fix: optimize read-inline-info

[not-matthias]

No description provided.

[codspeed-hq]

CodSpeed Performance Report

Merging #4 will improve performances by ×6.9

_{Comparing cod-1573-v431-seems-to-cause-hangs (77811f8) with add-benchmarks (f16dd47)}

Summary

⚡ 12 improvements
✅ 56 untouched
🆕 4 new

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	test_valgrind[valgrind-3.25.1, python3 testdata/test.py, full-no-inline]	5.8 s	5.2 s	+10.94%
⚡	test_valgrind[valgrind-3.25.1.codspeed, echo Hello, World!, full-with-inline]	852.6 ms	726.4 ms	+17.38%
⚡	test_valgrind[valgrind-3.25.1.codspeed, echo Hello, World!, inline]	786.8 ms	660.6 ms	+19.11%
⚡	test_valgrind[valgrind-3.25.1.codspeed, ls bench.py, full-with-inline]	931.9 ms	787.9 ms	+18.29%
⚡	test_valgrind[valgrind-3.25.1.codspeed, ls bench.py, inline]	849.7 ms	707.3 ms	+20.14%
⚡	test_valgrind[valgrind-3.25.1.codspeed, stress-ng --cpu 1 --timeout 1s, full-no-inline]	7.1 s	3.1 s	×2.3
⚡	test_valgrind[valgrind-3.25.1.codspeed, stress-ng --cpu 1 --timeout 1s, full-with-inline]	3.8 s	3.4 s	+10.15%
⚡	test_valgrind[valgrind-3.25.1.codspeed, stress-ng --cpu 1 --timeout 1s, inline]	2.7 s	2.2 s	+20.26%
⚡	test_valgrind[valgrind-3.25.1.codspeed, stress-ng --cpu 4 --timeout 1s, full-with-inline]	3.8 s	3.4 s	+14.15%
⚡	test_valgrind[valgrind-3.25.1.codspeed, stress-ng --cpu 4 --timeout 1s, inline]	2.7 s	2.3 s	+19.44%
⚡	test_valgrind[valgrind-3.25.1.codspeed, testdata/take_strings-aarch64 varbinview_non_null, full-with-inline]	49.5 s	7.7 s	×6.4
⚡	test_valgrind[valgrind-3.25.1.codspeed, testdata/take_strings-aarch64 varbinview_non_null, inline]	49.4 s	7.1 s	×6.9
🆕	test_valgrind[valgrind-3.26.0, python3 testdata/test.py, full-no-inline]	N/A	5.3 s	N/A
🆕	test_valgrind[valgrind-3.26.0, python3 testdata/test.py, full-with-inline]	N/A	5.5 s	N/A
🆕	test_valgrind[valgrind-3.26.0, python3 testdata/test.py, inline]	N/A	3.9 s	N/A
🆕	test_valgrind[valgrind-3.26.0, python3 testdata/test.py, no-inline]	N/A	3.6 s	N/A

[Copilot]

Pull Request Overview

This PR optimizes inline function lookups in Valgrind's debug info system by replacing O(n) linear search with a hash table cache backed by binary search fallback. The implementation uses the Verstable hash table library and adds benchmarking infrastructure using CodSpeed.

• Added hash table-based caching for inline function address lookups
• Integrated Verstable (v2.2.1) hash table library for efficient O(1) lookups
• Implemented hybrid lookup strategy: hash table cache with binary search fallback
• Added CodSpeed benchmarking workflow and test infrastructure

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
coregrind/m_debuginfo/verstable.h	Added Verstable v2.2.1 hash table library for efficient key-value storage
coregrind/m_debuginfo/storage.c	Build hash table during inltab canonicalization to cache addr_lo→index mappings
coregrind/m_debuginfo/priv_storage.h	Added inltab_lookup field to DebugInfo struct for hash table storage
coregrind/m_debuginfo/priv_inltab_lookup.h	Public API for inline table lookup with hash table and binary search
coregrind/m_debuginfo/inltab_lookup.c	Implementation of hybrid lookup using hash table cache and binary search fallback
coregrind/m_debuginfo/debuginfo.c	Updated VG_(get_inline_fnname) to use new hash table lookup API
bench/testdata/take_strings	Git LFS binary test data for benchmarks
bench/bench.py	Python benchmark script using pytest-codspeed
.github/workflows/codspeed.yml	CodSpeed benchmark CI workflow
.github/workflows/ci.yml	Updated CI to skip installing package docs
.gitattributes	Git LFS configuration for test data

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull Request Overview

Copilot reviewed 12 out of 12 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[GuillaumeLagrange]

olgtm, will approve once we have at lesat tested teh impact compared to baseline valgrind.

Comments are pretty minor

[art049]

Code changes lgtm! Got some feedback on the devX though

[art049]

We also need a few addtional binaries to bench IMO

[not-matthias]

We also need a few addtional binaries to bench IMO

I've added ls and echo to the test binaries, lmk if you want anyting else as well. I've also modified the bench script a bit to allow passing an arbitrary cmd which will be useful for cmds with args.

EDIT: Also added caching and disabled the take-strings benchmark for 3.25 since it takes a super long time.

[GuillaumeLagrange]

lgtm, very good work 🤗