Add diagnosis to failing hostap CI tests

Author: Frauschi

.github/workflows/hostap-vm.yml

@@ -340,6 +340,10 @@ jobs:
             KVMARGS="-cpu host"
           EOF
           git config --global --add safe.directory $GITHUB_WORKSPACE/hostap
+          # Persist logs for tests that fail (after retry) so they survive
+          # the per-batch `rm -r` and can be inspected/uploaded.
+          PERSISTENT_LOGS=$GITHUB_WORKSPACE/failed-test-logs
+          mkdir -p "$PERSISTENT_LOGS"
           # Run tests in increments of 200 to not stall out the parallel-vm script
           while mapfile -t -n 200 ary && ((${#ary[@]})); do
             TESTS=$(printf '%s\n' "${ary[@]}" | tr '\n' ' ')
@@ -349,28 +353,48 @@ jobs:
               # Let's re-run the failing tests. We gather the failed tests from the log file.
               FAILED_TESTS=$(grep 'failed tests' /tmp/hwsim-test-logs/*-parallel.log | sed 's/failed tests: //' | tr ' ' '\n' | sort | uniq | tr '\n' ' ')
               printf 'failed tests: %s\n' "$FAILED_TESTS"
-              ./vm/parallel-vm.py ${{ env.hostap_debug_flags }} --nocurses $(nproc) $FAILED_TESTS
+              RETRY_RES=0
+              ./vm/parallel-vm.py ${{ env.hostap_debug_flags }} --nocurses $(nproc) $FAILED_TESTS || RETRY_RES=$?
+              if [ "$RETRY_RES" -ne "0" ]; then
+                # Retry also had failures: snapshot relevant logs for diagnosis.
+                # Capture every test that was reported failing in this batch
+                # (initial run + retry) so we have full context.
+                ALL_FAILED=$(grep 'failed tests' /tmp/hwsim-test-logs/*-parallel.log 2>/dev/null | sed 's/.*failed tests: //' | tr ' ' '\n' | sort | uniq | tr '\n' ' ')
+                DEST="$PERSISTENT_LOGS/batch-$(date +%s%N)"
+                mkdir -p "$DEST"
+                cp /tmp/hwsim-test-logs/*-parallel.log "$DEST/" 2>/dev/null || true
+                for test in $ALL_FAILED; do
+                  for f in $(find /tmp/hwsim-test-logs -name "${test}.log"); do
+                    rel=${f#/tmp/hwsim-test-logs/}
+                    mkdir -p "$DEST/$(dirname "$rel")"
+                    cp "$f" "$DEST/$rel"
+                  done
+                done
+              fi
             fi
             rm -r /tmp/hwsim-test-logs
           done < $GITHUB_WORKSPACE/wolfssl/.github/workflows/hostap-files/configs/${{ matrix.config.hostap_ref }}/tests
 
-      - name: show errors
+      - name: Show failed test logs
         if: ${{ failure() && steps.testing.outcome == 'failure' }}
-        run: grep -riP 'fail|error' /tmp/hwsim-test-logs/latest
-
-      # The logs are quite big. It hasn't been useful so far so let's not waste
-      # precious gh space.
-      #- name: zip logs
-      #  if: ${{ failure() && steps.testing.outcome == 'failure' }}
-      #  working-directory: hostap/tests/hwsim/
-      #  run: |
-      #    rm /tmp/hwsim-test-logs/latest
-      #    zip -9 -r logs.zip /tmp/hwsim-test-logs
-      #
-      #- name: Upload failure logs
-      #  if: ${{ failure() && steps.testing.outcome == 'failure' }}
-      #  uses: actions/upload-artifact@v4
-      #  with:
-      #    name: hostap-logs-${{ env.our_job_run_id }}
-      #    path: hostap/tests/hwsim/logs.zip
-      #    retention-days: 5
+        run: |
+          DIR=$GITHUB_WORKSPACE/failed-test-logs
+          if [ ! -d "$DIR" ] || [ -z "$(ls -A "$DIR" 2>/dev/null)" ]; then
+            echo "No persisted failure logs found."
+            exit 0
+          fi
+          echo "=== Captured failure logs ==="
+          find "$DIR" -name "*.log" | sort
+          for f in $(find "$DIR" -name "*.log" -not -name "*-parallel.log" | sort); do
+            echo "----- $f -----"
+            tail -100 "$f"
+          done
+
+      - name: Upload failure logs
+        if: ${{ failure() && steps.testing.outcome == 'failure' }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: hostap-logs-${{ env.our_job_run_id }}
+          path: failed-test-logs
+          retention-days: 5
+          if-no-files-found: ignore