Matthew Wilcox wrote:
>
> On Wed, Jun 12, 2002 at 10:40:07AM +0100, Alan Cox wrote:
> > > SUS v3 does not offer any enlightenment. But it seems reasonable that
> > > processes which share a files_struct should share locks. After all,
> > > if one process closes the fd, they'll remove locks belonging to the
> > > other process.
> > >
> > > Here's a patch generated against 2.4; it also applies to 2.5.
> > > Please apply.
> >
> > This seems horribly inappropriate for 2.4 as it may break apps
>
> I have no problem with withdrawing the request for 2.4. It does mean that
> it's almost impossible to write an M:N threading library implementation.
> This doesn't concern me too much; I just want you to be aware this is
> the tradeoff you're making.
>
> I would still like to see it in 2.5.
Yes, it's needed for M:N threading library. Here is scenario: Task A
holds a lock and waiting for some event in library, now task B tries
to acquire that lock and waits in kernel and this can create a deadlock.
These tasks are created with CLONE_THREAD (for M:N) flag.
This change (removing pid check) may cause problem for 1:1 (linuxthreads),
where each task has unique pid and tgid. Again, whether that's a right
behavior or not is questionable.
However, with CLONE_THREAD flag, all tasks shares "tgid" value with unique
pid and that's why I suggested earlier to change the "fl_pid" from "pid"
to "tgid" and it works for both the cases (M:N and 1:1).
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Jun 15 2002 - 22:00:26 EST