path: root/sys-apps/gawk/files/gawk-5.4.1-update-minrx.patch

https://github.com/mikehaertel/minrx/pull/68
https://cgit.git.savannah.gnu.org/cgit/gawk.git/commit/?h=gawk-5.4-stable&id=1a61f93cd203dca8b0bc94f04a3eff80effae152

From 1a61f93cd203dca8b0bc94f04a3eff80effae152 Mon Sep 17 00:00:00 2001
From: "Arnold D. Robbins" <arnold@skeeve.com>
Date: Sat, 8 Aug 2026 22:24:22 +0300
Subject: Update minrx.c from upstream: memory management improvements.

---
 support/minrx.c   | 201 ++++++++++++++++++++++++++++++++++++++++++------------
 4 files changed, 174 insertions(+), 42 deletions(-)

diff --git a/support/minrx.c b/support/minrx.c
index 1e03fb99..5e268213 100644
--- a/support/minrx.c
+++ b/support/minrx.c
@@ -185,15 +185,6 @@ cowvec_allocator_construct(COWVec_Allocator *a, size_t length)
 	a->length = length;
 }
 
-static void
-cowvec_allocator_destruct(COWVec_Allocator *a)
-{
-	for (COWVec_Storage *s = a->freelist, *sfreelink = (COWVec_Storage *) NULL; s != (COWVec_Storage *) NULL; s = sfreelink) {
-		sfreelink = s->u.freelink;
-		free(s);
-	}
-}
-
 static size_t
 cowvec_storage_get(const COWVec_Storage *cvs, size_t i)
 {
@@ -1272,6 +1263,28 @@ nodelist_empty(void)
 	return r;
 }
 
+// Scratch space for an execution, kept with the compiled Regexp so that
+// the next execution can have it again.  Everything in it is sized from
+// the Regexp alone, so it need only be built once; building it afresh
+// for every execution otherwise dominates the cost of matching a short
+// subject.
+//
+// Only one execution can have it at a time, so an execution that finds
+// it taken -- a concurrent or a nested one -- falls back to its own.
+typedef struct Scratch Scratch;
+struct Scratch {
+	bool ready;
+#ifdef HAVE_PTHREADS
+	pthread_mutex_t inuse;
+#else
+	bool inuse;
+#endif
+	COWVec_Storage *freelist;
+	QSet epsq;
+	QVec epsv;
+	QVec mcsv[2];
+};
+
 typedef struct Regexp Regexp;
 struct Regexp {
 	WConv_Encoding enc;
@@ -1287,8 +1300,89 @@ struct Regexp {
 	FirstBytes firstbytes;
 	int32_t firstunique;
 	bool anchored;
+	Scratch scratch;
 };
 
