[PATCH v2 6/9] phy: qcom: Add M31 based eUSB2 PHY driver

[Konrad Dybcio]

On 3/19/25 8:03 PM, Wesley Cheng wrote:
> Hi Konrad,
> 
> On 3/11/2025 4:19 AM, Konrad Dybcio wrote:
>> On 3/4/25 10:56 PM, Melody Olvera wrote:
>>> From: Wesley Cheng <quic_wcheng at quicinc.com>
>>>
>>> SM8750 utilizes an eUSB2 PHY from M31.  Add the initialization
>>> sequences to bring it out of reset and into an operational state.  This
>>> differs to the M31 USB driver, in that the M31 eUSB2 driver will
>>> require a connection to an eUSB2 repeater.  This PHY driver will handle
>>> the initialization of the associated eUSB2 repeater when required.
>>>
>>> Signed-off-by: Wesley Cheng <quic_wcheng at quicinc.com>
>>> Signed-off-by: Melody Olvera <quic_molvera at quicinc.com>
>>> ---
>>
>> [...]
>>
>>> +static int msm_m31_eusb2_write_readback(void __iomem *base, u32 offset,
>>> +					const u32 mask, u32 val)
>>> +{
>>> +	u32 write_val;
>>> +	u32 tmp;
>>> +
>>> +	tmp = readl_relaxed(base + offset);
>>> +	tmp &= ~mask;
>>> +	write_val = tmp | val;
>>> +
>>> +	writel_relaxed(write_val, base + offset);
>>> +
>>> +	tmp = readl_relaxed(base + offset);
>>> +	tmp &= mask;
>>> +
>>> +	if (tmp != val) {
>>> +		pr_err("write: %x to offset: %x FAILED\n", val, offset);
>>> +		return -EINVAL;
>>> +	}
>>> +
>>> +	return 0;
>>
>> Is there a reason we need to read back every write?
>>
>> Does this have to do with some funny write buffering?
>>
> 
> Probably because its just a form of write synchronization, since we're
> using the relaxed variants.  If desired I can switch to just using writel
> and remove the readback.

non-relaxed variants are defined something like:


writel(foo) {
	writel_relaxed(foo);
	wmb();
}

with readbacks enforcing much stronger ordering (via a data/address
dependency) than a barrier, i.e. if you write to an address and read back the
register, the write must have arrived at the destination hardware (which is
not a given otherwise, see:

2f8cf2c3f3e3 ("clk: qcom: reset: Ensure write completion on reset de/assertion")

Konrad