Enforced a coverage floor on the merged report (#667)

The coverage summary reported a percentage and could not fail. Coverage
could fall from 99.97% to anything at all and every check stayed green,
against an AGENTS.md that asks for 100% test coverage -- a stated
requirement measured with a gauge that had no failure mode.

CodeCoverageSummary already takes thresholds and fail_below_min; neither
was set. Both are now, through a new coverage_thresholds input on the
template, because the two suites do not sit at the same figure: ThreadX
99, SMP 98.

Three things were probed against the pinned action on a runner before
picking those numbers, using the real merged.xml files from the dev push
run of #666.

The floor compares the line rate and nothing else. That mattered because
branch coverage is around 78% in both suites while line coverage is
98.8-100%, so a floor aimed at the line figure would have been an
immediate red wall had it tested branches or the lower of the two. The
ThreadX report at 100.00% lines and 77.67% branches clears a floor of 99.

The thresholds are whole numbers. '99.9 100' -- the value this was meant
to be -- is rejected with 'System.ArgumentException - Threshold parameter
set incorrectly.', and the step fails whether or not fail_below_min is
set. So the choice is 99 or 100 with nothing between.

100 would fail on a race. tx_thread_system_resume.c:529 is reached by
timing rather than by construction and flaps between runs of the same
green tree, which is why #666 left it; 4502/4503 fails a floor of 100 and
clears one of 99. A coverage gate that goes red on a coin toss is how
coverage gates get switched off.

SMP is 5114/5178 lines, 98.76%, with 64 uncovered lines across 11 files
of common_smp/src -- #666 closed the equivalent gaps in common/src only.
A shared floor of 99 would have failed that job on every run while
ThreadX passed.

One limit is recorded in the file rather than fixed: an empty report
reads as 100%. gcovr writes line-rate="1.0" beside lines-valid="0" when
it finds no data, and the action prints 'Line Rate = 100% (0 / 0)' and
passes any floor. The check for that is the emptiness assertion #664 put
in each suite's coverage.sh, not this one.

Also corrected two stale filenames in the deploy job's comment: since
#665 each coverage artifact carries merged.xml, not
default_build_coverage.xml. Verified on the runner.

Assisted-by: Claude Opus 5 <noreply@anthropic.com>

.github/workflows/regression_template.yml

@@ -38,6 +38,33 @@ on:
         default: 'merged'
         required: false
         type: string
+      # The lower and upper threshold percentages handed to
+      # CodeCoverageSummary, lower first. The lower one is a hard floor: the
+      # step below sets fail_below_min, so a run whose merged line rate falls
+      # under it turns the job red.
+      #
+      # Both must be whole numbers. Probed against the pinned action on a
+      # runner, 26 Aug 2026: '99.9 100' is rejected with
+      # 'System.ArgumentException - Threshold parameter set incorrectly.' and
+      # the step fails whether or not fail_below_min is set, because the action
+      # parses each half with an integer parse. So a floor can be 99 or 100 and
+      # nothing between, and the intended 99.9 is not expressible.
+      #
+      # The floor is compared against the LINE rate only -- not the branch
+      # rate, and not the lower of the two. Probed on the same run: the
+      # ThreadX report at 100.00% lines and 77.67% branches passes a floor of
+      # 99. Worth knowing, because branch coverage is around 78% in both
+      # suites while AGENTS.md asks for 100%, and a floor set from the
+      # headline line figure says nothing about it.
+      #
+      # Each suite sets its own in regression_test.yml, because they do not
+      # sit at the same figure. The default here is the action's own, low
+      # enough to be no floor at all in practice -- a caller that collects
+      # coverage is expected to name its number.
+      coverage_thresholds:
+        default: '50 75'
+        required: false
+        type: string
       skip_deploy:
         default: false
         required: false
@@ -128,6 +155,18 @@ jobs:
     # so a cancelled run still stops promptly -- the same idiom the
     # deploy_code_coverage job below already uses. The ${{ }} is required: a bare
     # ! opens a YAML tag, and the expression will not parse without it.
+    #
+    # fail_below_min turns the summary into a gate. Until now coverage could
+    # fall from any figure to any other and no check went red, against an
+    # AGENTS.md that asks for 100% -- a stated requirement measured with a
+    # gauge that could not fail.
+    #
+    # One thing the floor does not defend, and it is the likeliest way for
+    # coverage to break: an empty report reads as 100%. gcovr writes
+    # line-rate="1.0" beside lines-valid="0" when it finds no data, and the
+    # action reports 'Line Rate = 100% (0 / 0)' and passes any floor --
+    # probed on a runner, 26 Aug 2026. The check that catches that is the
+    # emptiness assertion in each suite's coverage.sh, not this one.
     - name: Generate Code Coverage Results Summary
       if: ${{ !cancelled() && (!inputs.skip_coverage) }}
       uses: irongut/CodeCoverageSummary@51cc3a756ddcd398d447c044c02cb6aa83fdae95 # v1.3.0
@@ -137,6 +176,8 @@ jobs:
         badge: true
         hide_complexity: true
         output: file
+        thresholds: ${{ inputs.coverage_thresholds }}
+        fail_below_min: true
 
