4bug@amsat.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20180117131821.18700-2-f4bug@amsat.org> Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC PATCH 1/3] compiler: add QEMU_WARN_NONNULL_ARGS() List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Philippe =?utf-8?Q?Mathieu-Daud=C3=A9?= Cc: =?utf-8?Q?Marc-Andr=C3=A9?= Lureau , Peter Maydell , qemu-devel@nongnu.org, Stefan Weil , Luiz Capitulino , Stefan Hajnoczi , Paolo Bonzini , Eric Blake On Wed, Jan 17, 2018 at 10:18:19AM -0300, Philippe Mathieu-Daud=C3=A9 wro= te: > Signed-off-by: Philippe Mathieu-Daud=C3=A9 > --- > include/qemu/compiler.h | 2 ++ > 1 file changed, 2 insertions(+) >=20 > diff --git a/include/qemu/compiler.h b/include/qemu/compiler.h > index 340e5fdc09..d9b2489391 100644 > --- a/include/qemu/compiler.h > +++ b/include/qemu/compiler.h > @@ -26,6 +26,8 @@ > =20 > #define QEMU_WARN_UNUSED_RESULT __attribute__((warn_unused_result)) > =20 > +#define QEMU_WARN_NONNULL_ARGS(args...) __attribute__((nonnull(args))) If we take this, it should come with a warning attached to it, because it has really nasty behaviour with GCC. Consider code like: void foo(void *bar) __attribute__((nonnull(1))); ... void foo(void *bar) { if (!bar) return; } GCC may or may not warn you about passing NULL for the 'bar' parameter, but it will none the less assume nothing passes NULL, and thus remove the 'if (!bar)' conditional during optimization. IOW, adding nonnull annotations can actually make your code less robust :-( After having a number of crashes in libvirt caused by gcc optimizing out checks for NULL, we now only define nonnull when running under static analysis (coverity) and not when compiling normally. https://libvirt.org/git/?p=3Dlibvirt.git;a=3Dblob;f=3Dsrc/internal.h;h=3D= 5895030415968d72200599e8a59bbf01ffc2d5a3;hb=3DHEAD#l162 The 2 functions you've added nonnull attrs to look safe enough, but people might unwittingly use this elsewhere in QEMU in future not realizing the side-effect it has. Regards, Daniel --=20 |: https://berrange.com -o- https://www.flickr.com/photos/dberran= ge :| |: https://libvirt.org -o- https://fstop138.berrange.c= om :| |: https://entangle-photo.org -o- https://www.instagram.com/dberran= ge :|