very unfinished FFI

This commit is contained in:
dzaima 2022-05-11 22:28:18 +03:00
parent 409524d958
commit a601ba6a76
10 changed files with 874 additions and 30 deletions

89
include/bqnffi.h Normal file
View File

@ -0,0 +1,89 @@
#include<stddef.h>
#include<stdint.h>
#include<stdbool.h>
typedef uint64_t BQNV;
#ifdef __cplusplus
extern "C" {
#endif
void bqn_free(BQNV v);
double bqn_toF64 (BQNV v); // includes bqn_free(v)
uint32_t bqn_toChar(BQNV v); // includes bqn_free(v)
double bqn_readF64 (BQNV v); // doesn't include bqn_free(v)
uint32_t bqn_readChar(BQNV v); // doesn't include bqn_free(v)
// invoke BQN function
BQNV bqn_call1(BQNV f, BQNV x);
BQNV bqn_call2(BQNV f, BQNV w, BQNV x);
// evaluate BQN code in a fresh environment
BQNV bqn_eval(BQNV src);
BQNV bqn_evalCStr(char* str); // evaluates the null-terminated UTF8-encoded str; equal to `BQNV s = bqn_makeUTF8Str(str, strlen(str)); result = bqn_eval(s); bqn_free(s);`
// read array data
size_t bqn_bound(BQNV a); // aka product of shape, ×´≢a
size_t bqn_rank(BQNV a);
void bqn_shape(BQNV a, size_t* buf); // writes bqn_rank(a) items in buf
BQNV bqn_pick(BQNV a, size_t pos);
// read all elements of `a` into the specified buffer
void bqn_readI8Arr (BQNV a, int8_t* buf);
void bqn_readI16Arr(BQNV a, int16_t* buf);
void bqn_readI32Arr(BQNV a, int32_t* buf);
void bqn_readF64Arr(BQNV a, double* buf);
void bqn_readC8Arr (BQNV a, uint8_t* buf);
void bqn_readC16Arr(BQNV a, uint16_t* buf);
void bqn_readC32Arr(BQNV a, uint32_t* buf);
void bqn_readObjArr(BQNV a, BQNV* buf);
// create objects
BQNV bqn_makeF64(double d);
BQNV bqn_makeChar(uint32_t c);
BQNV bqn_makeI8Arr (size_t rank, size_t* shape, int8_t* data);
BQNV bqn_makeI16Arr(size_t rank, size_t* shape, int16_t* data);
BQNV bqn_makeI32Arr(size_t rank, size_t* shape, int32_t* data);
BQNV bqn_makeF64Arr(size_t rank, size_t* shape, double* data);
BQNV bqn_makeC8Arr (size_t rank, size_t* shape, uint8_t* data);
BQNV bqn_makeC16Arr(size_t rank, size_t* shape, uint16_t* data);
BQNV bqn_makeC32Arr(size_t rank, size_t* shape, uint32_t* data);
BQNV bqn_makeObjArr(size_t rank, size_t* shape, BQNV* data); // frees the taken elements of data
BQNV bqn_makeI8Vec (size_t len, int8_t* data);
BQNV bqn_makeI16Vec(size_t len, int16_t* data);
BQNV bqn_makeI32Vec(size_t len, int32_t* data);
BQNV bqn_makeF64Vec(size_t len, double* data);
BQNV bqn_makeC8Vec (size_t len, uint8_t* data);
BQNV bqn_makeC16Vec(size_t len, uint16_t* data);
BQNV bqn_makeC32Vec(size_t len, uint32_t* data);
BQNV bqn_makeObjVec(size_t len, BQNV* data); // frees the taken elements of data
BQNV bqn_makeUTF8Str(size_t len, char* str);
typedef BQNV (*bqn_boundFn1)(BQNV obj, BQNV x);
typedef BQNV (*bqn_boundFn2)(BQNV obj, BQNV w, BQNV x);
// when called, 1st arg to `f` will be `obj`
BQNV bqn_makeBoundFn1(bqn_boundFn1 f, BQNV obj);
BQNV bqn_makeBoundFn2(bqn_boundFn2 f, BQNV obj);
// direct (zero copy) array item access
typedef enum { elt_i8, elt_i16, elt_i32, elt_f64, elt_c8, elt_c16, elt_c32, elt_unk } BQNElType;
BQNElType bqn_directType(BQNV a);
// can only use the functions below if bqn_elType returns the corresponding type
// a valid implementation of bqn_elType would be to always return elt_unk, thus disallowing the use of direct access entirely
int8_t* bqn_directI8 (BQNV a);
int16_t* bqn_directI16(BQNV a);
int32_t* bqn_directI32(BQNV a);
double* bqn_directF64(BQNV a);
uint8_t* bqn_directC8 (BQNV a);
uint16_t* bqn_directC16(BQNV a);
uint32_t* bqn_directC32(BQNV a);
#ifdef __cplusplus
}
#endif

