fix: align runner behaviours (#1338)

Contains a number of small bug fixes where jobs behaved differently from GitHub or failed silently.

1. Hooks fail on a refused `::set-env::` or `::add-path::` and no longer inherit the last step's identity
1. Docker build, exec and wait errors are returned instead of swallowed, token-only registry credentials work, logs of `--tty` services decode
1. Env files split lines like GitHub on each OS, so a carriage return is kept on Linux and CRLF ends a line on Windows, empty heredoc names or delimiters are rejected
1. Post steps run after their pre step even when main was skipped, composites see undeclared inputs, `runs.args` wins without `with.args` being parsed
1. Docker actions skip evaluating defaults of inputs given in `with` and read `entryPoint` and `args` in any case, a lowercase `dockerfile` is found, undeclared inputs get GitHub's warning
1. A user-set `CI` is kept, `GITHUB_REPOSITORY_ID` and the other missing `GITHUB_*` variables are exported
1. `timeout-minutes` can read the step's own `env`, a step-level `ACTIONS_ALLOW_UNSECURE_COMMANDS: false` beats a job-level `true` while the runner's own opt-in still wins, `pwsh` quotes the script path
1. A step `env:` given as one expression such as `${{ fromJSON(matrix.env) }}` is applied, `hashFiles` fails with GitHub's messages on script errors and after 120 seconds, a job image without node still gets an empty hash
1. A step whose command files cannot be reset fails instead of importing stale ones, a timeout between steps fails the job instead of reporting success
1. One named volume can be mounted at several paths, host-mode PTY steps keep their last output when they fail, host scratch directories are removed when job setup fails
1. Cancelling a step right after it starts kills its whole process tree
1. Cache prefix restores pick the newest entry created within the same second, also on S3, cache commits with a wrong size are rejected
1. Artifact chunks are written at their offset, download URLs are escaped, listing a run without artifacts no longer crashes
1. Poll backoff reaches `fetch_interval_max`, proxy passwords are masked in more forms, secrets are also masked in step names and in their quoted form in the runner log, a failed final report is logged
1. On Windows hosts a differently cased unsecure-commands opt-in or proxy variable counts, and an early cancel no longer leaks a job object handle
1. Interactive `register` prompts for a missing token and stops on SIGTERM, `exec --list --detect-event` no longer panics without events, `config set` works on `cache.s3` keys

Reviewed-on: https://gitea.com/gitea/runner/pulls/1338
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>

act/artifactcache/handler.go

@@ -15,6 +15,7 @@ import (
 	"encoding/json/v2"
 	"errors"
 	"fmt"
+	"io"
 	"net"
 	"net/http"
 	"net/http/httputil"
@@ -710,9 +711,10 @@ func (h *Handler) reserve(w http.ResponseWriter, r *http.Request) {
 	}
 	defer db.Close()
 
-	now := time.Now().Unix()
-	cache.CreatedAt = now
-	cache.UsedAt = now
+	now := time.Now()
+	cache.CreatedAt = now.Unix()
+	cache.CreatedAtNanos = now.Nanosecond()
+	cache.UsedAt = now.Unix()
 	if err := h.insertCache(db, cache); err != nil {
 		h.responseJSON(w, r, 500, err)
 		return
@@ -778,6 +780,17 @@ func (h *Handler) commit(w http.ResponseWriter, r *http.Request) {
 		h.responseJSON(w, r, 400, err)
 		return
 	}
+	var request struct {
+		Size *int64 `json:"size"`
+	}
+	body, err := io.ReadAll(io.LimitReader(r.Body, 1<<20))
+	if err == nil && len(body) > 0 {
+		err = json.Unmarshal(body, &request)
+	}
+	if err != nil {
+		h.responseJSON(w, r, http.StatusBadRequest, err)
+		return
+	}
 
 	db, err := h.openDB()
 	if err != nil {
@@ -805,6 +818,14 @@ func (h *Handler) commit(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
+	if request.Size != nil {
+		if *request.Size < 0 || cache.Size >= 0 && cache.Size != *request.Size {
+			h.responseJSON(w, r, http.StatusBadRequest, errors.New("invalid cache size"))
+			return
+		}
+		cache.Size = *request.Size
+	}
+
 	db.Close()
 
 	if err := h.commitCache(cache); err != nil {
@@ -1067,7 +1088,7 @@ func findCache(db *cacheDB, repo string, keys []string, version string) (*Cache,
 		if err != nil {
 			return nil, fmt.Errorf("find cache: %w", err)
 		}
-		if cache := latest(caches, func(cache *Cache) int64 { return cache.CreatedAt }); cache != nil {
+		if cache := latest(caches, (*Cache).createdAtNano); cache != nil {
 			return cache, nil
 		}
 	}
@@ -1078,7 +1099,9 @@ func latest(caches []*Cache, stamp func(*Cache) int64) *Cache {
 	if len(caches) == 0 {
 		return nil
 	}
-	return slices.MaxFunc(caches, func(left, right *Cache) int { return cmp.Compare(stamp(left), stamp(right)) })
+	return slices.MaxFunc(caches, func(left, right *Cache) int {
+		return cmp.Or(cmp.Compare(stamp(left), stamp(right)), cmp.Compare(left.ID, right.ID))
+	})
 }
 
