From: Robin Murphy <robin.murphy-5wv7dgnIgG8@public.gmane.org> To: Marek Szyprowski <m.szyprowski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>, will.deacon-5wv7dgnIgG8@public.gmane.org, joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org, iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org Cc: devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, punit.agrawal-5wv7dgnIgG8@public.gmane.org, Inki Dae <inki.dae-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>, thunder.leizhen-hv44wF8Li93QT0dZR+AlfA@public.gmane.org Subject: Re: [PATCH v7 22/22] iommu/dma: Avoid PCI host bridge windows Date: Wed, 14 Sep 2016 14:25:48 +0100 [thread overview] Message-ID: <bbdc42fa-ea35-945f-3e2a-e0ab03fc997d@arm.com> (raw) In-Reply-To: <dc9f945f-2756-ab70-d061-9fdc7c5afdee-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> On 14/09/16 13:35, Marek Szyprowski wrote: > Hi Robin, > > > On 2016-09-14 13:10, Robin Murphy wrote: >> On 14/09/16 11:55, Marek Szyprowski wrote: >>> On 2016-09-12 18:14, Robin Murphy wrote: >>>> With our DMA ops enabled for PCI devices, we should avoid allocating >>>> IOVAs which a host bridge might misinterpret as peer-to-peer DMA and >>>> lead to faults, corruption or other badness. To be safe, punch out >>>> holes >>>> for all of the relevant host bridge's windows when initialising a DMA >>>> domain for a PCI device. >>>> >>>> CC: Marek Szyprowski <m.szyprowski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> >>>> CC: Inki Dae <inki.dae-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> >>>> Reported-by: Lorenzo Pieralisi <lorenzo.pieralisi-5wv7dgnIgG8@public.gmane.org> >>>> Signed-off-by: Robin Murphy <robin.murphy-5wv7dgnIgG8@public.gmane.org> >>> I don't know much about PCI and their IOMMU integration, but can't we >>> use >>> the direct mapping region feature of iommu core for it? There are >>> already >>> iommu_get_dm_regions(), iommu_put_dm_regions() and >>> iommu_request_dm_for_dev() >>> functions for handling them... >> It's rather the opposite problem - in the direct-mapping case, we're >> making sure the iommu_domain has translations installed for the given >> IOVAs (which are also the corresponding physical address) before it goes >> live, whereas what we need to do here is make sure the these addresses >> never get used as IOVAs at all, because any attempt to do so them will >> likely go wrong. Thus we carve them out of the iova_domain such that >> they will never get near an actual IOMMU API call. >> >> This is a slightly generalised equivalent of e.g. amd_iommu.c's >> init_reserved_iova_ranges(). > > Hmmm. Each dm_region have protection parameter. Can't we reuse them to > create prohibited/reserved regions by setting it to 0 (no read / no write) > and mapping to physical 0 address? That's just a quick idea, because > dm_regions and the proposed code for pci looks a bit similar for me... It might look similar, but at different levels (iommu_domain vs. iova_domain) and with the opposite intent. The dm_region prot flag is just the standard flag as passed to iommu_map() - trying to map a region with no access in an empty pagetable isn't going to achieve anything anyway (it's effectively unmapping addresses that are already unmapped). But for this case, even if you _did_ map something in the pagetable (i.e. the iommu_domain), it wouldn't make any difference, because the thing we're mitigating against is handing out addresses which are going to cause a device's accesses to blow up inside the PCI root complex without ever even reaching the IOMMU. In short, dm_regions are about "these addresses are already being used for DMA, so make sure the IOMMU API doesn't block them", whereas reserved ranges are about "these addresses are unusable for DMA, so make sure the DMA API can't allocate them". Robin. > > [...] > > Best regards
