From 5c55d8692fb7ada98a40f2a2ee60262fba3765a6 Mon Sep 17 00:00:00 2001
From: Nuno Cruces <ncruces@users.noreply.github.com>
Date: Fri, 4 Jul 2025 14:58:48 +0100
Subject: [PATCH] Fix bitset.

---
 sqlite3/libc/build.sh     |   1 -
 sqlite3/libc/libc.wasm    | Bin 3071 -> 2991 bytes
 sqlite3/libc/libc.wat     | 818 ++++++++++++++++++--------------------
 sqlite3/libc/libc_test.go |  14 +-
 sqlite3/libc/string.h     |  96 +++--
 value.go                  |   2 +-
 6 files changed, 434 insertions(+), 497 deletions(-)

diff --git a/sqlite3/libc/build.sh b/sqlite3/libc/build.sh
index ea68bfa..4e13232 100755
--- a/sqlite3/libc/build.sh
+++ b/sqlite3/libc/build.sh
@@ -13,7 +13,6 @@ trap 'rm -f libc.c libc.tmp' EXIT
 cat << EOF > libc.c
 #include <stdlib.h>
 #include <string.h>
-#include <strings.h>
 EOF
 
 "$WASI_SDK/clang" --target=wasm32-wasi -std=c23 -g0 -O2 \
diff --git a/sqlite3/libc/libc.wasm b/sqlite3/libc/libc.wasm
index 43a150b96899c31db21840dfccef93f68c5b7e75..688e81451ab4a1bd18888934a90d00eb44ba06ba 100755
GIT binary patch
delta 829
zcma))&1w`u5P<uqrh9hVIQb<wMz*E{0SzJGQ9Mi^0v=uR2<EWtUd$Rq$YFBvC<uM>
zl1GS$58w+#Pk8`+0737zYjw}8D;_j+=;@)Ws;j<lW}9y}Utfx=wRfK+6fcBm_^GIo
zxu?gL`!BFpiY0DjjodAdEe{NsmEp+}GbsH~tuKe8N*EL)<j5vt^U8nE)_FYIpmj^_
zNXqj+vw<I}Qke6996#LtAqAXf4e~w4tUiF+vBIQYDn+m2xa#?js(hqrYjgK`FtvT`
zQ?TQx9Xz-CWcq-<z70?Y0}ON8PHm3)JSLx{h;&1x{k|^ynr5MiB`QDE<(AFDjkCX=
zFbp>lF!L|<ZUr@paO7q)`OG!gJ0VUjA*v7=U8~GJf03`P&`^_c5pP!!ENcHj-`e=2
zWLq$v`Jiv_r+D)Qh!QlE)pDYSXcL>5i%ZokL<#o+bto)^?eLhy9a6BQ?wf>j(fi^e
zo7^=Lf2SNI+2olU<Rk2wZhhpkj<g}sPVK1ZMn~?Xgw~e4igtRg!%yZqK{`p)sN1k!
zXUFaEam6V;j$Q!;r412oWdYqzKj?o@%Vg~Coj7z<)?mJv7_McjE3tBzAW)Gxg!f<j
GtNa%WV8@04

delta 1104
zcmb_bPm9w)6wkc*mzuUIqAUcZFH=1zEZ#f_LUJg0Dhmsq1qDlsVC$lX9^%=HO3osR
z7eSE4s~<u9414zE7Z~b$lVs7Nf?xuj>G0m0-@km^+wQ&i0`JZpf0b(VOaN430|iFq
z!<B>jMD1S^1Cbz<qeU)c(QE*Pk5rUP6yY4}u^wksCgunjEf>j4nzRvBJE?xH$Mq9^
zIi;>q4yd?GA=G#D`TDcIIDGpr<rnJv#$9=%KyGw}Vn2$UnX00U(j?%{H7)5XkD=hV
z{$R3Oh7O!9PsVH4!ydJi0!^tNJ9D(YLI1vwQTjwhXlc847HwA(FGEGRCpY!bX0aJW
z2`bd}aWBjK5G8`|r#p|I)!*#ZYB~^oNJt0GilOnTQ-0#4ze|wm&U@u{D<72=#z!j`
zq}H9(VWS(VJE7ZN2GI59jqXHxxgZ7Lmyx%3W;VW&e~oYL6=a3)Ki_%$?61^rUJExS
z?-i&@a%O@Pl33A(nViOmnLEp7U$+E1b+MoL184mq6<fkv1;bhaIAkFwwhLQmtt_6l
zO_@J6GIioIUl8}sd?^cd(*`#KBw+6q`%1y1-^N*B_jTFYOUa<|8~_zUqEIzux^=&-
z1bqP1&SY{<QrN@@goT*KM2u;gKL}~b`rl;45B-7f3sCV0i>KYn(ew}ekfzP3Z$j%~
tMOvAkwF3SwCHn}iWKnK9$-=oLRR0YR?g9!X5TYE&ggMGjug;tmKLGW={u=-Q

