Received-SPF: pass (google.com: domain of linux-kernel+bounces-52553-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
From: James Gowans <jgowans@amazon.com>
To: <linux-kernel@vger.kernel.org>
CC: Eric Biederman <ebiederm@xmission.com>, <kexec@lists.infradead.org>,
	"Joerg Roedel" <joro@8bytes.org>, Will Deacon <will@kernel.org>,
	<iommu@lists.linux.dev>, Alexander Viro <viro@zeniv.linux.org.uk>, "Christian
 Brauner" <brauner@kernel.org>, <linux-fsdevel@vger.kernel.org>, Paolo Bonzini
	<pbonzini@redhat.com>, Sean Christopherson <seanjc@google.com>,
	<kvm@vger.kernel.org>, Andrew Morton <akpm@linux-foundation.org>,
	<linux-mm@kvack.org>, Alexander Graf <graf@amazon.com>, David Woodhouse
	<dwmw@amazon.co.uk>, "Jan H . Schoenherr" <jschoenh@amazon.de>, Usama Arif
	<usama.arif@bytedance.com>, Anthony Yznaga <anthony.yznaga@oracle.com>,
	Stanislav Kinsburskii <skinsburskii@linux.microsoft.com>,
	<madvenka@linux.microsoft.com>, <steven.sistare@oracle.com>,
	<yuleixzhang@tencent.com>
Subject: [RFC 10/18] iommu/intel: zap context table entries on kexec
Date: Mon, 5 Feb 2024 12:01:55 +0000
Message-ID: <20240205120203.60312-11-jgowans@amazon.com>
In-Reply-To: <20240205120203.60312-1-jgowans@amazon.com>
References: <20240205120203.60312-1-jgowans@amazon.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Type: text/plain

In the next commit the IOMMU shutdown function will be modified to not
actually shut down the IOMMU when doing a kexec. To prevent leaving DMA
mappings for non-persistent devices around during kexec we add a
function to the kexec flow which iterates though all IOMMU domains and
zaps the context entries for the devices belonging to those domain.

A list of domains for the IOMMU is added and maintained.
---
 drivers/iommu/intel/dmar.c  |  1 +
 drivers/iommu/intel/iommu.c | 34 ++++++++++++++++++++++++++++++----
 drivers/iommu/intel/iommu.h |  2 ++
 3 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/drivers/iommu/intel/dmar.c b/drivers/iommu/intel/dmar.c
index 23cb80d62a9a..00f69f40a4ac 100644
--- a/drivers/iommu/intel/dmar.c
+++ b/drivers/iommu/intel/dmar.c
@@ -1097,6 +1097,7 @@ static int alloc_iommu(struct dmar_drhd_unit *drhd)
 	iommu->segment = drhd->segment;
 
 	iommu->node = NUMA_NO_NODE;
+	INIT_LIST_HEAD(&iommu->domains);
 
 	ver = readl(iommu->reg + DMAR_VER_REG);
 	pr_info("%s: reg_base_addr %llx ver %d:%d cap %llx ecap %llx\n",
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 2dd3f055dbce..315c6b7f901c 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -1831,6 +1831,7 @@ static int domain_attach_iommu(struct dmar_domain *domain,
 		goto err_clear;
 	}
 	domain_update_iommu_cap(domain);
+	list_add(&domain->domains, &iommu->domains);
 
 	spin_unlock(&iommu->lock);
 	return 0;
@@ -3608,6 +3609,33 @@ static void intel_disable_iommus(void)
 		iommu_disable_translation(iommu);
 }
 
+void zap_context_table_entries(struct intel_iommu *iommu)
+{
+	struct context_entry *context;
+	struct dmar_domain *domain;
+	struct device_domain_info *device;
+	int bus, devfn;
+	u16 did_old;
+
+	list_for_each_entry(domain, &iommu->domains, domains) {
+		list_for_each_entry(device, &domain->devices, link) {
+			context = iommu_context_addr(iommu, device->bus, device->devfn, 0);
+			if (!context || !context_present(context))
+				continue;
+			context_domain_id(context);
+			context_clear_entry(context);
+			__iommu_flush_cache(iommu, context, sizeof(*context));
+			iommu->flush.flush_context(iommu,
+						   did_old,
+						   (((u16)bus) << 8) | devfn,
+						   DMA_CCMD_MASK_NOBIT,
+						   DMA_CCMD_DEVICE_INVL);
+			iommu->flush.flush_iotlb(iommu,	did_old, 0, 0,
+						 DMA_TLB_DSI_FLUSH);
+		}
+	}
+}
+
 void intel_iommu_shutdown(void)
 {
 	struct dmar_drhd_unit *drhd;
@@ -3620,10 +3648,8 @@ void intel_iommu_shutdown(void)
 
 	/* Disable PMRs explicitly here. */
 	for_each_iommu(iommu, drhd)
-		iommu_disable_protect_mem_regions(iommu);
-
-	/* Make sure the IOMMUs are switched off */
-	intel_disable_iommus();
+		zap_context_table_entries(iommu);
+	return
 
 	up_write(&dmar_global_lock);
 }
diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h
index a2338e398ba3..4a2f163a86f3 100644
--- a/drivers/iommu/intel/iommu.h
+++ b/drivers/iommu/intel/iommu.h
@@ -600,6 +600,7 @@ struct dmar_domain {
 	spinlock_t lock;		/* Protect device tracking lists */
 	struct list_head devices;	/* all devices' list */
 	struct list_head dev_pasids;	/* all attached pasids */
+	struct list_head domains;	/* all struct dmar_domains on this IOMMU */
 
 	struct dma_pte	*pgd;		/* virtual address */
 	int		gaw;		/* max guest address width */
@@ -700,6 +701,7 @@ struct intel_iommu {
 	void *perf_statistic;
 
 	struct iommu_pmu *pmu;
+	struct list_head domains;	/* all struct dmar_domains on this IOMMU */
 };
 
 /* PCI domain-device relationship */
-- 
2.40.1