From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Yigit, Ferruh" Subject: Re: VFIO no-iommu Date: Fri, 18 Dec 2015 10:43:10 +0000 Message-ID: <20151218104310.GA11371@sivlogin002.ir.intel.com> References: <566B4A50.9090607@6wind.com> <1449874953.20509.6.camel@redhat.com> <26FA93C7ED1EAA44AB77D62FBE1D27BA6747CE55@IRSMSX108.ger.corp.intel.com> <1450198398.6042.32.camel@redhat.com> <20151216040408.GA18363@sivlogin002.ir.intel.com> <1450240711.2674.11.camel@redhat.com> <1450285912.2674.22.camel@redhat.com> <1450370639.2674.93.camel@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Cc: "dev@dpdk.org" To: Alex Williamson Return-path: Received: from mga03.intel.com (mga03.intel.com [134.134.136.65]) by dpdk.org (Postfix) with ESMTP id B697A8D3A for ; Fri, 18 Dec 2015 11:43:14 +0100 (CET) Content-Disposition: inline In-Reply-To: <1450370639.2674.93.camel@redhat.com> List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Thu, Dec 17, 2015 at 09:43:59AM -0700, Alex Williamson wrote: <...> > > > > > >=20 > > > > > > Also I need to disable VFIO_CHECK_EXTENSION ioctl, because in > > > > > > vfio > > > > > > module, > > > > > > container->noiommu is not set before doing a > > > > > > vfio_group_set_container() > > > > > > and vfio_for_each_iommu_driver selects wrong driver. > > > > >=20 > > > > > Running CHECK_EXTENSION on a container without the group > > > > > attached is > > > > > only going to tell you what extensions vfio is capable of, not > > > > > necessarily what extensions are available to you with that > > > > > group. > > > > > Is this just a general dpdk- vfio ordering bug? > > > >=20 > > > > Yes, that is how VFIO was implemented in DPDK. I was under the > > > > impression that checking extension before assigning devices was > > > > the > > > > correct way to do things, so as to not to try anything we know > > > > would > > > > fail anyway. Does this imply that CHECK_EXTENSION needs to be > > > > called > > > > on both container and groups (or just on groups)? > > >=20 > > > Hmm, in Documentation/vfio.txt we do give the following algorithm: > > >=20 > > > =A0=A0=A0=A0=A0=A0=A0=A0if (ioctl(container, VFIO_GET_API_VERSION) = !=3D > > > VFIO_API_VERSION) > > > =A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0/* Unknown API vers= ion */ > > >=20 > > > =A0=A0=A0=A0=A0=A0=A0=A0if (!ioctl(container, VFIO_CHECK_EXTENSION, > > > VFIO_TYPE1_IOMMU)) > > > =A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0/* Doesn't support = the IOMMU driver we want. */ > > > =A0=A0=A0=A0=A0=A0=A0=A0... > > >=20 > > > That's just going to query each iommu driver and we can't yet say > > > whether > > > the group the user attaches to the container later will actually > > > support that > > > extension until we try to do it, that would come at VFIO_SET_IOMMU. > > > =A0So is > > > it perhaps a vfio bug that we're not advertising no-iommu until the > > > group is > > > attached? =A0After all, we are capable of it with just an empty > > > container, just > > > like we are with type1, but we're going to fail SET_IOMMU for the > > > wrong > > > combination. > > > =A0This is exactly the sort of thing that makes me glad we reverted > > > it without > > > feedback from a working user driver. =A0Thanks, > >=20 > > Whether it should be considered a "bug" in VFIO or "by design" is up > > to you, of course, but at least according to the VFIO documentation, > > we are meant to check for type 1 extension