On 9/1/23 20:11, Tom Rini wrote:
On Fri, Sep 01, 2023 at 10:48:10AM +0200, Marek Vasut wrote:
Test both 32bit and 64bit sandbox boards in CI.
Signed-off-by: Marek Vasut <marek.vasut+rene...@mailbox.org>
---
Cc: Heinrich Schuchardt <heinrich.schucha...@canonical.com>
Cc: Simon Glass <s...@chromium.org>
Cc: Tom Rini <tr...@konsulko.com>
---
DEPENDS:
- [v2] test: print: Fix hexdump test on 64bit systems
- test: dm: test-fdt: Use fdtdec_get_int() in dm_check_devices()
- configs: sandbox64: Enable legacy image format support
---
.gitlab-ci.yml | 17 +++++++++++++++++
1 file changed, 17 insertions(+)
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 6d7ffdd1bbc..a7282be3181 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -291,6 +291,23 @@ sandbox without LTO test.py:
BUILD_ENV: "NO_LTO=1"
<<: *buildman_and_testpy_dfn
+sandbox64 test.py:
+ variables:
+ TEST_PY_BD: "sandbox64"
+ <<: *buildman_and_testpy_dfn
+
+sandbox64 with clang test.py:
+ variables:
+ TEST_PY_BD: "sandbox64"
+ OVERRIDE: "-O clang-16"
+ <<: *buildman_and_testpy_dfn
+
+sandbox64 without LTO test.py:
+ variables:
+ TEST_PY_BD: "sandbox64"
+ BUILD_ENV: "NO_LTO=1"
+ <<: *buildman_and_testpy_dfn
I'm not sure it's worth building and testing sandbox64 without LTO as
well. The primary reason we do this here is to ensure that the NO_LTO
flag keeps working, rather than specific outputs from the tests, and we
should _probably_ change the current one to be a case where we just test
that the version test runs and passes.
It is faster to finish building if nothing else.
