[bug report] Bluetooth: Introduce Qualcomm WCNSS SMD based HCI driver

[Dan Carpenter <dan.carpenter@oracle.com>]

Hello Bjorn Andersson,

The patch 1511cc750c3d: "Bluetooth: Introduce Qualcomm WCNSS SMD
based HCI driver" from Aug 12, 2016, leads to the following static
checker warning:

	drivers/bluetooth/btqcomsmd.c:164 btqcomsmd_probe()
	warn: 'btq->acl_channel' isn't an ERR_PTR

drivers/bluetooth/btqcomsmd.c
   149  static int btqcomsmd_probe(struct platform_device *pdev)
   150  {
   151          struct btqcomsmd *btq;
   152          struct hci_dev *hdev;
   153          void *wcnss;
   154          int ret;
   155  
   156          btq = devm_kzalloc(&pdev->dev, sizeof(*btq), GFP_KERNEL);
   157          if (!btq)
   158                  return -ENOMEM;
   159  
   160          wcnss = dev_get_drvdata(pdev->dev.parent);
   161  
   162          btq->acl_channel = qcom_wcnss_open_channel(wcnss, "APPS_RIVA_BT_ACL",
   163                                                     btqcomsmd_acl_callback, btq);
   164          if (IS_ERR(btq->acl_channel))
   165                  return PTR_ERR(btq->acl_channel);

The qcom_wcnss_open_channel() returns error pointers if the CONFIG is
disabled, otherwise it returns NULL on error.  That seems like a design
mistake to me...  Anyway, it leads to a bug here.

   166  
   167          btq->cmd_channel = qcom_wcnss_open_channel(wcnss, "APPS_RIVA_BT_CMD",
   168                                                     btqcomsmd_cmd_callback, btq);
   169          if (IS_ERR(btq->cmd_channel))
   170                  return PTR_ERR(btq->cmd_channel);
   171  
   172          /* The local-bd-address property is usually injected by the

regards,
dan carpenter