| 1 |
KSelfTest arm64/signal/ |
| 2 |
======================= |
| 3 |
|
| 4 |
Signals Tests |
| 5 |
+++++++++++++ |
| 6 |
|
| 7 |
- Tests are built around a common main compilation unit: such shared main |
| 8 |
enforces a standard sequence of operations needed to perform a single |
| 9 |
signal-test (setup/trigger/run/result/cleanup) |
| 10 |
|
| 11 |
- The above mentioned ops are configurable on a test-by-test basis: each test |
| 12 |
is described (and configured) using the descriptor signals.h::struct tdescr |
| 13 |
|
| 14 |
- Each signal testcase is compiled into its own executable: a separate |
| 15 |
executable is used for each test since many tests complete successfully |
| 16 |
by receiving some kind of fatal signal from the Kernel, so it's safer |
| 17 |
to run each test unit in its own standalone process, so as to start each |
| 18 |
test from a clean slate. |
| 19 |
|
| 20 |
- New tests can be simply defined in testcases/ dir providing a proper struct |
| 21 |
tdescr overriding all the defaults we wish to change (as of now providing a |
| 22 |
custom run method is mandatory though) |
| 23 |
|
| 24 |
- Signals' test-cases hereafter defined belong currently to two |
| 25 |
principal families: |
| 26 |
|
| 27 |
- 'mangle_' tests: a real signal (SIGUSR1) is raised and used as a trigger |
| 28 |
and then the test case code modifies the signal frame from inside the |
| 29 |
signal handler itself. |
| 30 |
|
| 31 |
- 'fake_sigreturn_' tests: a brand new custom artificial sigframe structure |
| 32 |
is placed on the stack and a sigreturn syscall is called to simulate a |
| 33 |
real signal return. This kind of tests does not use a trigger usually and |
| 34 |
they are just fired using some simple included assembly trampoline code. |
| 35 |
|
| 36 |
- Most of these tests are successfully passing if the process gets killed by |
| 37 |
some fatal signal: usually SIGSEGV or SIGBUS. Since while writing this |
| 38 |
kind of tests it is extremely easy in fact to end-up injecting other |
| 39 |
unrelated SEGV bugs in the testcases, it becomes extremely tricky to |
| 40 |
be really sure that the tests are really addressing what they are meant |
| 41 |
to address and they are not instead falling apart due to unplanned bugs |
| 42 |
in the test code. |
| 43 |
In order to alleviate the misery of the life of such test-developer, a few |
| 44 |
helpers are provided: |
| 45 |
|
| 46 |
- a couple of ASSERT_BAD/GOOD_CONTEXT() macros to easily parse a ucontext_t |
| 47 |
and verify if it is indeed GOOD or BAD (depending on what we were |
| 48 |
expecting), using the same logic/perspective as in the arm64 Kernel signals |
| 49 |
routines. |
| 50 |
|
| 51 |
- a sanity mechanism to be used in 'fake_sigreturn_'-alike tests: enabled by |
| 52 |
default it takes care to verify that the test-execution had at least |
| 53 |
successfully progressed up to the stage of triggering the fake sigreturn |
| 54 |
call. |
| 55 |
|
| 56 |
In both cases test results are expected in terms of: |
| 57 |
- some fatal signal sent by the Kernel to the test process |
| 58 |
or |
| 59 |
- analyzing some final regs state |