From 652e7ba973f2c1e7afd82994474345dd095a146e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Uwe=20Kr=C3=BCger?=
 <45282134+UweKrueger@users.noreply.github.com>
Date: Sat, 12 Jun 2021 10:27:08 +0200
Subject: [PATCH] gc: extend optimized mode to strings (#10431)

---
 vlib/builtin/array_notd_gcboehm_opt.v | 10 +++++
 vlib/builtin/builtin.c.v              | 59 ++++++++++++++++++++++++++-
 vlib/builtin/string.v                 | 40 +++++++++++-------
 3 files changed, 94 insertions(+), 15 deletions(-)
 create mode 100644 vlib/builtin/array_notd_gcboehm_opt.v

diff --git a/vlib/builtin/array_notd_gcboehm_opt.v b/vlib/builtin/array_notd_gcboehm_opt.v
new file mode 100644
index 0000000000..170316733a
--- /dev/null
+++ b/vlib/builtin/array_notd_gcboehm_opt.v
@@ -0,0 +1,10 @@
+// dummy placeholder for functions from `array_d_gcboehm_opt.v`
+// that might be needed for compile time
+// `$if gcboehm_opt ? { ... } $else { ... }`
+
+module builtin
+
+// this is needed in `string.v`
+fn __new_array_noscan(mylen int, cap int, elm_size int) array {
+	return array{}
+}
diff --git a/vlib/builtin/builtin.c.v b/vlib/builtin/builtin.c.v
index e73c45cb1d..29737d1671 100644
--- a/vlib/builtin/builtin.c.v
+++ b/vlib/builtin/builtin.c.v
@@ -256,6 +256,59 @@ pub fn malloc(n int) &byte {
 	return res
 }
 
+[unsafe]
+pub fn malloc_noscan(n int) &byte {
+	if n <= 0 {
+		panic('> V malloc(<=0)')
+	}
+	$if vplayground ? {
+		if n > 10000 {
+			panic('allocating more than 10 KB at once is not allowed in the V playground')
+		}
+		if total_m > 50 * 1024 * 1024 {
+			panic('allocating more than 50 MB is not allowed in the V playground')
+		}
+	}
+	$if trace_malloc ? {
+		total_m += n
+		C.fprintf(C.stderr, c'v_malloc %6d total %10d\n', n, total_m)
+		// print_backtrace()
+	}
+	mut res := &byte(0)
+	$if prealloc {
+		return unsafe { prealloc_malloc(n) }
+	} $else $if gcboehm ? {
+		$if gcboehm_opt ? {
+			unsafe {
+				res = C.GC_MALLOC_ATOMIC(n)
+			}
+		} $else {
+			unsafe {
+				res = C.GC_MALLOC(n)
+			}
+		}
+	} $else $if freestanding {
+		mut e := Errno{}
+		res, e = mm_alloc(u64(n))
+		if e != .enoerror {
+			eprint('malloc() failed: ')
+			eprintln(e.str())
+			panic('malloc() failed')
+		}
+	} $else {
+		res = unsafe { C.malloc(n) }
+	}
+	if res == 0 {
+		panic('malloc($n) failed')
+	}
+	$if debug_malloc ? {
+		// Fill in the memory with something != 0, so it is easier to spot
+		// when the calling code wrongly relies on it being zeroed.
+		unsafe { C.memset(res, 0x88, n) }
+	}
+	return res
+}
+
 // v_realloc resizes the memory block `b` with `n` bytes.
 // The `b byteptr` must be a pointer to an existing memory block
 // previously allocated with `malloc`, `v_calloc` or `vcalloc`.
@@ -369,7 +422,11 @@ pub fn vcalloc_noscan(n int) &byte {
 		if n < 0 {
 			panic('calloc(<0)')
 		}
-		return unsafe { &byte(C.memset(C.GC_MALLOC_ATOMIC(n), 0, n)) }
+		return $if gcboehm_opt ? {
+			unsafe { &byte(C.memset(C.GC_MALLOC_ATOMIC(n), 0, n)) }
+		} $else {
+			unsafe { &byte(C.GC_MALLOC(n)) }
+		}
 	} $else {
 		return unsafe { vcalloc(n) }
 	}
diff --git a/vlib/builtin/string.v b/vlib/builtin/string.v
index d8c48fb729..78bb2bd05b 100644
--- a/vlib/builtin/string.v
+++ b/vlib/builtin/string.v
@@ -248,7 +248,7 @@ pub fn (a string) clone() string {
 		return ''
 	}
 	mut b := string{
-		str: unsafe { malloc(a.len + 1) }
+		str: unsafe { malloc_noscan(a.len + 1) }
 		len: a.len
 	}
 	unsafe {
@@ -303,7 +303,7 @@ pub fn (s string) replace(rep string, with string) string {
 	}
 	// Now we know the number of replacements we need to do and we can calc the len of the new string
 	new_len := s.len + idxs.len * (with.len - rep.len)
-	mut b := unsafe { malloc(new_len + 1) } // add space for the null byte at the end
+	mut b := unsafe { malloc_noscan(new_len + 1) } // add space for the null byte at the end
 	// Fill the new string
 	mut b_i := 0
 	mut s_idx := 0
@@ -406,7 +406,7 @@ pub fn (s string) replace_each(vals []string) string {
 		return s.clone()
 	}
 	idxs.sort2()
-	mut b := unsafe { malloc(new_len + 1) } // add space for 0 terminator
+	mut b := unsafe { malloc_noscan(new_len + 1) } // add space for 0 terminator
 	// Fill the new string
 	mut idx_pos := 0
 	mut cur_idx := idxs[idx_pos]
@@ -532,7 +532,7 @@ fn (s string) < (a string) bool {
 fn (s string) + (a string) string {
 	new_len := a.len + s.len
 	mut res := string{
-		str: unsafe { malloc(new_len + 1) }
+		str: unsafe { malloc_noscan(new_len + 1) }
 		len: new_len
 	}
 	for j in 0 .. s.len {
@@ -676,7 +676,7 @@ pub fn (s string) substr(start int, end int) string {
 		return s.clone()
 	}
 	mut res := string{
-		str: unsafe { malloc(len + 1) }
+		str: unsafe { malloc_noscan(len + 1) }
 		len: len
 	}
 	for i in 0 .. len {
@@ -948,7 +948,7 @@ pub fn (s string) ends_with(p string) bool {
 // TODO only works with ASCII
 pub fn (s string) to_lower() string {
 	unsafe {
-		mut b := malloc(s.len + 1)
+		mut b := malloc_noscan(s.len + 1)
 		for i in 0 .. s.len {
 			if s.str[i] >= `A` && s.str[i] <= `Z` {
 				b[i] = s.str[i] + 32
@@ -977,7 +977,7 @@ pub fn (s string) is_lower() bool {
 // Example: assert 'Hello V'.to_upper() == 'HELLO V'
 pub fn (s string) to_upper() string {
 	unsafe {
-		mut b := malloc(s.len + 1)
+		mut b := malloc_noscan(s.len + 1)
 		for i in 0 .. s.len {
 			if s.str[i] >= `a` && s.str[i] <= `z` {
 				b[i] = s.str[i] - 32
@@ -1262,7 +1262,11 @@ pub fn (s string) ustring() ustring {
 	mut res := ustring{
 		s: s // runes will have at least s.len elements, save reallocations
 		// TODO use VLA for small strings?
-		runes: __new_array(0, s.len, int(sizeof(int)))
+	}
+	$if gcboehm_opt ? {
+		res.runes = __new_array_noscan(0, s.len, int(sizeof(int)))
+	} $else {
+		res.runes = __new_array(0, s.len, int(sizeof(int)))
 	}
 	for i := 0; i < s.len; i++ {
 		char_len := utf8_char_len(unsafe { s.str[i] })
@@ -1282,7 +1286,11 @@ __global (
 
