Re: [PATCH] Ref counting for tty_struct and some locking clean up

[Randy.Dunlap]

On Fri, 7 Jul 2006 21:33:37 -0400 Jon Smirl wrote:

> On 7/7/06, Randy.Dunlap <rdunlap@xxxxxxxxxxxx> wrote:
> > On Fri, 7 Jul 2006 20:37:36 -0400 Jon Smirl wrote:
> >
> > > This patch removes tty_mutex and replaces it with a ref counted
> > > tty_struct. A new spinlock, tty_lock, serializes tty
> > > creation/destruction. This code thoroughly audits everyone taking
> > > references to tty_struct and tracks them. I have been running it
> > > locally for a few days without any issues.
> > >
> > > If there is a refcount problem it will trigger a WARN_ON. If you hit
> > > one, defining DEBUG_TTY_REFCOUNT will provide plenty of debug info to
> > > help locate the missing reference. I definitely found multiple locking
> > > holes in the old tty_mutex code but they they look hard to hit.
> > >
> > > This is just a first step, if this code proves to be stable I'll do
> > > another round of cleanup. There is room for a lot more work to be
> > > done. Ultimately there may not even be a need for a tty refcount if
> > > everything is locked correctly. But that is a big step and we should
> > > take some smaller ones first.
> > >
> > > This is meant for mm until it gets some testing - I don't want to
> > > break anyone's console. If gmail wraps it let me know and I'll send
> > > you an attachment.
> >
> > It has one wrapped line in it but does not apply cleanly to
> > 2.6.18-rc1 or to 2.6.17-mm6.  What was it diffed against?
> >
> > ---
> > ~Randy
> >
> 
> Diffed against the current git tree. Commits going into git keep
> making it have fuzz but none of them have broken it.

Which current git tree?  www.kernel.org/git/ lists a whole bunch
of them, then there are some not at kernel.org.

---
~Randy
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/