229. Unit Tests: EAL

This section describes the tests that are done to validate the EAL. Each test can be launched independently using the command line interface. These tests are implemented as a linuxapp environment application.

The complete test suite is launched automatically using a python-expect script (launched using make test) that sends commands to the application and checks the results. A test report is displayed on stdout.

229.1. Version

To Be Filled

229.2. Common

To Be Filled

229.3. Eal_fs

To Be Filled

229.4. Memory

  • Dump the mapped memory. The python-expect script checks that at least one line is dumped.
  • Check that memory size is different than 0.
  • Try to read all memory; it should not segfault.

229.5. PCI

  • Register a driver with a devinit() function.
  • Dump all PCI devices.
  • Check that the devinit() function is called at least once.

229.6. Per-lcore Variables and lcore Launch

  • Use rte_eal_mp_remote_launch() to call assign_vars() on every available lcore. In this function, a per-lcore variable is assigned to the lcore_id.
  • Use rte_eal_mp_remote_launch() to call display_vars() on every available lcore. The function checks that the variable is correctly set, or returns -1.
  • If at least one per-core variable was not correct, the test function returns -1.

229.7. Spinlock

  • There is a global spinlock and a table of spinlocks (one per lcore).
  • The test function takes all of these locks and launches the test_spinlock_per_core() function on each core (except the master).
    • The function takes the global lock, displays something, then releases the global lock.
    • The function takes the per-lcore lock, displays something, then releases the per-core lock.
  • The main function unlocks the per-lcore locks sequentially and waits between each lock. This triggers the display of a message for each core, in the correct order. The autotest script checks that this order is correct.

229.8. Rwlock

  • There is a global rwlock and a table of rwlocks (one per lcore).

  • The test function takes all of these locks and launches the test_rwlock_per_core() function on each core (except the master).

    • The function takes the global write lock, display something, then releases the global lock.
    • Then, it takes the per-lcore write lock, display something, and releases the per-core lock.
    • Finally, a read lock is taken during 100 ms, then released.
  • The main function unlocks the per-lcore locks sequentially and waits between each lock. This triggers the display of a message for each core, in the correct order.

    Then, it tries to take the global write lock and display the last message. The autotest script checks that the message order is correct.

229.9. Atomic Variables

  • The main test function performs three subtests. The first test checks that the usual inc/dec/add/sub functions are working correctly:

    • Initialize 32-bit and 64-bit atomic variables to specific values.
    • These variables are incremented and decremented on each core at the same time in test_atomic_usual().
    • The function checks that once all lcores finish their function, the value of the atomic variables are still the same.
  • The second test verifies the behavior of “test and set” functions.

    • Initialize 32-bit and 64-bit atomic variables to zero.
    • Invoke test_atomic_tas() on each lcore before doing anything else. The cores are awaiting synchronization using the while (rte_atomic32_read(&val) == 0) statement which is triggered by the main test function. Then all cores do an rte_atomicXX_test_and_set() at the same time. If it is successful, it increments another atomic counter.
    • The main function checks that the atomic counter was incremented twice only (one for 32-bit and one for 64-bit values).
  • Test “add/sub and return”

    • Initialize 32-bit and 64-bit atomic variables to zero.

    • Invoke test_atomic_addsub_return() on each lcore. Before doing anything else, the cores are waiting a synchro. Each lcore does this operation several times:

      tmp = atomic_add_return(&a, 1);
      atomic_add(&count, tmp);
      tmp = atomic_sub_return(&a, 1);
      atomic_sub(&count, tmp+1);
      
    • At the end of the test, the count value must be 0.

229.10. Prefetch

Just test that the macro can be called and validate the compilation. The test always return success.

229.11. Byteorder functions

Check the result of optimized byte swap functions for each size (16-, 32- and 64-bit).

229.12. Cycles Test

  • Loop N times and check that the timer always increments and never decrements during this loop.
  • Wait one second using rte_usleep() and check that the increment of cycles is correct with regard to the frequency of the timer.

229.13. Logs

  • Enable log types.
  • Set log level.
  • Execute logging functions with different types and levels; some should not be displayed.

229.14. Memzone

  • Search for three reserved zones or reserve them if they do not exist:
    • One is on any socket id.
    • The second is on socket 0.
    • The last one is on socket 1 (if socket 1 exists).
  • Check that the zones exist.
  • Check that the zones are cache-aligned.
  • Check that zones do not overlap.
  • Check that the zones are on the correct socket id.
  • Check that a lookup of the first zone returns the same pointer.
  • Check that it is not possible to create another zone with the same name as an existing zone.

229.15. Memcpy

Create two buffers, and initialize one with random values. These are copied to the second buffer and then compared to see if the copy was successful. The bytes outside the copied area are also checked to make sure they were not changed.

