Re: [PATCH 3/8] bdi: add a ->dev_name field to struct backing_dev_info

[Yufen Yu]

Hi,

On 2020/4/16 15:15, Christoph Hellwig wrote:
> Cache a copy of the name for the life time of the backing_dev_info
> structure so that we can reference it even after unregistering.
>
> Fixes: 68f23b89067f ("memcg: fix a crash in wb_workfn when a device disappears")
> Reported-by: Yufen Yu <yuyufen@xxxxxxxxxx>
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
> ---
>   include/linux/backing-dev-defs.h |  1 +
>   mm/backing-dev.c                 | 13 ++++++++++---
>   2 files changed, 11 insertions(+), 3 deletions(-)
>
> diff --git a/include/linux/backing-dev-defs.h b/include/linux/backing-dev-defs.h
> index 4fc87dee005a..249590bcccf7 100644
> --- a/include/linux/backing-dev-defs.h
> +++ b/include/linux/backing-dev-defs.h
> @@ -220,6 +220,7 @@ struct backing_dev_info {
>   	wait_queue_head_t wb_waitq;
>   
>   	struct device *dev;
> +	const char *dev_name;
>   	struct device *owner;
>   
>   	struct timer_list laptop_mode_wb_timer;
> diff --git a/mm/backing-dev.c b/mm/backing-dev.c
> index c2c44c89ee5d..4f6c05df72f9 100644
> --- a/mm/backing-dev.c
> +++ b/mm/backing-dev.c
> @@ -938,9 +938,15 @@ int bdi_register_va(struct backing_dev_info *bdi, const char *fmt, va_list args)
>   	if (bdi->dev)	/* The driver needs to use separate queues per device */
>   		return 0;
>   
> -	dev = device_create_vargs(bdi_class, NULL, MKDEV(0, 0), bdi, fmt, args);
> -	if (IS_ERR(dev))
> +	bdi->dev_name = kvasprintf(GFP_KERNEL, fmt, args);
> +	if (!bdi->dev_name)
> +		return -ENOMEM;
> +
> +	dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, bdi->dev_name);
> +	if (IS_ERR(dev)) {
> +		kfree(bdi->dev_name);
>   		return PTR_ERR(dev);
> +	}
>   
>   	cgwb_bdi_register(bdi);
>   	bdi->dev = dev;
> @@ -1034,6 +1040,7 @@ static void release_bdi(struct kref *ref)
>   	WARN_ON_ONCE(bdi->dev);
>   	wb_exit(&bdi->wb);
>   	cgwb_bdi_exit(bdi);
> +	kfree(bdi->dev_name);
>   	kfree(bdi);
>   }


When driver try to to re-register bdi but without release_bdi(), the old dev_name
will be cover directly by the newer in bdi_register_va(). So, I am not sure whether
it can cause memory leak for bdi->dev_name.

Thanks,
Yufen

>   
> @@ -1047,7 +1054,7 @@ const char *bdi_dev_name(struct backing_dev_info *bdi)
>   {
>   	if (!bdi || !bdi->dev)
>   		return bdi_unknown_name;
> -	return dev_name(bdi->dev);
> +	return bdi->dev_name;
>   }
>   EXPORT_SYMBOL_GPL(bdi_dev_name);