Re: [PATCH 08/13] Bluetooth: hci_sync: Fix return value of hci_reset_sync()

[Zijun Hu]

On 6/22/2026 11:35 PM, Luiz Augusto von Dentz wrote:
>> diff --git a/net/bluetooth/hci_sync.c b/net/bluetooth/hci_sync.c
>> index fce9f9526cb5..601d44ef975f 100644
>> --- a/net/bluetooth/hci_sync.c
>> +++ b/net/bluetooth/hci_sync.c
>> @@ -3678,8 +3678,10 @@ int hci_reset_sync(struct hci_dev *hdev)
>>
>>         err = __hci_cmd_sync_status(hdev, HCI_OP_RESET, 0, NULL,
>>                                     HCI_CMD_TIMEOUT);
>> +       if (err < 0)
>> +               return err;
>>
>> -       return err;
>> +       return -bt_to_errno(err);
>>  }
> There seem to be 2 consecutive changes to hci_reset_sync that conflict
> with each other, also the expectation should be that positive errors
> are HCI errors and negative errors are stack generated ones, so the
> callers should really check `err` and not `err < 0`.

  Hi Luiz,

  Checking the full call chains:

  - hci_reset_sync() → hci_init0_sync() [if (err)] → hci_unconf_init_sync() [if (err < 0)]
  - hci_reset_sync() → hci_init1_sync() [if (err)] → hci_init_sync() [if (err < 0)]

  A positive HCI status from hci_reset_sync() propagates through
  hci_init0_sync()/hci_init1_sync() but is then silently ignored by
  hci_unconf_init_sync()/hci_init_sync() due to their if (err < 0)
  checks.

  Will not change hci_reset_sync(). Any suggestion on how to fix it?