The self-contained C path (ekp-c-break-lines, ekp-c-load-hyphenator, ekp-c-hyphenate, ekp-c-set-spacing, ekp_paragraph.c, ekp_hyphen.c, ~1500 lines) was never used by ekp.el, diverged semantically from the real pipeline (no kinsoku, no protrusion, no two-pass emergency), and contained an exploitable heap overflow reachable from Lisp: ekp_para_create sized its box array as box_count * 2, but a long word hyphenates into arbitrarily many syllable boxes, overflowing the calloc'd buffer. Deleting the path deletes the bug class. Hardening of the live path: - thread pool: sized from the machine's core count instead of a hardcoded 8; created lazily on the first multi-paragraph batch (single-paragraph users never start worker threads); a full queue now blocks the submitter until a worker makes room — tasks were silently dropped before, degrading the batch to the Elisp fallback exactly when parallelism mattered most. - unified failure gate: a partial allocation used to silently drop kinsoku, hyphenation or protrusion data and continue with a subtly different layout; any allocation failure or pending Lisp signal (non-local exit from a bad element type) now fails the whole call, and ekp.el falls back to the Elisp engine. The Elisp bridge wraps both C entry points in condition-case, and a whole-batch nil no longer crashes the per-paragraph loop. - integer safety: every extracted pixel value is clamped to int32 instead of silently wrapping. - EKP_INFINITY (the unreachable-state sentinel) is now a real infinity: extremely degenerate paragraphs could legitimately accumulate demerits past the old 1e10 constant, making C consider reachable states dead and diverge from the Elisp engine. BREAKING: the four experimental module functions are gone; rebuild with make -C ekp_c clean all (version gate unchanged at 1.5). 92 ERT green; fuzz 300/300 byte-identical across engines. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
188 lines
5.4 KiB
C
188 lines
5.4 KiB
C
/*
|
|
* ekp_thread_pool.c - Work-stealing thread pool
|
|
*
|
|
* Copyright (C) 2024-2026 Kinney Zhang
|
|
* SPDX-License-Identifier: GPL-3.0-or-later
|
|
*
|
|
* This file is part of emacs-kp, which is free software: you can
|
|
* redistribute it and/or modify it under the terms of the GNU General
|
|
* Public License as published by the Free Software Foundation, either
|
|
* version 3 of the License, or (at your option) any later version.
|
|
* It is distributed WITHOUT ANY WARRANTY; see the GNU General Public
|
|
* License (COPYING) for details.
|
|
*
|
|
* Simple but effective: fixed thread count, lock-free queue would be
|
|
* overkill for our batch sizes. Keep it simple, stupid.
|
|
*/
|
|
|
|
#include "ekp_module.h"
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#if !defined(_WIN32)
|
|
#include <unistd.h>
|
|
#endif
|
|
|
|
#define QUEUE_CAPACITY 1024
|
|
|
|
static void *worker_thread(void *arg)
|
|
{
|
|
ekp_thread_pool_t *pool = (ekp_thread_pool_t *)arg;
|
|
|
|
while (1) {
|
|
pthread_mutex_lock(&pool->queue_lock);
|
|
|
|
/* Wait for work */
|
|
while (pool->queue_head == pool->queue_tail && !pool->shutdown) {
|
|
pthread_cond_wait(&pool->queue_cond, &pool->queue_lock);
|
|
}
|
|
|
|
if (pool->shutdown && pool->queue_head == pool->queue_tail) {
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
break;
|
|
}
|
|
|
|
/* Dequeue work */
|
|
void (*func)(void *) = pool->queue[pool->queue_head].func;
|
|
void *work_arg = pool->queue[pool->queue_head].arg;
|
|
bool was_full =
|
|
((pool->queue_tail + 1) % pool->queue_size) == pool->queue_head;
|
|
pool->queue_head = (pool->queue_head + 1) % pool->queue_size;
|
|
pool->active_count++;
|
|
if (was_full)
|
|
pthread_cond_broadcast(&pool->done_cond);