diff --git a/sqlite3/libc/libc.wat b/sqlite3/libc/libc.wat
index 27472d1..e8a2510 100644
--- a/sqlite3/libc/libc.wat
+++ b/sqlite3/libc/libc.wat
@@ -900,7 +900,7 @@
      (i8x16.all_true
       (local.tee $3
        (v128.load
-        (local.tee $2
+        (local.tee $1
          (i32.and
           (local.get $0)
           (i32.const -16)
@@ -912,7 +912,7 @@
     )
     (then
      (br_if $block
-      (local.tee $1
+      (local.tee $2
        (i32.and
         (i8x16.bitmask
          (i8x16.eq
@@ -932,37 +932,23 @@
      )
     )
    )
-   (local.set $1
-    (i32.add
-     (local.get $2)
-     (i32.const 16)
-    )
-   )
    (loop $label
-    (local.set $2
-     (i32.add
-      (local.get $2)
-      (i32.const 16)
-     )
-    )
-    (local.set $3
-     (v128.load
-      (local.get $1)
-     )
-    )
-    (local.set $1
-     (i32.add
-      (local.get $1)
-      (i32.const 16)
-     )
-    )
     (br_if $label
      (i8x16.all_true
-      (local.get $3)
+      (local.tee $3
+       (v128.load
+        (local.tee $1
+         (i32.add
+          (local.get $1)
+          (i32.const 16)
+         )
+        )
+       )
+      )
      )
     )
    )
-   (local.set $1
+   (local.set $2
     (i8x16.bitmask
      (i8x16.eq
       (local.get $3)
@@ -973,10 +959,10 @@
   )
   (i32.add
    (i32.ctz
-    (local.get $1)
+    (local.get $2)
    )
    (i32.sub
-    (local.get $2)
+    (local.get $1)
     (local.get $0)
    )
   )
@@ -991,7 +977,7 @@
      (local.tee $2
       (v128.or
        (i8x16.eq
-        (local.tee $2
+        (local.tee $3
          (v128.load
           (local.tee $4
            (i32.and
@@ -1004,7 +990,7 @@
         (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
        )
        (i8x16.eq
-        (local.get $2)
+        (local.get $3)
         (local.tee $3
          (i8x16.splat
           (local.get $1)
@@ -1034,24 +1020,22 @@
     )
    )
    (loop $label
-    (local.set $2
-     (v128.load offset=16
-      (local.get $4)
-     )
-    )
-    (local.set $4
-     (i32.add
-      (local.get $4)
-      (i32.const 16)
-     )
-    )
     (br_if $label
      (i32.eqz
       (v128.any_true
        (local.tee $2
         (v128.or
          (i8x16.eq
-          (local.get $2)
+          (local.tee $2
+           (v128.load
+            (local.tee $4
+             (i32.add
+              (local.get $4)
+              (i32.const 16)
+             )
+            )
+           )
+          )
           (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
          )
          (i8x16.eq
@@ -1087,7 +1071,7 @@
      (local.tee $2
       (v128.or
        (i8x16.eq
-        (local.tee $2
+        (local.tee $3
          (v128.load
           (local.tee $4
            (i32.and
@@ -1100,7 +1084,7 @@
         (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
        )
        (i8x16.eq
-        (local.get $2)
+        (local.get $3)
         (local.tee $3
          (i8x16.splat
           (local.get $1)
@@ -1130,24 +1114,22 @@
     )
    )
    (loop $label
-    (local.set $2
-     (v128.load offset=16
-      (local.get $4)
-     )
-    )
-    (local.set $4
-     (i32.add
-      (local.get $4)
-      (i32.const 16)
-     )
-    )
     (br_if $label
      (i32.eqz
       (v128.any_true
        (local.tee $2
         (v128.or
          (i8x16.eq
-          (local.get $2)
+          (local.tee $2
+           (v128.load
+            (local.tee $4
+             (i32.add
+              (local.get $4)
+              (i32.const 16)
+             )
+            )
+           )
+          )
           (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
          )
          (i8x16.eq
@@ -1203,15 +1185,15 @@
   (local $2 v128)
   (local $3 v128)
   (local $4 v128)
-  (local $5 i32)
+  (local $5 v128)
   (local $6 i32)
   (local $7 i32)
   (local $8 i32)
   (local $9 i32)
-  (local $scratch v128)
+  (local $10 i32)
   (if
    (i32.eqz
-    (local.tee $6
+    (local.tee $7
      (i32.load8_u
       (local.get $1)
      )
@@ -1223,326 +1205,295 @@
     )
    )
   )
-  (local.set $5
+  (local.set $6
    (i32.and
     (local.get $0)
     (i32.const -16)
    )
   )
-  (local.set $8
+  (local.set $9
    (i32.and
     (local.get $0)
     (i32.const 15)
    )
   )
-  (block $block3
-   (block $block1
-    (block $block
-     (if
-      (i32.load8_u offset=1
+  (if
+   (i32.load8_u offset=1
+    (local.get $1)
+   )
+   (then
+    (local.set $1
+     (i32.add
+      (local.get $1)
+      (i32.const 1)
+     )
+    )
+    (loop $label
+     (v128.store
+      (i32.const 4064)
+      (local.get $4)
+     )
+     (i32.store8
+      (local.tee $10
+       (i32.or
+        (local.tee $8
+         (i32.and
+          (local.get $7)
+          (i32.const 15)
+         )
+        )
+        (i32.const 4064)
+       )
+      )
+      (i32.or
+       (i32.load8_u
+        (local.get $10)
+       )
+       (i32.shl
+        (i32.const 1)
+        (local.tee $7
+         (i32.shr_u
+          (local.get $7)
+          (i32.const 4)
+         )
+        )
+       )
+      )
+     )
+     (v128.store
+      (i32.const 4080)
+      (local.get $3)
+     )
+     (i32.store8
+      (local.tee $8
+       (i32.or
+        (local.get $8)
+        (i32.const 4080)
+       )
+      )
+      (i32.or
+       (i32.load8_u
+        (local.get $8)
+       )
+       (i32.shl
+        (i32.const 1)
+        (i32.sub
+         (local.get $7)
+         (i32.const 8)
+        )
+       )
+      )
+     )
+     (local.set $7
+      (i32.load8_u
        (local.get $1)
       )
-      (then
-       (local.set $1
-        (i32.add
-         (local.get $1)
-         (i32.const 1)
-        )
-       )
-       (loop $label
-        (v128.store
-         (i32.const 4064)
-         (local.get $4)
-        )
-        (i32.store8
-         (local.tee $9
-          (i32.or
-           (local.tee $7
-            (i32.and
-             (local.get $6)
-             (i32.const 15)
-            )
-           )
-           (i32.const 4064)
-          )
-         )
-         (i32.or
