linux-kernel - Re: [PATCH v10 0/8] Add StarFive Camera Subsystem driver

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ed3e726e-4a33-4597-9617-5c380d767fbe@xs4all.nl>
Date:   Wed, 18 Oct 2023 11:56:08 +0200
From:   Hans Verkuil <hverkuil-cisco@...all.nl>
To:     Jack Zhu <jack.zhu@...rfivetech.com>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        Robert Foss <rfoss@...nel.org>,
        Todor Tomov <todor.too@...il.com>, bryan.odonoghue@...aro.org,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Conor Dooley <conor+dt@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Philipp Zabel <p.zabel@...gutronix.de>,
        Laurent Pinchart <laurent.pinchart@...asonboard.com>
Cc:     linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
        devicetree@...r.kernel.org, linux-staging@...ts.linux.dev,
        changhuang.liang@...rfivetech.com
Subject: Re: [PATCH v10 0/8] Add StarFive Camera Subsystem driver

On 18/10/2023 11:52, Jack Zhu wrote:
> 
> 
> On 2023/10/18 17:31, Hans Verkuil wrote:
>> On 18/10/2023 11:25, Jack Zhu wrote:
>>>
>>>
>>> On 2023/10/18 16:50, Hans Verkuil wrote:
>>>> Hi Jack,
>>>>
>>>> On 18/10/2023 04:37, Jack Zhu wrote:
>>>>
>>>> <snip>
>>>>
>>>>>>> --------------------------------------------------------------------------------
>>>>>>> Compliance test for device /dev/v4l-subdev1:
>>>>>>>
>>>>>>> Driver Info:
>>>>>>> 	Driver version   : 6.6.0
>>>>>>> 	Capabilities     : 0x00000000
>>>>>>
>>>>>> But this does not appear for v4l-subdev1.
>>>>>>
>>>>>> I can't really tell why it doesn't show that. Can you debug a little bit?
>>>>>> The code is in v4l2-compliance.cpp, line 1086:
>>>>>>
>>>>>> ent_id = mi_media_info_for_fd(media_fd, node.g_fd(), &is_invalid, &node.function);
>>>>>>
>>>>>> The mi_media_info_for_fd() function calls ioctl(media_fd, MEDIA_IOC_DEVICE_INFO, &mdinfo),
>>>>>> and that fails for some reason. It could be that media_fd is invalid (would be weird).
>>>>>>
>>>>>> This could well be a v4l2-compliance bug that you hit with this driver.
>>>>>>
>>>>>
>>>>> On the test board, /dev/v4l-subdev1 is imx219, and the corresponding directory is
>>>>> /sys/dev/char/81:3/device. Media0 does not exist in this directory. Therefore, the media_fd
>>>>> obtained through mi_get_media_fd(node.g_fd(), node.bus_info) is invalid.
>>>>>
>>>>> I don't know why media0 does not exist in /sys/dev/char/81:3/device?
>>>>>
>>>>
>>>> Can you try again with this v4l2-compliance patch?
>>>>
>>>> I need to dig a bit deeper as to why media0 is missing, but for now try this.
>>>>
>>>> Regards,
>>>>
>>>> 	Hans
>>>>
>>>> diff --git a/utils/v4l2-compliance/v4l2-compliance.cpp b/utils/v4l2-compliance/v4l2-compliance.cpp
>>>> index 7169eefe..29475d6b 100644
>>>> --- a/utils/v4l2-compliance/v4l2-compliance.cpp
>>>> +++ b/utils/v4l2-compliance/v4l2-compliance.cpp
>>>> @@ -968,7 +968,7 @@ err:
>>>>  }
>>>>
>>>>  void testNode(struct node &node, struct node &node_m2m_cap, struct node &expbuf_node, media_type type,
>>>> -	      unsigned frame_count, unsigned all_fmt_frame_count)
>>>> +	      unsigned frame_count, unsigned all_fmt_frame_count, int parent_media_fd)
>>>>  {
>>>>  	struct node node2;
>>>>  	struct v4l2_capability vcap = {};
>>>> @@ -997,8 +997,12 @@ void testNode(struct node &node, struct node &node_m2m_cap, struct node &expbuf_
>>>>  		memset(&vcap, 0, sizeof(vcap));
>>>>  	}
>>>>
>>>> -	if (!node.is_media())
>>>> -		media_fd = mi_get_media_fd(node.g_fd(), node.bus_info);
>>>> +	if (!node.is_media()) {
>>>> +		if (parent_media_fd >= 0)
>>>> +			media_fd = parent_media_fd;
>>>> +		else
>>>> +			media_fd = mi_get_media_fd(node.g_fd(), node.bus_info);
>>>> +	}
>>>>
>>>>  	int fd = node.is_media() ? node.g_fd() : media_fd;
>>>>  	if (fd >= 0) {
>>>> diff --git a/utils/v4l2-compliance/v4l2-compliance.h b/utils/v4l2-compliance/v4l2-compliance.h
>>>> index 7caf254b..c47f25f5 100644
>>>> --- a/utils/v4l2-compliance/v4l2-compliance.h
>>>> +++ b/utils/v4l2-compliance/v4l2-compliance.h
>>>> @@ -308,7 +308,7 @@ int check_ustring(const __u8 *s, int len);
>>>>  int check_0(const void *p, int len);
>>>>  int restoreFormat(struct node *node);
>>>>  void testNode(struct node &node, struct node &node_m2m_cap, struct node &expbuf_node, media_type type,
>>>> -	      unsigned frame_count, unsigned all_fmt_frame_count);
>>>> +	      unsigned frame_count, unsigned all_fmt_frame_count, int parent_media_fd = -1);
>>>>  std::string stream_from(const std::string &pixelformat, bool &use_hdr);
>>>>
>>>>  // Media Controller ioctl tests
>>>>
>>>
>>> From the log, there is no change.
>>
>> Oops, my mistake. Also apply this change:
>>
>> diff --git a/utils/v4l2-compliance/v4l2-test-media.cpp b/utils/v4l2-compliance/v4l2-test-media.cpp
>> index 0195ac58..52ab7fb8 100644
>> --- a/utils/v4l2-compliance/v4l2-test-media.cpp
>> +++ b/utils/v4l2-compliance/v4l2-test-media.cpp
>> @@ -612,7 +612,7 @@ void walkTopology(struct node &node, struct node &expbuf_node,
>>  		}
>>
>>  		testNode(test_node, test_node, expbuf_node, type,
>> -			 frame_count, all_fmt_frame_count);
>> +			 frame_count, all_fmt_frame_count, node.g_fd());
>>  		test_node.close();
>>  	}
>>  }
>>
> 
> Can see relevant Info in the log.

Great! Can you do one more thing? Please run 'v4l2-compliance -m /dev/media0 --verbose'
and mail the output to me. It's pretty big, so just email it to me, without CCs.

I want to take a closer look at the output to see why this patch is needed.

Regards,

	Hans