Re: 答复: [PATCH] bus: mhi: core: Introduce sysfs ul chan id for mhi chan device

[Jeffrey Hugo <jhugo@codeaurora.org>]

On 10/27/2020 7:18 PM, Carl Yin(殷张成) wrote:
> Hi Jeffery and Hemant:
> 
> On Wednesday, October 28, 2020 6:44 AM, hemantk wrote:
>> On 10/27/20 8:06 AM, Jeffrey Hugo wrote:
>> Hi Carl,
>>
>> On 10/27/20 8:06 AM, Jeffrey Hugo wrote:
>>> On 10/27/2020 3:43 AM, carl.yin@quectel.com wrote:
>>>> From: "carl.yin" <carl.yin@quectel.com>
>>>>
>>>> User space software like ModemManager can identify the function of
>>>> the mhi chan device by ul_chan_id.
>>>>
>>>> Signed-off-by: carl.yin <carl.yin@quectel.com>
>>>> ---
>>>>    Documentation/ABI/stable/sysfs-bus-mhi | 10 ++++++++++
>>>>    drivers/bus/mhi/core/init.c            | 15 +++++++++++++++
>>>>    2 files changed, 25 insertions(+)
>>>>
>>>> diff --git a/Documentation/ABI/stable/sysfs-bus-mhi
>>>> b/Documentation/ABI/stable/sysfs-bus-mhi
>>>> index ecfe766..6d52768 100644
>>>> --- a/Documentation/ABI/stable/sysfs-bus-mhi
>>>> +++ b/Documentation/ABI/stable/sysfs-bus-mhi
>>>> @@ -19,3 +19,13 @@ Description:    The file holds the OEM PK Hash
>>>> value of the endpoint device
>>>>            read without having the device power on at least once, the
>>>> file
>>>>            will read all 0's.
>>>>    Users:        Any userspace application or clients interested in
>>>> device info.
>>>> +
>>>> +What:        /sys/bus/mhi/devices/.../ul_chan_id
>>>> +Date:        November 2020
>>>> +KernelVersion:    5.10
>>>> +Contact:    Carl Yin <carl.yin@quectel.com>
>>>> +Description:    The file holds the uplink chan id of the mhi chan
>>>> device.
>>>> +        User space software like ModemManager can identify the
>>>> function of
>>>> +        the mhi chan device. If the mhi device is not a chan device,
>>>> +        eg mhi controller device, the file read -1.
>>>> +Users:        Any userspace application or clients interested in
>>>> device info.
>>>> diff --git a/drivers/bus/mhi/core/init.c
>>>> b/drivers/bus/mhi/core/init.c index c6b43e9..ac4aa5c 100644
>>>> --- a/drivers/bus/mhi/core/init.c
>>>> +++ b/drivers/bus/mhi/core/init.c
>>>> @@ -105,9 +105,24 @@ static ssize_t oem_pk_hash_show(struct device
>>>> *dev,
>>>>    }
>>>>    static DEVICE_ATTR_RO(oem_pk_hash);
>>>> +static ssize_t ul_chan_id_show(struct device *dev,
>>>> +                struct device_attribute *attr,
>>>> +                char *buf)
>>>> +{
>>>> +    struct mhi_device *mhi_dev = to_mhi_device(dev);
>>>> +    int ul_chan_id = -1;
>>>> +
>>>> +    if (mhi_dev->ul_chan)
>>>> +        ul_chan_id = mhi_dev->ul_chan_id;
>>>> +
>>>> +    return snprintf(buf, PAGE_SIZE, "%d\n", ul_chan_id); } static
>>>> +DEVICE_ATTR_RO(ul_chan_id);
>>>> +
>>>>    static struct attribute *mhi_dev_attrs[] = {
>>>>        &dev_attr_serial_number.attr,
>>>>        &dev_attr_oem_pk_hash.attr,
>>>> +    &dev_attr_ul_chan_id.attr,
>>>>        NULL,
>>>>    };
>>>>    ATTRIBUTE_GROUPS(mhi_dev);
>>>>
>>>
>>> NACK
>>>
>>> Channel ID is a device specific detail.  Userspace should be basing
>>> decisions on the channel name.
>>>
>> I agree with Jeff, why do you need to know the channel id, if you need to poll for
>> any device node to get created you can try to open the device node from user
>> space and wait until the device gets opened.
>> Are you trying to wait for EDL channels to get started using UCI ?
> [carl.yin] In my opinion, mhi chan id is something like 'bInterfaceNumber' of USB device.
> A USB device and several USB interfaces, and a mhi devices have 128 mhi chans.
> Chan id is a physical attribute of one mhi chan.
> 
> Next is the udev info of one mhi chan:
> # udevadm info -a /dev/mhi_0000\:03\:00.0_EDL
>    looking at parent device '/devices/pci0000:00/0000:00:1d.0/0000:03:00.0/0000:03:00.0_EDL':
>      KERNELS=="0000:03:00.0_EDL"
>      SUBSYSTEMS=="mhi"
>      DRIVERS=="mhi_uci"
>      ATTRS{serial_number}=="Serial Number: 2644481182"
>      ATTRS{ul_chan_id}=="34"
> 
> If no ul_chan_id, the udev ruler will be ' KERNEL=="*_EDL" '

I have several usecases where this works just fine today.

> With ul_chan_id, the udev ruler will be ' ATTRS{ul_chan_id}=="34"'

This breaks when there is some new device that has the EDL channel on 
some different chan_id, like 7.  The above does not.  Additionally if 
there is a different device that is using chan_id 34 for a different 
purpose, say Diag, then your udev rule also breaks.

The name of the channel is the interface to the channel.  Not the 
chan_id.  This holds true within the kernel, and should be the same for 
userspace.  I still oppose this change.

-- 
Jeffrey Hugo
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.