fix: harden small core containers

dict.c

@@ -64,13 +64,15 @@ dictopen(char *path)
 	Biobuf *b;
 	Str key;
 	char *line, *tab;
-	int len;
+	int len, lineno;
 
 	b = Bopen(path, OREAD);
 	if(b == nil)
 		die("can't open: %s", path);
 	h = hmapalloc(4096);
+	lineno = 0;
 	while((line = Brdstr(b, '\n', 1)) != nil){
+		lineno++;
 		len = strlen(line);
 		if(len > 0 && line[len-1] == '\r')
 			line[--len] = '\0';
@@ -84,6 +86,8 @@ dictopen(char *path)
 			continue;
 		}
 		*tab = '\0';
+		if(utflen(line) > Maxrunes)
+			die("dictionary key too long: %s:%d", path, lineno);
 		sinit(&key, line, tab - line);
 		hmapset(&h, &key, tab+1, len - (tab - line) - 1);
 		free(line);
@@ -101,7 +105,10 @@ dictinit(char *dir)
 	for(i = 0; i < nlang; i++){
 		if(langs[i].dictname == nil)
 			continue;
-		snprint(path, sizeof(path), "%s/%s.dict", dir, langs[i].dictname);
+		if(snprint(path, sizeof(path), "%s/%s.dict", dir,
+		   langs[i].dictname) >= (int)sizeof path)
+			die("dictionary path too long: %s/%s.dict",
+				dir, langs[i].dictname);
 		langs[i].dict = dictopen(path);
 	}
 }

hash.c

@@ -25,7 +25,7 @@ hmapalloc(int nbuckets)
 	int nsz;
 
 	if(nbuckets < 1)
-		die("hmapalloc: no buckets");
+		return nil;
 	nsz = Tagsize;
 	store = emalloc(sizeof(*h) + nbuckets * nsz);
 	h = store;
@@ -54,6 +54,8 @@ hmapget(Hmap *h, Str *key)
 	Hnode *n;
 	uchar *v;
 
+	if(h == nil || key == nil || key->n < 0 || key->n > Maxrunes)
+		return nil;
 	v = h->nodes + (hash(key) % h->nbs) * h->nsz;
 	for(;;){
 		n = (Hnode*)v;
@@ -67,7 +69,7 @@ hmapget(Hmap *h, Str *key)
 }
 
 static char*
-sdup(Str *s)
+sdup(Str *s, int *len)
 {
 	char buf[Maxutf];
 	char *p;
@@ -77,6 +79,7 @@ sdup(Str *s)
 	p = emalloc(n + 1);
 	memmove(p, buf, n);
 	p[n] = '\0';
+	*len = n;
 	return p;
 }
 
@@ -121,6 +124,10 @@ hmapset(Hmap **store, Str *key, const char *val, int vlen)
 	int next;
 	vlong diff;
 
+	if(store == nil || *store == nil || key == nil ||
+	   key->n < 0 || key->n > Maxrunes || vlen < 0 ||
+	   (vlen > 0 && val == nil))
+		return;
 	newval = memdup(val, vlen);
 	h = *store;
 	v = h->nodes + (hash(key) % h->nbs) * h->nsz;
@@ -151,8 +158,7 @@ hmapset(Hmap **store, Str *key, const char *val, int vlen)
 	n = (Hnode*)v;
 replace:
 	if(n->filled == 0){
-		n->key = sdup(key);
-		n->klen = strlen(n->key);
+		n->key = sdup(key, &n->klen);
 		n->filled = 1;
 	}
 	n->next = next;

main.c

@@ -67,12 +67,18 @@ threadmain(int argc, char **argv)
 	dictresc = chancreate(sizeof(Dictres), 0);
 	mapinit(argv[1]);
 	dictinit(argv[1]);
-	proccreate(drawthread, nil, 16384);
-	proccreate(srvthread, nil, 16384);
-	proccreate(ibusthread, nil, 32768);
-	proccreate(waylandthread, nil, 32768);
-	threadcreate(dictthread, nil, 16384);
-	threadcreate(imthread, nil, 16384);
+	if(proccreate(drawthread, nil, 16384) < 0)
+		die("can't create draw worker");
+	if(proccreate(srvthread, nil, 16384) < 0)
+		die("can't create server worker");
+	if(proccreate(ibusthread, nil, 32768) < 0)
+		die("can't create IBus worker");
+	if(proccreate(waylandthread, nil, 32768) < 0)
+		die("can't create Wayland worker");
+	if(threadcreate(dictthread, nil, 16384) < 0)
+		die("can't create dictionary worker");
+	if(threadcreate(imthread, nil, 16384) < 0)
+		die("can't create input worker");
 
