Hi,
Here is the final version. If it is ok, I will test the code and
publish.
if (enable_lcore_rx_distributor){
// rx and distributor combined, 3 fixed function cores (stat, TX, at
least 1 worker)
min_cores = 4;
num_workers = rte_lcore_count() - 3;
}
else{
// separate rx and distributor, 3 fixed function cores (stat, TX, at
least 1 worker)
min_cores = 5;
num_workers = rte_lcore_count() - 4;
}
On 28.06.2022 14:25, Hunt, David wrote:
On 28/06/2022 12:06, Omer Yamac wrote:
Hi David,
I have one more question. When I was working on new patch, I just want
to make sure what we are doing.
On 27.06.2022 18:51, Hunt, David wrote:
Hi Ömer,
I've a few comments:
On 21/06/2022 21:15, Abdullah Ömer Yamaç wrote:
--clipped--
@@ -39,6 +39,7 @@ volatile uint8_t quit_signal_rx;
volatile uint8_t quit_signal_dist;
volatile uint8_t quit_signal_work;
unsigned int power_lib_initialised;
+bool enable_lcore_rx_distributor;
static volatile struct app_stats {
struct {
--clipped--
@@ -724,7 +794,12 @@ main(int argc, char *argv[])
if (ret < 0)
rte_exit(EXIT_FAILURE, "Invalid distributor
parameters\n");
- if (rte_lcore_count() < 5)
+ if (enable_lcore_rx_distributor)
+ num_workers = rte_lcore_count() - 3;
+ else
+ num_workers = rte_lcore_count() - 4;
+
This could be "num_workers = rte_lcore_count() - (4 -
enable_lcore_rx_distributor)".
For the "if-else" case of enable_lcore_rx_distributor, we will reduce
the line of codes; but I am not sure about that change. Because the
type of the variable is bool and we are using arithmetic operation on
that variable. I think it is a little bit harder for people to
understand operation. Am I right? I can suggest one more solution. We
may change the data type to "unsigned int" or Is it okay to leave as
before?
--clipped--
Hi Ömer,
You raise a good point about readability. Let's leave it as you had
it originally. Maybe just add a couple of one-line comments? "rx and
distributor combined, 3 fixed function cores" and "separate rx and
distributor, 4 fixed function cores?
Rgds,
Dave.
