linux-kernel - [PATCH v2 08/15] dma-iommu: Add iommu_reconfig_dev_group

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1620656249-68890-9-git-send-email-john.garry@huawei.com>
Date:   Mon, 10 May 2021 22:17:22 +0800
From:   John Garry <john.garry@...wei.com>
To:     <joro@...tes.org>, <will@...nel.org>, <robin.murphy@....com>,
        <hch@....de>, <m.szyprowski@...sung.com>
CC:     <iommu@...ts.linux-foundation.org>, <baolu.lu@...ux.intel.com>,
        <linux-kernel@...r.kernel.org>, <linux-scsi@...r.kernel.org>,
        <linuxarm@...wei.com>, <sai.praneeth.prakhya@...el.com>,
        John Garry <john.garry@...wei.com>
Subject: [PATCH v2 08/15] dma-iommu: Add iommu_reconfig_dev_group_dma()

Add a function to reconfigure the IOMMU group for a device, if necessary.

IOVAs are cached in power-of-2 granules, so there is no point in allocating
a new IOMMU domain if the current range is suitable.

Signed-off-by: John Garry <john.garry@...wei.com>
---
 drivers/iommu/dma-iommu.c | 25 +++++++++++++++++++++++++
 include/linux/dma-iommu.h |  4 ++++
 2 files changed, 29 insertions(+)

diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
index f6d3302bb829..4fb82c554ede 100644
--- a/drivers/iommu/dma-iommu.c
+++ b/drivers/iommu/dma-iommu.c
@@ -315,6 +315,31 @@ static bool dev_is_untrusted(struct device *dev)
 	return dev_is_pci(dev) && to_pci_dev(dev)->untrusted;
 }
 
+void iommu_reconfig_dev_group_dma(struct device *dev)
+{
+	struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
+	struct iommu_dma_cookie *cookie = domain->iova_cookie;
+	unsigned long shift, iova_len;
+	struct iova_domain *iovad;
+	size_t max_opt_dma_size;
+
+	if (!cookie || cookie->type != IOMMU_DMA_IOVA_COOKIE)
+		return;
+
+	max_opt_dma_size = iommu_group_get_max_opt_dma_size(dev->iommu_group);
+	if (!max_opt_dma_size)
+		return;
+
+	iovad = &cookie->iovad;
+	shift = iova_shift(iovad);
+	iova_len = max_opt_dma_size >> shift;
+
+	if (iova_domain_len_is_cached(iovad, iova_len))
+		return;
+
+	iommu_realloc_dev_group(dev);
+}
+
 /**
  * iommu_dma_init_domain - Initialise a DMA mapping domain
  * @domain: IOMMU domain previously prepared by iommu_get_dma_cookie()
diff --git a/include/linux/dma-iommu.h b/include/linux/dma-iommu.h
index 6e75a2d689b4..097398b76dcc 100644
--- a/include/linux/dma-iommu.h
+++ b/include/linux/dma-iommu.h
@@ -20,6 +20,7 @@ void iommu_put_dma_cookie(struct iommu_domain *domain);
 
 /* Setup call for arch DMA mapping code */
 void iommu_setup_dma_ops(struct device *dev, u64 dma_base, u64 size);
+void iommu_reconfig_dev_group_dma(struct device *dev);
 
 /* The DMA API isn't _quite_ the whole story, though... */
 /*
@@ -53,6 +54,9 @@ static inline void iommu_setup_dma_ops(struct device *dev, u64 dma_base,
 		u64 size)
 {
 }
+static inline void iommu_reconfig_dev_group_dma(struct device *dev)
+{
+}
 
 static inline int iommu_get_dma_cookie(struct iommu_domain *domain)
 {
-- 
2.26.2