+static bool
+scratch_construct(Scratch *sc, size_t nnode)
+{
+	if (sc->ready)
+		return true;
+	if (!qset_construct(&sc->epsq, nnode)
+	    || !qvec_construct(&sc->epsv, nnode)
+	    || !qvec_construct(&sc->mcsv[0], nnode)
+	    || !qvec_construct(&sc->mcsv[1], nnode)) {
+		qvec_destruct(&sc->mcsv[1]);
+		qvec_destruct(&sc->mcsv[0]);
+		qvec_destruct(&sc->epsv);
+		qset_destruct(&sc->epsq);
+		return false;
+	}
+	sc->ready = true;
+	return true;
+}
+
+static void
+scratch_destruct(Scratch *sc)
+{
+	if (sc->ready) {
+		qvec_destruct(&sc->mcsv[1]);
+		qvec_destruct(&sc->mcsv[0]);
+		qvec_destruct(&sc->epsv);
+		qset_destruct(&sc->epsq);
+		sc->ready = false;
+	}
+	for (COWVec_Storage *s = sc->freelist, *next = (COWVec_Storage *) NULL; s != (COWVec_Storage *) NULL; s = next) {
+		next = s->u.freelink;
+		free(s);
+	}
+	sc->freelist = (COWVec_Storage *) NULL;
+}
+
+// Claim the scratch space for an execution, or report it already taken.
+static bool
+scratch_acquire(Scratch *sc)
+{
+#ifdef HAVE_PTHREADS
+	return pthread_mutex_trylock(&sc->inuse) == 0;
+#else
+	if (sc->inuse)
+		return false;
+	sc->inuse = true;
+	return true;
+#endif
+}
+
+static void
+scratch_release(Scratch *sc)
+{
+#ifdef HAVE_PTHREADS
+	pthread_mutex_unlock(&sc->inuse);
+#else
+	sc->inuse = false;
+#endif
+}
+
+// For the Regexp's own scratch space, as opposed to the one an execution
+// falls back to, which is private to it and needs neither of these.
+static void
+scratch_init(Scratch *sc)
+{
+	memset(sc, 0, sizeof *sc);
+#ifdef HAVE_PTHREADS
+	pthread_mutex_init(&sc->inuse, (const pthread_mutexattr_t *) NULL);
+#endif
+}
+
+static void
+scratch_fini(Scratch *sc)
+{
+	scratch_destruct(sc);
+#ifdef HAVE_PTHREADS
+	pthread_mutex_destroy(&sc->inuse);
+#endif
+}
+
 static NInt
 satmul(NInt x, NInt y)
 {
@@ -1879,6 +1973,7 @@ compile(Compile *c)
 	memset(&r->firstbytes, 0, sizeof r->firstbytes);
 	r->firstunique = -1;
 	r->anchored = false;
+	scratch_init(&r->scratch);
 	int err;
 	if ((err = setjmp(c->errjmp)) != 0) {
 		c = vc, r = vr;
@@ -1944,13 +2039,16 @@ struct Execute {
 	COWVec_Allocator allocator;
 	COWVec best;
 	NInt bestmincount; // note mincounts are negated so this means +infinity
-	QSet epsq;
-	QVec epsv;
+	Scratch *scratch;	// the scratch space this execution is using
+	QSet *epsq;		// aliases of the members of *scratch, cached
+	QVec *epsv;		// here to save an indirection on every use
+	QVec *mcsv;
+	Scratch own;		// used where the Regexp's is already taken
 	bool exiting;
 };
 
 static bool
-execute_construct(Execute *e, const Regexp *r, minrx_regexec_flags_t flags, const char *bp, const char *ep)
+execute_construct(Execute *e, Regexp *r, minrx_regexec_flags_t flags, const char *bp, const char *ep)
 {
 	e->r = r;
 	e->flags = flags;
@@ -1965,21 +2063,47 @@ execute_construct(Execute *e, const Regexp *r, minrx_regexec_flags_t flags, cons
 	cowvec_allocator_construct(&e->allocator, e->nestoff + r->nmin);
 	cowvec_construct(&e->best, (COWVec_Allocator *) NULL);
 	e->bestmincount = 0;
-	if (!qset_construct(&e->epsq, r->nnode) || !qvec_construct(&e->epsv, r->nnode)) {
-		qset_destruct(&e->epsq);
+	if (scratch_acquire(&r->scratch)) {
+		e->scratch = &r->scratch;
+	} else {
+		// Private to this execution, so only what scratch_construct()
+		// reads needs setting here; it fills in the rest.
+		e->own.ready = false;
+		e->own.freelist = (COWVec_Storage *) NULL;
+		e->scratch = &e->own;
+	}
+	if (!scratch_construct(e->scratch, r->nnode)) {
+		if (e->scratch != &e->own)
+			scratch_release(e->scratch);
 		return false;
 	}
+	e->epsq = &e->scratch->epsq;
+	e->epsv = &e->scratch->epsv;
+	e->mcsv = e->scratch->mcsv;
+	e->allocator.freelist = e->scratch->freelist;
 	e->exiting = r->anchored;
 	return true;
 }
 
+// 'reusable' is false where the execution has failed part way through and
+// may have left the scratch space in a state the next one, which would
+// otherwise inherit it, should not build upon.
 static void
-execute_destruct(Execute *e)
+execute_destruct(Execute *e, bool reusable)
 {
-	qvec_destruct(&e->epsv);
-	qset_destruct(&e->epsq);
+	Scratch *sc = e->scratch;
 	cowvec_destruct(&e->best);
-	cowvec_allocator_destruct(&e->allocator);
+	// Drain, but keep the storage for the next execution.
+	qvec_clear(&e->mcsv[1]);
+	qvec_clear(&e->mcsv[0]);
+	qvec_clear(e->epsv);
+	while (!qset_empty(e->epsq))
+		(void) qset_remove(e->epsq);
+	sc->freelist = e->allocator.freelist;
+	if (!reusable || sc == &e->own)
+		scratch_destruct(sc);
+	if (sc != &e->own)
+		scratch_release(sc);
 }
 
