fix: make merge base fetching more robust
Several static check scripts rely on identifying the Git merge base via
get_merge_base, which may fail if the local Git history is too shallow.
This causes issues when checking patches on CI or LTS branches with
limited fetch depth.
To address this:
- Reuse a single evaluated merge_base instead of invoking get_merge_base
multiple times.
- If merge_base is not found initially, attempt to fetch more history
from the remote using the appropriate GERRIT_REFSPEC.
- Exit with an error if a merge base cannot be found even after
fetching.
This improves reliability of static checks when working on partial
clones or CI environments where fetch depth is restricted.
Change-Id: Icccec0eb2f29d254e54bbd6b639f6c1ef11291a3
Signed-off-by: Harrison Mutai <harrison.mutai@arm.com>
diff --git a/script/static-checks/static-checks-cpu-erratum-order.sh b/script/static-checks/static-checks-cpu-erratum-order.sh
index 89b5fa0..8631c48 100755
--- a/script/static-checks/static-checks-cpu-erratum-order.sh
+++ b/script/static-checks/static-checks-cpu-erratum-order.sh
@@ -14,7 +14,7 @@
TEST_CASE="Checking ascending order of CPU ERRATUM and CVE in the patch series"
echo "# $TEST_CASE"
"$CI_ROOT/script/static-checks/static-checks-cpu-erratum-order.py" --tree "$1" \
- --patch --from-ref $(get_merge_base) &> "$LOG_FILE"
+ --patch --from-ref ${merge_base} &> "$LOG_FILE"
else
TEST_CASE="Checking ascending order of CPU ERRATUM and CVE in the entire source tree"
echo "# $TEST_CASE"