Hi Ibtisam,
> -----Original Message-----
> From: dev <dev-boun...@dpdk.org> On Behalf Of Ibtisam Tariq
> Sent: Wednesday, July 22, 2020 1:50 PM
> To: Doherty, Declan <declan.dohe...@intel.com>
> Cc: dev@dpdk.org; Ibtisam Tariq <ibtisam.ta...@emumba.com>
> Subject: [dpdk-dev] [PATCH 2/2] examples/l2fwd-crypto: Free resources on
> exit
>
> When exiting the application, resources should be cleared.
>
> Signed-off-by: Ibtisam Tariq <ibtisam.ta...@emumba.com>
> ---
> examples/l2fwd-crypto/main.c | 31 ++++++++++++++++++++++++++++++-
> 1 file changed, 30 insertions(+), 1 deletion(-)
>
> diff --git a/examples/l2fwd-crypto/main.c b/examples/l2fwd-crypto/main.c
> index 827da9b3e..5b00e2ade 100644
> --- a/examples/l2fwd-crypto/main.c
> +++ b/examples/l2fwd-crypto/main.c
> @@ -18,6 +18,7 @@
> #include <getopt.h>
> #include <fcntl.h>
> #include <unistd.h>
> +#include <signal.h>
>
> #include <rte_string_fns.h>
> #include <rte_atomic.h>
> @@ -47,6 +48,8 @@
> #include <rte_cryptodev_scheduler.h>
> #endif
>
> +static volatile bool force_quit;
> +
> enum cdev_type {
> CDEV_TYPE_ANY,
> CDEV_TYPE_HW,
> @@ -838,7 +841,7 @@ l2fwd_main_loop(struct l2fwd_crypto_options
> *options)
> * so user can see the crypto information.
> */
> prev_tsc = rte_rdtsc();
> - while (1) {
> + while (!force_quit) {
>
> cur_tsc = rte_rdtsc();
>
> @@ -1738,8 +1741,12 @@ check_all_ports_link_status(uint32_t port_mask)
> printf("\nChecking link status");
> fflush(stdout);
> for (count = 0; count <= MAX_CHECK_TIME; count++) {
> + if (force_quit)
> + return;
> all_ports_up = 1;
> RTE_ETH_FOREACH_DEV(portid) {
> + if (force_quit)
> + return;
> if ((port_mask & (1 << portid)) == 0)
> continue;
> memset(&link, 0, sizeof(link));
> @@ -2662,6 +2669,16 @@ reserve_key_memory(struct
> l2fwd_crypto_options *options)
> options->aad.phys_addr = rte_malloc_virt2iova(options->aad.data);
> }
>
> +static void
> +signal_handler(int signum)
> +{
> + if (signum == SIGINT || signum == SIGTERM) {
> + printf("\n\nSignal %d received, preparing to exit...\n",
> + signum);
> + force_quit = true;
> + }
> +}
> +
> int
> main(int argc, char **argv)
> {
> @@ -2681,6 +2698,10 @@ main(int argc, char **argv)
> argc -= ret;
> argv += ret;
>
> + force_quit = false;
> + signal(SIGINT, signal_handler);
> + signal(SIGTERM, signal_handler);
> +
> /* reserve memory for Cipher/Auth key and IV */
> reserve_key_memory(&options);
>
> @@ -2807,6 +2828,14 @@ main(int argc, char **argv)
> if (rte_eal_wait_lcore(lcore_id) < 0)
> return -1;
> }
> + RTE_ETH_FOREACH_DEV(portid) {
> + printf("Closing port %d...", portid);
> + rte_eth_dev_stop(portid);
> + rte_eth_dev_close(portid);
> + printf(" Done\n");
> + }
> + rte_eal_cleanup();
> + printf("Bye...\n");
>
> return 0;
> }
> --
> 2.17.1
Thanks for the patch and sorry for not reviewing for so long.
In addition to Ethdev resources, some Cryptodev resource may need to be cleaned
too, e.g. draining all pending crypto operations from Cryptodev queue and
stop/close the device before closing the application. For SW Crypto PMDs and
QAT not doing the above steps are fine. But I am not sure about if it is a
necessary step for other vendor's PMDs.
Other than that the patch looks good to me.
Akhil what do you think?
Regards,
Fan