-          (i32.load8_u
-           (local.get $9)
-          )
-          (i32.shl
-           (i32.const 1)
-           (local.tee $6
-            (i32.shr_u
-             (local.get $6)
-             (i32.const 4)
-            )
-           )
-          )
-         )
-        )
-        (v128.store
-         (i32.const 4080)
-         (local.get $3)
-        )
-        (i32.store8
-         (local.tee $7
-          (i32.or
-           (local.get $7)
-           (i32.const 4080)
-          )
-         )
-         (i32.or
-          (i32.load8_u
-           (local.get $7)
-          )
-          (i32.shl
-           (i32.const 1)
-           (i32.sub
-            (local.get $6)
-            (i32.const 8)
-           )
-          )
-         )
-        )
-        (local.set $6
-         (i32.load8_u
-          (local.get $1)
-         )
-        )
-        (local.set $4
-         (v128.load
-          (i32.const 4064)
-         )
-        )
-        (local.set $3
-         (v128.load
-          (i32.const 4080)
-         )
-        )
-        (local.set $1
-         (i32.add
-          (local.get $1)
-          (i32.const 1)
-         )
-        )
-        (br_if $label
-         (local.get $6)
-        )
-       )
-       (br_if $block
-        (i32.eqz
-         (i8x16.all_true
-          (local.tee $2
-           (i8x16.eq
-            (v128.and
-             (v128.or
-              (i8x16.swizzle
-               (local.get $4)
+     )
+     (local.set $4
+      (v128.load
+       (i32.const 4064)
+      )
+     )
+     (local.set $3
+      (v128.load
+       (i32.const 4080)
+      )
+     )
+     (local.set $1
+      (i32.add
+       (local.get $1)
+       (i32.const 1)
+      )
+     )
+     (br_if $label
+      (local.get $7)
+     )
+    )
+    (block $block
+     (if
+      (i32.eqz
+       (i8x16.all_true
+        (local.tee $2
+         (i8x16.eq
+          (v128.and
+           (v128.or
+            (i8x16.swizzle
+             (local.get $3)
+             (v128.xor
+              (local.tee $5
                (v128.and
                 (local.tee $2
                  (v128.load
-                  (local.get $5)
+                  (local.get $6)
                  )
                 )
                 (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
                )
               )
-              (i8x16.swizzle
-               (local.get $3)
-               (v128.and
-                (local.get $2)
-                (v128.const i32x4 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f)
-               )
-              )
-             )
-             (local.tee $2
-              (i8x16.swizzle
-               (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
-               (i8x16.shr_u
-                (local.get $2)
-                (i32.const 4)
-               )
-              )
+              (v128.const i32x4 0x80808080 0x80808080 0x80808080 0x80808080)
              )
             )
+            (i8x16.swizzle
+             (local.get $4)
+             (local.get $5)
+            )
+           )
+           (local.tee $2
+            (i8x16.swizzle
+             (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
+             (i8x16.shr_u
+              (local.get $2)
+              (i32.const 4)
+             )
+            )
+           )
+          )
+          (local.get $2)
+         )
+        )
+       )
+      )
+      (then
+       (br_if $block
+        (local.tee $1
+         (i32.and
+          (i32.xor
+           (i8x16.bitmask
             (local.get $2)
            )
+           (i32.const 65535)
+          )
+          (i32.shl
+           (i32.const -1)
+           (local.get $9)
           )
          )
         )
        )
-       (br $block1)
       )
      )
-     (block $block2
-      (if
-       (i32.eqz
-        (i8x16.all_true
-         (local.tee $3
-          (i8x16.eq
-           (v128.load
-            (local.get $5)
-           )
-           (local.tee $4
-            (i8x16.splat
-             (local.get $6)
-            )
-           )
-          )
-         )
-        )
-       )
-       (then
-        (br_if $block2
-         (local.tee $1
-          (i32.and
-           (i32.xor
-            (i8x16.bitmask
+     (loop $label1
+      (br_if $label1
+       (i8x16.all_true
+        (local.tee $2
+         (i8x16.eq
+          (v128.and
+           (v128.or
+            (i8x16.swizzle
              (local.get $3)
-            )
-            (i32.const 65535)
-           )
-           (i32.shl
-            (i32.const -1)
-            (local.get $8)
-           )
-          )
-         )
-        )
-       )
-      )
-      (local.set $1
-       (i32.add
-        (local.get $5)
-        (i32.const 16)
-       )
-      )
-      (loop $label1
-       (local.set $5
-        (i32.add
-         (local.get $5)
-         (i32.const 16)
-        )
-       )
-       (br_if $label1
-        (i8x16.all_true
-         (local.tee $3
-          (i8x16.eq
-           (block (result v128)
-            (local.set $scratch
-             (v128.load
-              (local.get $1)
+             (v128.xor
+              (local.tee $5
+               (v128.and
+                (local.tee $2
+                 (v128.load
+                  (local.tee $6
+                   (i32.add
+                    (local.get $6)
+                    (i32.const 16)
+                   )
+                  )
+                 )
+                )
+                (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
+               )
+              )
+              (v128.const i32x4 0x80808080 0x80808080 0x80808080 0x80808080)
              )
             )
-            (local.set $1
-             (i32.add
-              (local.get $1)
-              (i32.const 16)
+            (i8x16.swizzle
+             (local.get $4)
+             (local.get $5)
+            )
+           )
+           (local.tee $2
+            (i8x16.swizzle
+             (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
+             (i8x16.shr_u
