Re: [PATCH net-next v3 1/2] dpll: zl3073x: update all DPLL channels on ref_sync_set

[Ivan Vecera <ivecera@redhat.com>]

Sahisko findings and my replies...

 > Is the loss of the error return here intentional?
 > ...
 > isn't the pair being already armed exactly the reason the sync input
 > must not stay selectable?

Yes, this is intentional. The ref sync pair is configured by the
preceding zl3073x_ref_state_set() call and is fully functional at
that point - the priority exclusion is a separate, advisory step
recommended by the datasheet. It prevents the automatic selection
algorithm from picking the sync source as a clock reference, which
would be counterproductive but does not affect the sync pair itself.

A failure in zl3073x_chan_state_set() here means a mailbox or bus
error, which is highly unlikely during normal operation and would
indicate a serious hardware problem that surfaces through other
error paths as well. Propagating it to userspace would falsely
suggest that the ref sync pair was not configured, when in fact it
was. The dev_warn is sufficient to flag the condition for debugging.

The old code propagated the error only because it was a single
return statement at the end of the function - not because the
priority update was considered essential for the operation to
succeed.

Ivan