This patch fixes segmentation fault in pmd_cyclecount_bench_ops
function in case when state->opts->nb_descriptors is not
natural multiple of burst size.
To reproduce run: dpdk-test-crypto-perf with params:
--ptest pmd-cyclecount --pmd-cyclecount-delay-ms 5 \
--devtype crypto_qat --optype cipher-then-auth \
--cipher-algo aes-cbc --cipher-op encrypt \
--cipher-key-sz 16 --cipher-iv-sz 16 \
--auth-algo sha2-256-hmac \
--auth-op generate --auth-key-sz 64 --digest-sz 32 \
--total-ops 10000 --burst-sz 255 --buffer-sz 1024 --silent
Fixes:
96dfeb609be1 ("app/crypto-perf: add new PMD benchmarking mode")
Cc: stable@dpdk.org
Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
for (cur_iter_op = 0; cur_iter_op < iter_ops_needed;
cur_iter_op += test_burst_size) {
- uint32_t burst_size = RTE_MIN(state->opts->total_ops - cur_op,
+ uint32_t burst_size = RTE_MIN(iter_ops_needed - cur_iter_op,
test_burst_size);
struct rte_crypto_op **ops = &state->ctx->ops[cur_iter_op];