Skip to contentSkip to conversation

Bluetooth: ISO: cleanup bound BISes

PATCH In review Iulia Tanasescu 2 participants 3 messages Jun 7, 2023

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover Bluetooth: ISO: cleanup bound BISes

This patch fixes the fact that if a broadcast socket is closed while it is still in BT_CONNECT state (that is before the BIG was successfully created), the connection will not be properly cleaned up - the bis_cleanup callback will not be called.

If the socket is closed, even if the BIG was not yet created, the advertising set should still be terminated, and the conn should be removed from the hash list, so that when the BIG complete event arrives, no active connections will be found and the BIG will be terminated also.

Iulia Tanasescu (1):
Bluetooth: ISO: cleanup bound BISes

 net/bluetooth/iso.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

base-commit: 817efd3cad7481ce2ee25fac5108afecbad56228

Signature · 1 line
2.34.1
#1 lore raw

PATCH 1/1 Bluetooth: ISO: cleanup bound BISes

For some broadcast scenarios, the user might open a nonblocking socket and might call connect without defer setup - this will queue the commands for adding an advertising instance and creating a BIG.

If the socket is closed before the LE BIG Complete event arrives (connection is still pending), the conn should be properly cleaned up:
it should be removed from the hash list, the adv set should be removed, and the BIG will be terminated once the LE BIG Complete event arrives, since no active connections will be found at that point.

---
 net/bluetooth/iso.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)
net/bluetooth/iso.c +7 -5
index b9a008fd10b1..261e26ec8f16 100644
@@ -655,13 +655,15 @@ static void __iso_sock_close(struct sock *sk)
 			iso_conn_defer_reject(iso_pi(sk)->conn->hcon);
 		iso_chan_del(sk, ECONNRESET);
 		break;
+
 	case BT_CONNECT:
-		/* In case of DEFER_SETUP the hcon would be bound to CIG which
-		 * needs to be removed so just call hci_conn_del so the cleanup
-		 * callback do what is needed.
+		/* In case hcon is bound to a CIG/BIG, just call
+		 * hci_conn_del so the cleanup callback does what
+		 * is needed.
 		 */
-		if (test_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags) &&
-		    iso_pi(sk)->conn->hcon) {
+		if ((!bacmp(&iso_pi(sk)->dst, BDADDR_ANY) ||
+		     test_bit(BT_SK_DEFER_SETUP, &bt_sk(sk)->flags)) &&
+		     iso_pi(sk)->conn->hcon) {
 			hci_conn_del(iso_pi(sk)->conn->hcon);
 			iso_pi(sk)->conn->hcon = NULL;
 		}
Signature · 1 line
2.34.1
#2 lore raw

This is automated email and please do not reply to this email!

Dear submitter,

Thank you for submitting the patches to the linux bluetooth mailing list. This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=754765

---Test result---

Test Summary:

CheckPatch                    PASS      0.71 seconds
GitLint                       PASS      0.35 seconds
SubjectPrefix                 PASS      0.13 seconds
BuildKernel                   PASS      33.93 seconds
CheckAllWarning               PASS      37.72 seconds
CheckSparse                   PASS      42.03 seconds
CheckSmatch                   PASS      115.65 seconds
BuildKernel32                 PASS      32.34 seconds
TestRunnerSetup               PASS      443.80 seconds
TestRunner_l2cap-tester       PASS      16.49 seconds
TestRunner_iso-tester         PASS      22.45 seconds
TestRunner_bnep-tester        PASS      5.39 seconds
TestRunner_mgmt-tester        PASS      110.42 seconds
TestRunner_rfcomm-tester      PASS      8.62 seconds
TestRunner_sco-tester         PASS      7.92 seconds
TestRunner_ioctl-tester       PASS      9.15 seconds
TestRunner_mesh-tester        PASS      6.80 seconds
TestRunner_smp-tester         PASS      7.84 seconds
TestRunner_userchan-tester    PASS      5.67 seconds
IncrementalBuild              PASS      29.24 seconds

---
Regards,
Linux Bluetooth