View File

@ -37,32 +37,48 @@ c:
@${MAKE} custom=1 run_incremental_0
# compiler setup
i_CC = clang
i_PIE = -no-pie
i_LD_LIBS = -lm
OUTPUT = BQN
i_CC := clang
i_PIE := -no-pie
i_LD_LIBS := -lm
i_FFI := 2
i_singeli := 0
OUTPUT := BQN
ifeq ($(origin CC),command line)
i_CC = $(CC)
custom = 1
endif
ifeq ($(origin PIE),command line)
i_PIE = $(PIE)
custom = 1
endif
ifeq ($(origin LD_LIBS),command line)
i_LD_LIBS = $(LD_LIBS)
i_CC := $(CC)
custom = 1
endif
ifeq ($(origin singeli),command line)
i_singeli = $(singeli)
i_singeli := $(singeli)
custom = 1
endif
ifeq ($(origin FFI),command line)
i_FFI := $(FFI)
custom = 1
endif
ifneq ($(i_FFI),0)
i_LD_LIBS += -ldl -rdynamic
endif
ifeq ($(i_FFI),2)
i_LD_LIBS += -lffi
endif
ifeq ($(origin LD_LIBS),command line)
i_LD_LIBS := $(LD_LIBS)
custom = 1
endif
ifeq ($(origin PIE),command line)
i_PIE := $(PIE)
custom = 1
endif
ifeq ($(origin f),command line)
custom = 1
else
f:=
endif
ifeq ($(origin lf),command line)
custom = 1
else
lf:=
endif
ifeq ($(origin CCFLAGS),command line)
custom = 1
@ -78,14 +94,7 @@ else
NOWARN = -Wno-parentheses
endif
ifeq (${i_singeli}, 1)
SINGELIFLAGS = '-DSINGELI'
else
i_singeli = 0
SINGELIFLAGS =
endif
ALL_CC_FLAGS = -std=gnu11 -Wall -Wno-unused-function -fms-extensions $(CCFLAGS) $(f) $(i_f) $(SINGELIFLAGS) $(NOWARN)
ALL_CC_FLAGS = -std=gnu11 -Wall -Wno-unused-function -fms-extensions $(CCFLAGS) $(f) $(i_f) $(NOWARN) -DSINGELI=$(i_singeli) -DFFI=$(i_FFI)
ALL_LD_FLAGS = $(LDFLAGS) $(lf) $(i_lf) $(i_PIE) $(i_LD_LIBS)
ifneq (${manualJobs},1)
@ -152,7 +161,7 @@ ${bd}/%.o: src/core/%.c
@echo $< | cut -c 5-
@$(CC_INC) $@.d -o $@ -c $<
base: ${addprefix ${bd}/, load.o main.o rtwrap.o vm.o ns.o nfns.o}
base: ${addprefix ${bd}/, load.o main.o rtwrap.o vm.o ns.o nfns.o ffi.o}
${bd}/%.o: src/%.c
@echo $< | cut -c 5-
@$(CC_INC) $@.d -o $@ -c $<

