Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <5f4a78e7-91dc-4d82-c46f-866f3492b674@xs4all.nl>
Date:   Thu, 17 Aug 2023 13:41:37 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.12.0
Subject: Re: [PATCH] media: vb2: frame_vector.c: replace WARN_ONCE with a
 comment
Content-Language: en-US, nl
From:   Hans Verkuil <hverkuil-cisco@xs4all.nl>
To:     Linux Media Mailing List <linux-media@vger.kernel.org>
Cc:     Marek Szyprowski <m.szyprowski@samsung.com>,
        Tomasz Figa <tfiga@chromium.org>,
        Linux Kernel <linux-kernel@vger.kernel.org>,
        Jan Kara <jack@suse.cz>, David Hildenbrand <david@redhat.com>,
        Linus Torvalds <torvalds@linux-foundation.org>
References: <d5e1ee76-75b3-26cb-23ae-cf6ab40597b7@xs4all.nl>
In-Reply-To: <d5e1ee76-75b3-26cb-23ae-cf6ab40597b7@xs4all.nl>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 17/08/2023 12:41, Hans Verkuil wrote:
> The WARN_ONCE was issued also in cases that had nothing to do with VM_IO
> (e.g. if the start address was just a random value and uaccess fails with
> -EFAULT).
> 
> There are no reports of WARN_ONCE being issued for actual VM_IO cases, so
> just drop it and instead add a note to the comment before the function.
> 
> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

I forgot to add:

Reported-by: Yikebaer Aizezi <yikebaer61@gmail.com>


> ---
> diff --git a/drivers/media/common/videobuf2/frame_vector.c b/drivers/media/common/videobuf2/frame_vector.c
> index 0f430ddc1f67..fd87747be9b1 100644
> --- a/drivers/media/common/videobuf2/frame_vector.c
> +++ b/drivers/media/common/videobuf2/frame_vector.c
> @@ -31,6 +31,10 @@
>   * different type underlying the specified range of virtual addresses.
>   * When the function isn't able to map a single page, it returns error.
>   *
> + * Note that get_vaddr_frames() cannot follow VM_IO mappings. It used
> + * to be able to do that, but that could (racily) return non-refcounted
> + * pfns.
> + *
>   * This function takes care of grabbing mmap_lock as necessary.
>   */
>  int get_vaddr_frames(unsigned long start, unsigned int nr_frames, bool write,
> @@ -59,8 +63,6 @@ int get_vaddr_frames(unsigned long start, unsigned int nr_frames, bool write,
>  	if (likely(ret > 0))
>  		return ret;
> 
> -	/* This used to (racily) return non-refcounted pfns. Let people know */
> -	WARN_ONCE(1, "get_vaddr_frames() cannot follow VM_IO mapping");
>  	vec->nr_frames = 0;
>  	return ret ? ret : -EFAULT;
>  }
>