|
|
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
|
|
/* Execute */
|
|
if (func)
|
|
func(work_arg);
|
|
|
|
/* Mark done */
|
|
pthread_mutex_lock(&pool->queue_lock);
|
|
pool->active_count--;
|
|
if (pool->active_count == 0 && pool->queue_head == pool->queue_tail) {
|
|
pthread_cond_signal(&pool->done_cond);
|
|
}
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
}
|
|
|
|
return NULL;
|
|
}
|
|
|
|
size_t ekp_pool_default_threads(void)
|
|
{
|
|
long n = 0;
|
|
#if defined(_SC_NPROCESSORS_ONLN)
|
|
n = sysconf(_SC_NPROCESSORS_ONLN);
|
|
#endif
|
|
if (n <= 0)
|
|
n = 4;
|
|
if (n > EKP_THREAD_POOL_MAX)
|
|
n = EKP_THREAD_POOL_MAX;
|
|
return (size_t)n;
|
|
}
|
|
|
|
ekp_thread_pool_t *ekp_pool_create(size_t num_threads)
|
|
{
|
|
if (num_threads == 0)
|
|
num_threads = ekp_pool_default_threads();
|
|
if (num_threads > EKP_THREAD_POOL_MAX)
|
|
num_threads = EKP_THREAD_POOL_MAX;
|
|
|
|
ekp_thread_pool_t *pool = calloc(1, sizeof(*pool));
|
|
if (!pool)
|
|
return NULL;
|
|
|
|
pool->queue_size = QUEUE_CAPACITY;
|
|
pool->queue = calloc(pool->queue_size, sizeof(pool->queue[0]));
|
|
if (!pool->queue) {
|
|
free(pool);
|
|
return NULL;
|
|
}
|
|
|
|
pthread_mutex_init(&pool->queue_lock, NULL);
|
|
pthread_cond_init(&pool->queue_cond, NULL);
|
|
pthread_cond_init(&pool->done_cond, NULL);
|
|
|
|
/* Start worker threads */
|
|
for (size_t i = 0; i < num_threads; i++) {
|
|
if (pthread_create(&pool->threads[i], NULL, worker_thread, pool) != 0) {
|
|
/* Cleanup on failure */
|
|
pool->shutdown = true;
|
|
pthread_cond_broadcast(&pool->queue_cond);
|
|
for (size_t j = 0; j < i; j++) {
|
|
pthread_join(pool->threads[j], NULL);
|
|
}
|
|
pthread_mutex_destroy(&pool->queue_lock);
|
|
pthread_cond_destroy(&pool->queue_cond);
|
|
pthread_cond_destroy(&pool->done_cond);
|
|
free(pool->queue);
|
|
free(pool);
|
|
return NULL;
|
|
}
|
|
}
|
|
|
|
pool->thread_count = num_threads;
|
|
return pool;
|
|
}
|
|
|
|
void ekp_pool_destroy(ekp_thread_pool_t *pool)
|
|
{
|
|
if (!pool)
|
|
return;
|
|
|
|
pthread_mutex_lock(&pool->queue_lock);
|
|
pool->shutdown = true;
|
|
pthread_cond_broadcast(&pool->queue_cond);
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
|
|
for (size_t i = 0; i < pool->thread_count; i++) {
|
|
pthread_join(pool->threads[i], NULL);
|
|
}
|
|
|
|
pthread_mutex_destroy(&pool->queue_lock);
|
|
pthread_cond_destroy(&pool->queue_cond);
|
|
pthread_cond_destroy(&pool->done_cond);
|
|
free(pool->queue);
|
|
free(pool);
|
|
}
|
|
|
|
void ekp_pool_submit(ekp_thread_pool_t *pool, void (*func)(void *), void *arg)
|
|
{
|
|
if (!pool || !func)
|
|
return;
|
|
|
|
pthread_mutex_lock(&pool->queue_lock);
|
|
|
|
/* Queue full: wait for a worker to make room. Dropping the task
|
|
* here used to silently degrade the batch to the Elisp fallback
|
|
* exactly when parallelism mattered most. */
|
|
while (((pool->queue_tail + 1) % pool->queue_size) == pool->queue_head
|
|
&& !pool->shutdown) {
|
|
pthread_cond_wait(&pool->done_cond, &pool->queue_lock);
|
|
}
|
|
if (pool->shutdown) {
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
return;
|
|
}
|
|
|
|
pool->queue[pool->queue_tail].func = func;
|
|
pool->queue[pool->queue_tail].arg = arg;
|
|
pool->queue_tail = (pool->queue_tail + 1) % pool->queue_size;
|
|
|
|
pthread_cond_signal(&pool->queue_cond);
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
}
|
|
|
|
void ekp_pool_wait(ekp_thread_pool_t *pool)
|
|
{
|
|
if (!pool)
|
|
return;
|
|
|
|
pthread_mutex_lock(&pool->queue_lock);
|
|
while (pool->active_count > 0 || pool->queue_head != pool->queue_tail) {
|
|
pthread_cond_wait(&pool->done_cond, &pool->queue_lock);
|
|
}
|
|
pthread_mutex_unlock(&pool->queue_lock);
|
|
}
|