On 3/16/2021 4:37 PM, Ferruh Yigit wrote:
On 3/16/2021 1:39 PM, Thomas Monjalon wrote:
16/03/2021 14:05, Ferruh Yigit:
On 3/12/2021 12:12 PM, Qi Zhang wrote:
PMDs use RTE_LIBRTE_<PMD_NAME>_DEBUG_RX|TX as compile option to wrap
data path debug code. As .config has been removed since the meson build,
It is not friendly for new DPDK users to notice those debug options.

The patch introduces below compile options for specific Rx/Tx data path
debug, so PMD can choose to reuse them to avoid maintain their own.

- RTE_LIBRTE_ETHDEV_DEBUG_RX
- RTE_LIBRTE_ETHDEV_DEBUG_TX

Also, all the compile options are documented on the overview page, so
users can easily find them.

<...>

+
+- ``RTE_LIBRTE_ETHDEV_DEBUG`` (default **disabled**)
+
+  Compile with debug code on data path.

What is data path if not Rx or Tx?

+
+- ``RTE_LIBRTE_ETHDEV_DEBUG_RX`` (default **disabled**)
+
+  Compile with debug code on Rx data path.
+
+- ``RTE_LIBRTE_ETHDEV_DEBUG_TX`` (default **disabled**)
+
+  Compile with debug code on Tx data path.

In general, I think "LIBRTE_" is redundant and useless as macro prefix.



'RTE_LIBRTE_ETHDEV_DEBUG' already exits, it enables datapath debug without distinguishing Rx or Tx.

When we have _RX and _TX macro variants now, it may be possible to
1- get rid of 'RTE_LIBRTE_ETHDEV_DEBUG' macro and continue with fine grained RX & TX
2- Use 'RTE_LIBRTE_ETHDEV_DEBUG' as an alias to enable both RX & TX
3- Keep 'RTE_LIBRTE_ETHDEV_DEBUG' for ethdev layer datapath debug, and RX & TX variants for PMDs.

I think (3) can be more backward compatible, and can be helpful to separate ethdev layer and PMD debugging, but no so strong opinion.

Hi Qi,

Reminder of above discussion, it is not addressed in v4. What do you think option 3 above?

Reply via email to