     - name: Write Code Coverage Summary
       if: ${{ !cancelled() && (!inputs.skip_coverage) }}
@@ -239,13 +280,15 @@ jobs:
     # merge-multiple puts the contents of both coverage artifacts directly
     # into coverage_report rather than under a directory named for each
     # artifact. Each one holds a single directory named for its suite,
-    # ThreadX or SMP, renamed from default_build_coverage by the "Prepare
-    # Coverage GitHub Pages" step, so the merge yields exactly the two
-    # suite directories the deploy expects, and the artifact name -- with
-    # its timestamp -- stops appearing in the published path at all.
+    # ThreadX or SMP, renamed from merged by the "Prepare Coverage GitHub
+    # Pages" step, so the merge yields exactly the two suite directories the
+    # deploy expects, and the artifact name -- with its timestamp -- stops
+    # appearing in the published path at all.
     #
-    # The two artifacts also each carry a default_build_coverage.xml, and
-    # the merge means one overwrites the other. That file is consumed by
+    # The two artifacts also each carry a merged.xml -- it was
+    # default_build_coverage.xml until #665 instrumented every configuration
+    # and unioned them -- and the merge means one overwrites the other.
+    # Verified on the runner: both hold merged.xml. That file is consumed by
     # CodeCoverageSummary back in run_tests and is not read here, so this is
     # untidy rather than wrong.
     - uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8.0.1

.github/workflows/regression_test.yml

@@ -37,6 +37,17 @@ jobs:
       cmake_path: ./test/tx/cmake
       result_affix: ThreadX
       skip_deploy: true
+      # 99 is the tightest floor available below this suite's figure. The
+      # merged report reads 4503/4503 lines -- 100.00% -- on the dev push run
+      # of #666, and two local runs of the same green tree came in at 100.00%
+      # and 99.98%: tx_thread_system_resume.c:529 is reached by timing rather
+      # than by construction and flaps between runs, and #666 could not reach
+      # it deliberately. So a floor of 100 is a check that fails on a race --
+      # probed against the pinned action, 4502/4503 fails a floor of 100 and
+      # clears one of 99 -- and the thresholds are integers, so there is
+      # nothing between the two. Ratchet this to 100 if 529 is ever covered
+      # by construction.
+      coverage_thresholds: '99 100'
   smp:
     permissions:
       contents: read
@@ -52,6 +63,15 @@ jobs:
       cmake_path: ./test/smp/cmake
       result_affix: SMP
       skip_deploy: true
+      # Lower than ThreadX's, because this suite is: the merged report reads
+      # 5114/5178 lines -- 98.76% -- on the same run, with 64 uncovered lines
+      # across 11 files of common_smp/src. #666 closed the equivalent gaps in
+      # common/src only, so common_smp still carries them, plus a family of
+      # its own in the *_delete.c teardown paths and tx_byte_pool_search.c.
+      # A shared floor of 99 would therefore red-wall this job on every run
+      # while ThreadX passed -- probed against the pinned action. Raise this
+      # towards 99 as those lines are covered.
+      coverage_thresholds: '98 100'
   freertos:
     permissions:
       contents: read