keyboard: Improve kbpress reliability for automation

The "kbpress" command had a few issues if you wanted to reliably use
it for automation.  Specifically it was not possible to guarantee how
much time would pass between the press of a key and the release of a
key.  Sometimes you might press and release before the key was
officially "there" and sometimes you might get a press and hold of a
key.

Fix this:
1. Make it so that kbpress with no press/release parameter gives a
   press and release (and guarantees that the press / release will
   actually take effect).
2. Make it so that kbpress guarantees that when it finishes that the
   key has actually been pressed or released.

BRANCH=pit
BUG=chrome-os-partner:24249
TEST=kbtype is (https://chromium-review.googlesource.com/178680) reliable
TEST=make -j32 BOARD=bds tests && make BOARD=bds runtests
TEST=Pick Ibe00a796bde7d06416889b621359671a2f68e162 and test.

Change-Id: Ia213ab2e8d8da273e3ac4876d97d5452df88f47d
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/178983
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
(cherry picked from commit 247650ecc90385417f5dcb2d60bb6ae1e5cfa32f)
Reviewed-on: https://chromium-review.googlesource.com/179325
Reviewed-by: Randall Spangler <rspangler@chromium.org>

1 files changed, 6 insertions, 2 deletions

diff --git a/test/kb_scan.c b/test/kb_scan.c
index a273a20466..e0e73d17f5 100644
--- a/test/kb_scan.c
+++ b/test/kb_scan.c
@@ -253,10 +253,14 @@ static int debounce_test(void)
 
 static int simulate_key_test(void)
 {
+	int old_count;
+
+	old_count = fifo_add_count;
 	host_command_simulate(1, 1, 1);
-	TEST_ASSERT(expect_keychange() == EC_SUCCESS);
+	TEST_ASSERT(fifo_add_count > old_count);
+	old_count = fifo_add_count;
 	host_command_simulate(1, 1, 0);
-	TEST_ASSERT(expect_keychange() == EC_SUCCESS);
+	TEST_ASSERT(fifo_add_count > old_count);
 
 	return EC_SUCCESS;
 }