Re: [PATCH] i2c: core: Lock address during client device instantiation

[Heiner Kallweit <hkallweit1@gmail.com>]

On 16.08.2024 11:23, Wolfram Sang wrote:
> Hi Heiner,
> 
>> I didn't find a single dts(i) with a 10bit i2c device. Because you said
> 
> In all those years, I didn't even find a device supporting 10 bit
> addresses. And I really looked especially for them. Some controllers
> offer 10-bit support in target mode, but that's all I found.
> 
I found LM8330 which supports 10 bit addressing. However the upper three
bits of supported addresses are always zero, so there's no benefit in
using 10 bit addressing.

> Also, there is only one user of I2C_CLIENT_TEN, and this is only a hack
> to allow instantiating the device at address 0x00. No actual 10-bit
> usage involved. I will remove this as a first step, because this hack is
> terrible anyhow.
> 
>> that you're tempted to remove 10 bit support: Could the device tree part
>> be a starting point?
> 
> I don't really have a roadmap how to deprecate 10 bit support. Because
> it is exported to userspace, the first question is if we can deprecate
> it, after all. But not much bandwidth even for that, currently.
> 
Yes, removing UAPI functionality may be tricky.
What I meant was that as a starting point we could replace the following in
of_i2c_get_board_info() with an error message stating that 10 bit mode
support has been removed.

	if (addr & I2C_TEN_BIT_ADDRESS) {
		addr &= ~I2C_TEN_BIT_ADDRESS;
		info->flags |= I2C_CLIENT_TEN;
	}

> Happy hacking,
> 
>    Wolfram
> 
Heiner