 pub fn (s string) ustring_tmp() ustring {
 	if g_ustring_runes.len == 0 {
-		g_ustring_runes = __new_array(0, 128, int(sizeof(int)))
+		$if gcboehm_opt ? {
+			g_ustring_runes = __new_array_noscan(0, 128, int(sizeof(int)))
+		} $else {
+			g_ustring_runes = __new_array(0, 128, int(sizeof(int)))
+		}
 	}
 	mut res := ustring{
 		s: s
@@ -1311,7 +1319,11 @@ fn (u ustring) < (a ustring) bool {
 fn (u ustring) + (a ustring) ustring {
 	mut res := ustring{
 		s: u.s + a.s
-		runes: __new_array(0, u.s.len + a.s.len, int(sizeof(int)))
+	}
+	$if gcboehm_opt ? {
+		res.runes = __new_array_noscan(0, u.s.len + a.s.len, int(sizeof(int)))
+	} $else {
+		res.runes = __new_array(0, u.s.len + a.s.len, int(sizeof(int)))
 	}
 	mut j := 0
 	for i := 0; i < u.s.len; i++ {
@@ -1586,7 +1598,7 @@ pub fn (a []string) join(sep string) string {
 	len -= sep.len
 	// Allocate enough memory
 	mut res := string{
-		str: unsafe { malloc(len + 1) }
+		str: unsafe { malloc_noscan(len + 1) }
 		len: len
 	}
 	mut idx := 0
@@ -1621,7 +1633,7 @@ pub fn (s string) reverse() string {
 		return s.clone()
 	}
 	mut res := string{
-		str: unsafe { malloc(s.len + 1) }
+		str: unsafe { malloc_noscan(s.len + 1) }
 		len: s.len
 	}
 	for i := s.len - 1; i >= 0; i-- {
@@ -1676,7 +1688,7 @@ pub fn (s string) repeat(count int) string {
 	} else if count == 1 {
 		return s.clone()
 	}
-	mut ret := unsafe { malloc(s.len * count + 1) }
+	mut ret := unsafe { malloc_noscan(s.len * count + 1) }
 	for i in 0 .. count {
 		for j in 0 .. s.len {
 			unsafe {
@@ -1753,7 +1765,7 @@ pub fn (s string) strip_margin_custom(del byte) string {
 	}
 	// don't know how much space the resulting string will be, but the max it
 	// can be is this big
-	mut ret := unsafe { malloc(s.len + 1) }
+	mut ret := unsafe { malloc_noscan(s.len + 1) }
 	mut count := 0
 	for i := 0; i < s.len; i++ {
 		if s[i] in [10, 13] {