WARNING: multiple messages have this Message-ID (diff)
From: robin.murphy@arm.com (Robin Murphy) To: linux-arm-kernel@lists.infradead.org Subject: [PATCH v7 22/22] iommu/dma: Avoid PCI host bridge windows Date: Wed, 14 Sep 2016 14:25:48 +0100 [thread overview] Message-ID: <bbdc42fa-ea35-945f-3e2a-e0ab03fc997d@arm.com> (raw) In-Reply-To: <dc9f945f-2756-ab70-d061-9fdc7c5afdee@samsung.com> On 14/09/16 13:35, Marek Szyprowski wrote: > Hi Robin, > > > On 2016-09-14 13:10, Robin Murphy wrote: >> On 14/09/16 11:55, Marek Szyprowski wrote: >>> On 2016-09-12 18:14, Robin Murphy wrote: >>>> With our DMA ops enabled for PCI devices, we should avoid allocating >>>> IOVAs which a host bridge might misinterpret as peer-to-peer DMA and >>>> lead to faults, corruption or other badness. To be safe, punch out >>>> holes >>>> for all of the relevant host bridge's windows when initialising a DMA >>>> domain for a PCI device. >>>> >>>> CC: Marek Szyprowski <m.szyprowski@samsung.com> >>>> CC: Inki Dae <inki.dae@samsung.com> >>>> Reported-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> >>>> Signed-off-by: Robin Murphy <robin.murphy@arm.com> >>> I don't know much about PCI and their IOMMU integration, but can't we >>> use >>> the direct mapping region feature of iommu core for it? There are >>> already >>> iommu_get_dm_regions(), iommu_put_dm_regions() and >>> iommu_request_dm_for_dev() >>> functions for handling them... >> It's rather the opposite problem - in the direct-mapping case, we're >> making sure the iommu_domain has translations installed for the given >> IOVAs (which are also the corresponding physical address) before it goes >> live, whereas what we need to do here is make sure the these addresses >> never get used as IOVAs at all, because any attempt to do so them will >> likely go wrong. Thus we carve them out of the iova_domain such that >> they will never get near an actual IOMMU API call. >> >> This is a slightly generalised equivalent of e.g. amd_iommu.c's >> init_reserved_iova_ranges(). > > Hmmm. Each dm_region have protection parameter. Can't we reuse them to > create prohibited/reserved regions by setting it to 0 (no read / no write) > and mapping to physical 0 address? That's just a quick idea, because > dm_regions and the proposed code for pci looks a bit similar for me... It might look similar, but at different levels (iommu_domain vs. iova_domain) and with the opposite intent. The dm_region prot flag is just the standard flag as passed to iommu_map() - trying to map a region with no access in an empty pagetable isn't going to achieve anything anyway (it's effectively unmapping addresses that are already unmapped). But for this case, even if you _did_ map something in the pagetable (i.e. the iommu_domain), it wouldn't make any difference, because the thing we're mitigating against is handing out addresses which are going to cause a device's accesses to blow up inside the PCI root complex without ever even reaching the IOMMU. In short, dm_regions are about "these addresses are already being used for DMA, so make sure the IOMMU API doesn't block them", whereas reserved ranges are about "these addresses are unusable for DMA, so make sure the DMA API can't allocate them". Robin. > > [...] > > Best regards
