Reuse the body of child_attach_to_its_parent for a test attaching to a
stopped parent.
As a workaround to ATF framework (atf-run), run this test in a subprocess.
The raise(SIGSTOP) call breaks the pipe from atf-run, that is polling
whether a process is in order.
According to code inspection the poll(2) call in safe_poll() receives
EINTR and this tears down the test.
This test passes.
These tests are used only when a distribution is built without MKSANITIZER
and without MKLIBCSANITIZER. They build the ubsan.c micro-implementation
in userland as a standalone library (a .c file linked into tests).
The code generation part emits symbols that are handled by uUBSan.
Tested with Clang amd64 + i386 and GCC amd64 + i386.
There are two sets of tests: C code generation and C++ code generation.
There are various little differences that are handled with proper ifdefs.
Clang 7svn has significantly more checks for Undefined Behavior than GCC
5.x. Some of the tests are or might be amd64 and/or i386 specific and are
marked with proper ifdefs and comments according to the compiler
documentation.
Do not change the signedness bit with a left shift operation.
Switch to unsigned integer to prevent this.
t_bitops.c:189:9, left shift of 1 by 31 places cannot be represented in type 'int'
Detected with micro-UBSan in the user mode.
Replace UB with implementation defined logic to check whether ssize_t can
wrap to a negative number.
t_types.c:63:7, signed integer overflow: 9223372036854775807 + 1 cannot be represented in type 'long int'
Detected with micro-UBSan in the user mode.
In the filename expansion test, don't insist on testing cases
of [^a] anything not an a) (etc) - while almost all shells allow
^ there (as in regular expressions) this is not standard sh/glob syntax
(! is used for that, not ^). Use of ^ as first char after '[' in a []
match has unspecified behaviour. So, check if the shell being
tested supports ^ used this way before running the tests of ^ in [].
This makes no difference at all when testing the NetBSD sh which
does allow ^ for that use case.
Fix an obvious (in hindsight) stupidity in the case pattern tests
(a cut/paste/forget-to-fix type error).
Many cleanups, including some additional sub-tests, for the var
substring matching (just a couple of actual fixes to the tests).
(This was the last added, and least polished part - still is.)
This makes no difference to the filename expansion tests (11 of 167
still fail), the fix reduces the failing case match tests from
14 of 261 to 13 of 261. For var substring tests, the failure
has changed from 6 of 87 to 8 of 99. (Some) Fixes to sh will appear
soon.
For comparison, using this version of the test, dash passes all case
and var substring tests, and (aside from the [^...] tests which it
does not implement) fails 4 of the filename tests. bash fails one
case test (a bash oddity in interpretation of the spec, which is unique
to bash amongst shells I have tested) and one filename expansion (all
var substring tests pass). Other shells are much worse (though the
freebsd shell results were coloured by the version of their shell I
tested still having the [[:xxx:]] always matches '[' bug, which is
fixed in later versions of their shell)
helper/child processes, so ATF timeout does not help).
Kamil knows about the SMP race in traced exit path and said he'll handle
it.
I would have skipped them on SMP machines only, but got lost in the
macro labyrinth - kamil, please feel free to improve ;-)
These tests are known to be faulty. These fatures aren't ready.
They tend to break to unrelated changes to the codebase.
Base Clang and GCC are in general dated to use with sanitizers as these
features are being fixed on HEAD versions of both compilers (mostly Clang).
These tests are known to be faulty. These fatures aren't ready.
They tend to break to unrelated changes to the codebase and require.
Base Clang and GCC are in general dated to use with sanitizers as these
features are being fixed on HEAD versions of both compilers (mostly Clang).
patterns, and variable expansion substring matching)
Currently (2018-07-10) all 3 sub-tests fail (sh bugs...)
Expect to see 14 (of 261) case matching sub-tests fail, 11 (of 167) filename
expansion (glob) sub-tests fail, and 6 (of 87) var substring sub-tests fail.
Also expect those numbers to reduce as sh bugs are fixed.
Assert that -9223372036854775808 % -1 and -9223372036854775808 / -1 return
message about overflow / underflow detected.
These tests pass correctly.
Sponsored by <The NetBSD Foundation>
If we write a byte character into a pointer, a compiler can emit a
read-modify-write operation, especially when a CPU cannot access directly
a character wide address.
In this scenario calling mmap(2) with PROT_WRITE, without PROT_READ will
emit unexpected trap.
There are two possible workarounds for this issue:
- write register wide memory without rmw sequence,
- mark the region with additional protection PROT_READ
Both work for NetBSD/alpha.
Go for the latter as perhaps more safe for dump compilers emitting rmw
sequences.
Investigated by <martin>
to bytes_transfer() which asserted that the size is < the size of 512 AuxInfo
structs. On a 32 bit system, an AuxInfo is 8 bytes, and 512 AuxInfo
structs is thus 4096 bytes, and 4096 is not < 4096, so the assertion
failed. It would probably work as a <= assert, but I am not confident
enough with this test case to make that call, so instead increase the
array size to be 513 instead, 4096 is < 513*8 so happiness should return
and all should be right with the world. At least for this one test.
(There was no problem on 64 bit systems as AuxInfo is bigger there.)
Add new tests:
- traceme_crash_trap
- traceme_crash_segv
- traceme_crash_ill (temporarily disabled)
- traceme_crash_fpe
- traceme_crash_bus
- unrelated_tracer_sees_crash_trap
- unrelated_tracer_sees_crash_segv
- unrelated_tracer_sees_crash_ill (temporarily disabled)
- unrelated_tracer_sees_crash_fpe
- unrelated_tracer_sees_crash_bus
These tests verify two subtle kernel paths that differ internally:
- for a tracer that is a parent of a traced program (and not vforked)
real parent == tracer
- for a tracer that is an unrelated process of a traced program
real parent != tracer
Regressions in the signal code can cause to break one of the scenarios, and
keep working the other one. It might be propagated to a scenario like:
- attaching to a process with a debugger (over pid)
- starting a process inside a debugger
We can assert on the ATF level that both will be functional.
While there, cover all the crash signals as testing is done practically
for free. While the Machine Independent signal routing code for crash
signals is mostly the same in the kernel, we can verify whether Machine
Dependent parts works as expected emitting the proper signal number and
signal code (si_code).
There have been already caught an unexpected behavior on Alpha, that
instead of SIGBUS we receive SIGSEGV.
Another benefit is that it's easier to catch races, as they pop up more
frequently. (This already helped to catch bugs more quickly)
These tests will help assert correctness of future kernel changes in the
signal routines.
All tests pass.
Sponsored by <The NetBSD Foundation>
This functionality now works.
Correct the code path in this test.
Add additional sanity check with PT_GET_SIGINFO and validating si_code.
Sponsored by <The NetBSD Foundation>
A function that returns a value that is not used, can be optimized out by
a compiler. Try to fool the compiler with calling libc functions that shall
trigger a division by zero crash.
Sponsored by <The NetBSD Foundation>
These crash signals are crucial for proper handling of abnormal conditions
in a program. The additional purpose of these tests it to assure the proper
handling of these signals for the coming ptrace(2)-related changes in the
signal routing code.
Add a stub for ILL scenarios.
All tests pass (on amd64).
The shell ATF script contains duplicated code. There should be a way to
deduplicate it, without rewrite to C.
Sponsored by <The NetBSD Foundation>
Make the initial buffer larger to stop triggering the runtime assert about
its size.
Its size is probably larger than all real use cases, but it's good enough
for the purpose of ATF tests and immune of potential AUXV size.
Sponsored by <The NetBSD Foundation>
christos