[PATCH v1] Bluetooth: hci_release_dev: disable delayed devcoredump work

[PATCH v1] Bluetooth: hci_release_dev: disable delayed devcoredump work

[Aby Sam Ross]

It is not necessary that the pending delayed hci devcoredump timeout
work, hdev->dump.dump_timeout, submitted to the hdev->workqueue by the
bluetooth devcoredump state machine,
	hci_devcd_rx()
		hci_devcd_handle_pkt_init()
will be reset by it or by the timeout func hci_devcd_timeout(), using
hci_devcd_reset(), before destroying the workqueue or before the hci
device is freed up in hci_release_dev().

In this bug the active delayed devcoredump timeout work's timer object
is active when the memory associated with the hci device is freed up in
hci_release_dev() causing the ODEBUG WARNING.

Make sure that the delayed devcoredump timeout work is disabled before
the hdev->workqueue is destroyed and before the hdev memory is freed in
hci_release_dev().

Tested the change with the syzbot reproducer that uses vhci device
locally on x86_64 and on syzbot portal as well. Ran kselftest with net
target.

Fixes: 9695ef876fd1 ("Bluetooth: Add support for hci devcoredump")
Reported-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=b170dbf55520ebf5969a
Tested-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
Signed-off-by: Aby Sam Ross <abysamross@gmail.com>
---
 net/bluetooth/hci_core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index 01f8ceeb1c0c..1c7ee2a33337 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -2747,6 +2747,9 @@ void hci_release_dev(struct hci_dev *hdev)
 	kfree_const(hdev->hw_info);
 	kfree_const(hdev->fw_info);
 
+	if (hdev->dump.supported)
+		disable_delayed_work_sync(&hdev->dump.dump_timeout);
+
 	destroy_workqueue(hdev->workqueue);
 	destroy_workqueue(hdev->req_workqueue);
 
-- 
2.53.0

RE: [v1] Bluetooth: hci_release_dev: disable delayed devcoredump work

[bluez.test.bot]

[-- Attachment #1: Type: text/plain, Size: 2833 bytes --]

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=1070558

---Test result---

Test Summary:
CheckPatch                    PENDING   0.39 seconds
GitLint                       PENDING   0.25 seconds
SubjectPrefix                 PASS      0.12 seconds
BuildKernel                   PASS      26.40 seconds
CheckAllWarning               PASS      28.76 seconds
CheckSparse                   PASS      30.08 seconds
BuildKernel32                 PASS      25.51 seconds
TestRunnerSetup               PASS      569.23 seconds
TestRunner_l2cap-tester       PASS      28.00 seconds
TestRunner_iso-tester         FAIL      32.58 seconds
TestRunner_bnep-tester        PASS      6.38 seconds
TestRunner_mgmt-tester        FAIL      115.69 seconds
TestRunner_rfcomm-tester      PASS      9.48 seconds
TestRunner_sco-tester         FAIL      14.16 seconds
TestRunner_ioctl-tester       PASS      10.13 seconds
TestRunner_mesh-tester        FAIL      12.48 seconds
TestRunner_smp-tester         PASS      8.47 seconds
TestRunner_userchan-tester    PASS      6.71 seconds
IncrementalBuild              PENDING   0.48 seconds

Details
##############################
Test: CheckPatch - PENDING
Desc: Run checkpatch.pl script
Output:

##############################
Test: GitLint - PENDING
Desc: Run gitlint
Output:

##############################
Test: TestRunner_iso-tester - FAIL
Desc: Run iso-tester with test-runner
Output:
BUG: KASAN: slab-use-after-free in le_read_features_complete+0x7e/0x2b0
Total: 141, Passed: 141 (100.0%), Failed: 0, Not Run: 0
##############################
Test: TestRunner_mgmt-tester - FAIL
Desc: Run mgmt-tester with test-runner
Output:
Total: 494, Passed: 489 (99.0%), Failed: 1, Not Run: 4

Failed Test Cases
Read Exp Feature - Success                           Failed       0.107 seconds
##############################
Test: TestRunner_sco-tester - FAIL
Desc: Run sco-tester with test-runner
Output:
WARNING: possible circular locking dependency detected
BUG: sleeping function called from invalid context at net/core/sock.c:3782
Total: 30, Passed: 30 (100.0%), Failed: 0, Not Run: 0
##############################
Test: TestRunner_mesh-tester - FAIL
Desc: Run mesh-tester with test-runner
Output:
Total: 10, Passed: 8 (80.0%), Failed: 2, Not Run: 0

Failed Test Cases
Mesh - Send cancel - 1                               Timed out    2.773 seconds
Mesh - Send cancel - 2                               Timed out    1.993 seconds
##############################
Test: IncrementalBuild - PENDING
Desc: Incremental build with the patches in the series
Output:



---
Regards,
Linux Bluetooth

Re: [PATCH v1] Bluetooth: hci_release_dev: disable delayed devcoredump work

[Luiz Augusto von Dentz]

