4 This is a harness to test the atomicity of certain operations, and to
5 make sure the compiler does not introduce data races in a
6 multi-threaded environment.
8 The basic premise is that we set up testcases such that the thing we
9 want test, say an atomic instruction which stores a double word is in
10 a function of its own. We then run this testcase within GDB,
11 controlled by a gdb script (simulate-thread.gdb). The gdb script will
12 break on the function to be tested, and then single step through every
13 machine instruction in the function. We set this up so GDB can make a
14 couple of inferior function calls before and after each of these
15 single step instructions for a couple of purposes:
17 1. One of the calls simulates another thread running in the
18 process which changes or access memory.
20 2. The other calls are used to verify that we always get the
23 For example, in the case of an atomic store, anyone looking at the
24 memory associated with an atomic variable should never see any in
25 between states. If you have an atomic long long int, and it starts
26 with the value 0, and you write the value MAX_LONG_LONG, any other
27 thread looking at that variable should never see anything other than 0
28 or MAX_LONG_LONG. If you implement the atomic write as a sequence of
29 2 stores, it is possible for another thread to read the location after
30 the first store, but before the second one is complete. That thread
31 would then see an in-between state (one word would still be 0).
33 We simulate this in the testcase by having GDB step through the
34 program, instruction by instruction, and after each step, making an
35 inferior function call which looks at the value of the atomic variable
36 and verifies that it sees either 0 or MAX_LONG_LONG. If it sees any
37 other value, it fails the testcase.
39 This way, we are *sure* there is no in between state because we
40 effectively acted like an OS and switched to another thread after
41 every single instruction of the routine is executed and looked at the
44 We use the same idea to test for data races to see if an illegal load
45 has been hoisted, or that two parallel bitfield writes don't overlap
48 Below is a skeleton of how a test should look like. For more details,
49 look at the tests themselves.
55 /* { dg-options "-some-flags" } */
56 /* { dg-final { simulate-thread } } */
58 /* NOTE: Any failure must be indicated by displaying "FAIL:". */
60 #include "simulate-thread.h"
62 /* Called before each instruction, simulating another thread executing. */
63 void simulate_thread_other_threads()
67 /* Called after each instruction. Returns 1 if any inconsistency is
68 found, 0 otherwise. */
69 int simulate_thread_step_verify()
73 printf("FAIL: reason\n");
79 /* Called at the end of the program (simulate_thread_fini == 1). Verifies
80 the state of the program and returns 1 if any inconsistency is
81 found, 0 otherwise. */
82 int simulate_thread_final_verify()
86 printf("FAIL: reason\n");
92 /* The gdb script will break on simulate_thread_main(), so make sure
93 GCC does not inline it, thus making the break point fail. */
94 __attribute__((noinline))
95 void simulate_thread_main()
103 /* Perform any setup code that will run outside of the testing
104 harness. Put code here that you do NOT want to be interrupted on
105 an instruction basis. E.g., setup code, and system library
108 /* Do un-instrumented stuff. */
111 /* Start the instrumented show. */
112 simulate_thread_main();
114 /* Must be called at the end of the test. */
115 simulate_thread_done();