 	threadexits(nil);
 }

str.c

@@ -7,6 +7,8 @@ sinit(Str *s, char *src, int n)
 	int len;
 
 	s->n = 0;
+	if(n < 0 || (n > 0 && src == nil))
+		return;
 	while(n > 0 && s->n < Maxrunes){
 		if(!fullrune(src, n))
 			break;
@@ -28,8 +30,9 @@ sclear(Str *s)
 void
 sputr(Str *s, Rune r)
 {
+	/* Str is a capped value; appends at capacity leave it unchanged. */
 	if(s->n >= Maxrunes)
-		die("sputr overflow");
+		return;
 	s->r[s->n++] = r;
 }
 

strans.c

@@ -724,6 +724,12 @@ keystroke(u32int ks, u32int mod, Str *com)
 	}
 	if(im.l->trans == nil)
 		return 0;
+	if(im.pre.n >= Maxrunes ||
+	   (isjp(&im) && im.pre.n + im.raw.n >= Maxrunes)){
+		commit(com);
+		reset();
+		return 0;
+	}
 	if(im.l->lang == LangKO)
 		ks = kokey(ks, mod);
 	n = dotrans(ks, com);
@@ -858,7 +864,9 @@ mapinit(char *dir)
 	for(i = 0; i < nelem(langs); i++){
 		if(langs[i].mapname == nil)
 			continue;
-		snprint(path, sizeof(path), "%s/%s.map", dir, langs[i].mapname);
+		if(snprint(path, sizeof(path), "%s/%s.map", dir,
+		   langs[i].mapname) >= (int)sizeof path)
+			die("map path too long: %s/%s.map", dir, langs[i].mapname);
 		langs[i].map = trieopen(path);
 	}
 }

tests/hash_test.c

@@ -76,3 +76,37 @@ hmap_long_utf8_keys(struct ct *t)
 cleanup:
 	hmapfree(h);
 }
+
+void
+hmap_binary_keys_and_invalid_lengths(struct ct *t)
+{
+	Hmap *h;
+	Hnode *n;
+	Str key, other;
+
+	h = hmapalloc(1);
+	CT_CHECK(t, h != nil);
+	key.n = 3;
+	key.r[0] = 'a';
+	key.r[1] = 0;
+	key.r[2] = 'b';
+	other = key;
+	other.r[2] = 'c';
+	hmapset(&h, &key, "one", 3);
+	hmapset(&h, &other, "two", 3);
+	n = hmapget(h, &key);
+	if(CT_CHECK(t, n != nil)){
+		CT_EQ_INT(t, 3, n->klen);
+		CT_EQ_MEM(t, "one", n->val, n->vlen);
+	}
+	n = hmapget(h, &other);
+	if(CT_CHECK(t, n != nil))
+		CT_EQ_MEM(t, "two", n->val, n->vlen);
+	hmapset(&h, &key, "bad", -1);
+	hmapset(&h, &key, nil, 1);
+	n = hmapget(h, &key);
+	if(CT_CHECK(t, n != nil))
+		CT_EQ_MEM(t, "one", n->val, n->vlen);
+	CT_EQ_PTR(t, nil, hmapalloc(0));
+	hmapfree(h);
+}

tests/str_test.c

@@ -82,3 +82,21 @@ str_utf8_capacity(struct ct *t)
 	CT_EQ_STR(t, "😀", buf);
 	CT_EQ_INT(t, 'Z', buf[5]);
 }
+
+void
+str_invalid_and_full_appends(struct ct *t)
+{
+	Str s;
+	int i;
+
+	sinit(&s, nil, 1);
+	CT_EQ_INT(t, 0, s.n);
+	sinit(&s, "x", -1);
+	CT_EQ_INT(t, 0, s.n);
+	for(i = 0; i < Maxrunes; i++)
+		s.r[i] = 'a';
+	s.n = Maxrunes;
+	sputr(&s, 'z');
+	CT_EQ_INT(t, Maxrunes, s.n);
+	CT_EQ_INT(t, 'a', s.r[Maxrunes-1]);
+}

