feat(go): enrich snapshots with package ownership

This commit is contained in:
Luna 2026-07-22 03:50:55 -07:00
parent 6355fb403d
commit 88d7a6e3b8
No known key found for this signature in database
8 changed files with 149 additions and 54 deletions

View file

@ -1,8 +1,8 @@
# Go Port Handoff # Go Port Handoff
Saved: 2026-07-22T02:34:00-07:00 Saved: 2026-07-22T02:48:00-07:00
Branch: `main` Branch: `main`
Base commit: `eff31b3` (`feat(go): enrich snapshots asynchronously`) Base commit: `6355fb4` (`feat(go): expose retained incident views`)
Status: implemented and green, but uncommitted Status: implemented and green, but uncommitted
## Worktree warning ## Worktree warning
@ -11,9 +11,9 @@ The checkout is intentionally dirty and contains work from multiple related
continuations. Do not reset, clean, or broadly restage it. continuations. Do not reset, clean, or broadly restage it.
- The validation-sidecar, incident-persistence, bounded-enrichment, local- - The validation-sidecar, incident-persistence, bounded-enrichment, local-
assurance, and asynchronous-enrichment tranches are signed commits assurance, asynchronous-enrichment, and incident-reader tranches are signed
`f85c2e8`, `538d1d9`, `fd2bbba`, `1ab4add`, and `eff31b3`. The native commits `f85c2e8`, `538d1d9`, `fd2bbba`, `1ab4add`, `eff31b3`, and
incident-reader slice is uncommitted. `6355fb4`. The package-ownership enrichment slice is uncommitted.
- At this checkpoint, `git status --short` has 20 entries with untracked - At this checkpoint, `git status --short` has 20 entries with untracked
directories collapsed. directories collapsed.
- The Python GUI files and tests are separate pre-existing work. Preserve them - The Python GUI files and tests are separate pre-existing work. Preserve them
@ -103,8 +103,10 @@ static binary.
- A single 16-job asynchronous worker now streams SHA-256 only for regular - A single 16-job asynchronous worker now streams SHA-256 only for regular
executables at most 8 MiB and adds lstat file metadata. Slow I/O occurs executables at most 8 MiB and adds lstat file metadata. Slow I/O occurs
outside the snapshot lock; a full queue simply leaves optional fields unknown. outside the snapshot lock; a full queue simply leaves optional fields unknown.
- Package ownership, richer integrity anchors, and notification fan-out are not - The asynchronous worker now resolves pacman/dpkg/rpm package-owner strings
yet ported. with a two-second bound and cache; ownership lookup never holds a detector
lock or blocks alert capture.
- Richer integrity anchors and notification fan-out are not yet ported.
- `--incidents-list` and `--incident-show <id>` now read the sidecar's isolated - `--incidents-list` and `--incident-show <id>` now read the sidecar's isolated
state without starting the agent. The latter returns `enodia.incident.view.v1` state without starting the agent. The latter returns `enodia.incident.view.v1`
with the incident, available snapshots, and time-ordered timeline. with the incident, available snapshots, and time-ordered timeline.
@ -152,8 +154,8 @@ the suite still exits successfully.
## Resume here ## Resume here
1. Add bounded package-ownership and integrity-anchor collectors to the existing 1. Add bounded integrity-anchor collectors to the existing asynchronous worker
asynchronous worker without destructive response behavior. without destructive response behavior.
2. Add a read-only management API consumer for isolated Go snapshot/event state 2. Add a read-only management API consumer for isolated Go snapshot/event state
while keeping Python authoritative. while keeping Python authoritative.
3. Continue broader Phase 3 rule metadata/state parity before considering any 3. Continue broader Phase 3 rule metadata/state parity before considering any

View file

@ -116,10 +116,11 @@ the configured `incident_tracking`, `incident_window`, and
and one snapshot name idempotently. Hot-path enrichment derives bounded and one snapshot name idempotently. Hot-path enrichment derives bounded
process/parent context, lineage, remote-IP classification, and candidate paths process/parent context, lineage, remote-IP classification, and candidate paths
from already-captured data without new I/O. A 16-job asynchronous worker then from already-captured data without new I/O. A 16-job asynchronous worker then
adds streamed SHA-256 values for regular executables up to 8 MiB and file adds streamed SHA-256 values for regular executables up to 8 MiB, file metadata,
metadata; saturation leaves fields unknown rather than delaying alerts. Package and pacman/dpkg/rpm package-owner strings with a two-second query bound;
ownership, richer integrity anchors, and notifications remain future parity saturation leaves fields unknown rather than delaying alerts. Richer integrity
work. Snapshot text/JSON revisions append synchronized `enodia.hash_chain.v1` anchors and notifications remain future parity work. Snapshot text/JSON
revisions append synchronized `enodia.hash_chain.v1`
records to a private local `hash-chain.jsonl`. records to a private local `hash-chain.jsonl`.
After every successfully emitted status record, the service atomically updates After every successfully emitted status record, the service atomically updates