Hi Aby,

On Sun, Mar 22, 2026 at 5:09 PM Aby Sam Ross <abysamross@gmail.com> wrote:
>
> It is not necessary that the pending delayed hci devcoredump timeout
> work, hdev->dump.dump_timeout, submitted to the hdev->workqueue by the
> bluetooth devcoredump state machine,
>         hci_devcd_rx()
>                 hci_devcd_handle_pkt_init()
> will be reset by it or by the timeout func hci_devcd_timeout(), using
> hci_devcd_reset(), before destroying the workqueue or before the hci
> device is freed up in hci_release_dev().
>
> In this bug the active delayed devcoredump timeout work's timer object
> is active when the memory associated with the hci device is freed up in
> hci_release_dev() causing the ODEBUG WARNING.
>
> Make sure that the delayed devcoredump timeout work is disabled before
> the hdev->workqueue is destroyed and before the hdev memory is freed in
> hci_release_dev().
>
> Tested the change with the syzbot reproducer that uses vhci device
> locally on x86_64 and on syzbot portal as well. Ran kselftest with net
> target.
>
> Fixes: 9695ef876fd1 ("Bluetooth: Add support for hci devcoredump")
> Reported-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=b170dbf55520ebf5969a
> Tested-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
> Signed-off-by: Aby Sam Ross <abysamross@gmail.com>
> ---
>  net/bluetooth/hci_core.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
> index 01f8ceeb1c0c..1c7ee2a33337 100644
> --- a/net/bluetooth/hci_core.c
> +++ b/net/bluetooth/hci_core.c
> @@ -2747,6 +2747,9 @@ void hci_release_dev(struct hci_dev *hdev)
>         kfree_const(hdev->hw_info);
>         kfree_const(hdev->fw_info);
>
> +       if (hdev->dump.supported)
> +               disable_delayed_work_sync(&hdev->dump.dump_timeout);
> +
>         destroy_workqueue(hdev->workqueue);
>         destroy_workqueue(hdev->req_workqueue);
>
> --
> 2.53.0
>

https://sashiko.dev/#/patchset/20260322210849.68743-1-abysamross%40gmail.com

Both points seems valid, that perhaps the dump shouldn't be attached
to hdev object since it maybe necessary to unregister the hdev as part
of devcoredump handling, anyway if it is required then upon unregister
it shall actually cleanup the dump object as well.

-- 
Luiz Augusto von Dentz

Re: [PATCH v1] Bluetooth: hci_release_dev: disable delayed devcoredump work

[Aby Sam Ross]

On Tue, Mar 24, 2026 at 04:42:21PM -0400, Luiz Augusto von Dentz wrote:
> Hi Aby,
> 
> On Sun, Mar 22, 2026 at 5:09 PM Aby Sam Ross <abysamross@gmail.com> wrote:
> >
> > It is not necessary that the pending delayed hci devcoredump timeout
> > work, hdev->dump.dump_timeout, submitted to the hdev->workqueue by the
> > bluetooth devcoredump state machine,
> >         hci_devcd_rx()
> >                 hci_devcd_handle_pkt_init()
> > will be reset by it or by the timeout func hci_devcd_timeout(), using
> > hci_devcd_reset(), before destroying the workqueue or before the hci
> > device is freed up in hci_release_dev().
> >
> > In this bug the active delayed devcoredump timeout work's timer object
> > is active when the memory associated with the hci device is freed up in
> > hci_release_dev() causing the ODEBUG WARNING.
> >
> > Make sure that the delayed devcoredump timeout work is disabled before
> > the hdev->workqueue is destroyed and before the hdev memory is freed in
> > hci_release_dev().
> >
> > Tested the change with the syzbot reproducer that uses vhci device
> > locally on x86_64 and on syzbot portal as well. Ran kselftest with net
> > target.
> >
> > Fixes: 9695ef876fd1 ("Bluetooth: Add support for hci devcoredump")
> > Reported-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
> > Closes: https://syzkaller.appspot.com/bug?extid=b170dbf55520ebf5969a
> > Tested-by: syzbot+b170dbf55520ebf5969a@syzkaller.appspotmail.com
> > Signed-off-by: Aby Sam Ross <abysamross@gmail.com>
> > ---
> >  net/bluetooth/hci_core.c | 3 +++
> >  1 file changed, 3 insertions(+)
> >
> > diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
> > index 01f8ceeb1c0c..1c7ee2a33337 100644
> > --- a/net/bluetooth/hci_core.c
> > +++ b/net/bluetooth/hci_core.c
> > @@ -2747,6 +2747,9 @@ void hci_release_dev(struct hci_dev *hdev)
> >         kfree_const(hdev->hw_info);
> >         kfree_const(hdev->fw_info);
> >
> > +       if (hdev->dump.supported)
> > +               disable_delayed_work_sync(&hdev->dump.dump_timeout);
> > +
> >         destroy_workqueue(hdev->workqueue);
> >         destroy_workqueue(hdev->req_workqueue);
> >
> > --
> > 2.53.0
> >
> 
> https://sashiko.dev/#/patchset/20260322210849.68743-1-abysamross%40gmail.com
> 
> Both points seems valid, that perhaps the dump shouldn't be attached
> to hdev object since it maybe necessary to unregister the hdev as part
> of devcoredump handling, anyway if it is required then upon unregister
> it shall actually cleanup the dump object as well.
> 
> -- 
> Luiz Augusto von Dentz

Hi Luiz,

From what I see hci_unregister_dev() is called by all bluetooth drivers that use
hci_devcd_init(). Please correct me if I am wrong.

So we can either make hci_devcd_free() non-static and then call it from
hci_unregister_dev()
OR
Do vfree(hdev->dump.head) followed by disabling the delayed devcoredump timeout
work in hci_unregister_dev()

And then destroy the workqueue(dump_rx)?