The new iavf_timesync_read_time() callback converts every error returned by iavf_phc_get_time() to -EIO. The underlying AdminQ path can return more specific errors, including -ENOTSUP and -EINVAL, so this prevents callers from distinguishing an unsupported operation from an actual I/O failure.
Could the original error be propagated instead?
ret = iavf_phc_get_time(adapter, &time);
if (ret != 0)
return ret;

