[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 6 Jun 2019 16:54:04 -0300
From: Jason Gunthorpe <jgg@...pe.ca>
To: Ralph Campbell <rcampbell@...dia.com>
Cc: Felix Kuehling <Felix.Kuehling@....com>,
Philip Yang <Philip.Yang@....com>,
Alex Deucher <alexander.deucher@....com>, linux-mm@...ck.org,
linux-kernel@...r.kernel.org, John Hubbard <jhubbard@...dia.com>,
Ira Weiny <ira.weiny@...el.com>,
Dan Williams <dan.j.williams@...el.com>,
Arnd Bergmann <arnd@...db.de>,
Balbir Singh <bsingharora@...il.com>,
Dan Carpenter <dan.carpenter@...cle.com>,
Matthew Wilcox <willy@...radead.org>,
Souptick Joarder <jrdr.linux@...il.com>,
Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH 4/5] mm/hmm: hmm_vma_fault() doesn't always call
hmm_range_unregister()
On Thu, Jun 06, 2019 at 12:44:36PM -0700, Ralph Campbell wrote:
>
> On 6/6/19 7:50 AM, Jason Gunthorpe wrote:
> > On Mon, May 06, 2019 at 04:29:41PM -0700, rcampbell@...dia.com wrote:
> > > From: Ralph Campbell <rcampbell@...dia.com>
> > >
> > > The helper function hmm_vma_fault() calls hmm_range_register() but is
> > > missing a call to hmm_range_unregister() in one of the error paths.
> > > This leads to a reference count leak and ultimately a memory leak on
> > > struct hmm.
> > >
> > > Always call hmm_range_unregister() if hmm_range_register() succeeded.
> > >
> > > Signed-off-by: Ralph Campbell <rcampbell@...dia.com>
> > > Signed-off-by: Jérôme Glisse <jglisse@...hat.com>
> > > Cc: John Hubbard <jhubbard@...dia.com>
> > > Cc: Ira Weiny <ira.weiny@...el.com>
> > > Cc: Dan Williams <dan.j.williams@...el.com>
> > > Cc: Arnd Bergmann <arnd@...db.de>
> > > Cc: Balbir Singh <bsingharora@...il.com>
> > > Cc: Dan Carpenter <dan.carpenter@...cle.com>
> > > Cc: Matthew Wilcox <willy@...radead.org>
> > > Cc: Souptick Joarder <jrdr.linux@...il.com>
> > > Cc: Andrew Morton <akpm@...ux-foundation.org>
> > > include/linux/hmm.h | 3 ++-
> > > 1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > > diff --git a/include/linux/hmm.h b/include/linux/hmm.h
> > > index 35a429621e1e..fa0671d67269 100644
> > > +++ b/include/linux/hmm.h
> > > @@ -559,6 +559,7 @@ static inline int hmm_vma_fault(struct hmm_range *range, bool block)
> > > return (int)ret;
> > > if (!hmm_range_wait_until_valid(range, HMM_RANGE_DEFAULT_TIMEOUT)) {
> > > + hmm_range_unregister(range);
> > > /*
> > > * The mmap_sem was taken by driver we release it here and
> > > * returns -EAGAIN which correspond to mmap_sem have been
> > > @@ -570,13 +571,13 @@ static inline int hmm_vma_fault(struct hmm_range *range, bool block)
> > > ret = hmm_range_fault(range, block);
> > > if (ret <= 0) {
> > > + hmm_range_unregister(range);
> >
> > While this seems to be a clear improvement, it seems there is still a
> > bug in nouveau_svm.c around here as I see it calls hmm_vma_fault() but
> > never calls hmm_range_unregister() for its on stack range - and
> > hmm_vma_fault() still returns with the range registered.
> >
> > As hmm_vma_fault() is only used by nouveau and is marked as
> > deprecated, I think we need to fix nouveau, either by dropping
> > hmm_range_fault(), or by adding the missing unregister to nouveau in
> > this patch.
>
> I will send a patch for nouveau to use hmm_range_register() and
> hmm_range_fault() and do some testing with OpenCL.
wow, thanks, I'd like to also really like to send such a thing through
hmm.git - do you know who the nouveau maintainers are so we can
collaborate on patch planning this?
> I can also send a separate patch to then remove hmm_vma_fault()
> but I guess that should be after AMD's changes.
Let us wait to hear back from AMD how they can consume hmm.git - I'd
very much like to get everything done in one kernel cycle!
Regards,
Jason
Powered by blists - more mailing lists