Hi Brian,
On 04/05, Gix, Brian wrote:
> I think this needs some more work, in that it no longer correctly
> fails the mesh-test-crypto test when running:
>
> make distcheck
>
> It *almost* does in that it flags an error in red if a "verify" step
> fails, but it does not exit with a fail code. This is an important
> step, particularily with the dependance on kernel based crypto
> functions.
>
> My test methodology was to flip a bit in one of the test payloads
> (i.e., make it "not perfectly match" the specification sample data).
>
> As best I can tell, it is otherwise functional, but the unit tests are
> important.
Agreed, but as of 14301cf0d42ba3fc4de99c3bdf183045be733e0c running
unit/test-mesh-crypto with modified payloads, e.g.
diff --git a/unit/test-mesh-crypto.c b/unit/test-mesh-crypto.c
index 0043b0b10..a832b2000 100644
--- a/unit/test-mesh-crypto.c
+++ b/unit/test-mesh-crypto.c
@@ -591,7 +591,7 @@ static const struct mesh_crypto_test s8_3_11 = {
static const struct mesh_crypto_test s8_3_22 = {
.name = "8.3.22 Message #22",
- .app_key = "63964771734fbd76e3b40519d1d94a48",
+ .app_key = "63964771735fbd76e3b40519d1d94a48",
.net_key = "7dd7364cd842ad18c17c2b820c84c3d6",
.dev_key = "9d6dd0e96eb25dc19a40ed9914f8f03f",
.iv_index = 0x12345677,
also does not cause it to exit non-zero return code, it just prints red
"FAIL"s...
I thought it was done on purpose, as mesh tests don't seem to use the
src/shared/tester.h.
Also, after applying the patch onto the mentioned commit, I'm getting
all green runs. What system are you using? I'd like to try reproducing
your results.
regards
--
Michał Lowas-Rzechonek <michal.lowas-rzechonek@xxxxxxxxxxx>
Silvair http://silvair.com
Jasnogórska 44, 31-358 Krakow, POLAND
