Date: Tue, 18 Sep 2012 11:17:49 -1000 (HST) From: Jeff Roberson <jroberson@jroberson.net> To: Poul-Henning Kamp <phk@phk.freebsd.dk> Cc: Konstantin Belousov <kostikbel@gmail.com>, attilio@freebsd.org, freebsd-arch@freebsd.org, arch@freebsd.org Subject: Re: Aliasing issue with TAILQ on ppc64 ? Message-ID: <alpine.BSF.2.00.1209181116181.47770@desktop> In-Reply-To: <23196.1347999426@critter.freebsd.dk> References: <23196.1347999426@critter.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 18 Sep 2012, Poul-Henning Kamp wrote:
> In message <alpine.BSF.2.00.1209180858190.47770@desktop>, Jeff Roberson writes:
>
>> If you compile at low optimization levels I assume it goes away?
>
> I don't have access to the machine at this time, and in general
> I'm not particular focused on PPC64, since this is a general
> problem on any architecture, as I understand the C-language rules.
Have you verified that it actually happens on other architectures? It may
be a bug only in the ppc backend and not in bsd code.
>
>> Can you place asm volatile ("" ::: "memory"); between the two tailq macros
>> with the same optimization level?
>
> I tried earlier to put a printf("FOO") there, and that fixed it, which
> I why I diagnosed it as an aliasing issue in the first place.
I suggest this approach to further narrow the scope of the issue.
Jeff
>
> --
> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
> phk@FreeBSD.ORG | TCP/IP since RFC 956
> FreeBSD committer | BSD since 4.3-tahoe
> Never attribute to malice what can adequately be explained by incompetence.
>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.1209181116181.47770>