Autopkgtest: Don't track "ever passed" for kernel triggers

We trigger independent tests for every linux/linux-meta* reverse dependencies,
as they run under the triggering kernel. Thus "ever passed" is rather
meaningless for these as we don't want to track this on a per-trigger basis (as
it would be wrong for everything else but kernels). This led to a lot of false
regressions, as some DKMS modules only work on some kernel flavours.

The kernel team is doing per-kernel regression analysis of the test results, so
we don't need to duplicate this logic in britney. Thus effectively disable the
"Regression" state for kernel reverse dependencies, and rely on the kernel
test machinery to untag the tracking bug only if there are no actual
regressions.

autopkgtest.py

@@ -657,7 +657,13 @@ class AutoPackageTest(object):
                         result = 'PASS'
                     else:
                         # test failed, check ever_passed flag for that src/arch
-                        if ever_passed:
+                        # unless we got triggered from linux-meta*: we trigger
+                        # separate per-kernel tests for reverse test
+                        # dependencies, and we don't want to track per-trigger
+                        # ever_passed. This would be wrong for everything
+                        # except the kernel, and the kernel team tracks
+                        # per-kernel regressions already
+                        if ever_passed and not trigsrc.startswith('linux-meta') and trigsrc != 'linux':
                             result = 'REGRESSION'
                         else:
                             result = 'ALWAYSFAIL'

tests/test_autopkgtest.py

@@ -1310,7 +1310,7 @@ fancy 1 i386 linux-meta-lts-grumpy 1
              ('linux-image-64only', {'Source': 'linux-meta-64only', 'Architecture': 'amd64'}, None),
             ],
             {'linux-meta': (True, {'fancy 1': {'amd64': 'PASS', 'i386': 'PASS'}}),
-             'linux-meta-lts-grumpy': (False, {'fancy 1': {'amd64': 'RUNNING', 'i386': 'REGRESSION'}}),
+             'linux-meta-lts-grumpy': (False, {'fancy 1': {'amd64': 'RUNNING', 'i386': 'ALWAYSFAIL'}}),
              'linux-meta-64only': (True, {'fancy 1': {'amd64': 'PASS'}}),
             })
 
@@ -1342,7 +1342,7 @@ fancy 1 i386 linux-meta-lts-grumpy 1
             ],
             {'linux-meta': (True, {'fancy 1': {'amd64': 'PASS', 'i386': 'PASS'}}),
              # we don't have an explicit result for amd64, so the old one counts
-             'linux-meta-lts-grumpy': (False, {'fancy 1': {'amd64': 'REGRESSION', 'i386': 'REGRESSION'}}),
+             'linux-meta-lts-grumpy': (True, {'fancy 1': {'amd64': 'ALWAYSFAIL', 'i386': 'ALWAYSFAIL'}}),
              'linux-meta-64only': (True, {'fancy 1': {'amd64': 'PASS'}}),
             })