| Age | Commit message (Collapse) | Author |
|---|
|
50-line limit
Extract buildTUIRuntime and makeTUIEventLoopConfigurer from the monolithic
tuiTraceStarterFromRunTrace closure, and extract maybePrependFlamegraphConfigure
and finaliseTrace from runTraceWithContext. All functions are now <= 50 lines.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
Extract handleHelpOverlayKeyPress, handleQuitKeyPress, routeQuitAsEsc, and
handleDashboardShortcutKeys from the 89-line handleGlobalKeyPress to comply
with the project's 50-line function limit. All existing behaviour is preserved.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
Extract captureSnapshotInputs (lock-guarded state copy), buildSubSnapshots
(concurrent per-category builders), and populateSnapshotFields (scalar
field assignment) from the monolithic Snapshot method. The method itself
is now 18 lines; all three helpers are well under 30 lines each.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
Extract handleSearchInput and handleKeyNavigation from Update (111 lines → 33),
then split handleKeyNavigation into handleModeKey and handleMovementKey.
Extract frameCoordToTargetRow and findFrameAtRow from frameIndexAt (66 lines → 32).
All new helpers are ≤ 29 lines; behavior is preserved.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
Move the per-tick snapshot refresh off the Bubble Tea update goroutine,
add a frame ancestry index so navigation and filter helpers run in
O(subtree) instead of O(frames), skip refresh and animation while the
user is actively pressing keys, and memoize View() output. Keystrokes
(pause, filter, navigate) now land within one frame even when the live
trie ingests thousands of events per tick.
- new SnapshotTree() on LiveTrie bypasses JSON marshal+unmarshal
- RefreshFromLiveTrieCmd runs SnapshotTree + layout + ancestry on a
background goroutine, coalesced via refreshInFlight, and returns a
flameSnapshotReadyMsg the Update loop applies cheaply
- driveWindow gate (250 ms after last key press) skips refresh dispatch
and snaps frames directly to target without animation while the user
is driving
- View() caches its rendered string keyed on the inputs that affect
output; cache is bypassed during animation
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
|
Renders the next-tick countdown ("12s/30s") in the dashboard chrome so
users can see when aggregates will clear, and adds a dedicated H-help
line spelling out the cycle keys (off → 10s → 30s → 1m → 2m → 5m).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
|
Two follow-up refinements to the auto-reset timer added in 8da473a.
- Hotkey cycle now goes off -> 10s -> 30s -> 60s -> 2m -> 5m -> off,
giving the user finer control between 60s and 5m and a quicker
starting cadence.
- The timer now pauses while the TUI is blurred. SetFocused returns a
tea.Cmd that re-arms a fresh tick on focus regain, and bumps the
generation counter on every focus change so any tick scheduled before
blur is dropped on arrival. autoResetTickCmd and handleAutoResetTick
also gate on m.focused as defense in depth.
- Dashboard chrome shows 'auto-reset: 30s (paused)' while the timer is
enabled but blurred, distinguishing it from the disabled 'off' state.
Tests cover the full preset cycle (including custom-value passthrough)
and the pause-on-blur lifecycle: stale ticks ignored, current-gen ticks
ignored while blurred, focus regain re-arms and fires the reset, no-op
focus calls don't churn the generation counter, and the chrome label
flips to '(paused)' as expected.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
|
Live flamegraph trie and stats engine grow unboundedly during long
traces. Add a periodic auto-reset (same effect as the 'r' key) so they
stay bounded.
- New CLI flag -resetTimer=30s (default 30s, 0 disables).
- Hotkey I cycles the cadence: off -> 15s -> 30s -> 60s -> 5m -> off.
Custom intervals (e.g. -resetTimer=47s) advance to the first preset
greater than the current value, then wrap to off.
- autoResetTickMsg carries a generation counter so changing the cadence
drops in-flight ticks scheduled under the previous interval.
- Dashboard chrome shows 'auto-reset: 30s' or 'auto-reset: off'.
|
|
PrepareForTraceRestart was designed for the full-restart path, where the
dashboard's live-trie reference is rebound when TracingStartedMsg fires.
The in-place filter swap skips that message, leaving the flamegraph tab
stuck on 'Flame: waiting for data...' until the next real trace restart.