View File

@ -12,6 +12,7 @@
/* sysfn.c*/M(repr,"•Repr") M(fmt,"•Fmt") A(asrt,"!") A(casrt,"!") M(out,"•Out") M(show,"•Show") A(bqn,"•BQN") A(sh,"•SH") M(fromUtf8,"•FromUTF8") M(currentError,"•CurrentError") \
/* sysfn.c*/D(cmp,"•Cmp") A(hash,"•Hash") M(unixTime,"•UnixTime") M(monoTime,"•MonoTime") M(delay,"•Delay") M(makeRand,"•MakeRand") M(reBQN,"•ReBQN") M(exit,"•Exit") M(getLine,"•GetLine") \
/* sysfn.c*/M(fName,"•file.Name") \
/* ffi.c*/D(loadffi,"•LoadFFI") \
/* sysfn.c*/M(tRawMode,"•term.RawMode") M(tFlush,"•term.Flush") M(tCharB,"•term.CharB") M(tCharN,"•term.CharN") M(tOutRaw,"•term.OutRaw") M(tErrRaw,"•term.ErrRaw") \
/* inverse.c*/M(setInvReg, "(SetInvReg)") M(setInvSwap, "(SetInvSwap)") M(nativeInvReg, "(NativeInvReg)") M(nativeInvSwap, "(NativeInvSwap)") \
/*internal.c*/M(itype,"•internal.Type") M(elType,"•internal.ElType") M(refc,"•internal.Refc") M(isPure,"•internal.IsPure") A(info,"•internal.Info") M(heapDump,"•internal.HeapDump") \

View File

@ -239,7 +239,7 @@ B show_c1(B t, B x) {
return x;
}
static B vfyStr(B x, char* name, char* arg) {
B vfyStr(B x, char* name, char* arg) {
if (isAtm(x) || rnk(x)!=1) thrF("%U: %U must be a character vector", name, arg);
if (!elChr(TI(x,elType))) {
usz ia = a(x)->ia;
@ -1141,6 +1141,9 @@ B sys_c1(B t, B x) {
else if (eqStr(c, U"flines")) cr = m_nfn(fLinesDesc, inc(REQ_PATH));
else if (eqStr(c, U"import")) cr = m_nfn(importDesc, inc(REQ_PATH));
else if (eqStr(c, U"currenterror")) cr = inc(bi_currentError);
#if FFI
else if (eqStr(c, U"loadffi")) cr = inc(bi_loadffi);
#endif
else if (eqStr(c, U"state")) {
if (q_N(comp_currArgs)) thrM("No arguments present for •state");
cr = m_hVec3(inc(REQ_PATH), inc(REQ_NAME), inc(comp_currArgs));

