From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:41479) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YXPfH-0007Rd-Sq for qemu-devel@nongnu.org; Mon, 16 Mar 2015 03:42:20 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YXPfC-0007fo-PV for qemu-devel@nongnu.org; Mon, 16 Mar 2015 03:42:19 -0400 Received: from mx1.redhat.com ([209.132.183.28]:48241) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YXPfC-0007fb-KV for qemu-devel@nongnu.org; Mon, 16 Mar 2015 03:42:14 -0400 Received: from int-mx13.intmail.prod.int.phx2.redhat.com (int-mx13.intmail.prod.int.phx2.redhat.com [10.5.11.26]) by mx1.redhat.com (Postfix) with ESMTPS id E9419C1F78 for ; Mon, 16 Mar 2015 07:42:13 +0000 (UTC) Date: Mon, 16 Mar 2015 15:42:11 +0800 From: Fam Zheng Message-ID: <20150316074211.GC15098@ad.nay.redhat.com> References: <1426483910-24597-1-git-send-email-famz@redhat.com> <1426483910-24597-2-git-send-email-famz@redhat.com> <55068687.1020304@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <55068687.1020304@redhat.com> Subject: Re: [Qemu-devel] [PATCH v3 1/4] exec: Atomic access to bounce buffer List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Paolo Bonzini Cc: qemu-devel@nongnu.org On Mon, 03/16 08:30, Paolo Bonzini wrote: > > > On 16/03/2015 06:31, Fam Zheng wrote: > > There could be a race condition when two processes call > > address_space_map concurrently and both want to use the bounce buffer. > > > > Add an in_use flag in BounceBuffer to sync it. > > > > Signed-off-by: Fam Zheng > > --- > > exec.c | 5 ++++- > > 1 file changed, 4 insertions(+), 1 deletion(-) > > > > diff --git a/exec.c b/exec.c > > index e97071a..4080044 100644 > > --- a/exec.c > > +++ b/exec.c > > @@ -2483,6 +2483,7 @@ typedef struct { > > void *buffer; > > hwaddr addr; > > hwaddr len; > > + bool in_use; > > } BounceBuffer; > > > > static BounceBuffer bounce; > > @@ -2571,9 +2572,10 @@ void *address_space_map(AddressSpace *as, > > l = len; > > mr = address_space_translate(as, addr, &xlat, &l, is_write); > > if (!memory_access_is_direct(mr, is_write)) { > > - if (bounce.buffer) { > > + if (atomic_xchg(&bounce.in_use, true)) { > > return NULL; > > } > > + smp_mb(); > > smp_mb() not needed. OK, I was confused by the Linux documentation on atomic_xchg. Now I've looked at the right places, it is not needed. Thanks, Fam > > Ok with this change. > > Paolo > > > /* Avoid unbounded allocations */ > > l = MIN(l, TARGET_PAGE_SIZE); > > bounce.buffer = qemu_memalign(TARGET_PAGE_SIZE, l); > > @@ -2641,6 +2643,7 @@ void address_space_unmap(AddressSpace *as, void *buffer, hwaddr len, > > qemu_vfree(bounce.buffer); > > bounce.buffer = NULL; > > memory_region_unref(bounce.mr); > > + atomic_mb_set(&bounce.in_use, false); > > cpu_notify_map_clients(); > > } > > > >