On 2024-10-10 12:45, David Marchand wrote:
On Fri, Sep 20, 2024 at 12:57 PM Mattias Rönnblom <[email protected]> wrote:+ static int \ + run_parallel_test_and_modify ## size(void *arg) \ + { \ + struct parallel_test_and_set_lcore ## size *lcore = arg; \ + uint64_t deadline = rte_get_timer_cycles() + \ + PARALLEL_TEST_RUNTIME * rte_get_timer_hz(); \ + do { \ + bool old_value; \ + bool new_value = rte_rand() & 1; \ + bool use_assign = rte_rand() & 1; \ + \ + if (use_assign) \ + old_value = rte_bit_atomic_test_and_assign( \ + lcore->word, lcore->bit, new_value, \ + rte_memory_order_relaxed); \ + else \ + old_value = new_value ? \ + rte_bit_atomic_test_and_set( \ + lcore->word, lcore->bit, \ + rte_memory_order_relaxed) : \ + rte_bit_atomic_test_and_clear( \ + lcore->word, lcore->bit, \ + rte_memory_order_relaxed); \ + if (old_value != new_value) \ + lcore->flips++; \ + } while (rte_get_timer_cycles() < deadline); \ + \ + return 0; \ + } \ + \ + static int \ + test_bit_atomic_parallel_test_and_modify ## size(void) \ + { \ + unsigned int worker_lcore_id; \ + uint ## size ## _t word = 0; \ + unsigned int bit = rte_rand_max(size); \ + struct parallel_test_and_set_lcore ## size lmain = { \ + .word = &word, \ + .bit = bit \ + }; \ + struct parallel_test_and_set_lcore ## size lworker = { \ + .word = &word, \ + .bit = bit \ + }; \ + \ + if (rte_lcore_count() < 2) { \ + printf("Need multiple cores to run parallel test.\n"); \ + return TEST_SKIPPED; \ + } \ + \ + worker_lcore_id = rte_get_next_lcore(-1, 1, 0); \ + \ + int rc = rte_eal_remote_launch(run_parallel_test_and_modify ## size, \ + &lworker, worker_lcore_id); \ + TEST_ASSERT(rc == 0, "Worker thread launch failed"); \ + \ + run_parallel_test_and_modify ## size(&lmain); \ + \ + rte_eal_mp_wait_lcore(); \ + \ + uint64_t total_flips = lmain.flips + lworker.flips; \ + bool expected_value = total_flips % 2; \ + \ + TEST_ASSERT(expected_value == rte_bit_test(&word, bit), \ + "After %"PRId64" flips, the bit value " \ + "should be %d", total_flips, expected_value); \ + \ + uint64_t expected_word = 0; \ + rte_bit_assign(&expected_word, bit, expected_value); \ + \ + TEST_ASSERT(expected_word == word, "Untouched bits have " \ + "changed value"); \ + \ + return TEST_SUCCESS; \ + } + +GEN_TEST_BIT_PARALLEL_TEST_AND_MODIFY(32) +GEN_TEST_BIT_PARALLEL_TEST_AND_MODIFY(64)It appears this test failed once in the CI for an unrelated series (uAPI kernel header import): https://lab.dpdk.org/results/dashboard/testruns/logs/1385993/ + TestCase [ 0] : test_bit_access32 succeeded + TestCase [ 1] : test_bit_access64 succeeded + TestCase [ 2] : test_bit_access32 succeeded + TestCase [ 3] : test_bit_access64 succeeded + TestCase [ 4] : test_bit_v_access32 succeeded + TestCase [ 5] : test_bit_v_access64 succeeded + TestCase [ 6] : test_bit_atomic_access32 succeeded + TestCase [ 7] : test_bit_atomic_access64 succeeded + TestCase [ 8] : test_bit_atomic_v_access32 succeeded + TestCase [ 9] : test_bit_atomic_v_access64 succeeded + TestCase [10] : test_bit_atomic_parallel_assign32 succeeded + TestCase [11] : test_bit_atomic_parallel_assign64 succeeded + TestCase [12] : test_bit_atomic_parallel_test_and_modify32 failed + TestCase [13] : test_bit_atomic_parallel_test_and_modify64 succeeded + TestCase [14] : test_bit_atomic_parallel_flip32 succeeded + TestCase [15] : test_bit_atomic_parallel_flip64 succeeded + TestCase [16] : test_bit_relaxed_set succeeded + TestCase [17] : test_bit_relaxed_clear succeeded + TestCase [18] : test_bit_relaxed_test_set_clear succeeded EAL: Test assert test_bit_atomic_parallel_test_and_modify32 line 236 failed: After 1070523 flips, the bit value should be 1 Please have a look.
OK. Nothing obvious from what I can see in the code. Unrelated: why did you remove all empty lines in the "template" macros? Makes them much harder to read.
I've been unable to reproduce this on my Raptor Lake x86_64 with GCC 12.3 (CI machine used GCC 12.2).
I'll try if I have better luck on some other systems.