+              (local.get $2)
+              (i32.const 4)
              )
             )
-            (local.get $scratch)
            )
-           (local.get $4)
           )
+          (local.get $2)
          )
         )
        )
       )
-      (local.set $1
-       (i32.xor
-        (i8x16.bitmask
-         (local.get $3)
-        )
-        (i32.const 65535)
-       )
-      )
      )
-     (return
-      (i32.add
-       (i32.ctz
-        (local.get $1)
-       )
-       (i32.sub
-        (local.get $5)
-        (local.get $0)
+     (local.set $1
+      (i32.xor
+       (i8x16.bitmask
+        (local.get $2)
        )
+       (i32.const 65535)
       )
      )
     )
-    (br_if $block3
-     (local.tee $1
-      (i32.and
-       (i32.xor
-        (i8x16.bitmask
-         (local.get $2)
-        )
-        (i32.const 65535)
-       )
-       (i32.shl
-        (i32.const -1)
-        (local.get $8)
-       )
+    (return
+     (i32.add
+      (i32.ctz
+       (local.get $1)
+      )
+      (i32.sub
+       (local.get $6)
+       (local.get $0)
       )
      )
     )
    )
-   (local.set $1
-    (i32.add
-     (local.get $5)
-     (i32.const 16)
+  )
+  (block $block1
+   (if
+    (i32.eqz
+     (i8x16.all_true
+      (local.tee $3
+       (i8x16.eq
+        (v128.load
+         (local.get $6)
+        )
+        (local.tee $4
+         (i8x16.splat
+          (local.get $7)
+         )
+        )
+       )
+      )
+     )
+    )
+    (then
+     (br_if $block1
+      (local.tee $1
+       (i32.and
+        (i32.xor
+         (i8x16.bitmask
+          (local.get $3)
+         )
+         (i32.const 65535)
+        )
+        (i32.shl
+         (i32.const -1)
+         (local.get $9)
+        )
+       )
+      )
+     )
     )
    )
    (loop $label2
-    (local.set $2
-     (v128.load
-      (local.get $1)
-     )
-    )
-    (local.set $1
-     (i32.add
-      (local.get $1)
-      (i32.const 16)
-     )
-    )
-    (local.set $5
-     (i32.add
-      (local.get $5)
-      (i32.const 16)
-     )
-    )
     (br_if $label2
      (i8x16.all_true
-      (local.tee $2
+      (local.tee $3
        (i8x16.eq
-        (v128.and
-         (v128.or
-          (i8x16.swizzle
-           (local.get $4)
-           (v128.and
-            (local.get $2)
-            (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
-           )
-          )
-          (i8x16.swizzle
-           (local.get $3)
-           (v128.and
-            (local.get $2)
-            (v128.const i32x4 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f)
-           )
-          )
-         )
-         (local.tee $2
-          (i8x16.swizzle
-           (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
-           (i8x16.shr_u
-            (local.get $2)
-            (i32.const 4)
-           )
+        (v128.load
+         (local.tee $6
+          (i32.add
+           (local.get $6)
+           (i32.const 16)
           )
          )
         )
-        (local.get $2)
+        (local.get $4)
        )
       )
      )
@@ -1551,7 +1502,7 @@
    (local.set $1
     (i32.xor
      (i8x16.bitmask
-      (local.get $2)
+      (local.get $3)
      )
      (i32.const 65535)
     )
@@ -1562,22 +1513,23 @@
     (local.get $1)
    )
    (i32.sub
-    (local.get $5)
+    (local.get $6)
     (local.get $0)
    )
   )
  )
  (func $strcspn (param $0 i32) (param $1 i32) (result i32)
-  (local $2 i32)
-  (local $3 i32)
-  (local $4 i32)
-  (local $5 i32)
-  (local $6 v128)
-  (local $7 v128)
-  (local $8 v128)
+  (local $2 v128)
+  (local $3 v128)
+  (local $4 v128)
+  (local $5 v128)
+  (local $6 i32)
+  (local $7 i32)
+  (local $8 i32)
+  (local $9 i32)
   (block $block
    (if
-    (local.tee $3
+    (local.tee $7
      (i32.load8_u
       (local.get $1)
      )
@@ -1593,12 +1545,12 @@
    (block $block1
     (if
      (v128.any_true
-      (local.tee $7
+      (local.tee $3
        (v128.or
         (i8x16.eq
-         (local.tee $7
+         (local.tee $4
           (v128.load
-           (local.tee $2
+           (local.tee $6
             (i32.and
              (local.get $0)
              (i32.const -16)
@@ -1609,10 +1561,10 @@
          (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
         )
         (i8x16.eq
-         (local.get $7)
-         (local.tee $8
+         (local.get $4)
+         (local.tee $4
           (i8x16.splat
-           (local.get $3)
+           (local.get $7)
           )
          )
         )
@@ -1624,7 +1576,7 @@
        (local.tee $1
         (i32.and
          (i8x16.bitmask
-          (local.get $7)
+          (local.get $3)
          )
          (i32.shl
           (i32.const -1)
@@ -1639,29 +1591,27 @@
      )
     )
     (loop $label
-     (local.set $7
-      (v128.load offset=16
-       (local.get $2)
-      )
-     )
-     (local.set $2
-      (i32.add
-       (local.get $2)
-       (i32.const 16)
-      )
-     )
      (br_if $label
       (i32.eqz
        (v128.any_true
-        (local.tee $7
+        (local.tee $3
          (v128.or
           (i8x16.eq
-           (local.get $7)
+           (local.tee $3
+            (v128.load
+             (local.tee $6
+              (i32.add
+               (local.get $6)
+               (i32.const 16)
+              )
+             )
+            )
+           )
            (v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000)
           )
           (i8x16.eq
-           (local.get $7)
-           (local.get $8)
+           (local.get $3)
+           (local.get $4)
           )
          )
         )
@@ -1671,38 +1621,38 @@
     )
     (local.set $1
      (i8x16.bitmask
-      (local.get $7)
+      (local.get $3)
      )
     )
    )
    (return
-    (i32.sub
-     (i32.add
-      (local.get $2)
-      (i32.ctz
-       (local.get $1)
-      )
+    (i32.add
+     (i32.ctz
+      (local.get $1)
+     )
+     (i32.sub
+      (local.get $6)
+      (local.get $0)
      )
-     (local.get $0)
     )
    )
   )
-  (local.set $3
+  (local.set $8
    (i32.and
     (local.get $0)
