[PATCH] iommupt: Fix unlikely flows in increase_top()

[Jason Gunthorpe <jgg@nvidia.com>]

Since increase_top() does it's own READ_ONCE() on top_of_table, the
caller's prior READ_ONCE() could be inconsistent and the first time
through the loop we may actually already have the right level if two
threads are racing map.

In this case new_level will be left uninitialized.

Further all the exits from the loop have to either commit to the new top
or free any memory allocated so the early return must be a goto err_free.

Make it so the only break from the loop always sets new_level to the right
value and all other exits go to err_free. Use pts.level (the pts
represents the top we are stacking) within the loop instead of new_level.

Fixes: dcd6a011a8d5 ("iommupt: Add map_pages op")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Closes: https://lore.kernel.org/r/aRwgNW9PiW2j-Qwo@stanley.mountain
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/iommu/generic_pt/iommu_pt.h | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/drivers/iommu/generic_pt/iommu_pt.h b/drivers/iommu/generic_pt/iommu_pt.h
index 0e046fe0eea34c..032d04ec7b568d 100644
--- a/drivers/iommu/generic_pt/iommu_pt.h
+++ b/drivers/iommu/generic_pt/iommu_pt.h
@@ -683,8 +683,11 @@ static int increase_top(struct pt_iommu *iommu_table, struct pt_range *range,
 		top_range.va = range->va;
 		top_range.last_va = range->last_va;
 
-		if (!pt_check_range(&top_range) && map->leaf_level <= pts.level)
+		if (!pt_check_range(&top_range) &&
+		    map->leaf_level <= pts.level) {
+			new_level = pts.level;
 			break;
+		}
 
 		pts.level++;
 		if (pts.level > PT_MAX_TOP_LEVEL ||
@@ -693,17 +696,18 @@ static int increase_top(struct pt_iommu *iommu_table, struct pt_range *range,
 			goto err_free;
 		}
 
-		new_level = pts.level;
 		table_mem =
 			table_alloc_top(common, _pt_top_set(NULL, pts.level),
 					map->attrs.gfp, ALLOC_DEFER_COHERENT_FLUSH);
-		if (IS_ERR(table_mem))
-			return PTR_ERR(table_mem);
+		if (IS_ERR(table_mem)) {
+			ret = PTR_ERR(table_mem);
+			goto err_free;
+		}
 		iommu_pages_list_add(&free_list, table_mem);
 
 		/* The new table links to the lower table always at index 0 */
 		top_range.va = 0;
-		top_range.top_level = new_level;
+		top_range.top_level = pts.level;
 		pts.table_lower = pts.table;
 		pts.table = table_mem;
 		pt_load_single_entry(&pts);
@@ -735,7 +739,8 @@ static int increase_top(struct pt_iommu *iommu_table, struct pt_range *range,
 	 */
 	domain_lock = iommu_table->driver_ops->get_top_lock(iommu_table);
 	spin_lock_irqsave(domain_lock, flags);
-	if (common->top_of_table != top_of_table) {
+	if (common->top_of_table != top_of_table ||
+	    top_of_table == new_top_of_table) {
 		spin_unlock_irqrestore(domain_lock, flags);
 		ret = -EAGAIN;
 		goto err_free;

base-commit: 30dc4423a35507df9378a6eb1a7104f788f4bbb9
-- 
2.43.0