From mboxrd@z Thu Jan 1 00:00:00 1970 Subject: Re: [RFC][PATCH 3/5] mm: RCUify vma lookup From: Peter Zijlstra In-Reply-To: <20070306022319.GF23845@wotan.suse.de> References: <20070306013815.951032000@taijtu.programming.kicks-ass.net> <20070306014211.293824000@taijtu.programming.kicks-ass.net> <20070306022319.GF23845@wotan.suse.de> Content-Type: text/plain Date: Tue, 06 Mar 2007 13:31:49 +0100 Message-Id: <1173184309.6374.110.camel@twins> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: owner-linux-mm@kvack.org Return-Path: To: Nick Piggin Cc: linux-mm@kvack.org, Christoph Lameter , "Paul E. McKenney" , Ingo Molnar List-ID: On Tue, 2007-03-06 at 03:23 +0100, Nick Piggin wrote: > On Tue, Mar 06, 2007 at 02:38:18AM +0100, Peter Zijlstra wrote: > > +static void lock_vma(struct vm_area_struct *vma) > > +{ > > + wait_event(vma->vm_mm->mm_wq, (atomic_cmpxchg(&vma->vm_count, 1, 0) == 1)); > > +} > > + > > +static void unlock_vma(struct vm_area_struct *vma) > > +{ > > + BUG_ON(atomic_read(&vma->vm_count)); > > + atomic_set(&vma->vm_count, 1); > > +} > > This is a funny scheme you're trying to do in order to try to avoid > rwsems. Of course it is subject to writer starvation, so please just > use an rwsem per vma for this. [damn, he spotted it :-)] Yeah, I know :-( > If the -rt tree cannot do them properly, then it just has to turn them > into mutexes and take the hit itself. That is, unfortunately, still not acceptable. Take futexes for example, many threads 1 vma. > There is no benefit for the -rt tree to do this anyway, because you're > just re-introducing the fundamental problem that it has with rwsems > anyway (ie. poor priority inheritance). The thing is, we cannot make the whole VM realtime, that is just plain impossible. What we do try to do is to carve a niche where RT operation is possible. Like a preallocated mlocked arena. So mmap and all the other vma modifiers would fall outside, but faults and futexes would need to be inside the RT scope. Ingo, any ideas? perhaps just introduce a raw rwlock in -rt and somehow warn whenever an rt_task does a write lock? Full vma RCUification is hindered by the serialisation requirements; one cannot have two inconsistent views of the mm. > In this case I guess you still need some sort of refcount in order to force > the lookup into the slowpath, but please don't use it for locking. down_read_trylock on the vma rw lock would work I guess. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org