-    (i32.const -16)
+    (i32.const 15)
    )
   )
   (loop $label1
    (v128.store
     (i32.const 4080)
-    (local.get $7)
+    (local.get $4)
    )
    (i32.store8
     (i32.or
-     (local.tee $4
+     (local.tee $7
       (i32.and
-       (local.tee $2
+       (local.tee $6
         (i32.load8_u
          (local.get $1)
         )
@@ -1715,16 +1665,16 @@
     (i32.or
      (i32.load8_u
       (i32.or
-       (local.get $4)
+       (local.get $7)
        (i32.const 4080)
       )
      )
      (i32.shl
       (i32.const 1)
       (i32.sub
-       (local.tee $5
+       (local.tee $9
         (i32.shr_u
-         (local.get $2)
+         (local.get $6)
          (i32.const 4)
         )
        )
@@ -1735,22 +1685,22 @@
    )
    (v128.store
     (i32.const 4064)
-    (local.get $8)
+    (local.get $3)
    )
    (i32.store8
-    (local.tee $4
+    (local.tee $7
      (i32.or
-      (local.get $4)
+      (local.get $7)
       (i32.const 4064)
      )
     )
     (i32.or
      (i32.load8_u
-      (local.get $4)
+      (local.get $7)
      )
      (i32.shl
       (i32.const 1)
-      (local.get $5)
+      (local.get $9)
      )
     )
    )
@@ -1760,154 +1710,146 @@
      (i32.const 1)
     )
    )
-   (local.set $7
+   (local.set $4
     (v128.load
      (i32.const 4080)
     )
    )
-   (local.set $8
+   (local.set $3
     (v128.load
      (i32.const 4064)
     )
    )
    (br_if $label1
-    (local.get $2)
+    (local.get $6)
    )
   )
   (block $block2
    (if
     (v128.any_true
-     (local.tee $6
+     (local.tee $2
       (i8x16.eq
        (v128.and
         (v128.or
          (i8x16.swizzle
-          (local.get $8)
-          (v128.and
-           (local.tee $6
-            (v128.load
-             (local.get $3)
+          (local.get $4)
+          (v128.xor
+           (local.tee $5
+            (v128.and
+             (local.tee $2
+              (v128.load
+               (local.tee $6
+                (i32.and
+                 (local.get $0)
+                 (i32.const -16)
+                )
+               )
+              )
+             )
+             (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
             )
            )
-           (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
+           (v128.const i32x4 0x80808080 0x80808080 0x80808080 0x80808080)
           )
          )
          (i8x16.swizzle
-          (local.get $7)
-          (v128.and
-           (local.get $6)
-           (v128.const i32x4 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f)
-          )
+          (local.get $3)
+          (local.get $5)
          )
         )
-        (local.tee $6
+        (local.tee $2
          (i8x16.swizzle
           (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
           (i8x16.shr_u
-           (local.get $6)
+           (local.get $2)
            (i32.const 4)
           )
          )
         )
        )
-       (local.get $6)
+       (local.get $2)
       )
      )
     )
     (then
      (br_if $block2
-      (local.tee $2
+      (local.tee $1
        (i32.and
         (i8x16.bitmask
-         (local.get $6)
+         (local.get $2)
         )
         (i32.shl
          (i32.const -1)
-         (i32.and
-          (local.get $0)
-          (i32.const 15)
-         )
+         (local.get $8)
         )
        )
       )
      )
     )
    )
-   (local.set $2
-    (i32.add
-     (local.get $3)
-     (i32.const 16)
-    )
-   )
    (loop $label2
-    (local.set $6
-     (v128.load
-      (local.get $2)
-     )
-    )
-    (local.set $2
-     (i32.add
-      (local.get $2)
-      (i32.const 16)
-     )
-    )
-    (local.set $3
-     (i32.add
-      (local.get $3)
-      (i32.const 16)
-     )
-    )
     (br_if $label2
      (i32.eqz
       (v128.any_true
-       (local.tee $6
+       (local.tee $2
         (i8x16.eq
          (v128.and
           (v128.or
            (i8x16.swizzle
-            (local.get $8)
-            (v128.and
-             (local.get $6)
-             (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
+            (local.get $4)
+            (v128.xor
+             (local.tee $5
+              (v128.and
+               (local.tee $2
+                (v128.load
+                 (local.tee $6
+                  (i32.add
+                   (local.get $6)
+                   (i32.const 16)
+                  )
+                 )
+                )
+               )
+               (v128.const i32x4 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f 0x8f8f8f8f)
+              )
+             )
+             (v128.const i32x4 0x80808080 0x80808080 0x80808080 0x80808080)
             )
            )
            (i8x16.swizzle
-            (local.get $7)
-            (v128.and
-             (local.get $6)
-             (v128.const i32x4 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f 0x0f0f0f0f)
-            )
+            (local.get $3)
+            (local.get $5)
            )
           )
-          (local.tee $6
+          (local.tee $2
            (i8x16.swizzle
             (v128.const i32x4 0x08040201 0x80402010 0x08040201 0x80402010)
             (i8x16.shr_u
-             (local.get $6)
+             (local.get $2)
              (i32.const 4)
             )
            )
           )
          )
-         (local.get $6)
+         (local.get $2)
         )
        )
       )
      )
     )
    )
-   (local.set $2
+   (local.set $1
     (i8x16.bitmask
-     (local.get $6)
+     (local.get $2)
     )
    )
   )
   (i32.add
    (i32.ctz
-    (local.get $2)
+    (local.get $1)
    )
    (i32.sub
-    (local.get $3)
+    (local.get $6)
     (local.get $0)
    )
   )
diff --git a/sqlite3/libc/libc_test.go b/sqlite3/libc/libc_test.go
index 4a1bb5b..1232cc0 100644
--- a/sqlite3/libc/libc_test.go
+++ b/sqlite3/libc/libc_test.go
@@ -408,7 +408,7 @@ func Test_strspn(t *testing.T) {
 				fill(memory[ptr:ptr+max(pos, length)], 5)
 				memory[ptr+pos] = 7
 				memory[ptr+length] = 0
-				memory[128] = 3
+				memory[128] = 7 | 128
 				memory[129] = 5
 