View file

@ -46,6 +46,10 @@ func TestRunAttachesLiveBaselineLifecycle(t *testing.T) {
cfg := config.Default() cfg := config.Default()
cfg.LogDir = t.TempDir() cfg.LogDir = t.TempDir()
cfg.BaselineGrace = 0 cfg.BaselineGrace = 0
// This fixture exercises listener-baseline lifecycle only. Avoid inheriting
// the host's live persistence paths (for example /etc/passwd), whose mtime
// would otherwise add an unrelated environment-dependent alert.
cfg.Detectors["persistence"] = false
calls := 0 calls := 0
capture := func() (model.State, error) { capture := func() (model.State, error) {
calls++ calls++

View file

@ -47,17 +47,23 @@ type SlowResult struct {
Files map[string]map[string]any Files map[string]map[string]any
} }
func CollectSlow(processes []Process, paths []string) SlowResult { func CollectSlow(processes []Process, paths []string, owner ...func(string) string) SlowResult {
result := SlowResult{Processes: map[int]map[string]any{}, Files: map[string]map[string]any{}} result := SlowResult{Processes: map[int]map[string]any{}, Files: map[string]map[string]any{}}
packageOwner := func(string) string { return "" }
if len(owner) > 0 && owner[0] != nil {
packageOwner = owner[0]
}
for _, process := range processes { for _, process := range processes {
exe := cleanPath(process.Exe) exe := cleanPath(process.Exe)
if exe == "" { if exe == "" {
continue continue
} }
result.Processes[process.PID] = map[string]any{"exe_sha256": boundedHash(exe)} result.Processes[process.PID] = map[string]any{"exe_sha256": boundedHash(exe), "package_owner": packageOwner(exe)}
} }
for _, path := range paths { for _, path := range paths {
result.Files[path] = fileMetadata(path) metadata := fileMetadata(path)
metadata["package_owner"] = packageOwner(path)
result.Files[path] = metadata
} }
return result return result
} }

View file

@ -3,6 +3,8 @@
package enrich package enrich
import ( import (
"os"
"path/filepath"
"testing" "testing"
"codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/model" "codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/model"
@ -54,3 +56,21 @@ func TestBuildCapsAttackerControlledIndicators(t *testing.T) {
t.Fatal("indicator caps were not applied") t.Fatal("indicator caps were not applied")
} }
} }
func TestCollectSlowAddsInjectedPackageOwnership(t *testing.T) {
directory := t.TempDir()
exe := filepath.Join(directory, "payload")
if err := os.WriteFile(exe, []byte("payload"), 0o700); err != nil {
t.Fatal(err)
}
result := CollectSlow([]Process{{PID: 42, Exe: exe}}, []string{exe}, func(path string) string {
if path == exe {
return "example-package 1.0-1"
}
return ""
})
if result.Processes[42]["package_owner"] != "example-package 1.0-1" ||
result.Files[exe]["package_owner"] != "example-package 1.0-1" {
t.Fatalf("result=%#v", result)
}
}

View file

@ -14,57 +14,91 @@ import (
) )
var ( var (
mu sync.Mutex toolOnce sync.Once
tool *string tool string
cache = map[string]bool{} cacheMu sync.Mutex
cache = map[string]string{}
seen = map[string]bool{}
) )
func packageTool() string { func packageTool() string {
if tool != nil { toolOnce.Do(func() {
return *tool
}
found := ""
for _, name := range []string{"pacman", "dpkg", "rpm"} { for _, name := range []string{"pacman", "dpkg", "rpm"} {
if _, err := exec.LookPath(name); err == nil { if _, err := exec.LookPath(name); err == nil {
found = name tool = name
break break
} }
} }
tool = &found })
return found return tool
} }
func query(name, path string) bool { func queryOwner(name, path string) string {
ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second) ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
defer cancel() defer cancel()
var output []byte
var err error
switch name { switch name {
case "pacman": case "pacman":
output, err := exec.CommandContext(ctx, "pacman", "-Qo", path).Output() output, err = exec.CommandContext(ctx, "pacman", "-Qo", path).Output()
return err == nil && strings.Contains(string(output), "owned by")
case "dpkg": case "dpkg":
output, err := exec.CommandContext(ctx, "dpkg", "-S", path).Output() output, err = exec.CommandContext(ctx, "dpkg", "-S", path).Output()
return err == nil && strings.Contains(string(output), ":")
case "rpm": case "rpm":
output, err := exec.CommandContext(ctx, "rpm", "-qf", path).Output() output, err = exec.CommandContext(ctx, "rpm", "-qf", path).Output()
return err == nil && !strings.Contains(string(output), "not owned")
} }
return false if err != nil {
return ""
}
return parseOwner(name, string(output))
}
func parseOwner(name, output string) string {
output = strings.TrimSpace(output)
switch name {
case "pacman":
_, owner, found := strings.Cut(output, " is owned by ")
if found {
return strings.TrimSpace(owner)
}
case "dpkg":
owner, _, found := strings.Cut(output, ":")
if found {
return strings.TrimSpace(owner)
}
case "rpm":
if output != "" && !strings.Contains(strings.ToLower(output), "not owned") {
return output
}
}
return ""
}
// PackageOwner returns the owning installed-package description, or an empty
// string. The command is bounded and runs without holding the cache lock.
func PackageOwner(path string) string {
path = strings.ReplaceAll(path, " (deleted)", "")
if path == "" || path == "(deleted)" {
return ""
}
cacheMu.Lock()
if seen[path] {
owner := cache[path]
cacheMu.Unlock()
return owner
}
cacheMu.Unlock()
owner := ""
if name := packageTool(); name != "" {
owner = queryOwner(name, path)
}
cacheMu.Lock()
cache[path], seen[path] = owner, true
cacheMu.Unlock()
return owner
} }
// IsPackageOwned reports whether path belongs to an installed package. // IsPackageOwned reports whether path belongs to an installed package.
// Results are cached because package ownership is stable within a sweep run. // Results are cached because package ownership is stable within a sweep run.
func IsPackageOwned(path string) bool { func IsPackageOwned(path string) bool {
if path == "" || path == "(deleted)" { return PackageOwner(path) != ""
return false
}
path = strings.ReplaceAll(path, " (deleted)", "")
mu.Lock()
defer mu.Unlock()
if owned, seen := cache[path]; seen {
return owned
}
name := packageTool()
owned := name != "" && query(name, path)
cache[path] = owned
return owned
} }

