[PATCH 6.1.y 0/2] blk-mq: backport CVE-2023-54227 to 6.1.y

[PATCH 6.1.y 0/2] blk-mq: backport CVE-2023-54227 to 6.1.y

[Artem Dinaburg]

Hi Greg, Sasha, and block maintainers,

I am continuing with the smaller CVE backports still missing from
6.1.y. This series handles CVE-2023-54227. The first patch frees excess tag
sets after a queue-count reduction. Its immediate upstream follow-up moves
that free past hctx resizing so the repair itself cannot race into a UAF;
they need to travel together.

Both changes are already inherited by v6.6 and every later mainline
release.

Could you please queue the pair for 6.1.y?

Thanks,
Artem Dinaburg

AI assistance: An LLM helped find the missing CVE fix, adapt
it, and prepare the draft; I reviewed the code and build output.

Chengming Zhou (2):
  blk-mq: fix tags leak when shrink nr_hw_queues
  blk-mq: fix tags UAF when shrinking q->nr_hw_queues

 block/blk-mq.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

-- 
2.39.5

[PATCH 6.1.y 1/2] blk-mq: fix tags leak when shrink nr_hw_queues

[Artem Dinaburg]

From: Chengming Zhou <zhouchengming@bytedance.com>

[ Upstream commit e1dd7bc93029024af5688253b0c05181d6e01f8e ]

Although we don't need to realloc set->tags[] when shrink nr_hw_queues,
we need to free them. Or these tags will be leaked.

How to reproduce:
1. mount -t configfs configfs /mnt
2. modprobe null_blk nr_devices=0 submit_queues=8
3. mkdir /mnt/nullb/nullb0
4. echo 1 > /mnt/nullb/nullb0/power
5. echo 4 > /mnt/nullb/nullb0/submit_queues
6. rmdir /mnt/nullb/nullb0

In step 4, will alloc 9 tags (8 submit queues and 1 poll queue), then
in step 5, new_nr_hw_queues = 5 (4 submit queues and 1 poll queue).
At last in step 6, only these 5 tags are freed, the other 4 tags leaked.

Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20230821095602.70742-1-chengming.zhou@linux.dev
Signed-off-by: Jens Axboe <axboe@kernel.dk>

[ Backport to 6.1.y: use this tree's cur_nr_hw_queues snapshot when
  freeing excess tag sets. ]
Assisted-by: LLM
Signed-off-by: Artem Dinaburg <artem@trailofbits.com>
---
Stable submission note: patch 1/2 in the CVE-2023-54227 series for Linux
6.1.y.

CVE: CVE-2023-54227

Build: This patch was included in the x86_64 allmodconfig and
CONFIG_WERROR=y build. It produced no new warnings or errors.

 block/blk-mq.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 8a9d9e3db1668a..b6215bee3d3c45 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -4509,9 +4509,13 @@ static int blk_mq_realloc_tag_set_tags(struct blk_mq_tag_set *set,
 				  int cur_nr_hw_queues, int new_nr_hw_queues)
 {
 	struct blk_mq_tags **new_tags;
+	int i;
 
-	if (cur_nr_hw_queues >= new_nr_hw_queues)
+	if (cur_nr_hw_queues >= new_nr_hw_queues) {
+		for (i = new_nr_hw_queues; i < cur_nr_hw_queues; i++)
+			__blk_mq_free_map_and_rqs(set, i);
 		return 0;
+	}
 
 	new_tags = kcalloc_node(new_nr_hw_queues, sizeof(struct blk_mq_tags *),
 				GFP_KERNEL, set->numa_node);
-- 
2.39.5

[PATCH 6.1.y 2/2] blk-mq: fix tags UAF when shrinking q->nr_hw_queues

[Artem Dinaburg]

From: Chengming Zhou <zhouchengming@bytedance.com>

[ Upstream commit 6be6d112419713334ddd9c01f219ca16adaa4c76 ]

When nr_hw_queues shrink, we free the excess tags before realloc'ing
hw_ctxs for each queue. During that resize, we may need to access those
tags, like blk_mq_tag_idle(hctx) will access queue shared tags.

This can cause a slab use-after-free, as reported by KASAN. Fix it by
moving the releasing of excess tags to the end.

Fixes: e1dd7bc93029 ("blk-mq: fix tags leak when shrink nr_hw_queues")
Reported-by: Yi Zhang <yi.zhang@redhat.com>
Closes: https://lore.kernel.org/all/CAHj4cs_CK63uoDpGBGZ6DN4OCTpzkR3UaVgK=LX8Owr8ej2ieQ@mail.gmail.com/
Cc: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20230908005702.2183908-1-chengming.zhou@linux.dev
Signed-off-by: Jens Axboe <axboe@kernel.dk>

[ Backport to 6.1.y follow-up to e1dd7bc93029: move those frees after
  hctx resizing using cur_nr_hw_queues, preventing its tag UAF. ]
Assisted-by: LLM
Signed-off-by: Artem Dinaburg <artem@trailofbits.com>
---
Stable submission note: patch 2/2 in the CVE-2023-54227 series for Linux
6.1.y.

CVE: CVE-2023-54227

Build: This patch was included in the x86_64 allmodconfig and
CONFIG_WERROR=y build. It produced no new warnings or errors.

 block/blk-mq.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/block/blk-mq.c b/block/blk-mq.c
index b6215bee3d3c45..bf69079f6bd517 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -4509,13 +4509,9 @@ static int blk_mq_realloc_tag_set_tags(struct blk_mq_tag_set *set,
 				  int cur_nr_hw_queues, int new_nr_hw_queues)
 {
 	struct blk_mq_tags **new_tags;
-	int i;
 
-	if (cur_nr_hw_queues >= new_nr_hw_queues) {
-		for (i = new_nr_hw_queues; i < cur_nr_hw_queues; i++)
-			__blk_mq_free_map_and_rqs(set, i);
+	if (cur_nr_hw_queues >= new_nr_hw_queues)
 		return 0;
-	}
 
 	new_tags = kcalloc_node(new_nr_hw_queues, sizeof(struct blk_mq_tags *),
 				GFP_KERNEL, set->numa_node);
@@ -4804,7 +4800,8 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 {
 	struct request_queue *q;
 	LIST_HEAD(head);
-	int prev_nr_hw_queues;
+	int prev_nr_hw_queues = set->nr_hw_queues;
+	int i;
 
 	lockdep_assert_held(&set->tag_list_lock);
 
@@ -4831,7 +4828,6 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 		blk_mq_sysfs_unregister_hctxs(q);
 	}
 
-	prev_nr_hw_queues = set->nr_hw_queues;
 	if (blk_mq_realloc_tag_set_tags(set, set->nr_hw_queues, nr_hw_queues) <
 	    0)
 		goto reregister;
@@ -4869,6 +4865,10 @@ static void __blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set,
 
 	list_for_each_entry(q, &set->tag_list, tag_set_list)
 		blk_mq_unfreeze_queue(q);
+
+	/* Free the excess tags when nr_hw_queues shrink. */
+	for (i = set->nr_hw_queues; i < prev_nr_hw_queues; i++)
+		__blk_mq_free_map_and_rqs(set, i);
 }
 
 void blk_mq_update_nr_hw_queues(struct blk_mq_tag_set *set, int nr_hw_queues)
-- 
2.39.5