 // findExactCache returns the entry for exactly this key and version, or (nil, nil) if there is
@@ -1097,7 +1120,7 @@ func findExactCache(db *cacheDB, repo, key, version string, complete bool) (*Cac
 	}
 	stamp := func(cache *Cache) int64 { return cache.UsedAt }
 	if complete {
-		stamp = func(cache *Cache) int64 { return cache.CreatedAt }
+		stamp = (*Cache).createdAtNano
 	}
 	return latest(caches, stamp), nil
 }
@@ -1267,7 +1290,9 @@ func (h *Handler) evictSuperseded(db *cacheDB) {
 		groups[group] = append(groups[group], cache)
 	}
 	for _, caches := range groups {
-		slices.SortFunc(caches, func(left, right *Cache) int { return cmp.Compare(left.CreatedAt, right.CreatedAt) })
+		slices.SortFunc(caches, func(left, right *Cache) int {
+			return cmp.Or(cmp.Compare(left.createdAtNano(), right.createdAtNano()), cmp.Compare(left.ID, right.ID))
+		})
 		for _, cache := range caches[:len(caches)-1] {
 			if inUse(cache) {
 				continue

act/artifactcache/handler_test.go

@@ -15,6 +15,7 @@ import (
 	"net/http/httptest"
 	"os"
 	"path/filepath"
+	"strconv"
 	"strings"
 	"testing"
 	"testing/synctest"
@@ -302,6 +303,37 @@ func TestHandler(t *testing.T) {
 		}
 	})
 
+	t.Run("chunked commit validates size omitted at reservation", func(t *testing.T) {
+		for _, size := range []int64{5, 10} {
+			t.Run(strconv.FormatInt(size, 10), func(t *testing.T) {
+				id := reserveCache(t, testClient, base, strings.ToLower(t.Name()), version, 0)
+				req, err := http.NewRequest(http.MethodPatch, fmt.Sprintf("%s/caches/%d", base, id), strings.NewReader("cache"))
+				require.NoError(t, err)
+				req.Header.Set("Content-Range", "bytes 0-4/*")
+				resp, err := testClient.Do(req)
+				require.NoError(t, err)
+				resp.Body.Close()
+				require.Equal(t, http.StatusOK, resp.StatusCode)
+
+				resp, err = testClient.Post(fmt.Sprintf("%s/caches/%d", base, id), "application/json", io.MultiReader(strings.NewReader(fmt.Sprintf(`{"size":%d}`, size))))
+				require.NoError(t, err)
+				resp.Body.Close()
+				if size == 5 {
+					require.Equal(t, http.StatusOK, resp.StatusCode)
+				} else {
+					require.Equal(t, http.StatusInternalServerError, resp.StatusCode)
+				}
+
+				db, err := handler.openDB()
+				require.NoError(t, err)
+				defer db.Close()
+				cache, err := db.get(id)
+				require.NoError(t, err)
+				require.Equal(t, size == 5, cache.Complete)
+			})
+		}
+	})
+
 	t.Run("get with bad id", func(t *testing.T) {
 		resp, err := testClient.Get(base + "/artifacts/invalid_id")
 		require.NoError(t, err)
@@ -709,6 +741,20 @@ func TestHandler_FindProtectsFromEviction(t *testing.T) {
 	assert.Equal(t, []string{wanted.Key}, keptKeys(t, handler, []*Cache{wanted, other}), "the entry just promised to a job survives, the next least recently used goes instead")
 }
 
+func TestHandler_PrefixUsesNewestReservationWithinSecond(t *testing.T) {
+	created := time.Now().Add(-time.Hour).Unix()
+	older := &Cache{Repo: testRepo, Key: "deps-old", Version: "v", Complete: true, CreatedAt: created}
+	newer := &Cache{Repo: testRepo, Key: "deps-new", Version: "v", Complete: true, CreatedAt: created}
+	handler := newTestHandler(t, Policy{}, older, newer)
+
+	db, err := handler.openDB()
+	require.NoError(t, err)
+	defer db.Close()
+	cache, err := findCache(db, testRepo, []string{"deps-"}, "v")
+	require.NoError(t, err)
+	require.Equal(t, newer, cache)
+}
+
 func TestHandler_evictOnCommit(t *testing.T) {
 	full := &Cache{Repo: testRepo, Key: "full", Version: "v", Complete: true, Size: 4 * miB, UsedAt: time.Now().Add(-time.Hour).Unix()}
 	handler := newTestHandler(t, Policy{RepoSizeLimit: 4 * miB}, full)

