linux-kernel - [PATCH 2/5] iommu/amd: Hold the domain lock when calling __map

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: <1568137765-20278-3-git-send-email-sironi@amazon.de>
Date:   Tue, 10 Sep 2019 19:49:22 +0200
From:   Filippo Sironi <sironi@...zon.de>
To:     sironi@...zon.de, joro@...tes.org,
        iommu@...ts.linux-foundation.org, linux-kernel@...r.kernel.org
Subject: [PATCH 2/5] iommu/amd: Hold the domain lock when calling __map_single

__map_single makes several calls to __domain_flush_pages, which
traverses the device list that is protected by the domain lock.

Also, this is in line with the comment on top of __map_single, which
says that the domain lock should be held when calling.

Signed-off-by: Filippo Sironi <sironi@...zon.de>
---
 drivers/iommu/amd_iommu.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
index f026a8c2b218..8e3664821b3c 100644
--- a/drivers/iommu/amd_iommu.c
+++ b/drivers/iommu/amd_iommu.c
@@ -2482,6 +2482,8 @@ static dma_addr_t map_page(struct device *dev, struct page *page,
 	struct protection_domain *domain;
 	struct dma_ops_domain *dma_dom;
 	u64 dma_mask;
+	unsigned long flags;
+	dma_addr_t dma_addr;
 
 	domain = get_domain(dev);
 	if (PTR_ERR(domain) == -EINVAL)
@@ -2492,7 +2494,10 @@ static dma_addr_t map_page(struct device *dev, struct page *page,
 	dma_mask = *dev->dma_mask;
 	dma_dom = to_dma_ops_domain(domain);
 
-	return __map_single(dev, dma_dom, paddr, size, dir, dma_mask);
+	spin_lock_irqsave(&domain->lock, flags);
+	dma_addr = __map_single(dev, dma_dom, paddr, size, dir, dma_mask);
+	spin_unlock_irqrestore(&domain->lock, flags);
+	return dma_addr;
 }
 
 /*
@@ -2663,6 +2668,7 @@ static void *alloc_coherent(struct device *dev, size_t size,
 	struct protection_domain *domain;
 	struct dma_ops_domain *dma_dom;
 	struct page *page;
+	unsigned long flags;
 
 	domain = get_domain(dev);
 	if (PTR_ERR(domain) == -EINVAL) {
@@ -2692,8 +2698,10 @@ static void *alloc_coherent(struct device *dev, size_t size,
 	if (!dma_mask)
 		dma_mask = *dev->dma_mask;
 
+	spin_lock_irqsave(&domain->lock, flags);
 	*dma_addr = __map_single(dev, dma_dom, page_to_phys(page),
 				 size, DMA_BIDIRECTIONAL, dma_mask);
+	spin_unlock_irqrestore(&domain->lock, flags);
 
 	if (*dma_addr == DMA_MAPPING_ERROR)
 		goto out_free;
-- 
2.7.4