ACPI code is changing as bit too fast to make
it worth it to require everyone to check
it against expected output.
Make test run depend on an ACPI_TEST_EXPECTED_AML
set in environment, so interested people can
run it without affecting everyone.
Signed-off-by: Michael S. Tsirkin <m...@redhat.com>
---
I'd like to prevent tests failing for everyone each time
we change some ACPI code.
This is an inelegant way to do this, would be better to emit
a warning and keep going.
tests/acpi-test.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/tests/acpi-test.c b/tests/acpi-test.c
index b5ab70a..c164382 100644
--- a/tests/acpi-test.c
+++ b/tests/acpi-test.c
@@ -22,6 +22,7 @@
#define MACHINE_PC "pc"
#define MACHINE_Q35 "q35"
+#define ACPI_TEST_EXPECTED_AML "TEST_ACPI_TEST_AML"
#define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
/* DSDT and SSDTs format */
@@ -572,7 +573,10 @@ static void test_acpi_one(const char *params, test_data
*data)
test_acpi_dsdt_table(data);
test_acpi_ssdt_tables(data);
- if (iasl) {
+ if (getenv(ACPI_TEST_EXPECTED_AML)) {
+
+ g_assert(iasl);
+
if (getenv(ACPI_REBUILD_EXPECTED_AML)) {
dump_aml_files(data, true);
} else {
--
MST