Re-bind via SetLiveTrie immediately after PrepareForTraceRestart in both
applyGlobalFilter and undoGlobalFilter.
|
|
Changing the global filter used to call stopTrace + beginTraceCmd, which
detached and re-attached every tracepoint and re-loaded the BPF object.
On heavily loaded I/O systems that took several seconds and showed an
'Attaching tracepoints...' overlay each time. The probe set never depends
on the global filter (ShouldIAttachTracepoint only reads CLI regex flags),
so the restart was gratuitous.
Now the eventloop holds its filter behind atomic.Pointer with SetFilter /
Filter accessors, and the trace starter registers el.SetFilter via the
runtime bindings as a SetLiveFilterSetter callback. applyGlobalFilter
and undoGlobalFilter call runtime.applyLiveFilter first; only if no
trace is running do they fall back to the full restart path.
|
|
- Bubbles, treemap, icicle, and the live flamegraph 'b' cycle now include
syscall duration (sum) as a third metric alongside events and bytes.
Statsengine snapshots expose TotalLatencyNs to support this.
- AttachAll takes an optional warn callback. Production passes one so older
kernels that lack newer tracepoints log a warning and keep going instead
of aborting startup.
- Dockerfile.el8 + scripts/build-with-docker-el8.sh + mage buildDockerEl8
produce ior.el8, a static binary built against Rocky Linux 8 glibc for
RHEL/Rocky/Alma 8 hosts.
- README.md documents installing mage and the new el8 target.
|
|
|
|
|
|
Introduces a Docker-based build path so ior can be compiled on any
Linux host without a native Rocky 9 toolchain setup:
- Dockerfile: Rocky 9 minimal image with Go (version from ARG, default
from go.mod), static libelf/libzstd built from source, libbpfgo at
v0.9.2-libbpf-1.5.1, and mage; CMD runs mage generate + mage all
against the repo root mounted as a volume.
- scripts/build-with-docker.sh: reads GO_VERSION from go.mod, passes it
as --build-arg to docker build, mounts tracefs and BTF into the
container, writes the binary to the repo root.
- Magefile.go: adds BuildDocker target that wraps the script.
- README.md: simplified to the two build paths (Docker + native) with
links to docs/; removed GOTOOLCHAIN=auto throughout.
- docs/build-rocky-linux-9.md: full manual Rocky 9 steps, libbpfgo
toolchain setup/rollback, compile-once-run-everywhere explanation,
and timing semantics.
- docs/tui-reference.md: complete TUI hotkey reference, recording mode
details, and the .ior.zst vs Parquet trade-off table.
- AGENTS.md: removed GOTOOLCHAIN=auto from all build commands.
- internal/c/generated_tracepoints.c: regenerated against the host kernel.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
|
The BPF handler generator emitted struct trace_event_raw_sys_enter/
trace_event_raw_sys_exit (the BTF-blessed aliases). RHEL 9 carries an
rt-tree backport that adds preempt_lazy_count to struct trace_entry,
which widens those aliases by 8 bytes and shifts args/ret. The actual
tracepoint context the kernel hands the program is still
syscall_trace_enter / syscall_trace_exit, where the offsets did not
move. Programs typed against the wider alias read past max_ctx_offset
and the verifier rejects the attach with EACCES.
Switching the generator to emit syscall_trace_enter/exit lines up with
the real context on RHEL 9 (and is identical on every other distro,
since the two structs only diverge there). Same fix bcc shipped in
iovisor/bcc#4920 and inspektor-gadget did in inspektor-gadget#2546.
Field accesses (ctx->args[N], ctx->ret) are unchanged.
Verified end-to-end on Rocky Linux 9.7 stock 5.14.0-611.5.1.el9_7
(no kernel-ml needed) and Fedora 6.19. README rewritten accordingly:
drops the elrepo kernel-ml step and the trailing 'permission denied'
troubleshooting paragraph; adds a historical note explaining why the
old workaround existed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
|
runTraceWithContext and runHeadlessParquet opened cpu/mem/exec-trace
profile files via setupProfiling, then relied on the shutdown watcher
goroutine (registered later) to close them. If any intervening step
(newEventLoop, recorder.Start) returned an error, the watcher was never
registered and the profile fds leaked until process exit. Add a
defer profiling.stop(logln) immediately after setupProfiling so the fds
are released on every return path. profiling.stop is idempotent via
sync.Once so the watcher and defer can both run harmlessly.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
tui.Model (task 429)
tui.Model had 33 fields mixing keyboard tracking, filter chain, process
selection, and UI presentation concerns (SRP violation).
