From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751656AbcEKIkT (ORCPT ); Wed, 11 May 2016 04:40:19 -0400 Received: from mail-wm0-f52.google.com ([74.125.82.52]:37542 "EHLO mail-wm0-f52.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751364AbcEKIkQ (ORCPT ); Wed, 11 May 2016 04:40:16 -0400 Subject: Re: [PATCH v9 5/7] vfio/type1: also check IRQ remapping capability at msi domain To: Robin Murphy , Alex Williamson References: <1462362858-2925-1-git-send-email-eric.auger@linaro.org> <1462362858-2925-6-git-send-email-eric.auger@linaro.org> <20160509164942.5af8b30d@t450s.home> <57320803.8070506@linaro.org> <57321952.2020206@arm.com> Cc: eric.auger@st.com, will.deacon@arm.com, joro@8bytes.org, tglx@linutronix.de, jason@lakedaemon.net, marc.zyngier@arm.com, christoffer.dall@linaro.org, linux-arm-kernel@lists.infradead.org, patches@linaro.org, linux-kernel@vger.kernel.org, Bharat.Bhushan@freescale.com, pranav.sawargaonkar@gmail.com, p.fedin@samsung.com, iommu@lists.linux-foundation.org, Jean-Philippe.Brucker@arm.com, julien.grall@arm.com, yehuday@marvell.com From: Eric Auger Message-ID: <5732EF96.4070001@linaro.org> Date: Wed, 11 May 2016 10:38:46 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.6.0 MIME-Version: 1.0 In-Reply-To: <57321952.2020206@arm.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Robin, Alex, On 05/10/2016 07:24 PM, Robin Murphy wrote: > Hi Eric, > > On 10/05/16 17:10, Eric Auger wrote: >> Hi Alex, >> On 05/10/2016 12:49 AM, Alex Williamson wrote: >>> On Wed, 4 May 2016 11:54:16 +0000 >>> Eric Auger wrote: >>> >>>> On x86 IRQ remapping is abstracted by the IOMMU. On ARM this is >>>> abstracted >>>> by the msi controller. vfio_safe_irq_domain allows to check whether >>>> interrupts are "safe" for a given device. They are if the device does >>>> not use MSI >>> >>> Are we sure we're not opening a security hole here? An MSI is simply a >>> DMA write, so really whether or not a device uses MSI is irrelevant. >>> If it can generate a DMA to the MSI doorbell then we need to be >>> protected and I think we pretty much need to assume that devices are >>> DMA capable. Do the MSI domain checks cover this? >> Let me try to rephrase: we check the device is not attached to an MSI >> controller (I think this is the semantic of dev_get_msi_domain(dev)). >> >> If it is not, we don't have to care about MSI isolation: there will be >> no IOMMU binding between the device and any MSI doorbell. If it is we >> check the msi domain is backed by an MSI controller able to perform MSI >> isolation. >> >> So effectively "usage of MSIs" is improper - since it is decided after >> the group attachment anyway - and the commit message should rather >> state "if the device is linked to an MSI controller" (dt msi-parent >> notion I think). > > Hmm, I think Alex has a point here - on a GICv2m I can happily fire > arbitrary MSIs from _a shell_ (using /dev/mem), and the CPUs definitely > aren't in an MSI domain, so I don't think it's valid to assume that a > device using only wired interrupts, therefore with no connection to any > MSI controller, isn't still capable of maliciously spewing DMA all over > any and every doorbell region in the system. Sorry but I still don't get the point. For the device to reach the doorbell there must be an IOMMU mapping. - if the device is not attached to an MSI domain, there won't be any doorbell iommu mapping built by this series, so no risk, right? The device will be allowed to reach only memory iommu mapped by userspace with VFIO DMA MAP standard API. Of course if the userspace can mmap all the host PA that's a more general issue, right? - If the device is attached to an MSI domain (msi-parent link), 2 cases: 1) the MSI controller advertises MSI isolation (ITS cases), no risk 2) the MSI controller does not advertise MSI isolation (GICv2m), there is a security hole. a) by default we reject the device attachment b) if the userspace