 				got := call(strspn, uint64(ptr), 129)
@@ -434,7 +434,7 @@ func Test_strspn(t *testing.T) {
 		clear(memory)
 		fill(memory[ptr:ptr+length], 5)
 		memory[len(memory)-1] = 7
-		memory[128] = 3
+		memory[128] = 7 | 128
 		memory[129] = 5
 
 		got := call(strspn, uint64(ptr), 129)
@@ -462,7 +462,7 @@ func Test_strcspn(t *testing.T) {
 				fill(memory[ptr:ptr+max(pos, length)], 5)
 				memory[ptr+pos] = 7
 				memory[ptr+length] = 0
-				memory[128] = 3
+				memory[128] = 5 | 128
 				memory[129] = 7
 
 				got := call(strcspn, uint64(ptr), 129)
@@ -488,7 +488,7 @@ func Test_strcspn(t *testing.T) {
 		clear(memory)
 		fill(memory[ptr:ptr+length], 5)
 		memory[len(memory)-1] = 7
-		memory[128] = 3
+		memory[128] = 5 | 128
 		memory[129] = 7
 
 		got := call(strcspn, uint64(ptr), 129)
@@ -761,8 +761,10 @@ func Fuzz_strspn(f *testing.F) {
 		}
 
 		if uint32(got) != uint32(want) {
-			t.Errorf("strspn(%q, %q) = %d, want %d",
-				s, chars, uint32(got), uint32(want))
+			t.Errorf("strspn(%v, %v) = %d, want %d",
+				[]byte(memory[ptr1:ptr1+len(s)]),
+				[]byte(memory[ptr2:ptr2+len(chars)]),
+				uint32(got), uint32(want))
 		}
 	})
 }
diff --git a/sqlite3/libc/string.h b/sqlite3/libc/string.h
index 91a395e..b48beeb 100644
--- a/sqlite3/libc/string.h
+++ b/sqlite3/libc/string.h
@@ -3,9 +3,7 @@
 #ifndef _WASM_SIMD128_STRING_H
 #define _WASM_SIMD128_STRING_H
 
-#include <ctype.h>
 #include <stdint.h>
-#include <strings.h>
 #include <wasm_simd128.h>
 #include <__macro_PAGESIZE.h>
 
