Variety of changes (got Cuik compiling with it)

bake.cmd

@@ -2,12 +2,12 @@
 :: Compile UTF-8 resource into .obj file
 :: this .obj file has to be linked to the executable using it, NOT archived
 :: together with ciabatta.lib.
-windres -o utf8.obj utf8.rc
+rem windres -o utf8.obj utf8.rc
-ld -relocatable -o libwinsane.obj utf8.obj
 
 :: Compile chkstk
 nasm src\_win\chkstk.asm -ochkstk.o -fwin64
 
 :: Compile the rest of the party
-clang -Wall src\ciabatta.c -o ciabatta.obj -c -DCIABATTA_WIN -I inc -I src\_win -nodefaultlibs -g -mfma
+rem clang -Wall src\ciabatta.c -o ciabatta.obj -c -DCIABATTA_WIN -I inc -I src\_win -nodefaultlibs -g -mfma
+cuik src\ciabatta.c -o ciabatta.obj -c -DCIABATTA_WIN -I inc -I src\_win -nostdlib
 lib /nologo /out:ciabatta.lib chkstk.o ciabatta.obj

src/_win/cmdline.c

@@ -108,18 +108,9 @@ static int cmdline_to_argv8(const wchar_t *cmd, char **argv) {
     return argc;
 }
 
-static wchar_t *get_wcmdline() {
-    // That's right, that's where windows hid the command line
-    TEB *teb = (TEB *)__readgsqword(offsetof(NT_TIB, Self));
-    PEB *peb = teb->ProcessEnvironmentBlock;
-    RTL_USER_PROCESS_PARAMETERS *params = peb->ProcessParameters;
-    UNICODE_STRING command_line_str = params->CommandLine;
-    return command_line_str.Buffer;
-}
-
 static char **get_command_args(int *argc_ptr) {
     static char *argv_buffer[CMDLINE_ARGV_MAX];
-    wchar_t *cmdline = get_wcmdline();
+    wchar_t *cmdline = GetCommandLineW();
     *argc_ptr = cmdline_to_argv8(cmdline, argv_buffer);
     return argv_buffer;
 }

src/_win/cookie.c

@@ -1,10 +1,10 @@
 
 u64 __security_cookie;
 
-static void __security_init_cookie() {
+void __security_init_cookie() {
     // They say it's a random number so I generated
     // one using numbergenerator.org
-    __security_cookie = 0xb26e04cc62ba48a;
+    __security_cookie = 0xb26e04cc62ba48aULL;
 }
 
 void __security_check_cookie(u64 retrieved) {

src/_win/entry.c

@@ -4,8 +4,8 @@
 #pragma weak WinMain
 
 int main(int argc, char** argv);
-int wmain(int argc, wchar_t** argv, wchar_t **envp);
+// int wmain(int argc, wchar_t** argv, wchar_t **envp);
-int WinMain(HINSTANCE inst, HINSTANCE pinst, LPSTR cmdline, int showcmd);
+// int WinMain(HINSTANCE inst, HINSTANCE pinst, LPSTR cmdline, int showcmd);
 
 _Noreturn void mainCRTStartup() {
     _setup_eh();
@@ -24,7 +24,7 @@ _Noreturn void mainCRTStartup() {
     exit(exit_code);
 }
 
-_Noreturn void WinMainCRTStartup() {
+/*_Noreturn void WinMainCRTStartup() {
     _setup_eh();
     _setup_heap();
     _setup_timer();
@@ -39,4 +39,4 @@ _Noreturn void WinMainCRTStartup() {
     int exit_code = WinMain(inst, 0, cmdline, SW_SHOWDEFAULT);
 
     exit(exit_code);
-}
+}*/

src/_win/win.h

@@ -12,4 +12,3 @@ static void _setup_eh();
 static void _setup_heap();
 static void _setup_io();
 static void _close_io();
-static void __security_init_cookie();

src/ciabatta.c

@@ -33,7 +33,7 @@
 // Which introduces compiler errors. Actually does it really matter? I would
 // need to check again
 #undef __STDC_HOSTED__
-#include <immintrin.h>
+// #include <immintrin.h>
 #include <xmmintrin.h>
 
 #include "intrin.h"

src/math/basic.c

@@ -75,6 +75,28 @@ long double fminl(long double x, long double y) {
     return fmin(x, y);
 }
 
+#ifdef __CUIK__
+#warning "Cuik doesn't support the FMA intrinsics... fix that NeGate"
+double fma(double x, double y, double z) {
+    return (x * y) + z;
+}
+
+float fmaf(float x, float y, float z) {
+    return (x * y) + z;
+}
+
+long double fmal(long double x, long double y, long double z) {
+    return (x * y) + z;
+}
+
+double sqrt(double x) {
+    return 0.0;
+}
+
+float sqrtf(float x) {
+    return 0.0;
+}
+#else
 double fma(double x, double y, double z) {
     __m128d xd = _mm_set_sd(x);
     __m128d yd = _mm_set_sd(y);
@@ -115,4 +137,4 @@ float sqrtf(float x) {
 long double sqrtl(long double x) {
     return sqrt(x);
 }
-
+#endif

test/test_threads.c

@@ -7,7 +7,7 @@ int cnt;
 
 int f(void* thr_data)
 {
-    for(int n = 0; n < 1000; ++n) {
+    for(int n = 0; n < 100000; ++n) {
         atomic_fetch_add_explicit(&acnt, 1, memory_order_relaxed); // atomic
         ++cnt; // undefined behavior, in practice some updates missed
     }