chore(pack): sync turbopack canary

[fireairforce]

Summary

The bundled Turbopack fork was behind the current vercel/next.js canary and was missing upstream runtime, filesystem, HMR, task scheduling, and build-performance improvements.

• sync the next.js submodule to upstream canary 76e24273c77
• preserve Utoo-specific watcher, runtime, HMR, wasm, and lightningcss behavior in submodule commit 8abf2310ba6
• adapt pack-napi to the new backing-storage options API without changing existing behavior
• update 8 runtime source map snapshots affected by the upstream runtime changes

Test Plan

• cargo check
• cargo fmt --check
• cargo clippy --all-targets -- -D warnings --no-deps
• cargo clippy -p pack-api -- -D warnings
• cargo nextest r -p pack-tests — 133 passed, 41 skipped
• cargo build --features plugin -p pack-napi --profile release-local
• run ut build:local from packages/utoo-web
• npm test --workspace shared-runtime-async-module

[github-actions]

📊 Performance Benchmark Report (with-antd)

Utoopack Performance Report

Report ID: utoopack_performance_report_20260825_095455
Generated: 2026-08-25 09:54:55
Trace File: trace_antd.json (0.3GB, 0.80M spans)
Test Project: examples/with-antd

---

Executive Summary

Metric	Value	Assessment
Total Wall Time	5,831.6 ms	Baseline
Total Thread Work (de-duped)	17,401.4 ms	Non-overlapping busy time
Effective Parallelism	3.0x	thread_work / wall_time
Working Threads	10	Threads with actual spans
Thread Utilization	29.8%	⚠️ Suboptimal
Total Spans	800,874	All B/E + X events
Meaningful Spans (>= 10us)	187,065	(23.4% of total)
Tracing Noise (< 10us)	613,809	(76.6% of total)

---

Build Phase Timeline

Shows when each build phase is active and how much CPU it consumes.
Self-Time is the time spent exclusively in that phase (excluding children).

Phase	Spans	Inclusive (ms)	Self-Time (ms)	Wall Range (ms)
Resolve	41,918	5,501.8	1,480.8	2,578.2
Parse	8,130	1,220.0	912.1	5,114.8
Analyze	120,311	33,923.5	7,154.5	5,036.9
Chunk	5,334	5,278.9	1,024.1	2,198.8
Codegen	9,086	1,841.5	1,214.1	1,914.5
Emit	33	36.1	18.1	7.5
Other	2,253	6,833.4	4,092.1	5,831.6

---

Workload Distribution by Diagnostic Tier

Category	Spans	Inclusive (ms)	% Work	Self-Time (ms)	% Self
P0: Scheduling & Resolution	162,523	39,889.2	229.2%	8,852.8	50.9%
P1: I/O & Heavy Tasks	2,834	102.1	0.6%	84.0	0.5%
P2: Architecture (Locks/Memory)	0	0.0	0.0%	0.0	0.0%
P3: Asset Pipeline	20,851	8,358.2	48.0%	3,162.3	18.2%
P4: Bridge/Interop	0	0.0	0.0%	0.0	0.0%
Other	857	6,285.7	36.1%	3,796.6	21.8%

---

Top 20 Tasks by Self-Time

Self-time is the exclusive duration: time spent in the task itself, not in sub-tasks.
This is the most accurate indicator of where CPU cycles are actually spent.

Self (ms)	Inclusive (ms)	Count	Avg Self (us)	P95 Self (ms)	Max Self (ms)	% Work	Task Name	Top Caller
3,651.3	20,433.0	82,740	44.1	0.1	12.8	21.0%	module	module (60%)
2,652.7	2,790.7	18	147371.7	323.1	580.7	15.2%	save snapshot	persist (6%)
1,626.3	2,444.1	3,835	424.1	1.5	176.7	9.3%	analyze ecmascript module	analyze ecmascript module (44%)
1,132.9	10,119.4	25,977	43.6	0.1	38.1	6.5%	process module	process module (78%)
984.0	2,651.0	22,685	43.4	0.1	6.1	5.7%	internal resolving	internal resolving (73%)
879.6	4,992.5	3,930	223.8	0.2	106.7	5.1%	chunking	chunking (49%)
855.3	1,163.3	5,964	143.4	0.5	66.9	4.9%	parse ecmascript	parse ecmascript (66%)
670.9	724.7	6,904	97.2	0.4	5.5	3.9%	precompute code generation	generate merged code (49%)
629.7	1,928.7	603	1044.3	2.4	265.0	3.6%	generate merged code	chunking (69%)
628.9	726.0	5,511	114.1	0.3	112.5	3.6%	compute async module info	compute async module info (49%)
487.6	2,841.5	18,598	26.2	0.0	3.5	2.8%	resolving	module (61%)
381.0	381.0	329	1158.0	0.9	229.9	2.2%	generate source map	code generation (83%)
353.4	712.2	130	2718.1	4.8	207.5	2.0%	emit code	generate merged code (42%)
282.2	455.5	1,074	262.7	0.1	167.7	1.6%	write all entrypoints to disk	write all entrypoints to disk (12%)
162.3	735.9	1,853	87.6	0.2	35.1	0.9%	code generation	code generation (84%)
140.1	281.6	1,335	105.0	0.1	26.0	0.8%	compute async chunks	compute async chunks (47%)
139.1	139.1	9	15460.3	53.1	54.6	0.8%	blocking	save snapshot (67%)
68.7	74.5	467	147.1	0.0	25.4	0.4%	compute binding usage info	compute binding usage info (45%)
56.7	56.7	2,163	26.2	0.0	3.1	0.3%	read file	parse ecmascript (91%)
32.6	59.4	1,765	18.5	0.0	11.7	0.2%	collect mergeable modules	collect mergeable modules (100%)

---

Critical Path Analysis

The longest sequential dependency chains that determine wall-clock time.
Focus on reducing the depth of these chains to improve parallelism.

Rank	Self-Time (ms)	Depth	Path
1	635.3	3	persist → save snapshot → blocking
2	475.8	6	chunking → generate merged code → emit code → emit code → emit code → read file
3	271.5	4	chunking → generate merged code → emit code → generate source map
4	176.8	2	process module → analyze ecmascript module
5	150.6	2	save snapshot → blocking

---

Batching Candidates

High-volume tasks dominated by a single parent. If the parent can batch them,
it drastically reduces scheduler overhead.

Task Name	Count	Top Caller (Attribution)	Avg Self	P95 Self	Total Self
process module	25,977	process module (78%)	43.6 us	0.06 ms	1,132.9 ms
internal resolving	22,685	internal resolving (73%)	43.4 us	0.07 ms	984.0 ms

---

Duration Distribution

Range	Count	Percentage
<10us	613,809	76.6%
10us-100us	120,595	15.1%
100us-1ms	57,346	7.2%
1ms-10ms	8,917	1.1%
10ms-100ms	177	0.0%
>100ms	30	0.0%

---

Action Items

1. [P0] Focus on tasks with the highest Self-Time — these are where CPU cycles are actually spent.
2. [P0] Use Batching Candidates to identify callers that should use try_join or reduce #[turbo_tasks::function] granularity.
3. [P1] Check Build Phase Timeline for phases with disproportionate wall range vs. self-time (= serialization).
4. [P1] Inspect P95 Self (ms) for heavy monolith tasks. Focus on long-tail outliers, not averages.
5. [P1] Review Critical Paths — reducing the longest chain depth directly improves wall-clock time.
6. [P2] If Thread Utilization < 60%, investigate scheduling gaps (lock contention or deep dependency chains).

Report generated by Utoopack Performance Analysis Agent