@@ -90,15 +88,14 @@ void *memchr(const void *s, int c, size_t n) {
 
   // memchr must behave as if it reads characters sequentially
   // and stops as soon as a match is found.
-  // Aligning ensures loads beyond the first match are safe.
-  // Casting through uintptr_t makes this implementation-defined,
-  // rather than undefined behavior.
+  // Aligning ensures out of bounds loads are safe.
   uintptr_t align = (uintptr_t)s % sizeof(v128_t);
-  const v128_t *v = (v128_t *)((uintptr_t)s - align);
-  const v128_t vc = wasm_i8x16_splat(c);
+  uintptr_t addr = (uintptr_t)s - align;
+  v128_t vc = wasm_i8x16_splat(c);
 
   for (;;) {
-    const v128_t cmp = wasm_i8x16_eq(*v, vc);
+    v128_t v = *(v128_t *)addr;
+    v128_t cmp = wasm_i8x16_eq(v, vc);
     // Bitmask is slow on AArch64, any_true is much faster.
     if (wasm_v128_any_true(cmp)) {
       // Clear the bits corresponding to align (little-endian)
@@ -114,7 +111,7 @@ void *memchr(const void *s, int c, size_t n) {
         // That's a match, unless it is beyond the end of the object.
         // Recall that we decremented n, so less-than-or-equal-to is correct.
         size_t ctz = __builtin_ctz(mask);
-        return ctz - align <= n ? (char *)v + ctz : NULL;
+        return ctz - align <= n ? (char *)s + (addr - (uintptr_t)s + ctz) : NULL;
       }
     }
     // Decrement n; if it overflows we're done.
@@ -122,7 +119,7 @@ void *memchr(const void *s, int c, size_t n) {
       return NULL;
     }
     align = 0;
-    v++;
+    addr += sizeof(v128_t);
   }
 }
 
@@ -155,16 +152,15 @@ void *memrchr(const void *s, int c, size_t n) {
 __attribute__((weak))
 size_t strlen(const char *s) {
   // strlen must stop as soon as it finds the terminator.
-  // Aligning ensures loads beyond the terminator are safe.
-  // Casting through uintptr_t makes this implementation-defined,
-  // rather than undefined behavior.
+  // Aligning ensures out of bounds loads are safe.
   uintptr_t align = (uintptr_t)s % sizeof(v128_t);
-  const v128_t *v = (v128_t *)((uintptr_t)s - align);
+  uintptr_t addr = (uintptr_t)s - align;
 
   for (;;) {
+    v128_t v = *(v128_t *)addr;
     // Bitmask is slow on AArch64, all_true is much faster.
-    if (!wasm_i8x16_all_true(*v)) {
-      const v128_t cmp = wasm_i8x16_eq(*v, (v128_t){});
+    if (!wasm_i8x16_all_true(v)) {
+      const v128_t cmp = wasm_i8x16_eq(v, (v128_t){});
       // Clear the bits corresponding to align (little-endian)
       // so we can count trailing zeros.
       int mask = wasm_i8x16_bitmask(cmp) >> align << align;
@@ -175,25 +171,24 @@ size_t strlen(const char *s) {
       // it's as if we didn't find anything.
       if (mask) {
         // Find the offset of the first one bit (little-endian).
-        return (char *)v - s + __builtin_ctz(mask);
+        return addr - (uintptr_t)s + __builtin_ctz(mask);
       }
     }
     align = 0;
-    v++;
+    addr += sizeof(v128_t);
   }
 }
 
 static char *__strchrnul(const char *s, int c) {
-  // strchrnul must stop as soon as it finds the terminator.
-  // Aligning ensures loads beyond the terminator are safe.
-  // Casting through uintptr_t makes this implementation-defined,
-  // rather than undefined behavior.
+  // strchrnul must stop as soon as a match is found.
+  // Aligning ensures out of bounds loads are safe.
   uintptr_t align = (uintptr_t)s % sizeof(v128_t);
-  const v128_t *v = (v128_t *)((uintptr_t)s - align);
-  const v128_t vc = wasm_i8x16_splat(c);
+  uintptr_t addr = (uintptr_t)s - align;
+  v128_t vc = wasm_i8x16_splat(c);
 
   for (;;) {
-    const v128_t cmp = wasm_i8x16_eq(*v, (v128_t){}) | wasm_i8x16_eq(*v, vc);
+    v128_t v = *(v128_t *)addr;
+    const v128_t cmp = wasm_i8x16_eq(v, (v128_t){}) | wasm_i8x16_eq(v, vc);
     // Bitmask is slow on AArch64, any_true is much faster.
     if (wasm_v128_any_true(cmp)) {
       // Clear the bits corresponding to align (little-endian)
@@ -206,11 +201,11 @@ static char *__strchrnul(const char *s, int c) {
       // it's as if we didn't find anything.
       if (mask) {
         // Find the offset of the first one bit (little-endian).
-        return (char *)v + __builtin_ctz(mask);
+        return (char *)s + (addr - (uintptr_t)s + __builtin_ctz(mask));
       }
     }
     align = 0;
-    v++;
+    addr += sizeof(v128_t);
   }
 }
 
@@ -269,19 +264,19 @@ static void __wasm_v128_setbit(__wasm_v128_bitmap256_t *bitmap, int i) {
 
 __attribute__((always_inline))
 static v128_t __wasm_v128_chkbits(__wasm_v128_bitmap256_t bitmap, v128_t v) {
+  v128_t hi_nibbles = wasm_u8x16_shr(v, 4);
+  v128_t bitmask_lookup = wasm_u8x16_const(1, 2, 4, 8, 16, 32, 64, 128,  //
+                                           1, 2, 4, 8, 16, 32, 64, 128);
+  v128_t bitmask = wasm_i8x16_relaxed_swizzle(bitmask_lookup, hi_nibbles);
+
   v128_t indices_0_7 = v & wasm_u8x16_const_splat(0x8f);
-  v128_t indices_8_15 = (v & wasm_u8x16_const_splat(0x80)) ^ indices_0_7;
+  v128_t indices_8_15 = indices_0_7 ^ wasm_u8x16_const_splat(0x80);
 
   v128_t row_0_7 = wasm_i8x16_swizzle(bitmap.l, indices_0_7);
   v128_t row_8_15 = wasm_i8x16_swizzle(bitmap.h, indices_8_15);
 
   v128_t bitsets = row_0_7 | row_8_15;
 