 inline static void
@@ -2001,7 +2125,7 @@ execute_add(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WChar
 			qvi.newnsp->gen = e->gen;
 		}
 	} else {
-		QVecInsert qvi = qvec_insert(&e->epsv, k, nsp);
+		QVecInsert qvi = qvec_insert(e->epsv, k, nsp);
 		if (qvi.newly) {
 			nstate_construct_copy(qvi.newnsp, nsp);
 		} else {
@@ -2012,7 +2136,7 @@ execute_add(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WChar
 				return;
 		}
 		qvi.newnsp->gen = e->gen;
-		qset_insert(&e->epsq, k);
+		qset_insert(e->epsq, k);
 	}
 }
 
@@ -2036,7 +2160,7 @@ execute_add_1(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 			cowvec_put(&qvi.newnsp->substack, nstk - 1, arg1);
 		}
 	} else {
-		QVecInsert qvi = qvec_insert(&e->epsv, k, nsp);
+		QVecInsert qvi = qvec_insert(e->epsv, k, nsp);
 		if (qvi.newly) {
 			nstate_construct_copy(qvi.newnsp, nsp);
 		} else {
@@ -2048,7 +2172,7 @@ execute_add_1(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 		}
 		qvi.newnsp->gen = e->gen;
 		cowvec_put(&qvi.newnsp->substack, nstk - 1, arg1);
-		qset_insert(&e->epsq, k);
+		qset_insert(e->epsq, k);
 	}
 }
 
@@ -2073,7 +2197,7 @@ execute_add_2(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 			cowvec_put(&qvi.newnsp->substack, nstk - 1, arg2);
 		}
 	} else {
-		QVecInsert qvi = qvec_insert(&e->epsv, k, nsp);
+		QVecInsert qvi = qvec_insert(e->epsv, k, nsp);
 		if (qvi.newly) {
 			nstate_construct_copy(qvi.newnsp, nsp);
 		} else {
@@ -2086,7 +2210,7 @@ execute_add_2(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 		qvi.newnsp->gen = e->gen;
 		cowvec_put(&qvi.newnsp->substack, nstk - 2, arg1);
 		cowvec_put(&qvi.newnsp->substack, nstk - 1, arg2);
-		qset_insert(&e->epsq, k);
+		qset_insert(e->epsq, k);
 	}
 }
 
@@ -2112,7 +2236,7 @@ execute_add_3(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 			cowvec_put(&qvi.newnsp->substack, nstk - 1, arg3);
 		}
 	} else {
-		QVecInsert qvi = qvec_insert(&e->epsv, k, nsp);
+		QVecInsert qvi = qvec_insert(e->epsv, k, nsp);
 		if (qvi.newly) {
 			nstate_construct_copy(qvi.newnsp, nsp);
 		} else {
@@ -2126,7 +2250,7 @@ execute_add_3(Execute *e, QVec *ncsv, NInt k, NInt nstk, const NState *nsp, WCha
 		cowvec_put(&qvi.newnsp->substack, nstk - 3, arg1);
 		cowvec_put(&qvi.newnsp->substack, nstk - 2, arg2);
 		cowvec_put(&qvi.newnsp->substack, nstk - 1, arg3);
-		qset_insert(&e->epsq, k);
+		qset_insert(e->epsq, k);
 	}
 }
 
@@ -2148,8 +2272,8 @@ execute_epsclosure(Execute *e, QVec *ncsv, WChar wcnext)
 	const Node *nodes = e->nodes;
 	bool (*is_word)(WChar) = e->r->enc == Byte ? is_word_byte : is_word_wide;
 	do {
-		NInt k = qset_remove(&e->epsq);
-		NState *nsp = qvec_lookup(&e->epsv, k);
+		NInt k = qset_remove(e->epsq);
+		NState *nsp = qvec_lookup(e->epsv, k);
 		if (cowvec_valid(&e->best) && nsp->boff > cowvec_get(&e->best, e->suboff + 0))
 			continue;
 		const Node *np = &e->nodes[k];
@@ -2305,7 +2429,7 @@ execute_epsclosure(Execute *e, QVec *ncsv, WChar wcnext)
 			abort();
 			break;
 		}
-	} while (!qset_empty(&e->epsq));
+	} while (!qset_empty(e->epsq));
 }
 
