NFC: reorganize the functions in nci_request
authorLin Ma <linma@zju.edu.cn>
Mon, 15 Nov 2021 14:56:00 +0000 (22:56 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Nov 2021 08:48:40 +0000 (09:48 +0100)
[ Upstream commit 86cdf8e38792545161dbe3350a7eced558ba4d15 ]

There is a possible data race as shown below:

thread-A in nci_request()       | thread-B in nci_close_device()
                                | mutex_lock(&ndev->req_lock);
test_bit(NCI_UP, &ndev->flags); |
...                             | test_and_clear_bit(NCI_UP, &ndev->flags)
mutex_lock(&ndev->req_lock);    |
                                |

This race will allow __nci_request() to be awaked while the device is
getting removed.

Similar to commit e2cb6b891ad2 ("bluetooth: eliminate the potential race
condition when removing the HCI controller"). this patch alters the
function sequence in nci_request() to prevent the data races between the
nci_close_device().

Signed-off-by: Lin Ma <linma@zju.edu.cn>
Fixes: 6a2968aaf50c ("NFC: basic NCI protocol implementation")
Link: https://lore.kernel.org/r/20211115145600.8320-1-linma@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
net/nfc/nci/core.c

index 82ab39d..39994db 100644 (file)
@@ -144,12 +144,15 @@ inline int nci_request(struct nci_dev *ndev,
 {
        int rc;
 
-       if (!test_bit(NCI_UP, &ndev->flags))
-               return -ENETDOWN;
-
        /* Serialize all requests */
        mutex_lock(&ndev->req_lock);
-       rc = __nci_request(ndev, req, opt, timeout);
+       /* check the state after obtaing the lock against any races
+        * from nci_close_device when the device gets removed.
+        */
+       if (test_bit(NCI_UP, &ndev->flags))
+               rc = __nci_request(ndev, req, opt, timeout);
+       else
+               rc = -ENETDOWN;
        mutex_unlock(&ndev->req_lock);
 
        return rc;