640
src/ffi.c Normal file
View File

@ -0,0 +1,640 @@
#include "core.h"
#if FFI
#include "../include/bqnffi.h"
#include "utils/utf.h"
#include "utils/cstr.h"
#include "nfns.h"
#include <dlfcn.h>
// base interface defs for when GC stuff needs to be added in
static B getB(BQNV v) {
return b(v);
}
static BQNV makeX(B x) {
return x.u;
}
void bqn_free(BQNV v) {
dec(getB(v));
}
static void freeTagged (BQNV v) { }
#define DIRECT_BQNV 1
double bqn_toF64 (BQNV v) { double r = o2fu(getB(v)); freeTagged(v); return r; }
uint32_t bqn_toChar(BQNV v) { uint32_t r = o2cu(getB(v)); freeTagged(v); return r; }
double bqn_readF64 (BQNV v) { return o2fu(getB(v)); }
uint32_t bqn_readChar(BQNV v) { return o2cu(getB(v)); }
BQNV bqn_call1(BQNV f, BQNV x) {
return makeX(c1(getB(f), inc(getB(x))));
}
BQNV bqn_call2(BQNV f, BQNV w, BQNV x) {
return makeX(c2(getB(f), inc(getB(w)), inc(getB(x))));
}
BQNV bqn_eval(BQNV src) {
return makeX(bqn_exec(inc(getB(src)), bi_N, bi_N));
}
BQNV bqn_evalCStr(char* str) {
return makeX(bqn_exec(fromUTF8l(str), bi_N, bi_N));
}
size_t bqn_bound(BQNV a) { return a(getB(a))->ia; }
size_t bqn_rank(BQNV a) { return rnk(getB(a)); }
void bqn_shape(BQNV a, size_t* buf) { B b = getB(a);
ur r = rnk(b);
usz* sh = a(b)->sh;
for (usz i = 0; i < r; i++) buf[i] = sh[i];
}
BQNV bqn_pick(BQNV a, size_t pos) {
return makeX(IGet(getB(a),pos));
}
// TODO copy directly with some mut.h thing
void bqn_readI8Arr (BQNV a, i8* buf) { B c = toI8Any (inc(getB(a))); memcpy(buf, i8any_ptr (c), a(c)->ia * 1); dec(c); }
void bqn_readI16Arr(BQNV a, i16* buf) { B c = toI16Any(inc(getB(a))); memcpy(buf, i16any_ptr(c), a(c)->ia * 2); dec(c); }
void bqn_readI32Arr(BQNV a, i32* buf) { B c = toI32Any(inc(getB(a))); memcpy(buf, i32any_ptr(c), a(c)->ia * 4); dec(c); }
void bqn_readF64Arr(BQNV a, f64* buf) { B c = toF64Any(inc(getB(a))); memcpy(buf, f64any_ptr(c), a(c)->ia * 8); dec(c); }
void bqn_readC8Arr (BQNV a, u8* buf) { B c = toC8Any (inc(getB(a))); memcpy(buf, c8any_ptr (c), a(c)->ia * 1); dec(c); }
void bqn_readC16Arr(BQNV a, u16* buf) { B c = toC16Any(inc(getB(a))); memcpy(buf, c16any_ptr(c), a(c)->ia * 2); dec(c); }
void bqn_readC32Arr(BQNV a, u32* buf) { B c = toC32Any(inc(getB(a))); memcpy(buf, c32any_ptr(c), a(c)->ia * 4); dec(c); }
void bqn_readObjArr(BQNV a, BQNV* buf) { B b = getB(a);
usz ia = a(b)->ia;
B* p = arr_bptr(b);
if (p!=NULL) {
for (usz i = 0; i < ia; i++) buf[i] = makeX(inc(p[i]));
} else {
SGet(b)
for (usz i = 0; i < ia; i++) buf[i] = makeX(Get(b, i));
}
}
BQNV bqn_makeF64(double d) { return makeX(m_f64(d)); }
BQNV bqn_makeChar(uint32_t c) { return makeX(m_c32(c)); }
static usz calcIA(size_t rank, size_t* shape) {
if (rank>UR_MAX) thrM("Rank too large");
usz r = 1;
for (size_t i = 0; i < rank; i++) if (mulOn(r, shape[i])) thrM("Size too large");
return r;
}
static void copyBData(B* r, BQNV* data, usz ia) {
for (size_t i = 0; i < ia; i++) {
BQNV c = data[i];
#if DIRECT_BQNV
r[i] = getB(c);
#else
r[i] = inc(getB(c));
bqn_free(c);
#endif
}
}
#define CPYSH(R) usz* sh = arr_shAlloc(R, r0); \
