[PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask in vmbus_device_register()

[PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask in vmbus_device_register()

[Haoxiang Li]

child_device_obj->device cannot perform DMA properly if dma_set_mask()
returns non-zero. Add check for dma_set_mask() and return the error if it
fails. child_device_obj->device is not initialized here, so use kfree() to
free child_device_obj->device.

Fixes: 3a5469582c24 ("Drivers: hv: vmbus: Fix initialization of device object in vmbus_device_register()")
Signed-off-by: Haoxiang Li <make24@iscas.ac.cn>
---
 drivers/hv/vmbus_drv.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
index 12a707ab73f8..daa584eaa2af 100644
--- a/drivers/hv/vmbus_drv.c
+++ b/drivers/hv/vmbus_drv.c
@@ -1897,7 +1897,13 @@ int vmbus_device_register(struct hv_device *child_device_obj)
 
 	child_device_obj->device.dma_parms = &child_device_obj->dma_parms;
 	child_device_obj->device.dma_mask = &child_device_obj->dma_mask;
-	dma_set_mask(&child_device_obj->device, DMA_BIT_MASK(64));
+	ret = dma_set_mask(&child_device_obj->device, DMA_BIT_MASK(64));
+
+	if (ret) {
+		pr_err("64-bit DMA enable failed\n");
+		kfree(&child_device_obj->device);
+		return ret;
+	}
 
 	/*
 	 * Register with the LDM. This will kick off the driver/device
-- 
2.25.1

RE: [PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask in vmbus_device_register()

[Michael Kelley]

From: Haoxiang Li <make24@iscas.ac.cn> Sent: Sunday, June 30, 2024 7:31 PM
> 
> child_device_obj->device cannot perform DMA properly if dma_set_mask()
> returns non-zero. Add check for dma_set_mask() and return the error if it
> fails. child_device_obj->device is not initialized here, so use kfree() to
> free child_device_obj->device.
> 
> Fixes: 3a5469582c24 ("Drivers: hv: vmbus: Fix initialization of device object in
> vmbus_device_register()")
> Signed-off-by: Haoxiang Li <make24@iscas.ac.cn>
> ---
>  drivers/hv/vmbus_drv.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
> index 12a707ab73f8..daa584eaa2af 100644
> --- a/drivers/hv/vmbus_drv.c
> +++ b/drivers/hv/vmbus_drv.c
> @@ -1897,7 +1897,13 @@ int vmbus_device_register(struct hv_device
> *child_device_obj)
> 
>  	child_device_obj->device.dma_parms = &child_device_obj->dma_parms;
>  	child_device_obj->device.dma_mask = &child_device_obj->dma_mask;
> -	dma_set_mask(&child_device_obj->device, DMA_BIT_MASK(64));
> +	ret = dma_set_mask(&child_device_obj->device, DMA_BIT_MASK(64));
> +
> +	if (ret) {
> +		pr_err("64-bit DMA enable failed\n");
> +		kfree(&child_device_obj->device);
> +		return ret;
> +	}

I commented on a previous attempt to check the return value from
dma_set_mask(). See [1]. See also commit f92a4b50f0bd that adds
some detail about the error handling in vmbus_device_register(). It's
all rather messy. I have not personally worked through the details, so
I'm not sure of what the right cleanup is. But I don't think it is
sufficient to just do the kfree() as your patch proposes. Or if I'm wrong,
and that *is* sufficient, please provides details on the reasoning
why, and let's make sure to record that for future reference! :-)

Michael

[1] https://lore.kernel.org/linux-hyperv/20230607014310.19850-1-jiasheng@iscas.ac.cn/

Re: [PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask in vmbus_device_register()

[Markus Elfring]

> child_device_obj->device cannot perform DMA properly if dma_set_mask()
> returns non-zero. …

Another wording suggestion:
  Direct memory access can not be properly performed any more
  after a dma_set_mask() call failed.


See also:
https://elixir.bootlin.com/linux/v6.10-rc6/source/kernel/dma/mapping.c#L804


>      … child_device_obj->device is not initialized here, so use kfree() to
> free child_device_obj->device.

How did you come to the conclusion that meaningful data processing
would still be possible according to such information?


…
> Signed-off-by: Haoxiang Li <make24@iscas.ac.cn>

I find it interesting that another personal name is presented here.
I noticed that some patches were published with the name “Ma Ke” previously.
How will requirements be resolved for the Developer's Certificate of Origin?
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.10-rc6#n398


Would you like to append parentheses to another function name in the summary phrase?

Regards,
Markus