tests/test.h

@@ -17,9 +17,12 @@ Str	shownpre(Im*);
 void	str_init_utf8(struct ct*);
 void	str_edit_and_alias(struct ct*);
 void	str_utf8_capacity(struct ct*);
+void	str_invalid_and_full_appends(struct ct*);
 void	hmap_set_replace_and_grow(struct ct*);
 void	hmap_long_utf8_keys(struct ct*);
+void	hmap_binary_keys_and_invalid_lengths(struct ct*);
 void	trie_exact_prefix_and_duplicate(struct ct*);
+void	trie_optional_outputs_and_invalid_lengths(struct ct*);
 void	production_maps_load(struct ct*);
 void	transmap_states(struct ct*);
 void	korean_sequences(struct ct*);

tests/trie_test.c

@@ -34,6 +34,29 @@ cleanup:
 	trieclose(trie);
 }
 
+void
+trie_optional_outputs_and_invalid_lengths(struct ct *t)
+{
+	char *v;
+	Trie *trie;
+	int n;
+
+	trie = trieopen("data/trie.map");
+	CT_CHECK(t, trieget(trie, "a", 1, nil) != nil);
+	v = nil;
+	CT_CHECK(t, trielookup(trie, "a", 1, &v, nil));
+	CT_CHECK(t, v != nil);
+	n = -1;
+	CT_CHECK(t, trielookup(trie, "a", 1, nil, &n));
+	CT_EQ_INT(t, 5, n);
+	CT_CHECK(t, trielookup(trie, "dupli", 5, nil, nil));
+	CT_EQ_PTR(t, nil, trieget(trie, nil, 1, &n));
+	CT_EQ_PTR(t, nil, trieget(trie, "a", -1, &n));
+	CT_CHECK(t, !trielookup(trie, nil, 1, &v, &n));
+	CT_CHECK(t, !trielookup(trie, "a", -1, &v, &n));
+	trieclose(trie);
+}
+
 void
 production_maps_load(struct ct *t)
 {

tests/unit_test.c

@@ -66,9 +66,12 @@ static const struct ct_test tests[] = {
 	{ "str/init-utf8", str_init_utf8 },
 	{ "str/edit-and-alias", str_edit_and_alias },
 	{ "str/utf8-capacity", str_utf8_capacity },
+	{ "str/invalid-full-appends", str_invalid_and_full_appends },
 	{ "hmap/set-replace-grow", hmap_set_replace_and_grow },
 	{ "hmap/long-utf8-keys", hmap_long_utf8_keys },
+	{ "hmap/binary-invalid-lengths", hmap_binary_keys_and_invalid_lengths },
 	{ "trie/exact-prefix-duplicate", trie_exact_prefix_and_duplicate },
+	{ "trie/optional-invalid-lengths", trie_optional_outputs_and_invalid_lengths },
 	{ "map/production-lifecycle", production_maps_load },
 	{ "transmap/states", transmap_states },
 	{ "hangul/sequences", korean_sequences },

trie.c

@@ -118,6 +118,8 @@ trieget(Trie *t, char *key, int klen, int *vlen)
 	int ni;
 	int i;
 
+	if(t == nil || klen < 0 || (klen > 0 && key == nil))
+		return nil;
 	ni = t->root;
 	for(i = 0; i < klen; i++){
 		ni = find(t, ni, key[i]);
@@ -126,7 +128,8 @@ trieget(Trie *t, char *key, int klen, int *vlen)
 	}
 	if(t->nodes[ni].val == nil)
 		return nil;
-	*vlen = t->nodes[ni].vlen;
+	if(vlen != nil)
+		*vlen = t->nodes[ni].vlen;
 	return t->nodes[ni].val;
 }
 
@@ -136,15 +139,19 @@ trielookup(Trie *t, char *key, int klen, char **val, int *vlen)
 	int ni;
 	int i;
 
+	if(t == nil || klen < 0 || (klen > 0 && key == nil))
+		return 0;
 	ni = t->root;
 	for(i = 0; i < klen; i++){
 		ni = find(t, ni, key[i]);
 		if(ni < 0)
 			return 0;
 	}
-	if(val != nil && t->nodes[ni].val != nil){
-		*val = t->nodes[ni].val;
-		*vlen = t->nodes[ni].vlen;
+	if(t->nodes[ni].val != nil){
+		if(val != nil)
+			*val = t->nodes[ni].val;
+		if(vlen != nil)
+			*vlen = t->nodes[ni].vlen;
 	}
 	return 1;
 }