next prev parent reply other threads:[~2016-09-14 13:25 UTC|newest] Thread overview: 104+ messages / expand[flat|nested] mbox.gz Atom feed top 2016-09-12 16:13 [PATCH v7 00/22] Generic DT bindings for PCI IOMMUs and ARM SMMU Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <cover.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-12 16:13 ` [PATCH v7 01/22] Docs: dt: add PCI IOMMU map bindings Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 02/22] of/irq: Break out msi-map lookup (again) Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 03/22] iommu/of: Handle iommu-map property for PCI Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 04/22] iommu: Introduce iommu_fwspec Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <742a71630de502ac5a7a8641c6ed368d8409324d.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-13 9:54 ` [PATCH v7.1 " Robin Murphy 2016-09-13 9:54 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 05/22] Docs: dt: document ARM SMMUv3 generic binding usage Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <2273645f1fa5c76b6b98b5fd03804ab8b55a7691.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-20 14:46 ` Rob Herring 2016-09-20 14:46 ` Rob Herring 2016-09-12 16:13 ` [PATCH v7 06/22] iommu/arm-smmu: Fall back to global bypass Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 07/22] iommu/arm-smmu: Implement of_xlate() for SMMUv3 Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 08/22] iommu/arm-smmu: Support non-PCI devices with SMMUv3 Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 09/22] iommu/arm-smmu: Set PRIVCFG in stage 1 STEs Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 10/22] iommu/arm-smmu: Handle stream IDs more dynamically Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 11/22] iommu/arm-smmu: Consolidate stream map entry state Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 12/22] iommu/arm-smmu: Keep track of S2CR state Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 13/22] iommu/arm-smmu: Refactor mmu-masters handling Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <046d2d21f988d6ece916fc45b0af0804a7f200f2.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-14 14:21 ` [PATCH v7.1 " Robin Murphy 2016-09-14 14:21 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 14/22] iommu/arm-smmu: Streamline SMMU data lookups Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 15/22] iommu/arm-smmu: Add a stream map entry iterator Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 16/22] iommu/arm-smmu: Intelligent SMR allocation Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 17/22] iommu/arm-smmu: Convert to iommu_fwspec Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 18/22] Docs: dt: document ARM SMMU generic binding usage Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 19/22] iommu/arm-smmu: Wire up generic configuration support Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <228dc6c675f10ae7481640d4ef2f4960c170621f.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-14 14:26 ` [PATCH v7.1 " Robin Murphy 2016-09-14 14:26 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 20/22] iommu/arm-smmu: Set domain geometry Robin Murphy 2016-09-12 16:13 ` Robin Murphy 2016-09-12 16:13 ` [PATCH v7 21/22] iommu/dma: Add support for mapping MSIs Robin Murphy 2016-09-12 16:13 ` Robin Murphy [not found] ` <2273af20d844bd618c6a90b57e639700328ebf7f.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-10-05 7:00 ` Nipun Gupta 2016-10-05 7:00 ` Nipun Gupta [not found] ` <DB6PR0402MB2694B2E5AE266F138784D2C2E6C40-2mNvjAGDOPn2WJ5A9zev/o3W/0Ik+aLCnBOFsp37pqbUKgpGm//BTAC/G2K4zDHf@public.gmane.org> 2016-10-05 9:55 ` Robin Murphy 2016-10-05 9:55 ` Robin Murphy [not found] ` <6ec9519b-01df-3be8-2967-7556bd306909-5wv7dgnIgG8@public.gmane.org> 2016-10-05 11:31 ` Nipun Gupta 2016-10-05 11:31 ` Nipun Gupta 2016-09-12 16:14 ` [PATCH v7 22/22] iommu/dma: Avoid PCI host bridge windows Robin Murphy 2016-09-12 16:14 ` Robin Murphy [not found] ` <5f7bfee298f98d29a35933d3e0252d32b83d62b8.1473695704.git.robin.murphy-5wv7dgnIgG8@public.gmane.org> 2016-09-14 10:55 ` Marek Szyprowski 2016-09-14 10:55 ` Marek Szyprowski [not