and then attach devices, > > so it would be expected to get VFIO_NOIOMMU_IOMMU marked as supported > > even without any devices attached to the container (just like we get > > type 1 as supported without any devices attached). Having said that, > > if it was meant to attach devices first and then check the > > extensions, then perhaps the documentation should also point out that > > fact (or perhaps I missed that detail in my readings of the docs, in > > which case my apologies). >=20 > Hi Anatoly, >=20 > Does the below patch make it behave more like you'd expect. =A0This > applies to v4.4-rc4, I'd fold this into the base patch if we > reincorporate it to a future kernel. =A0Thanks, >=20 > Alex >=20 > commit 88d4dcb6b77624965f0b45b5cd305a2b4a105c94 > Author: Alex Williamson > Date:=A0=A0=A0Wed Dec 16 19:02:01 2015 -0700 >=20 > =A0=A0=A0=A0vfio: Fix no-iommu CHECK_EXTENSION > =A0=A0=A0=A0 > =A0=A0=A0=A0Previously the no-iommu iommu driver was only visible when = the > =A0=A0=A0=A0container had an attached no-iommu group.=A0=A0This means t= hat > =A0=A0=A0=A0CHECK_EXTENSION on and empty container couldn't report the = possibility > =A0=A0=A0=A0of using VFIO_NOIOMMU_IOMMU.=A0=A0We report TYPE1 whether o= r not the user > =A0=A0=A0=A0can make use of it with the group, so this is inconsistent.= =A0=A0Add the > =A0=A0=A0=A0no-iommu iommu to the list of iommu drivers when enabled vi= a module > =A0=A0=A0=A0option, but skip all the others if the container is attache= d to a > =A0=A0=A0=A0no-iommu groups.=A0=A0Note that tainting is now done with t= he "unsafe" > =A0=A0=A0=A0module callback rather than explictly within vfio. > =A0=A0=A0=A0 > =A0=A0=A0=A0Also fixes module option and module description name incons= istency. > =A0=A0=A0=A0 > =A0=A0=A0=A0Also make vfio_noiommu_ops const. > =A0=A0=A0=A0 > =A0=A0=A0=A0Signed-off-by: Alex Williamson Hi Alex, I got following crash with this update: [ +0.046973] BUG: unable to handle kernel NULL pointer dereference at = (null) [ +0.000031] IP: [] __list_add+V0x1f/0xc0 [ +0.000024] PGD 0=20 [ +0.000010] Oops: 0000 [#1] SMP=20 ... [ +0.000028] Call Trace: [ +0.000014] [] __mutex_lock_slowpath+0x91/0x110 [ +0.000022] [] mutex_lock+0x23/0x40 [ +0.000020] [] vfio_register_iommu_driver+0x40/0xc0 = [vfio] [ +0.000025] [] noiommu_param_set+0x51/0x60 [vfio] [ +0.000022] [] parse_args+0x1be/0x4a0 [ +0.000021] [] load_module+0xe30/0x2730 [ +0.000942] [] ? __symbol_put+0x60/0x60 [ +0.000921] [] ? kernel_read+0x50/0x80 [ +0.000917] [] SyS_finit_module+0xb9/0xf0 [ +0.000925] [] entry_SYSCALL_64_fastpath+0x12/0x71 > -static struct vfio_iommu_driver vfio_noiommu_driver =3D { > - .ops =3D &vfio_noiommu_ops, > +static int noiommu_param_set(const char *val, const struct kernel_para= m *kp) > +{ > + int ret; > + > + if (!val) > + val =3D "1"; > + > + ret =3D strtobool(val, kp->arg); > + if (ret) > + return ret; > + > + if (noiommu) > + ret =3D vfio_register_iommu_driver(&vfio_noiommu_ops); Is it possible that kernel_param_ops->set() called before module_init() t= hat initializes the mutex. https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/kern= el/module.c?id=3Drefs/tags/v4.4-rc5#n3517 > + else > + vfio_unregister_iommu_driver(&vfio_noiommu_ops); > + > + return ret; > +} > + > +static const struct kernel_param_ops noiommu_param_ops =3D { > + .flags =3D KERNEL_PARAM_OPS_FL_NOARG, > + .set =3D noiommu_param_set, > + .get =3D param_get_bool, > =A0}; > =A0 Thanks, ferruh