Replaced tn/bn prefixes with an actual dependency system in tests/benches

The previous system of relying on test name prefixes for ordering was
simple, but organizing tests by dependencies and topologically sorting
during compilation is 1. more flexible and 2. simplifies test names,
which get typed a lot.

Note these are not "hard" dependencies, each test suite should work fine
in isolation. These "after" dependencies just hint an ordering when all
tests are ran.

As such, it's worth noting the tests should NOT error of a dependency is
missing. This unfortunately makes it a bit hard to catch typos, but
allows faster compilation of a subset of tests.

---

To make this work the way tests are linked has changed from using custom
linker section (fun linker magic!) to a weakly linked array appended to
every source file (also fun linker magic!).

At least with this method test.py has strict control over the test
ordering, and doesn't depend on 1. the order in which the linker merges
sections, and 2. the order tests are passed to test.py. I didn't realize
the previous system was so fragile.

scripts/test.py

@@ -144,6 +144,14 @@ class TestCase:
                 k),
                 file=sys.stderr)
 
+    def __repr__(self):
+        return '<TestCase %s>' % self.name
+
+    def __lt__(self, other):
+        # sort by suite, lineno, and name
+        return ((self.suite, self.lineno, self.name)
+            < (other.suite, other.lineno, other.name))
+
 
 class TestSuite:
     # create a TestSuite object from a toml file
@@ -195,14 +203,17 @@ class TestSuite:
                     if not case_linenos or l < case_linenos[0][0]),
                 default=None)
 
+            self.after = config.pop('after', [])
+            if not isinstance(self.after, list):
+                self.after = [self.after]
+
             # a couple of these we just forward to all cases
             defines = config.pop('defines', {})
             in_ = config.pop('in', None)
             reentrant = config.pop('reentrant', False)
 
             self.cases = []
-            for name, case in sorted(cases.items(),
-                    key=lambda c: c[1].get('lineno')):
+            for name, case in cases.items():
                 self.cases.append(TestCase(config={
                     'name': name,
                     'path': path + (':%d' % case['lineno']
@@ -214,6 +225,9 @@ class TestSuite:
                     **case},
                     args=args))
 
+            # sort for consistency
+            self.cases.sort()
+
             # combine per-case defines
             self.defines = set.union(set(), *(
                 set(case.defines) for case in self.cases))
@@ -230,6 +244,14 @@ class TestSuite:
                 k),
                 file=sys.stderr)
 
+    def __repr__(self):
+        return '<TestSuite %s>' % self.name
+
+    def __lt__(self, other):
+        # sort by name
+        #
+        # note we override this with a topological sort during compilation
+        return self.name < other.name
 
 
 def compile(test_paths, **args):
@@ -251,7 +273,29 @@ def compile(test_paths, **args):
 
     # load the suites
     suites = [TestSuite(path, args) for path in paths]
-    suites.sort(key=lambda s: s.name)
+
+    # sort suites by:
+    # 1. topologically by "after" dependencies
+    # 2. lexicographically for consistency
+    pending = co.OrderedDict((suite.name, suite)
+            for suite in sorted(suites))
+    suites = []
+    while pending:
+        pending_ = co.OrderedDict()
+        for suite in pending.values():
+            if not any(after in pending for after in suite.after):
+                suites.append(suite)
+            else:
+                pending_[suite.name] = suite
+
+        if len(pending_) == len(pending):
+            print('%serror:%s cycle detected in suite ordering, %s' % (
+                '\x1b[01;31m' if args['color'] else '',
+                '\x1b[m' if args['color'] else '',
+                ', '.join(suite.name for suite in pending.values())))
+            sys.exit(-1)
+
+        pending = pending_
 
     # check for name conflicts, these will cause ambiguity problems later
     # when running tests
@@ -420,12 +464,6 @@ def compile(test_paths, **args):
                         f.writeln()
 
                 # create suite struct
-                #
-                # note we place this in the custom test_suites section with
-                # minimum alignment, otherwise GCC ups the alignment to
-                # 32-bytes for some reason
-                f.writeln('__attribute__((section("_test_suites"), '
-                    'aligned(1)))')
                 f.writeln('const struct test_suite __test__%s__suite = {'
                     % suite.name)
                 f.writeln(4*' '+'.name = "%s",' % suite.name)
@@ -533,6 +571,26 @@ def compile(test_paths, **args):
                                     f.writeln('#endif')
                                 f.writeln()
 
+                # declare our test suites
+                #
+                # by declaring these as weak we can write these to every
+                # source file without issue, eventually one of these copies
+                # will be linked
+                for suite in suites:
+                    f.writeln('extern const struct test_suite '
+                        '__test__%s__suite;' % suite.name);
+                f.writeln()
+
+                f.writeln('__attribute__((weak))')
+                f.writeln('const struct test_suite *const test_suites[] = {');
+                for suite in suites:
+                    f.writeln(4*' '+'&__test__%s__suite,' % suite.name);
+                f.writeln('};')
+                f.writeln('__attribute__((weak))')
+                f.writeln('const size_t test_suite_count = %d;' % len(suites))
+                f.writeln()
+
+
 def find_runner(runner, id=None, **args):
     cmd = runner.copy()