act/artifactcache/handler_v2.go

@@ -100,15 +100,16 @@ func (h *Handler) v2CreateCacheEntry(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	now := time.Now().Unix()
+	now := time.Now()
 	cache := &Cache{
-		Repo:      cred.Repo,
-		Key:       req.Key,
-		Version:   req.Version,
-		Owner:     owner,
-		Size:      -1, // the size is only known at finalize time
-		CreatedAt: now,
-		UsedAt:    now,
+		Repo:           cred.Repo,
+		Key:            req.Key,
+		Version:        req.Version,
+		Owner:          owner,
+		Size:           -1, // the size is only known at finalize time
+		CreatedAt:      now.Unix(),
+		CreatedAtNanos: now.Nanosecond(),
+		UsedAt:         now.Unix(),
 	}
 	if err := h.insertCache(db, cache); err != nil {
 		h.twirpError(w, r, twirpInternal, err)

act/artifactcache/model.go

@@ -4,6 +4,8 @@
 
 package artifactcache
 
+import "time"
+
 type Request struct {
 	Key     string `json:"key" `
 	Version string `json:"version"`
@@ -11,13 +13,18 @@ type Request struct {
 }
 
 type Cache struct {
-	ID        uint64 `json:"id"`
-	Repo      string `json:"repo"`
-	Key       string `json:"key"`
-	Version   string `json:"version"`
-	Size      int64  `json:"cacheSize"`
-	Owner     string `json:"owner"` // hashed bearer, so a job's own retry is not another job's reservation
-	Complete  bool   `json:"complete"`
-	UsedAt    int64  `json:"usedAt"`
-	CreatedAt int64  `json:"createdAt"`
+	ID             uint64 `json:"id"`
+	Repo           string `json:"repo"`
+	Key            string `json:"key"`
+	Version        string `json:"version"`
+	Size           int64  `json:"cacheSize"`
+	Owner          string `json:"owner"` // hashed bearer, so a job's own retry is not another job's reservation
+	Complete       bool   `json:"complete"`
+	UsedAt         int64  `json:"usedAt"`
+	CreatedAt      int64  `json:"createdAt"`
+	CreatedAtNanos int    `json:"createdAtNanos,omitzero"`
+}
+
+func (cache *Cache) createdAtNano() int64 {
+	return cache.CreatedAt*int64(time.Second) + int64(cache.CreatedAtNanos)
 }

act/artifactcache/s3_storage_test.go

@@ -8,6 +8,7 @@ import (
 	"io"
 	"net/http"
 	"net/http/httptest"
+	"strconv"
 	"strings"
 	"sync"
 	"sync/atomic"
@@ -51,14 +52,33 @@ func TestHandler_S3SharesEntriesAndDropsExpiredOnes(t *testing.T) {
 	saver, restorer := startTestHandler(t, opts), startTestHandler(t, opts)
 	entry := map[string]any{"key": "deps", "version": "v1"}
 
+	started := time.Now()
 	finalized, _ := saveV2(t, saver, "deps", "v1", []byte("archive"))
 	require.Equal(t, true, finalized["ok"])
 	assert.EqualValues(t, 1, uploads.Load())
+	id, err := strconv.ParseUint(finalized["entry_id"].(string), 10, 64)
+	require.NoError(t, err)
+	cache, err := saver.s3.fetch(t.Context(), id)
+	require.NoError(t, err)
+	assert.WithinRange(t, time.Unix(cache.CreatedAt, int64(cache.CreatedAtNanos)), started, time.Now())
 	require.NoError(t, restorer.importMetadata(t.Context()))
 	found := v2Call(t, restorer, testClient, "GetCacheEntryDownloadURL", entry)
 	require.Equal(t, true, found["ok"])
 	assert.Equal(t, []byte("archive"), getURL(t, found["signed_download_url"].(string)))
 
+	for _, cache := range []*Cache{
+		{ID: 2, Repo: testRepo, Key: "deps-old", Version: "v1", Complete: true, CreatedAt: cache.CreatedAt},
+		{ID: 1, Repo: testRepo, Key: "deps-new", Version: "v1", Complete: true, CreatedAt: cache.CreatedAt, CreatedAtNanos: 1},
+	} {
+		require.NoError(t, saver.s3.publish(cache))
+		objects.Store(saver.s3.blobKey(cache.ID), []byte(cache.Key))
+	}
+	require.NoError(t, restorer.importMetadata(t.Context()))
+	found = v2Call(t, restorer, testClient, "GetCacheEntryDownloadURL", map[string]any{"key": "deps-", "version": "v1"})
+	require.Equal(t, true, found["ok"])
+	assert.Equal(t, "deps-new", found["matched_key"])
+	assert.Equal(t, []byte("deps-new"), getURL(t, found["signed_download_url"].(string)))
+
 	objects.Clear()
 	require.NoError(t, restorer.importMetadata(t.Context()))
 	assert.Equal(t, true, v2Call(t, restorer, testClient, "CreateCacheEntry", entry)["ok"])