if (sh) for (size_t i = 0; RARE(i < r0); i++) sh[i] = sh0[i];
BQNV bqn_makeI8Arr (size_t r0, size_t* sh0, i8* data) { usz ia=calcIA(r0,sh0); i8* rp; Arr* r = m_i8arrp (&rp,ia); CPYSH(r); memcpy(rp,data,ia*1); return makeX(taga(r)); }
BQNV bqn_makeI16Arr(size_t r0, size_t* sh0, i16* data) { usz ia=calcIA(r0,sh0); i16* rp; Arr* r = m_i16arrp(&rp,ia); CPYSH(r); memcpy(rp,data,ia*2); return makeX(taga(r)); }
BQNV bqn_makeI32Arr(size_t r0, size_t* sh0, i32* data) { usz ia=calcIA(r0,sh0); i32* rp; Arr* r = m_i32arrp(&rp,ia); CPYSH(r); memcpy(rp,data,ia*4); return makeX(taga(r)); }
BQNV bqn_makeF64Arr(size_t r0, size_t* sh0, f64* data) { usz ia=calcIA(r0,sh0); f64* rp; Arr* r = m_f64arrp(&rp,ia); CPYSH(r); memcpy(rp,data,ia*8); return makeX(taga(r)); }
BQNV bqn_makeC8Arr (size_t r0, size_t* sh0, u8* data) { usz ia=calcIA(r0,sh0); u8* rp; Arr* r = m_c8arrp (&rp,ia); CPYSH(r); memcpy(rp,data,ia*1); return makeX(taga(r)); }
BQNV bqn_makeC16Arr(size_t r0, size_t* sh0, u16* data) { usz ia=calcIA(r0,sh0); u16* rp; Arr* r = m_c16arrp(&rp,ia); CPYSH(r); memcpy(rp,data,ia*2); return makeX(taga(r)); }
BQNV bqn_makeC32Arr(size_t r0, size_t* sh0, u32* data) { usz ia=calcIA(r0,sh0); u32* rp; Arr* r = m_c32arrp(&rp,ia); CPYSH(r); memcpy(rp,data,ia*4); return makeX(taga(r)); }
BQNV bqn_makeObjArr(size_t r0, size_t* sh0, BQNV* data) { usz ia=calcIA(r0,sh0); HArr_p r = m_harrUp(ia); CPYSH((Arr*)r.c); copyBData(r.a,data,ia); return makeX(r.b); }
BQNV bqn_makeI8Vec (size_t len, i8* data) { i8* rp; B r = m_i8arrv (&rp,len); memcpy(rp,data,len*1); return makeX(r); }
BQNV bqn_makeI16Vec(size_t len, i16* data) { i16* rp; B r = m_i16arrv(&rp,len); memcpy(rp,data,len*2); return makeX(r); }
BQNV bqn_makeI32Vec(size_t len, i32* data) { i32* rp; B r = m_i32arrv(&rp,len); memcpy(rp,data,len*4); return makeX(r); }
BQNV bqn_makeF64Vec(size_t len, f64* data) { f64* rp; B r = m_f64arrv(&rp,len); memcpy(rp,data,len*8); return makeX(r); }
BQNV bqn_makeC8Vec (size_t len, u8* data) { u8* rp; B r = m_c8arrv (&rp,len); memcpy(rp,data,len*1); return makeX(r); }
BQNV bqn_makeC16Vec(size_t len, u16* data) { u16* rp; B r = m_c16arrv(&rp,len); memcpy(rp,data,len*2); return makeX(r); }
BQNV bqn_makeC32Vec(size_t len, u32* data) { u32* rp; B r = m_c32arrv(&rp,len); memcpy(rp,data,len*4); return makeX(r); }
BQNV bqn_makeObjVec(size_t len, BQNV* data) { HArr_p r = m_harrUv(len); copyBData(r.a,data,len ); return makeX(r.b); }
BQNV bqn_makeUTF8Str(size_t len, char* str) { return makeX(fromUTF8(str, len)); }
typedef struct BoundFn {
struct NFn;
void* w_c1;
void* w_c2;
} BoundFn;
NFnDesc* boundFnDesc;
NFnDesc* foreignFnDesc;
B boundFn_c1(B t, B x) { BoundFn* c = c(BoundFn,t); return getB(((bqn_boundFn1)c->w_c1)(makeX(inc(c->obj)), makeX(x))); }
B boundFn_c2(B t, B w, B x) { BoundFn* c = c(BoundFn,t); return getB(((bqn_boundFn2)c->w_c2)(makeX(inc(c->obj)), makeX(w), makeX(x))); }
typedef BQNV (*bqn_foreignFn1)(BQNV x);
typedef BQNV (*bqn_foreignFn2)(BQNV w, BQNV x);
