[PATCH v2] mailbox: test: really ignore optional memory resources

[PATCH v2] mailbox: test: really ignore optional memory resources

[Wolfram Sang]

Memory resources are optional but if the resource is empty
devm_platform_get_and_ioremap_resource() prints an error nonetheless.
Refactor the code to check the resources locally first and process them
only if they are present. The -EBUSY error message of ioremap_resource()
is still kept because it is correct. The comment which explains that a
plain ioremap() is tried as a workaround is turned into a info message.
So, a user will be informed about it, too.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---

Change since v1:
* don't fail if second region is empty but use the first region instead
* refactor into separate function for consistent handling of both
  regions and to make the logic in probe() more obvious

 drivers/mailbox/mailbox-test.c | 40 ++++++++++++++++++++--------------
 1 file changed, 24 insertions(+), 16 deletions(-)

diff --git a/drivers/mailbox/mailbox-test.c b/drivers/mailbox/mailbox-test.c
index 3a28ab5c42e5..309610ba43a1 100644
--- a/drivers/mailbox/mailbox-test.c
+++ b/drivers/mailbox/mailbox-test.c
@@ -355,11 +355,30 @@ mbox_test_request_channel(struct platform_device *pdev, const char *name)
 	return channel;
 }
 
-static int mbox_test_probe(struct platform_device *pdev)
+static void *mbox_test_ioremap(struct platform_device *pdev, unsigned int res_num)
 {
-	struct mbox_test_device *tdev;
 	struct resource *res;
 	resource_size_t size;
+	void *mmio = NULL;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, res_num);
+	if (res) {
+		mmio = devm_ioremap_resource(&pdev->dev, res);
+		if (PTR_ERR(mmio) == -EBUSY) {
+			dev_info(&pdev->dev, "trying workaround with plain ioremap\n");
+			size = resource_size(res);
+			mmio = devm_ioremap(&pdev->dev, res->start, size);
+		} else if (IS_ERR(mmio)) {
+			mmio = NULL;
+		}
+	}
+
+	return mmio;
+}
+
+static int mbox_test_probe(struct platform_device *pdev)
+{
+	struct mbox_test_device *tdev;
 	int ret;
 
 	tdev = devm_kzalloc(&pdev->dev, sizeof(*tdev), GFP_KERNEL);
@@ -367,23 +386,12 @@ static int mbox_test_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	/* It's okay for MMIO to be NULL */
-	tdev->tx_mmio = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
-	if (PTR_ERR(tdev->tx_mmio) == -EBUSY) {
-		/* if reserved area in SRAM, try just ioremap */
-		size = resource_size(res);
-		tdev->tx_mmio = devm_ioremap(&pdev->dev, res->start, size);
-	} else if (IS_ERR(tdev->tx_mmio)) {
-		tdev->tx_mmio = NULL;
-	}
+	tdev->tx_mmio = mbox_test_ioremap(pdev, 0);
 
 	/* If specified, second reg entry is Rx MMIO */
-	tdev->rx_mmio = devm_platform_get_and_ioremap_resource(pdev, 1, &res);
-	if (PTR_ERR(tdev->rx_mmio) == -EBUSY) {
-		size = resource_size(res);
-		tdev->rx_mmio = devm_ioremap(&pdev->dev, res->start, size);
-	} else if (IS_ERR(tdev->rx_mmio)) {
+	tdev->rx_mmio = mbox_test_ioremap(pdev, 1);
+	if (!tdev->rx_mmio)
 		tdev->rx_mmio = tdev->tx_mmio;
-	}
 
 	tdev->tx_channel = mbox_test_request_channel(pdev, "tx");
 	tdev->rx_channel = mbox_test_request_channel(pdev, "rx");
-- 
2.51.0

Re: [PATCH v2] mailbox: test: really ignore optional memory resources

[Geert Uytterhoeven]

Hi Wolfram,

On Mon, 23 Feb 2026 at 13:30, Wolfram Sang
<wsa+renesas@sang-engineering.com> wrote:
> Memory resources are optional but if the resource is empty
> devm_platform_get_and_ioremap_resource() prints an error nonetheless.
> Refactor the code to check the resources locally first and process them
> only if they are present. The -EBUSY error message of ioremap_resource()
> is still kept because it is correct. The comment which explains that a
> plain ioremap() is tried as a workaround is turned into a info message.
> So, a user will be informed about it, too.
>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Thanks for your patch!

> --- a/drivers/mailbox/mailbox-test.c
> +++ b/drivers/mailbox/mailbox-test.c
> @@ -355,11 +355,30 @@ mbox_test_request_channel(struct platform_device *pdev, const char *name)
>         return channel;
>  }
>
> -static int mbox_test_probe(struct platform_device *pdev)
> +static void *mbox_test_ioremap(struct platform_device *pdev, unsigned int res_num)
>  {
> -       struct mbox_test_device *tdev;
>         struct resource *res;
>         resource_size_t size;
> +       void *mmio = NULL;
> +
> +       res = platform_get_resource(pdev, IORESOURCE_MEM, res_num);
> +       if (res) {

if (!res)
        return NULL;

> +               mmio = devm_ioremap_resource(&pdev->dev, res);
> +               if (PTR_ERR(mmio) == -EBUSY) {
> +                       dev_info(&pdev->dev, "trying workaround with plain ioremap\n");
> +                       size = resource_size(res);

With the reduced indentation, you don't need a temporary to avoid the
next line being too long.

> +                       mmio = devm_ioremap(&pdev->dev, res->start, size);

return devm_ioremap(...);

> +               } else if (IS_ERR(mmio)) {
> +                       mmio = NULL;
> +               }
> +       }
> +
> +       return mmio;

return IS_ERR(mmio) ? NULL : mmio;


> +}

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v2] mailbox: test: really ignore optional memory resources

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 218 bytes --]

Hi Geert,

thank you for the review!

> if (!res)
>         return NULL;

I applied the 'single exit' paradigm, but your solution is a bit easier
to understand, I think. Will send v3 soon.

Happy hacking,

   Wolfram


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]