7.0, fedora-8 and "kernel: linux_sys_futex: unknown op 129"

[Roman Divacky]

On Wed, Mar 19, 2008 at 01:18:46PM +0100, Roman Divacky wrote:
> On Wed, Mar 19, 2008 at 01:01:13PM +0300, Boris Samorodov wrote:
> > On Wed, 19 Mar 2008 10:09:56 +0100 Roman Divacky wrote:
> > > On Tue, Mar 18, 2008 at 08:27:09PM -0400, Gary Stanley wrote:
> > > > At 08:17 PM 3/18/2008, Boris Samorodov wrote:
> > > > >Hi!
> > > > >
> > > > >
> > > > >I see some kernel messages "kernel: linux_sys_futex: unknown op 129"
> > > > >when launching print/acroread7 with emulators/linux_base-f8 (soon will
> > > > >appear at ports) and compat.linux.osrelease=2.6.16 at 7.0-RELEASE.
> > > > >
> > > > >Nothing seems to be broken but I don't like such messages. ;-)
> > > > >
> > > > >Is it hard to emplement one? Thanks!
> > > > 
> > > > Try this to silence them. They appear to be harmless for now.
> > 
> > > I'll commit a patch that silences this as it is really harmless...
> > 
> > Yes, it will be great, thanks.
> > And what about implementing the unknown linux_sys_futex 128-129? Is it
> > hard/real? The op no 128 is used many times. Actually, linux-firefox is
> > cycled on op 128 and is not launched.
> 
> thats FUTEX_PRIVATE_FLAG and if I understand it correctly we can just
> ignore this flag and pretend its just "normal" futex.
> 
> I'll post a patch soon.. anyone willing to test?

hm... I think our current implementation of futexes is "private"
so ignoring the flag is the right thing ;)

the super-correct solution would be to

if (!(flags & FUTEX_PRIVATE_FLAG))
   return (EINVAL);

but this is nonsense..