 #define WCNEXT(E, WCN) ((E)->wcprev = (WCN), (E)->off = wconv_off(&(E)->wconv), (WCN) = wconv_nextchr(&(E)->wconv))
@@ -2313,18 +2437,12 @@ execute_epsclosure(Execute *e, QVec *ncsv, WChar wcnext)
 static int
 execute(Execute *e, size_t nm, minrx_regmatch_t *rm)
 {
-	QVec mcsvs[2];
-	if (!qvec_construct(&mcsvs[0], e->r->nnode) || !qvec_construct(&mcsvs[1], e->r->nnode)) {
-		qvec_destruct(&mcsvs[0]);
-		return MINRX_REG_ESPACE;
-	}
+	QVec *mcsvs = e->mcsv;
 	NState nsinit;
 	nstate_construct(&nsinit, (COWVec_Allocator *) NULL);	// fake construction so "exception handler" will be safe
 	int err;
 	if ((err = setjmp(e->allocator.errjmp)) != 0) {
 		nstate_destruct(&nsinit);
-		qvec_destruct(&mcsvs[1]);
-		qvec_destruct(&mcsvs[0]);
 		return err;
 	}
 	nstate_construct(&nsinit, &e->allocator);		// real construction
@@ -2381,7 +2499,7 @@ execute(Execute *e, size_t nm, minrx_regmatch_t *rm)
 	for (size_t i = 0; i < e->r->nmin; ++i)
 		cowvec_put(&nsinit.substack, e->nestoff + i, 0);
 	execute_add(e, &mcsvs[0], 0, 0, &nsinit, wcnext);
-	if (!qset_empty(&e->epsq))
+	if (!qset_empty(e->epsq))
 		execute_epsclosure(e, &mcsvs[0], wcnext);
 	for (;;) { // unrolled to ping-pong roles of mcsvs[0]/[1]
 		if (wcnext == End)
@@ -2396,7 +2514,7 @@ execute(Execute *e, size_t nm, minrx_regmatch_t *rm)
 			nsinit.boff = e->off;
 			execute_add(e, &mcsvs[1], 0, 0, &nsinit, wcnext);
 		}
-		if (!qset_empty(&e->epsq))
+		if (!qset_empty(e->epsq))
 			execute_epsclosure(e, &mcsvs[1], wcnext);
 		if (qvec_empty(&mcsvs[1])) {
 			if (e->exiting)
@@ -2416,7 +2534,7 @@ execute(Execute *e, size_t nm, minrx_regmatch_t *rm)
 			nsinit.boff = e->off;
 			execute_add(e, &mcsvs[0], 0, 0, &nsinit, wcnext);
 		}
-		if (!qset_empty(&e->epsq))
+		if (!qset_empty(e->epsq))
 			execute_epsclosure(e, &mcsvs[0], wcnext);
 		if (qvec_empty(&mcsvs[0])) {
 			if (e->exiting)
@@ -2427,8 +2545,6 @@ execute(Execute *e, size_t nm, minrx_regmatch_t *rm)
 	}
 exit:
 	nstate_destruct(&nsinit);
-	qvec_destruct(&mcsvs[1]);
-	qvec_destruct(&mcsvs[0]);
 	if (cowvec_valid(&e->best)) {
 		if (rm) {
 			size_t nsub = MIN(nm, e->r->nsub);
@@ -2504,7 +2620,7 @@ minrx_regnexec(minrx_regex_t *rx, size_t ns, const char *s, size_t nm, minrx_reg
 	if (!execute_construct(&e, r, (minrx_regexec_flags_t) flags, s, s + ns))
 		return MINRX_REG_ESPACE;
 	int ret = execute(&e, nm, rm);
-	execute_destruct(&e);
+	execute_destruct(&e, ret == MINRX_REG_SUCCESS || ret == MINRX_REG_NOMATCH);
 	return ret;
 }
 
@@ -2513,6 +2629,7 @@ minrx_regfree(minrx_regex_t *rx)
 {
 	Regexp *r = (Regexp *) rx->re_regexp;
 	if (r) {
+		scratch_fini(&r->scratch);
 		if (r->firstcset) {
 			cset_destruct(r->firstcset);
 			free((void *) r->firstcset);
-- 
cgit v1.3