This is repeated for a number of different sizes and offsets, with the second buffer being cleared before each test.

229.16. Debug test

  • Call rte_dump_stack() and rte_dump_registers().

229.17. Alarm

  • Check that the callback for the alarm can to be called.
  • Check that it is not possible to set alarm with invalid time value.
  • Check that it is not possible to set alarm without a callback.
  • Check that it is not possible to cancel alarm without a callback pointer.
  • Check that multiple callbacks for the alarm can be called.
  • Check that the number of removed and unremoved alarms are correct.
  • Check that no callback is called if all alarm removed.
  • Check that it is not possible to cancel an alarm within the callback itself.
  • Check that the callback which is the head of all is able to be removed.
  • Check that all alarms for the same callback can be canceled.

229.18. CPU flags

  • Using the rte_cpu_get_flag_enabled() checks for CPU features from different CPUID tables
  • Checks if rte_cpu_get_flag_enabled() properly fails on trying to check for invalid feature

229.19. Errno

Performs validation on the error message strings provided by the rte_strerror() call, to ensure that suitable strings are returned for the rte-specific error codes, as well as ensuring that for standard error codes the correct error message is returned.

229.20. Interrupts

  • Check that the callback for the specific interrupt can be called.
  • Check that it is not possible to register a callback to an invalid interrupt handle.
  • Check that it is not possible to register no callback to an interrupt handle.
  • Check that it is not possible to unregister a callback to an invalid interrupt handle.
  • Check that multiple callbacks are registered to the same interrupt handle.
  • Check that it is not possible to unregister a callback with invalid parameter.
  • Check that it is not possible to enable an interrupt with invalid handle or wrong handle type.
  • Check that it is not possible to disable an interrupt with invalid handle or wrong handle type.

229.21. Multiprocess

Validates that a secondary DPDK instance can be run alongside a primary when the appropriate EAL command-line flags are passed. Also validates that secondary processes cannot interfere with primary processes by creating memory objects, such as mempools or rings.

229.22. String

Performs validation on the new string functions provided in rte_string_fns.h, ensuring that all values returned are NULL terminated, and that suitable errors are returned when called with invalid parameters.

229.23. Tailq

Validates that we can create and perform lookups on named tail queues within the EAL for various object types. Also ensures appropriate error codes are returned from the functions if invalid parameters are passed.

229.24. Devargs

To Be Filled

229.25. Kvargs

To Be Filled

229.26. Acl

To Be Filled

229.28. Hash

This does unit function test for hash features:

  • Average table utilization when disable extendable table function
  • Average table utilization when enable extendable table function, check could reach 100% utilization

229.29. Hash_perf

This does the performance test with a single thread, including the cases with and without extendable table:

  • Measure cycles for add, lookup, lookup_bulk, delete
  • With/without pre-computed hash values
  • For different key lengths

229.30. Hash_functions

This does unit test for hash functions:

  • Measure cycles for hashing
  • Jhash vs rte_hash_crc
  • For different key lenthgs, seeds

229.31. Hash_multiwriter

This does the performance and function test of multi-threads case – multiple writers.

Introduce scalable multi-writer Cuckoo Hash insertion based on a split cuckoo search and move operation using Intel TSX. It can do scalable hash insertion with 22 cores with little performance loss and negligible TSX abortion rate.

229.32. Hash_readwrite

This does the performance and function test of multi-threads case – multiple reader/writer.

Read-write concurrency support in rte_hash. A new flag value is added to indicate if read-write concurrency is needed during creation time. The new concurrency model is based on rte_rwlock. When Intel TSX is available and the users indicate to use it, the TM version of the rte_rwlock will be called. Both multi-writer and read-write concurrency are protected by the rte_rwlock instead of the x86 specific RTM instructions, so the x86 specific header rte_cuckoo_hash_x86.h is removed and the code is infused into the main .c file. A new rte_hash_count API is proposed to count how many keys are inserted into the hash table.

229.33. Hash_hash_readwrite_lf

This does the unit tests to check for hash lookup and bulk-lookup perf with lock-free enabled and with lock-free disabled. Unit tests performed with readers running in parallel with writers. Tests include:

  • Hash lookup on existing keys
    • Hash add causing NO key-shifts of existing keys in the table
  • Hash lookup on existing keys likely to be on shift-path
    • Hash add causing key-shifts of existing keys in the table
  • Hash lookup on existing keys NOT likely to be on shift-path
    • Hash add causing key-shifts of existing keys in the table
  • Hash lookup on non-existing keys
    • Hash add causing NO key-shifts of existing keys in the table
    • Hash add causing key-shifts of existing keys in the table
  • Hash lookup on keys likely to be on shift-path
    • Multiple writers causing key-shifts of existing keys in the table