linux-kernel - Re: [alsa-devel] [RFC PATCH 1/7] soundwire: Add sysfs support for master(s)

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: <b0059709-027e-26c4-25a1-bd55df7c507f@linux.intel.com>
Date:   Mon, 6 May 2019 21:24:17 -0500
From:   Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To:     Greg KH <gregkh@...uxfoundation.org>
Cc:     alsa-devel@...a-project.org, tiwai@...e.de,
        linux-kernel@...r.kernel.org, liam.r.girdwood@...ux.intel.com,
        vkoul@...nel.org, broonie@...nel.org,
        srinivas.kandagatla@...aro.org, jank@...ence.com, joe@...ches.com,
        Sanyog Kale <sanyog.r.kale@...el.com>
Subject: Re: [alsa-devel] [RFC PATCH 1/7] soundwire: Add sysfs support for
 master(s)


>> +int sdw_sysfs_bus_init(struct sdw_bus *bus)
>> +{
>> +	struct sdw_master_sysfs *master;
>> +	int err;
>> +
>> +	if (bus->sysfs) {
>> +		dev_err(bus->dev, "SDW sysfs is already initialized\n");
>> +		return -EIO;
>> +	}
>> +
>> +	master = kzalloc(sizeof(*master), GFP_KERNEL);
>> +	if (!master)
>> +		return -ENOMEM;
> 
> Why are you creating a whole new device to put all of this under?  Is
> this needed?  What will the sysfs tree look like when you do this?  Why
> can't the "bus" device just get all of these attributes and no second
> device be created?

I tried a quick hack and indeed we could simplify the code with 
something as simple as:

[attributes omitted]

static const struct attribute_group sdw_master_node_group = {
	.attrs = master_node_attrs,
	.name = "mipi-disco"
};

int sdw_sysfs_bus_init(struct sdw_bus *bus)
{
	return sysfs_create_group(&bus->dev->kobj, &sdw_master_node_group);
}

void sdw_sysfs_bus_exit(struct sdw_bus *bus)
{
	sysfs_remove_group(&bus->dev->kobj, &sdw_master_node_group);	
}

which gives me a simpler structure and doesn't require additional 
pretend-devices:

/sys/bus/acpi/devices/PRP00001:00/int-sdw.0/mipi-disco# ls
clock_gears
/sys/bus/acpi/devices/PRP00001:00/int-sdw.0/mipi-disco# more clock_gears
8086

The issue I have is that for the _show() functions, I don't see a way to 
go from the device argument to bus. In the example above I forced the 
output but would need a helper.

static ssize_t clock_gears_show(struct device *dev,
				struct device_attribute *attr, char *buf)
{
	struct sdw_bus *bus; // this is what I need to find from dev
	ssize_t size = 0;
	int i;

	return sprintf(buf, "%d \n", 8086);
}

my brain is starting to fry, but I don't see how container_of() would 
work here since the bus structure contains a pointer to the device. I 
don't also see a way to check for all devices for the bus_type soundwire.
For the slaves we do have a macro based on container_of(), so wondering 
if we made a mistake in the bus definition? Vinod, any thoughts?