Extract three focused sub-structs:
- keyboardState (kb): enhancements, lastEvent{ID,At,WasPress},
suppress{ID,Until} — 7 fields managed by keys_normalize.go
- filterState (filter): global filter, history slice, label stack
— 3 fields for the trace filter chain
- processState (proc): pid, tid, pickerReturn
— 3 fields for PID/TID selection and picker navigation
Model drops from 33 to 23 top-level fields. All field accesses in
tui.go, keys_normalize.go, and tui_test.go are updated accordingly.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
(task 427/ISP)
TraceRuntimeBindings mixed 4 setter methods (injecting data into TUI) with
4 getter methods (reading persistent TUI-owned state), violating ISP.
Split into two focused interfaces:
- RuntimePublisher: SetDashboardSnapshotSource, SetEventStreamSource,
SetLiveTrie, SetProbeManager — the write/inject side
- RuntimeState: StreamBuffer, Recorder, StreamSequencer, FilterEpoch
— the read/persistent-state side
TraceRuntimeBindings now embeds both, preserving all existing call sites.
RuntimePublisherFromContext() added so callers that only inject data do not
see the getter surface. Three such callers are narrowed: setupBPFModule,
tuiTestFlamesStarter, tuiTestLiveFlamesStarter.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
ior.go had 763 lines covering 9+ concerns. Follow the eventloop_*.go
pattern and extract into three focused files:
- ior_bpfsetup.go: libbpfTracepoint{Program,Module} adapter types,
setupBPFModule, setupBPFModuleError, setupEventChannel
- ior_profiling.go: profilingControl type, setupProfiling,
profilingFilesForMode, stop()
- ior_parquet_sink.go: headlessParquetSink type, runHeadlessParquet,
isHeadlessParquetMode, hasHeadlessParquetContentFilters,
headlessParquetTraceConfig; inline parquetMetadata one-liner
ior.go shrinks from 763 → 453 lines, retaining entry, dispatch,
TUI wiring, and core trace execution.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
(task 429)
flags.Get() (global mutable singleton) was called inside library packages,
coupling them to global state and making tests fragile (DIP violation).
- internal.Run() now takes an explicit flags.Config; main.go calls
flags.Get() once at the CLI boundary and passes it in.
- tui.Run(), RunWithTraceStarter(), RunTestFlamesWithTraceStarter() removed;
callers already used the WithConfig variants directly.
- tui.NewModel() preserved for test ergonomics but now uses flags.NewFlags()
(pure defaults) instead of flags.Get() (global state).
- Tests updated to use NewModelWithConfig() with explicit config structs
instead of flags.Set*() + NewModel(), eliminating all test-level
global-state mutation.
flags.Get() is now called only in cmd/ior/main.go, the correct boundary.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
responsibilities (task 428)
The eventLoop struct held 20+ fields across 5+ responsibilities (SRP violation).
Extract two cohesive sub-structs:
- pairTracker: enter/exit pair matching, age-based LRU pruning, and
DurationToPrev tracking. Replaces enterEvs/enterEvAges/prevPairTimes/
maxPendingEnterEvs/cacheAge fields with a single embedded value.
- fdTracker (extended): absorbs procFdCache/procFdAges/maxProcFdCacheSize,
moving all procfs-resolution cache logic (resolve, cache, prune, delete)
off eventLoop and onto the tracker that already owns the fd table.
eventLoop drops from 20 fields to 12. All methods that previously reached
into eventLoop fields now live on the struct that owns the data.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
(task 432)
The exitHandlers map keyed by reflect.Type was a reflection-based dispatch
table that incurred allocation and reflection overhead on every event pair
in the hot processing path. Replace it with a plain type switch in
handleTracepointExit, which the compiler resolves statically.
Removes: initExitHandlers, typeKey, mustBeType, newTypedExitHandler,
exitHandlerRegistry, the exitHandlers struct field, and the tracepointExitHandler
type alias — all dead after the switch. Fixes a pre-existing uint→uint64
mismatch in stats(). Updates tests to target the new default branch directly.
Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