overrides the safe interrupt option he accepts the security hole What am I missing? Best Regards Eric > > Robin. > >> Does it sound better? >> >> Regards >> >> Eric >>> >>>> or if the device uses MSI and the msi-parent controller >>>> supports IRQ remapping. >>>> >>>> Then we check at group level if all devices have safe interrupts: if >>>> not, >>>> we only allow the group to be attached if allow_unsafe_interrupts is >>>> set. >>>> >>>> At this point ARM sMMU still advertises IOMMU_CAP_INTR_REMAP. This is >>>> changed in next patch. >>>> >>>> Signed-off-by: Eric Auger >>>> >>>> --- >>>> v3 -> v4: >>>> - rename vfio_msi_parent_irq_remapping_capable into >>>> vfio_safe_irq_domain >>>> and irq_remapping into safe_irq_domains >>>> >>>> v2 -> v3: >>>> - protect vfio_msi_parent_irq_remapping_capable with >>>> CONFIG_GENERIC_MSI_IRQ_DOMAIN >>>> --- >>>> drivers/vfio/vfio_iommu_type1.c | 44 >>>> +++++++++++++++++++++++++++++++++++++++-- >>>> 1 file changed, 42 insertions(+), 2 deletions(-) >>>> >>>> diff --git a/drivers/vfio/vfio_iommu_type1.c >>>> b/drivers/vfio/vfio_iommu_type1.c >>>> index 4d3a6f1..2fc8197 100644 >>>> --- a/drivers/vfio/vfio_iommu_type1.c >>>> +++ b/drivers/vfio/vfio_iommu_type1.c >>>> @@ -37,6 +37,8 @@ >>>> #include >>>> #include >>>> #include >>>> +#include >>>> +#include >>>> >>>> #define DRIVER_VERSION "0.2" >>>> #define DRIVER_AUTHOR "Alex Williamson >>>> " >>>> @@ -777,6 +779,33 @@ static int vfio_bus_type(struct device *dev, >>>> void *data) >>>> return 0; >>>> } >>>> >>>> +/** >>>> + * vfio_safe_irq_domain: returns whether the irq domain >>>> + * the device is attached to is safe with respect to MSI isolation. >>>> + * If the irq domain is not an MSI domain, we return it is safe. >>>> + * >>>> + * @dev: device handle >>>> + * @data: unused >>>> + * returns 0 if the irq domain is safe, -1 if not. >>>> + */ >>>> +static int vfio_safe_irq_domain(struct device *dev, void *data) >>>> +{ >>>> +#ifdef CONFIG_GENERIC_MSI_IRQ_DOMAIN >>>> + struct irq_domain *domain; >>>> + struct msi_domain_info *info; >>>> + >>>> + domain = dev_get_msi_domain(dev); >>>> + if (!domain) >>>> + return 0; >>>> + >>>> + info = msi_get_domain_info(domain); >>>> + >>>> + if (!(info->flags & MSI_FLAG_IRQ_REMAPPING)) >>>> + return -1; >>>> +#endif >>>> + return 0; >>>> +} >>>> + >>>> static int vfio_iommu_replay(struct vfio_iommu *iommu, >>>> struct vfio_domain *domain) >>>> { >>>> @@ -870,7 +899,7 @@ static int vfio_iommu_type1_attach_group(void >>>> *iommu_data, >>>> struct vfio_group *group, *g; >>>> struct vfio_domain *domain, *d; >>>> struct bus_type *bus = NULL; >>>> - int ret; >>>> + int ret, safe_irq_domains; >>>> >>>> mutex_lock(&iommu->lock); >>>> >>>> @@ -893,6 +922,13 @@ static int vfio_iommu_type1_attach_group(void >>>> *iommu_data, >>>> >>>> group->iommu_group = iommu_group; >>>> >>>> + /* >>>> + * Determine if all the devices of the group have a safe irq >>>> domain >>>> + * with respect to MSI isolation >>>> + */ >>>> + safe_irq_domains = !iommu_group_for_each_dev(iommu_group, &bus, >>>> + vfio_safe_irq_domain); >>>> + >>>> /* Determine bus_type in order to allocate a domain */ >>>> ret = iommu_group_for_each_dev(iommu_group, &bus, vfio_bus_type); >>>> if (ret) >>>> @@ -920,8 +956,12 @@ static int vfio_iommu_type1_attach_group(void >>>> *iommu_data, >>>> INIT_LIST_HEAD(&domain->group_list); >>>> list_add(&group->next, &domain->group_list); >>>> >>>> + /* >>>> + * to advertise safe interrupts either the IOMMU or the MSI >>>> controllers >>>> + * must support IRQ remapping/interrupt translation >>>> + */ >>>> if (!allow_unsafe_interrupts && >>>> - !iommu_capable(bus, IOMMU_CAP_INTR_REMAP)) { >>>> + (!iommu_capable(bus, IOMMU_CAP_INTR_REMAP) && >>>> !safe_irq_domains)) { >>>> pr_warn("%s: No interrupt remapping support. Use the >>>> module param \"allow_unsafe_interrupts\" to enable VFIO IOMMU >>>> support on this platform\n", >>>> __func__); >>>> ret = -EPERM; >>> >> >