View file

@ -0,0 +1,19 @@
// SPDX-License-Identifier: GPL-3.0-or-later
package provenance
import "testing"
func TestParseOwnerMatchesSupportedPackageManagers(t *testing.T) {
cases := []struct{ tool, output, want string }{
{"pacman", "/usr/bin/x is owned by core/example 1.2-1", "core/example 1.2-1"},
{"dpkg", "example: /usr/bin/x", "example"},
{"rpm", "example-1.2-1.x86_64", "example-1.2-1.x86_64"},
{"rpm", "file /tmp/x is not owned by any package", ""},
}
for _, tc := range cases {
if got := parseOwner(tc.tool, tc.output); got != tc.want {
t.Fatalf("%s %q -> %q, want %q", tc.tool, tc.output, got, tc.want)
}
}
}

View file

@ -21,6 +21,7 @@ import (
"codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/enrich" "codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/enrich"
"codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/incident" "codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/incident"
"codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/model" "codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/model"
"codeberg.org/anassaeneroi/enodia-sentinal/go-agent/internal/provenance"
) )
const Schema = "enodia.alert.snapshot.v1" const Schema = "enodia.alert.snapshot.v1"
@ -289,7 +290,7 @@ func (s *Store) applySlowEnrichment(job slowJob) {
} }
processes := enrichmentProcesses(report.Processes) processes := enrichmentProcesses(report.Processes)
paths := enrichmentPaths(report.Enrichment) paths := enrichmentPaths(report.Enrichment)
updates := enrich.CollectSlow(processes, paths) updates := enrich.CollectSlow(processes, paths, provenance.PackageOwner)
// All slow I/O has completed. Take the lock only to merge into the newest // All slow I/O has completed. Take the lock only to merge into the newest
// same-second report, so a concurrent alert capture never loses evidence. // same-second report, so a concurrent alert capture never loses evidence.
@ -342,6 +343,7 @@ func mergeSlow(report *Report, updates enrich.SlowResult) {
pid, _ := item["pid"].(float64) pid, _ := item["pid"].(float64)
if update := updates.Processes[int(pid)]; update != nil { if update := updates.Processes[int(pid)]; update != nil {
item["exe_sha256"] = update["exe_sha256"] item["exe_sha256"] = update["exe_sha256"]
item["package_owner"] = nullToNil(update["package_owner"])
} }
} }
} }
@ -354,13 +356,20 @@ func mergeSlow(report *Report, updates enrich.SlowResult) {
path, _ := item["path"].(string) path, _ := item["path"].(string)
if update := updates.Files[path]; update != nil { if update := updates.Files[path]; update != nil {
for key, value := range update { for key, value := range update {
item[key] = value item[key] = nullToNil(value)
} }
} }
} }
} }
} }
func nullToNil(value any) any {
if owner, ok := value.(string); ok && owner == "" {
return nil
}
return value
}
func (s *Store) collectProcesses(alerts []model.Alert) []ProcessDetail { func (s *Store) collectProcesses(alerts []model.Alert) []ProcessDetail {
pids := alertPIDs(alerts) pids := alertPIDs(alerts)
result := make([]ProcessDetail, 0, len(pids)) result := make([]ProcessDetail, 0, len(pids))