Re: [PATCH net-next 1/4] dpll: zl3073x: add channel ToD and phase step operations

[Ivan Vecera <ivecera@redhat.com>]

On 7/10/26 2:56 PM, Vadim Fedorenko wrote:
> On 08/07/2026 18:05, Ivan Vecera wrote:
>> Add low-level DPLL channel operations for ToD read/write/adjust,
>> delta frequency offset write and output phase step. ToD operations
>> use a wait-before-write pattern to avoid blocking after each
>> operation. tod_adjust additionally waits for completion since callers
>> may follow with phase step operations.
>>
>> The tod_ready_wait helper selects the poll timeout based on the
>> current ToD command - write operations use a longer timeout (1000 ms)
>> than reads (30 ms).
>>
>> The ToD read captures system timestamps (ptp_system_timestamp) around
>> the HW command and completion poll to support cross-timestamping.
>>
>> Add output step-time mask invariant to zl3073x_chan and
>> zl3073x_chan_is_out_stepped() helper to check if an output
>> participates in step-time operations.
>>
> 
> [...]
> 
>> +/**
>> + * zl3073x_chan_tod_read - read ToD registers after issuing a command
>> + * @zldev: pointer to zl3073x device
>> + * @ch: DPLL channel index
>> + * @next_hz: if true, read predicted ToD at next 1 Hz; otherwise read 
>> current
>> + * @ts: timespec to store the result
>> + * @sts: optional system timestamp pair for cross-timestamping
>> + *
>> + * Context: Caller must serialize all zl3073x_chan_tod_* calls 
>> externally.
>> + * Return: 0 on success, <0 on error
>> + */
>> +int zl3073x_chan_tod_read(struct zl3073x_dev *zldev, u8 ch,
>> +              bool next_hz, struct timespec64 *ts,
>> +              struct ptp_system_timestamp *sts)
>> +{
>> +    u32 nsec;
>> +    u64 sec;
>> +    u8 cmd;
>> +    int rc;
>> +
>> +    if (next_hz)
>> +        cmd = ZL_DPLL_TOD_CTRL_CMD_RD_NEXT_1HZ;
>> +    else
>> +        cmd = ZL_DPLL_TOD_CTRL_CMD_RD_CURRENT;
>> +
>> +    /* Wait for any previous ToD operation to complete */
>> +    rc = zl3073x_chan_tod_ready_wait(zldev, ch);
>> +    if (rc)
>> +        return rc;
>> +
>> +    ptp_read_system_prets(sts);
>> +    rc = zl3073x_chan_tod_ctrl(zldev, ch, cmd);
>> +    if (rc)
>> +        return rc;
>> +
>> +    rc = zl3073x_chan_tod_ready_wait(zldev, ch);
>> +    if (rc)
>> +        return rc;
>> +    ptp_read_system_postts(sts);
> 
> AFAIU, this code means that the ToD value was somewhere between tod_ctrl
> command and tod_ready read value 0 of the register. How does it work
> with "predicted ToD at next 1 Hz"?

The ptp_system_timestamp is only used by the gettimex64 callback which
always calls tod_read with next_hz=false (current ToD). The next_hz=true
path is only used internally by tod_adjust() which passes sts=NULL.

So in practice, system timestamps are never captured around a predicted
ToD read. The function accepts both options through the same interface
for simplicity, but the sts parameter is only meaningful with
next_hz=false.

Thanks,
Ivan