Patch "iommu/vt-d: Fix PASID table allocation" has been added to the 4.4-stable tree

PATCH [email protected] 1 participant 1 message Nov 29, 2016

Subsystems (2)

Stable branch IOMMU

lore

Share & tools

lore

[email protected]

author maintainer Applied

2016-11-29 15:33 UTC #0 lore raw

This is a note to let you know that I've just added the patch titled

iommu/vt-d: Fix PASID table allocation

to the 4.4-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
iommu-vt-d-fix-pasid-table-allocation.patch
and it can be found in the queue-4.4 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

Quoted 1 line

From 910170442944e1f8674fd5ddbeeb8ccd1877ea98 Mon Sep 17 00:00:00 2001

From: David Woodhouse <[email protected]>
Date: Mon, 12 Sep 2016 10:49:11 +0800
Subject: iommu/vt-d: Fix PASID table allocation
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

From: David Woodhouse <[email protected]>

commit 910170442944e1f8674fd5ddbeeb8ccd1877ea98 upstream.

Somehow I ended up with an off-by-three error in calculating the size of the PASID and PASID State tables, which triggers allocations failures as those tables unfortunately have to be physically contiguous.

In fact, even the *correct* maximum size of 8MiB is problematic and is wont to lead to allocation failures. Since I have extracted a promise that this *will* be fixed in hardware, I'm happy to limit it on the current hardware to a maximum of 0x20000 PASIDs, which gives us 1MiB tables — still not ideal, but better than before.

Reported by Mika Kuoppala <[email protected]> and also by Xunlei Pang <[email protected]> who submitted a simpler patch to fix only the allocation (and not the free) to the "correct" limit... which was still problematic.

Signed-off-by David Woodhouse <[email protected]>

Signed-off-by Greg Kroah-Hartman <[email protected]>

---
 drivers/iommu/intel-svm.c   |   26 ++++++++++++++++----------
 include/linux/intel-iommu.h |    1 +
 2 files changed, 17 insertions(+), 10 deletions(-)

drivers/iommu/intel-svm.c +16 ~~-10~~

@@ -39,10 +39,18 @@ int intel_svm_alloc_pasid_tables(struct
 	struct page *pages;
 	int order;
 
-	order = ecap_pss(iommu->ecap) + 7 - PAGE_SHIFT;
-	if (order < 0)
-		order = 0;
+	/* Start at 2 because it's defined as 2^(1+PSS) */
+	iommu->pasid_max = 2 << ecap_pss(iommu->ecap);
 
+	/* Eventually I'm promised we will get a multi-level PASID table
+	 * and it won't have to be physically contiguous. Until then,
+	 * limit the size because 8MiB contiguous allocations can be hard
+	 * to come by. The limit of 0x20000, which is 1MiB for each of
+	 * the PASID and PASID-state tables, is somewhat arbitrary. */
+	if (iommu->pasid_max > 0x20000)
+		iommu->pasid_max = 0x20000;
+
+	order = get_order(sizeof(struct pasid_entry) * iommu->pasid_max);
 	pages = alloc_pages(GFP_KERNEL | __GFP_ZERO, order);
 	if (!pages) {
 		pr_warn("IOMMU: %s: Failed to allocate PASID table\n",

@@ -53,6 +61,8 @@ int intel_svm_alloc_pasid_tables(struct
 	pr_info("%s: Allocated order %d PASID table.\n", iommu->name, order);
 
 	if (ecap_dis(iommu->ecap)) {
+		/* Just making it explicit... */
+		BUILD_BUG_ON(sizeof(struct pasid_entry) != sizeof(struct pasid_state_entry));
 		pages = alloc_pages(GFP_KERNEL | __GFP_ZERO, order);
 		if (pages)
 			iommu->pasid_state_table = page_address(pages);

@@ -68,11 +78,7 @@ int intel_svm_alloc_pasid_tables(struct
 
 int intel_svm_free_pasid_tables(struct intel_iommu *iommu)
 {
-	int order;
-
-	order = ecap_pss(iommu->ecap) + 7 - PAGE_SHIFT;
-	if (order < 0)
-		order = 0;
+	int order = get_order(sizeof(struct pasid_entry) * iommu->pasid_max);
 
 	if (iommu->pasid_table) {
 		free_pages((unsigned long)iommu->pasid_table, order);

@@ -371,8 +377,8 @@ int intel_svm_bind_mm(struct device *dev
 		}
 		svm->iommu = iommu;
 
-		if (pasid_max > 2 << ecap_pss(iommu->ecap))
-			pasid_max = 2 << ecap_pss(iommu->ecap);
+		if (pasid_max > iommu->pasid_max)
+			pasid_max = iommu->pasid_max;
 
 		/* Do not use PASID 0 in caching mode (virtualised IOMMU) */
 		ret = idr_alloc(&iommu->pasid_idr, svm,

include/linux/intel-iommu.h +1 -0

@@ -429,6 +429,7 @@ struct intel_iommu {
 	struct page_req_dsc *prq;
 	unsigned char prq_name[16];    /* Name for PRQ interrupt */
 	struct idr pasid_idr;
+	u32 pasid_max;
 #endif
 	struct q_inval  *qi;            /* Queued invalidation info */
 	u32 *iommu_state; /* Store iommu states between suspend and resume.*/

Patches currently in stable-queue which might be from [email protected] are

queue-4.4/iommu-vt-d-fix-iommu-lookup-for-sr-iov-virtual-functions.patch queue-4.4/iommu-vt-d-fix-pasid-table-allocation.patch