Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 9 Aug 2022 21:01:55 +0000
From:   Carlos Llamas <cmllamas@google.com>
To:     Liam Howlett <liam.howlett@oracle.com>
Cc:     "linux-mm@kvack.org" <linux-mm@kvack.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        syzbot <syzbot+a7b60a176ec13cafb793@syzkaller.appspotmail.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Ondrej Mosnacek <omosnace@redhat.com>,
        "syzkaller-bugs@googlegroups.com" <syzkaller-bugs@googlegroups.com>,
        Minchan Kim <minchan@kernel.org>,
        Christian Brauner <brauner@kernel.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Hridya Valsaraju <hridya@google.com>,
        Joel Fernandes <joel@joelfernandes.org>,
        Martijn Coenen <maco@android.com>,
        Suren Baghdasaryan <surenb@google.com>,
        Todd Kjos <tkjos@android.com>,
        Matthew Wilcox <willy@infradead.org>,
        Arve =?iso-8859-1?B?SGr4bm5lduVn?= <arve@android.com>
Subject: Re: [PATCH] binder_alloc: Add missing mmap_lock calls when using the
 VMA
Message-ID: <YvLLQ/qasEJffdnn@google.com>
References: <20220809160618.1052539-1-Liam.Howlett@oracle.com>
 <YvKsI5pMbgQ5Irup@google.com>
 <20220809190211.ew4taam2lcomramw@revolver>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220809190211.ew4taam2lcomramw@revolver>
Precedence: bulk

On Tue, Aug 09, 2022 at 07:02:17PM +0000, Liam Howlett wrote:
> > 
> > do we need to hold on to the lock while we loop through the pages here?
> 
> I think we do?  Holding this lock will ensure the pages don't go away, I
> believe (looking at mm/rmap.c comments on locking at the top)?
> 
> In any case, this function is called from print_binder_proc_stats()
> which looks to be a debugfs/debugging call so I thought safer would be
> better than faster and with a potential race.

The pages are protected by alloc->mutex, so you could immediately
release the mmap lock after binder_alloc_get_vma() call. I agree this
is a debugging call so it would be nice to reduce contention.