[bug report] habanalabs: add h/w queues module

[bug report] habanalabs: add h/w queues module

[Dan Carpenter]

Hello Oded Gabbay,

The patch 9494a8dd8d22: "habanalabs: add h/w queues module" from Feb
16, 2019, leads to the following static checker warning:

	drivers/misc/habanalabs/goya/goya.c:3386 goya_test_cpu_queue()
	error: uninitialized symbol 'result'.

drivers/misc/habanalabs/goya/goya.c
    3367 int goya_test_cpu_queue(struct hl_device *hdev)
    3368 {
    3369 	struct armcp_packet test_pkt;
    3370 	long result;
    3371 	int rc;
    3372 
    3373 	/* cpu_queues_enable flag is always checked in send cpu message */
    3374 
    3375 	memset(&test_pkt, 0, sizeof(test_pkt));
    3376 
    3377 	test_pkt.ctl = ARMCP_PACKET_TEST << ARMCP_PKT_CTL_OPCODE_SHIFT;
    3378 	test_pkt.value = ARMCP_PACKET_FENCE_VAL;
    3379 
    3380 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &test_pkt,
    3381 			sizeof(test_pkt), HL_DEVICE_TIMEOUT_USEC, &result);
                                                                          ^^^^^^^
Presumably, result isn't set on all failure paths.

    3382 
    3383 	if (!rc)
    3384 		dev_info(hdev->dev, "queue test on CPU queue succeeded\n");
    3385 	else
--> 3386 		dev_err(hdev->dev, "CPU queue test failed (0x%08lX)\n", result);
                                                                                ^^^^^^
    3387 
    3388 	return rc;
    3389 }

regards,
dan carpenter

Re: [bug report] habanalabs: add h/w queues module

[Oded Gabbay]

On Wed, Feb 20, 2019 at 10:12 PM Dan Carpenter <dan.carpenter@oracle.com> wrote:
>
> Hello Oded Gabbay,
>
> The patch 9494a8dd8d22: "habanalabs: add h/w queues module" from Feb
> 16, 2019, leads to the following static checker warning:
>
>         drivers/misc/habanalabs/goya/goya.c:3386 goya_test_cpu_queue()
>         error: uninitialized symbol 'result'.
>
> drivers/misc/habanalabs/goya/goya.c
>     3367 int goya_test_cpu_queue(struct hl_device *hdev)
>     3368 {
>     3369        struct armcp_packet test_pkt;
>     3370        long result;
>     3371        int rc;
>     3372
>     3373        /* cpu_queues_enable flag is always checked in send cpu message */
>     3374
>     3375        memset(&test_pkt, 0, sizeof(test_pkt));
>     3376
>     3377        test_pkt.ctl = ARMCP_PACKET_TEST << ARMCP_PKT_CTL_OPCODE_SHIFT;
>     3378        test_pkt.value = ARMCP_PACKET_FENCE_VAL;
>     3379
>     3380        rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &test_pkt,
>     3381                        sizeof(test_pkt), HL_DEVICE_TIMEOUT_USEC, &result);
>                                                                           ^^^^^^^
> Presumably, result isn't set on all failure paths.
>
>     3382
>     3383        if (!rc)
>     3384                dev_info(hdev->dev, "queue test on CPU queue succeeded\n");
>     3385        else
> --> 3386                dev_err(hdev->dev, "CPU queue test failed (0x%08lX)\n", result);
>                                                                                 ^^^^^^
>     3387
>     3388        return rc;
>     3389 }
>
> regards,
> dan carpenter
Thanks!
Because this is the initial merge of the driver to the kernel, I
prefer not to change the code before the merge, so I'll queue the
patch in my tree and send it for -rc2
Oded