2023-06-07 01:53:20

by Jiasheng Jiang

Subject: [PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask

Add check for dma_set_mask() and return the error if it fails.

Fixes: 6bf625a4140f ("Drivers: hv: vmbus: Rework use of DMA_BIT_MASK(64)")
Signed-off-by: Jiasheng Jiang <[email protected]>
---
drivers/hv/vmbus_drv.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
index 1c65a6dfb9fa..68b7be2762ea 100644
--- a/drivers/hv/vmbus_drv.c
+++ b/drivers/hv/vmbus_drv.c
@@ -1908,7 +1908,11 @@ 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) {
+ put_device(&child_device_obj->device);
+ return ret;
+ }

/*
* Register with the LDM. This will kick off the driver/device
--
2.25.1

2023-06-08 15:24:48

by Michael Kelley (LINUX)

Subject: RE: [PATCH] Drivers: hv: vmbus: Add missing check for dma_set_mask

From: Jiasheng Jiang <[email protected]> Sent: Tuesday, June 6, 2023 6:43 PM
>
> Add check for dma_set_mask() and return the error if it fails.
>
> Fixes: 6bf625a4140f ("Drivers: hv: vmbus: Rework use of DMA_BIT_MASK(64)")
> Signed-off-by: Jiasheng Jiang <[email protected]>
> ---
> drivers/hv/vmbus_drv.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
> index 1c65a6dfb9fa..68b7be2762ea 100644
> --- a/drivers/hv/vmbus_drv.c
> +++ b/drivers/hv/vmbus_drv.c
> @@ -1908,7 +1908,11 @@ 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) {
> + put_device(&child_device_obj->device);

I don't think the put_device() call is correct here. The underlying kobj in
child_device_obj->device is not initialized until device_register() calls
device_initialize(). It's after device_initialize() is called that put_device()
must be used.

That said, I don't see that the memory for the child_device_obj gets
freed if we just do "return ret", though maybe I'm missing it. And there's
the matter of the memory allocated by dev_set_name(). Getting this
error path fully correct may be rather awkward. :-(

Michael

> + return ret;
> + }
>
> /*
> * Register with the LDM. This will kick off the driver/device
> --
> 2.25.1