block/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
blkdev_pr_read_keys() takes num_keys from userspace and uses it to
calculate the allocation size for keys_info via struct_size(). While
there is a check for SIZE_MAX (integer overflow), there is no upper
bound validation on the allocation size itself.
A malicious or buggy userspace can pass a large num_keys value that
doesn't trigger overflow but still results in an excessive allocation
attempt, causing a warning in the page allocator when the order exceeds
MAX_PAGE_ORDER.
Fix this by checking that keys_info_len does not exceed KMALLOC_MAX_SIZE
before attempting the allocation.
Fixes: 22a1ffea5f80 ("block: add IOC_PR_READ_KEYS ioctl")
Reported-by: syzbot+660d079d90f8a1baf54d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=660d079d90f8a1baf54d
Signed-off-by: Deepanshu Kartikey <kartikey406@gmail.com>
---
block/ioctl.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/block/ioctl.c b/block/ioctl.c
index 61feed686418..3e9e4257569f 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -443,7 +443,7 @@ static int blkdev_pr_read_keys(struct block_device *bdev, blk_mode_t mode,
return -EFAULT;
keys_info_len = struct_size(keys_info, keys, read_keys.num_keys);
- if (keys_info_len == SIZE_MAX)
+ if (keys_info_len == SIZE_MAX || keys_info_len > KMALLOC_MAX_SIZE)
return -EINVAL;
keys_info = kzalloc(keys_info_len, GFP_KERNEL);
--
2.43.0
On Fri, Dec 12, 2025 at 07:05:10AM +0530, Deepanshu Kartikey wrote: > keys_info_len = struct_size(keys_info, keys, read_keys.num_keys); > - if (keys_info_len == SIZE_MAX) > + if (keys_info_len == SIZE_MAX || keys_info_len > KMALLOC_MAX_SIZE) Well, bother checks for sure are redundant. But maybe this is also a good chance to pick a sane arbitrary limited instead of KMALLOC_MAX_SIZE. And if that is above KMALLOC_MAX_SIZE, switch to using kvzalloc.
On Fri, Dec 12, 2025 at 11:46 AM Christoph Hellwig <hch@infradead.org> wrote:
> Well, bother checks for sure are redundant. But maybe this is also
> a good chance to pick a sane arbitrary limited instead of
> KMALLOC_MAX_SIZE. And if that is above KMALLOC_MAX_SIZE, switch to
> using kvzalloc.
>
Thanks for the feedback.
How about limiting num_keys to 64K (1u << 16)? In practice, PR keys
are used for shared storage coordination and typical deployments have
only a handful of hosts, so this should be more than enough for any
realistic use case.
With a bounded num_keys, the SIZE_MAX check becomes unnecessary, so
I've removed it. Also switched to kvzalloc/kvfree to handle larger
allocations gracefully.
Something like below:
+/* Limit the number of keys to prevent excessive memory allocation */
+#define PR_KEYS_MAX_NUM (1u << 16)
+
static int blkdev_pr_read_keys(struct block_device *bdev, blk_mode_t mode,
struct pr_read_keys __user *arg)
{
...
if (copy_from_user(&read_keys, arg, sizeof(read_keys)))
return -EFAULT;
+ if (read_keys.num_keys > PR_KEYS_MAX_NUM)
+ return -EINVAL;
+
keys_info_len = struct_size(keys_info, keys, read_keys.num_keys);
- if (keys_info_len == SIZE_MAX)
- return -EINVAL;
- keys_info = kzalloc(keys_info_len, GFP_KERNEL);
+ keys_info = kvzalloc(keys_info_len, GFP_KERNEL);
if (!keys_info)
return -ENOMEM;
...
out:
- kfree(keys_info);
+ kvfree(keys_info);
return ret;
}
© 2016 - 2025 Red Hat, Inc.