syz-cluster: use base-commit hint from patch series for triage

If the series author provided a base-commit, use it. If applying the series against the base-commit fails, fall back to
the default method.

Author: pimyn-girgis

pkg/vcs/git.go

@@ -68,8 +68,8 @@ func filterEnv() []string {
 
 func (git *gitRepo) Poll(repo, branch string) (*Commit, error) {
 	git.Reset()
-	origin, err := git.Run("remote", "get-url", "origin")
-	if err != nil || strings.TrimSpace(string(origin)) != repo {
+	origin, err := git.GetURL("origin")
+	if err != nil || strings.TrimSpace(origin) != repo {
 		// The repo is here, but it has wrong origin (e.g. repo in config has changed), re-clone.
 		if err := git.clone(repo, branch); err != nil {
 			return nil, err
@@ -102,6 +102,14 @@ func (git *gitRepo) Poll(repo, branch string) (*Commit, error) {
 	return git.Commit(HEAD)
 }
 
+func (git Git) GetURL(branch string) (string, error) {
+	url, err := git.Run("remote", "get-url", branch)
+	if err != nil {
+		return "", err
+	}
+	return string(url), nil
+}
+
 func (git *gitRepo) isNetworkError(output []byte) bool {
 	// The list is not exhaustive and is meant to be extended over time.
 	return bytes.Contains(output, []byte("fatal: read error: Connection reset by peer"))
@@ -752,6 +760,24 @@ type BaseCommit struct {
 	Branches []string
 }
 
+func (git Git) TreeAndBranchForCommit(commit string) (tree, branch string, err error) {
+	const cutOffDays = 60
+	branchList, err := git.BranchesThatContain(commit, time.Now().Add(-time.Hour*24*cutOffDays))
+	if err == nil {
+		return "", "", fmt.Errorf("failed to query branches: %w", err)
+	}
+	for _, branch := range branchList {
+		if strings.Contains(branch.Branch, "/") {
+			strs := strings.SplitN(branch.Branch, "/", 2)
+			return strs[0], strs[1], nil
+		}
+	}
+	if len(branchList) != 0 {
+		return branchList[0].Branch, "", nil
+	}
+	return "", "", fmt.Errorf("failed to find a branch containing commit")
+}
+
 // BaseForDiff returns a list of commits that could have been the base commit
 // for the specified git patch.
 // The returned list is minimized to only contain the commits that are represented in different
@@ -830,7 +856,7 @@ func (git Git) BaseForDiff(diff []byte, tracer debugtracer.DebugTracer) ([]*Base
 		}
 		// Only focus on branches that are still alive.
 		const cutOffDays = 60
-		list, err := git.branchesThatContain(candidate, time.Now().Add(-time.Hour*24*cutOffDays))
+		list, err := git.BranchesThatContain(candidate, time.Now().Add(-time.Hour*24*cutOffDays))
 		if err != nil {
 			return nil, fmt.Errorf("failed to query branches: %w", err)
 		}
@@ -924,7 +950,7 @@ type branchCommit struct {
 	Commit string
 }
 
-func (git Git) branchesThatContain(commit string, since time.Time) ([]branchCommit, error) {
+func (git Git) BranchesThatContain(commit string, since time.Time) ([]branchCommit, error) {
 	output, err := git.Run(
 		"branch", "-a",
 		"--contains", commit,

syz-cluster/workflow/triage-step/main.go

@@ -106,9 +106,15 @@ func (triager *seriesTriager) GetVerdict(ctx context.Context, sessionID string)
 
 func (triager *seriesTriager) prepareFuzzingTask(ctx context.Context, series *api.Series, trees []*api.Tree,
 	target *triage.MergedFuzzConfig) (*api.FuzzTask, error) {
-	result, err := triager.selectFromBlobs(series, trees)
+	result, err := triager.selectFromBaseCommitHint(series.BaseCommitHint)
 	if err != nil {
-		return nil, fmt.Errorf("selection by blob failed: %w", err)
+		return nil, fmt.Errorf("selection by base-commit failed: %w", err)
+	}
+	if result == nil {
+		result, err = triager.selectFromBlobs(series, trees)
+		if err != nil {
+			return nil, fmt.Errorf("selection by blob failed: %w", err)
+		}
 	}
 	if result == nil {
 		result, err = triager.selectFromList(ctx, series, trees, target)
@@ -168,6 +174,29 @@ func (triager *seriesTriager) selectFromBlobs(series *api.Series, trees []*api.T
 	}, nil
 }
 
+func (triager *seriesTriager) selectFromBaseCommitHint(commit string) (*SelectResult, error) {
+	triager.Log("attempting to use the base commit provided by author")
+	git := triager.ops.Git
+	tree, branch, err := triager.ops.TreeAndBranchForCommit(commit)
+	if err != nil {
+		return nil, err
+	}
+	url, err := git.GetURL(tree + "/" + branch)
+	if err != nil {
+		return nil, err
+	}
+	return &SelectResult{
+		Tree: &api.Tree{
+			Name:       tree,
+			URL:        url,
+			Branch:     branch,
+			EmailLists: []string{""},
+		},
+		Commit: commit,
+		Arch:   fuzzArch,
+	}, nil
+}
+
 func (triager *seriesTriager) selectFromList(ctx context.Context, series *api.Series, trees []*api.Tree,
 	target *triage.MergedFuzzConfig) (*SelectResult, error) {
 	selectedTrees := triage.SelectTrees(series, trees)