-  v128_t hi_nibbles = wasm_u8x16_shr(v, 4);
-  v128_t bitmask_lookup = wasm_u8x16_const(1, 2, 4, 8, 16, 32, 64, 128,  //
-                                           1, 2, 4, 8, 16, 32, 64, 128);
-  v128_t bitmask = wasm_i8x16_relaxed_swizzle(bitmask_lookup, hi_nibbles);
-
   return wasm_i8x16_eq(bitsets & bitmask, bitmask);
 }
 
@@ -290,17 +285,16 @@ static v128_t __wasm_v128_chkbits(__wasm_v128_bitmap256_t bitmap, v128_t v) {
 __attribute__((weak))
 size_t strspn(const char *s, const char *c) {
   // strspn must stop as soon as it finds the terminator.
-  // Aligning ensures loads beyond the terminator are safe.
-  // Casting through uintptr_t makes this implementation-defined,
-  // rather than undefined behavior.
+  // Aligning ensures out of bounds loads are safe.
   uintptr_t align = (uintptr_t)s % sizeof(v128_t);
-  const v128_t *v = (v128_t *)((uintptr_t)s - align);
+  uintptr_t addr = (uintptr_t)s - align;
 
   if (!c[0]) return 0;
   if (!c[1]) {
-    const v128_t vc = wasm_i8x16_splat(*c);
+    v128_t vc = wasm_i8x16_splat(*c);
     for (;;) {
-      const v128_t cmp = wasm_i8x16_eq(*v, vc);
+      v128_t v = *(v128_t *)addr;
+      v128_t cmp = wasm_i8x16_eq(v, vc);
       // Bitmask is slow on AArch64, all_true is much faster.
       if (!wasm_i8x16_all_true(cmp)) {
         // Clear the bits corresponding to align (little-endian)
@@ -313,11 +307,11 @@ size_t strspn(const char *s, const char *c) {
         // it's as if we didn't find anything.
         if (mask) {
           // Find the offset of the first one bit (little-endian).
-          return (char *)v - s + __builtin_ctz(mask);
+          return addr - (uintptr_t)s + __builtin_ctz(mask);
         }
       }
       align = 0;
-      v++;
+      addr += sizeof(v128_t);
     }
   }
 
@@ -329,7 +323,8 @@ size_t strspn(const char *s, const char *c) {
   }
 
   for (;;) {
-    const v128_t cmp = __wasm_v128_chkbits(bitmap, *v);
+    v128_t v = *(v128_t *)addr;
+    v128_t cmp = __wasm_v128_chkbits(bitmap, v);
     // Bitmask is slow on AArch64, all_true is much faster.
     if (!wasm_i8x16_all_true(cmp)) {
       // Clear the bits corresponding to align (little-endian)
@@ -342,11 +337,11 @@ size_t strspn(const char *s, const char *c) {
       // it's as if we didn't find anything.
       if (mask) {
         // Find the offset of the first one bit (little-endian).
-        return (char *)v - s + __builtin_ctz(mask);
+        return addr - (uintptr_t)s + __builtin_ctz(mask);
       }
     }
     align = 0;
-    v++;
+    addr += sizeof(v128_t);
   }
 }
 
@@ -355,11 +350,9 @@ size_t strcspn(const char *s, const char *c) {
   if (!c[0] || !c[1]) return __strchrnul(s, *c) - s;
 
   // strcspn must stop as soon as it finds the terminator.
-  // Aligning ensures loads beyond the terminator are safe.
-  // Casting through uintptr_t makes this implementation-defined,
-  // rather than undefined behavior.
+  // Aligning ensures out of bounds loads are safe.
   uintptr_t align = (uintptr_t)s % sizeof(v128_t);
-  const v128_t *v = (v128_t *)((uintptr_t)s - align);
+  uintptr_t addr = (uintptr_t)s - align;
 
   __wasm_v128_bitmap256_t bitmap = {};
 
@@ -369,7 +362,8 @@ size_t strcspn(const char *s, const char *c) {
   } while (*c++);
 
   for (;;) {
-    const v128_t cmp = __wasm_v128_chkbits(bitmap, *v);
+    v128_t v = *(v128_t *)addr;
+    v128_t cmp = __wasm_v128_chkbits(bitmap, v);
     // Bitmask is slow on AArch64, any_true is much faster.
     if (wasm_v128_any_true(cmp)) {
       // Clear the bits corresponding to align (little-endian)
@@ -382,11 +376,11 @@ size_t strcspn(const char *s, const char *c) {
       // it's as if we didn't find anything.
       if (mask) {
         // Find the offset of the first one bit (little-endian).
-        return (char *)v - s + __builtin_ctz(mask);
+        return addr - (uintptr_t)s + __builtin_ctz(mask);
       }
     }
     align = 0;
-    v++;
+    addr += sizeof(v128_t);
   }
 }
 
diff --git a/value.go b/value.go
index 6753027..c89d536 100644
--- a/value.go
+++ b/value.go
@@ -57,7 +57,7 @@ func (v Value) Type() Datatype {
 	return Datatype(v.c.call("sqlite3_value_type", v.protected()))
 }
 
-// Type returns the numeric datatype of the value.
+// NumericType returns the numeric datatype of the value.
 //
 // https://sqlite.org/c3ref/value_blob.html
 func (v Value) NumericType() Datatype {