found] ` <ab8693f6-20d6-2a95-9f1f-0607e72bc012-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> 2016-09-14 11:10 ` Robin Murphy 2016-09-14 11:10 ` Robin Murphy [not found] ` <49c51c4f-cb00-445d-b8f8-b632babf2b3e-5wv7dgnIgG8@public.gmane.org> 2016-09-14 12:35 ` Marek Szyprowski 2016-09-14 12:35 ` Marek Szyprowski [not found] ` <dc9f945f-2756-ab70-d061-9fdc7c5afdee-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org> 2016-09-14 13:25 ` Robin Murphy [this message] 2016-09-14 13:25 ` Robin Murphy [not found] ` <bbdc42fa-ea35-945f-3e2a-e0ab03fc997d-5wv7dgnIgG8@public.gmane.org> 2016-09-15 7:08 ` Marek Szyprowski 2016-09-15 7:08 ` Marek Szyprowski 2016-09-13 12:14 ` [PATCH v7 00/22] Generic DT bindings for PCI IOMMUs and ARM SMMU Auger Eric 2016-09-13 12:14 ` Auger Eric [not found] ` <92f27a6b-9752-516d-3924-c552fc6a5ace-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-13 12:40 ` Robin Murphy 2016-09-13 12:40 ` Robin Murphy [not found] ` <e24821be-5cc4-52b3-f961-1eb32cf58293-5wv7dgnIgG8@public.gmane.org> 2016-09-13 12:57 ` Auger Eric 2016-09-13 12:57 ` Auger Eric 2016-09-14 8:41 ` Auger Eric 2016-09-14 8:41 ` Auger Eric [not found] ` <11ebd81e-2ea5-5ff3-35b3-be95f03e05bd-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-14 9:20 ` Will Deacon 2016-09-14 9:20 ` Will Deacon [not found] ` <20160914092051.GB19622-5wv7dgnIgG8@public.gmane.org> 2016-09-14 9:35 ` Auger Eric 2016-09-14 9:35 ` Auger Eric 2016-09-14 10:35 ` Robin Murphy 2016-09-14 10:35 ` Robin Murphy [not found] ` <d03ea5e7-59f1-8b49-4ba2-d05fc2030ebc-5wv7dgnIgG8@public.gmane.org> 2016-09-14 12:32 ` Auger Eric 2016-09-14 12:32 ` Auger Eric [not found] ` <04a0a682-4fdc-8d62-57cd-efdf730582c6-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-14 12:53 ` Robin Murphy 2016-09-14 12:53 ` Robin Murphy [not found] ` <c2645c5e-edd3-2b31-4311-0ca621a915e2-5wv7dgnIgG8@public.gmane.org> 2016-09-15 9:29 ` Auger Eric 2016-09-15 9:29 ` Auger Eric [not found] ` <4d87d5f2-0350-b5f8-ffc3-4e9377cf1f87-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-15 10:15 ` Robin Murphy 2016-09-15 10:15 ` Robin Murphy [not found] ` <fc4ce398-4eeb-f2ca-b964-e9f466be79c4-5wv7dgnIgG8@public.gmane.org> 2016-09-15 16:46 ` Auger Eric 2016-09-15 16:46 ` Auger Eric [not found] ` <1838c65d-5944-8946-781c-b420bea1acab-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-16 16:18 ` Robin Murphy 2016-09-16 16:18 ` Robin Murphy [not found] ` <f16db032-1905-9804-0607-fe007af72b0e-5wv7dgnIgG8@public.gmane.org> 2016-09-19 12:13 ` Auger Eric 2016-09-19 12:13 ` Auger Eric [not found] ` <48f3bc10-3966-7d50-d070-7ec7f0946c92-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 2016-09-19 12:24 ` Will Deacon 2016-09-19 12:24 ` Will Deacon [not found] ` <20160919122435.GD9005-5wv7dgnIgG8@public.gmane.org> 2016-09-19 12:41 ` Robin Murphy 2016-09-19 12:41 ` Robin Murphy [not found] ` <99ee0946-c7ff-e6e4-08c1-ff686ea1a8a5-5wv7dgnIgG8@public.gmane.org> 2016-09-19 14:17 ` Will Deacon 2016-09-19 14:17 ` Will Deacon
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bbdc42fa-ea35-945f-3e2a-e0ab03fc997d@arm.com \ --to=robin.murphy-5wv7dgnigg8@public.gmane.org \ --cc=devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \ --cc=inki.dae-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org \ --cc=iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org \ --cc=joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org \ --cc=linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org \ --cc=m.szyprowski-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org \ --cc=punit.agrawal-5wv7dgnIgG8@public.gmane.org \ --cc=thunder.leizhen-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \ --cc=will.deacon-5wv7dgnIgG8@public.gmane.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.