I am trying to get HW timestamping work in a dpdk multi-process environment.
To make HW timestamping work, i referenced the rxtx_callbacks example. It uses rte_eth_read_clock(). In the description of this function it says:

E.g, a simple heuristic to derivate the frequency would be: 
uint64_t start, end; rte_eth_read_clock(port, start); 
rte_delay_ms(100); 
rte_eth_read_clock(port, end); 
double freq = (end - start) * 10;

Compute a common reference with: 
uint64_t base_time_sec = current_time(); 
uint64_t base_clock; rte_eth_read_clock(port, base_clock);

Then, convert the raw mbuf timestamp with: 
base_time_sec + (double)(*timestamp_dynfield(mbuf) - base_clock) / freq;

I follow this approach, I get it working for single process. Then I modify the symmetric multi process example, so that each instance calculates these params (freq, base_time, base_clock). You can find it here
Apparantely the secondary process cannot call rte_eth_read_clock(), gives this error:

EAL: Detected CPU lcores: 128
EAL: Detected NUMA nodes: 2
EAL: Auto-detected process type: SECONDARY
EAL: Detected shared linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket_2429753_54915832f7e13
EAL: Selected IOVA mode 'VA'
EAL: VFIO support initialized
EAL: Probe PCI driver: mlx5_pci (15b3:101d) device: 0000:ca:00.0 (socket 1)
EAL: Probe PCI driver: mlx5_pci (15b3:101d) device: 0000:ca:00.1 (socket 1)
APP: Finished Process Init.
MLNX_DPDK: Cannot read device clock, err: Unknown error -95
EAL: Error - exiting with code: 1
  Cause: Cannot configure dpdk system
EAL: failed to send to (/var/run/dpdk/rte/mp_socket) due to Bad file descriptor
EAL: Cannot send message to primary

I then change it so that primary process shares the nic_freq, base_time, base_clock with secondary process. I then calculate the actual timestamp for each packet like as mentioned,

pkt_timestamp_ns = base_time_ns + (double)(*hwts_field(bufs[i]) - base_nic_clock) / nic_freq;

here, hwts_field is just like it is in rxtx_callbacks

static inline rte_mbuf_timestamp_t *
hwts_field(struct rte_mbuf *mbuf)
{
    return RTE_MBUF_DYNFIELD(mbuf, hwts_dynfield_offset, rte_mbuf_timestamp_t *);
}

Primary process is able to get correct timestamps. But Secondary process gets wrong value from hwts_field(). Here’s the output for secondary process

Received 1 packets on port 1 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1617275617280, base_clock: 533084928073076
Skipping
Received 1 packets on port 1 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1617275011072, base_clock: 533084928073076
Skipping
Received 1 packets on port 1 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1617274404864, base_clock: 533084928073076
Skipping
Received 1 packets on port 0 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1616080175104, base_clock: 533084928073076
Skipping
Received 1 packets on port 1 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1617273798656, base_clock: 533084928073076
Skipping
Received 1 packets on port 1 | q 0 in this burst
ts: 18446744073709551615
rx_ts: 1617273192448, base_clock: 533084928073076
Skipping

ts: calculated timestamp in ns
rx_ts: clock cycle at packet rx
base_clock: base clock cycle calculated at primary proc start.

The rx_ts is lower than base_clock! and ts equals 2^64-1

So, my two questions would be

1. Is there a better/different way to get packet timestamps in ns?
2. Why am I getting a lower clock value after I have read the base clock cycle?