Files

Moe Charm (CI) 1010a961fb Tiny: fix header/stride mismatch and harden refill paths

- Root cause: header-based class indexing (HEADER_CLASSIDX=1) wrote a 1-byte
  header during allocation, but linear carve/refill and initial slab capacity
  still used bare class block sizes. This mismatch could overrun slab usable
  space and corrupt freelists, causing reproducible SEGV at ~100k iters.

Changes
- Superslab: compute capacity with effective stride (block_size + header for
  classes 0..6; class7 remains headerless) in superslab_init_slab(). Add a
  debug-only bound check in superslab_alloc_from_slab() to fail fast if carve
  would exceed usable bytes.
- Refill (non-P0 and P0): use header-aware stride for all linear carving and
  TLS window bump operations. Ensure alignment/validation in tiny_refill_opt.h
  also uses stride, not raw class size.
- Drain: keep existing defense-in-depth for remote sentinel and sanitize nodes
  before splicing into freelist (already present).

Notes
- This unifies the memory layout across alloc/linear-carve/refill with a single
  stride definition and keeps class7 (1024B) headerless as designed.
- Debug builds add fail-fast checks; release builds remain lean.

Next
- Re-run Tiny benches (256/1024B) in debug to confirm stability, then in
  release. If any remaining crash persists, bisect with HAKMEM_TINY_P0_BATCH_REFILL=0
  to isolate P0 batch carve, and continue reducing branch-miss as planned.

2025-11-09 18:55:50 +09:00

2.5 KiB

Raw Blame History

HAKMEM Phase 7 + Pool TLS Phase 1.5b — Build & Run Cheatsheet

This document captures the stable build/run recipe used for recent benches.

One‑liner Build (recommended)

./build.sh <target>

# examples
./build.sh bench_mid_large_mt_hakmem
./build.sh bench_random_mixed_hakmem
./build.sh larson_hakmem

Enables at build time:

POOL_TLS_PHASE1=1 (Pool TLS Phase 1.5b)
HEADER_CLASSIDX=1 (Phase 7 header)
AGGRESSIVE_INLINE=1
PREWARM_TLS=1

Verify switches:

make print-flags

Optional safety/verbosity toggles:

HAKMEM_TINY_SAFE_FREE=1 — strict free validation (mincore on all frees). Slower but safest.
HAKMEM_DEBUG_VERBOSE=1 — enable verbose logs for Tiny header/free, etc.

Examples:

make clean && make HAKMEM_TINY_SAFE_FREE=1 POOL_TLS_PHASE1=1 HEADER_CLASSIDX=1 AGGRESSIVE_INLINE=1 PREWARM_TLS=1 bench_mid_large_mt_hakmem

Bench Recipes (used in reports)

Larson (Mixed)

./build.sh larson_hakmem
make larson_system
./larson_hakmem 2 8 128 1024 1 12345 1
./larson_hakmem 2 8 128 1024 1 12345 4
./larson_system 2 8 128 1024 1 12345 1
./larson_system 2 8 128 1024 1 12345 4

Pool TLS (8–52KB)

./build.sh bench_pool_tls_hakmem
make bench_pool_tls_system
./bench_pool_tls_hakmem 1 100000 256 42
./bench_pool_tls_hakmem 4 50000 256 42
./bench_pool_tls_system 1 100000 256 42
./bench_pool_tls_system 4 50000 256 42

Random Mixed (Tiny 128–1024B)

./build.sh bench_random_mixed_hakmem
make bench_random_mixed_system
for s in 128 256 512 1024; do \
  ./bench_random_mixed_hakmem 100000 $s 42; \
  ./bench_random_mixed_system 100000 $s 42; \
done

Mid‑Large MT (8–32KB)

./build.sh bench_mid_large_mt_hakmem
make bench_mid_large_mt_system
./bench_mid_large_mt_hakmem 1 100000 256 42
./bench_mid_large_mt_hakmem 4 50000 256 42
./bench_mid_large_mt_system 1 100000 256 42
./bench_mid_large_mt_system 4 50000 256 42

Mimalloc note (when comparing)

Direct‑link mimalloc benches require runtime path:

export LD_LIBRARY_PATH=$PWD/mimalloc-bench/extern/mi/out/release

Build hygiene

Always prefer ./build.sh over ad‑hoc make (prevents flag drift)
Check switches: make print-flags
Verify freshness: ./verify_build.sh <binary>
Arena (Pool TLS) ENV

You can tune the Pool TLS Arena growth via ENV vars:

# Initial chunk size in MB (default: 1)
export HAKMEM_POOL_TLS_ARENA_MB_INIT=2

# Maximum chunk size in MB (default: 8)
export HAKMEM_POOL_TLS_ARENA_MB_MAX=16

# Number of growth levels (default: 3 → 1→2→4→8MB)
export HAKMEM_POOL_TLS_ARENA_GROWTH_LEVELS=4

2.5 KiB Raw Blame History Unescape Escape