linux-kernel - [PATCH 5/5] iommu/amd: Do not flush Device Table in iommu_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]

Date:   Mon,  4 May 2020 14:54:13 +0200
From:   Joerg Roedel <joro@...tes.org>
To:     iommu@...ts.linux-foundation.org
Cc:     Qian Cai <cai@....pw>,
        Suravee Suthikulpanit <suravee.suthikulpanit@....com>,
        linux-kernel@...r.kernel.org, Joerg Roedel <jroedel@...e.de>
Subject: [PATCH 5/5] iommu/amd: Do not flush Device Table in iommu_map_page()

From: Joerg Roedel <jroedel@...e.de>

The flush of the Device Table Entries for the domain has already
happened in increase_address_space(), if necessary. Do no flush them
again in iommu_map_page().

Tested-by: Qian Cai <cai@....pw>
Signed-off-by: Joerg Roedel <jroedel@...e.de>
---
 drivers/iommu/amd_iommu.c | 27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
index 2ae1daac888a..1dc3718560d0 100644
--- a/drivers/iommu/amd_iommu.c
+++ b/drivers/iommu/amd_iommu.c
@@ -1446,15 +1446,18 @@ static bool increase_address_space(struct protection_domain *domain,
 {
 	struct domain_pgtable pgtable;
 	unsigned long flags;
-	bool ret = false;
+	bool ret = true;
 	u64 *pte, root;
 
 	spin_lock_irqsave(&domain->lock, flags);
 
 	amd_iommu_domain_get_pgtable(domain, &pgtable);
 
-	if (address <= PM_LEVEL_SIZE(pgtable.mode) ||
-	    WARN_ON_ONCE(pgtable.mode == PAGE_MODE_6_LEVEL))
+	if (address <= PM_LEVEL_SIZE(pgtable.mode))
+		goto out;
+
+	ret = false;
+	if (WARN_ON_ONCE(pgtable.mode == PAGE_MODE_6_LEVEL))
 		goto out;
 
 	pte = (void *)get_zeroed_page(gfp);
@@ -1499,19 +1502,15 @@ static u64 *alloc_pte(struct protection_domain *domain,
 	amd_iommu_domain_get_pgtable(domain, &pgtable);
 
 	while (address > PM_LEVEL_SIZE(pgtable.mode)) {
-		bool upd = increase_address_space(domain, address, gfp);
-
-		/* Read new values to check if update was successful */
-		amd_iommu_domain_get_pgtable(domain, &pgtable);
-
 		/*
 		 * Return an error if there is no memory to update the
 		 * page-table.
 		 */
-		if (!upd && (address > PM_LEVEL_SIZE(pgtable.mode)))
+		if (!increase_address_space(domain, address, gfp))
 			return NULL;
 
-		*updated = *updated || upd;
+		/* Read new values to check if update was successful */
+		amd_iommu_domain_get_pgtable(domain, &pgtable);
 	}
 
 
@@ -1719,7 +1718,13 @@ static int iommu_map_page(struct protection_domain *dom,
 		unsigned long flags;
 
 		spin_lock_irqsave(&dom->lock, flags);
-		update_domain(dom);
+		/*
+		 * Flush domain TLB(s) and wait for completion. Any Device-Table
+		 * Updates and flushing already happened in
+		 * increase_address_space().
+		 */
+		domain_flush_tlb_pde(dom);
+		domain_flush_complete(dom);
 		spin_unlock_irqrestore(&dom->lock, flags);
 	}
 
-- 
2.17.1