linux-kernel - [PATCH] iommu/amd: Ratelimit io-page-faults per device

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>] [day] [month] [year] [list]

Message-Id: <1495185692-18061-1-git-send-email-joro@8bytes.org>
Date:   Fri, 19 May 2017 11:21:32 +0200
From:   Joerg Roedel <joro@...tes.org>
To:     iommu@...ts.linux-foundation.org
Cc:     linux-kernel@...r.kernel.org, bp@...e.de,
        Joerg Roedel <jroedel@...e.de>
Subject: [PATCH] iommu/amd: Ratelimit io-page-faults per device

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

Misbehaving devices can cause an endless chain of
io-page-faults, flooding dmesg and making the system-log
unusable or even prevent the system from booting.

So ratelimit the error messages about io-page-faults in a
per-device basis.

Signed-off-by: Joerg Roedel <jroedel@...e.de>
---
 drivers/iommu/amd_iommu.c | 33 +++++++++++++++++++++++++++++----
 1 file changed, 29 insertions(+), 4 deletions(-)

diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
index 63cacf5..f574a0f 100644
--- a/drivers/iommu/amd_iommu.c
+++ b/drivers/iommu/amd_iommu.c
@@ -138,6 +138,8 @@ struct iommu_dev_data {
 					     PPR completions */
 	u32 errata;			  /* Bitmap for errata to apply */
 	bool use_vapic;			  /* Enable device to use vapic mode */
+
+	struct ratelimit_state rs;	  /* Ratelimit IOPF messages */
 };
 
 /*
@@ -253,6 +255,8 @@ static struct iommu_dev_data *alloc_dev_data(u16 devid)
 	list_add_tail(&dev_data->dev_data_list, &dev_data_list);
 	spin_unlock_irqrestore(&dev_data_list_lock, flags);
 
+	ratelimit_default_init(&dev_data->rs);
+
 	return dev_data;
 }
 
@@ -551,6 +555,30 @@ static void dump_command(unsigned long phys_addr)
 		pr_err("AMD-Vi: CMD[%d]: %08x\n", i, cmd->data[i]);
 }
 
+static void amd_iommu_report_page_fault(u16 devid, u16 domain_id,
+					u64 address, int flags)
+{
+	struct iommu_dev_data *dev_data = NULL;
+	struct pci_dev *pdev;
+
+	pdev = pci_get_bus_and_slot(PCI_BUS_NUM(devid), devid & 0xff);
+	if (pdev)
+		dev_data = get_dev_data(&pdev->dev);
+
+	if (dev_data && __ratelimit(&dev_data->rs)) {
+		dev_err(&pdev->dev, "IO_PAGE_FAULT domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+			domain_id, address, flags);
+	} else if (printk_ratelimit()) {
+		pr_err("IO_PAGE_FAULT device=%02x:%02x.%x "
+		       "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+			PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+			domain_id, address, flags);
+	}
+
+	if (pdev)
+		pci_dev_put(pdev);
+}
+
 static void iommu_print_event(struct amd_iommu *iommu, void *__evt)
 {
 	int type, devid, domid, flags;
@@ -586,10 +614,7 @@ static void iommu_print_event(struct amd_iommu *iommu, void *__evt)
 		dump_dte_entry(devid);
 		break;
 	case EVENT_TYPE_IO_FAULT:
-		printk("IO_PAGE_FAULT device=%02x:%02x.%x "
-		       "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
-		       PCI_BUS_NUM(devid), PCI_SLOT(devid), PCI_FUNC(devid),
-		       domid, address, flags);
+		amd_iommu_report_page_fault(devid, domid, address, flags);
 		break;
 	case EVENT_TYPE_DEV_TAB_ERR:
 		printk("DEV_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
-- 
2.7.4