B directFn_c1(B t, B x) { BoundFn* c = c(BoundFn,t); return getB(((bqn_foreignFn1)c->w_c1)( makeX(x))); }
B directFn_c2(B t, B w, B x) { BoundFn* c = c(BoundFn,t); return getB(((bqn_foreignFn2)c->w_c2)(makeX(w), makeX(x))); }
static B m_ffiFn(NFnDesc* desc, B obj, BB2B c1, BBB2B c2, void* wc1, void* wc2) {
BoundFn* r = mm_alloc(sizeof(BoundFn), t_nfn);
nfn_lateInit((NFn*)r, desc);
r->obj = obj;
r->c1 = c1;
r->c2 = c2;
r->w_c1 = wc1;
r->w_c2 = wc2;
return tag(r, FUN_TAG);
}
BQNV bqn_makeBoundFn1(bqn_boundFn1 f, BQNV obj) { return makeX(m_ffiFn(boundFnDesc, inc(getB(obj)), boundFn_c1, c2_bad, f, NULL)); }
BQNV bqn_makeBoundFn2(bqn_boundFn2 f, BQNV obj) { return makeX(m_ffiFn(boundFnDesc, inc(getB(obj)), c1_bad, boundFn_c2, NULL, f)); }
const static u8 typeMap[] = {
[el_bit] = elt_unk,
[el_B ] = elt_unk,
[el_i8 ] = elt_i8, [el_c8 ] = elt_c8,
[el_i16] = elt_i16, [el_c16] = elt_c16,
[el_i32] = elt_i32, [el_c32] = elt_c32,
[el_f64] = elt_f64,
};
BQNElType bqn_directType(BQNV a) {
B b = getB(a);
if (!isArr(b)) return elt_unk;
return typeMap[TI(b,elType)];
}
i8* bqn_directI8 (BQNV a) { return i8any_ptr (getB(a)); }
i16* bqn_directI16(BQNV a) { return i16any_ptr(getB(a)); }
i32* bqn_directI32(BQNV a) { return i32any_ptr(getB(a)); }
f64* bqn_directF64(BQNV a) { return f64any_ptr(getB(a)); }
u8* bqn_directC8 (BQNV a) { return c8any_ptr (getB(a)); }
u16* bqn_directC16(BQNV a) { return c16any_ptr(getB(a)); }
u32* bqn_directC32(BQNV a) { return c32any_ptr(getB(a)); }
void ffiFn_visit(Value* v) { mm_visit(((BoundFn*)v)->obj); }
DEF_FREE(ffiFn) { dec(((BoundFn*)x)->obj); }
#if FFI==2
#include <ffi.h>
#include "utils/mut.h"
typedef struct BQNFFIEnt {
B o;
ffi_type t;
u8 extra;
u8 extra2;
} BQNFFIEnt;
typedef struct BQNFFIType {
struct Value;
u8 ty;
usz ia;
BQNFFIEnt a[];
} BQNFFIType;
enum ScalarTy {
sty_void, sty_a,
sty_u8, sty_u16, sty_u32, sty_u64,
sty_i8, sty_i16, sty_i32, sty_i64,
sty_f32, sty_f64
};
static const u8 sty_w[] = {
[sty_void]=0, [sty_a]=sizeof(BQNV),
[sty_u8]=1, [sty_u16]=2, [sty_u32]=4, [sty_u64]=8,
[sty_i8]=1, [sty_i16]=2, [sty_i32]=4, [sty_i64]=8,
[sty_f32]=4, [sty_f64]=8
};
static const char* sty_names[] = {
[sty_void]="void", [sty_a]="a",
[sty_u8]="u8", [sty_u16]="u16", [sty_u32]="u32", [sty_u64]="u64",
[sty_i8]="i8", [sty_i16]="i16", [sty_i32]="i32", [sty_i64]="i64",
[sty_f32]="f32", [sty_f64]="f64"
};
enum CompoundTy {
cty_ptr,
cty_repr
};
static void printFFIType(FILE* f, B x) {
if (isC32(x)) fprintf(f, "%d", o2cu(x));
else fprint(f, x);
}
static B m_bqnFFIType(BQNFFIEnt** rp, u8 ty, usz ia) {
BQNFFIType* r = mm_alloc(fsizeof(BQNFFIType, a, BQNFFIEnt, ia), t_ffiType);
r->ty = ty;
r->ia = ia;
memset(r->a, 0, ia*sizeof(BQNFFIEnt));
*rp = r->a;
return tag(r, OBJ_TAG);
}
static u32 readUInt(u32** p) {
u32* c = *p;
u32 r = 0;
while (*c>='0' & *c<='9') {
if (r >= U32_MAX/10 - 10) thrM("FFI: number literal too large");
r = r*10 + *c-'0';
c++;
}
*p = c;
return r;
}
BQNFFIEnt ffi_parseTypeStr(u32** src, bool inPtr) { // parse actual type; res.extra: whether should parse a :abc; res.extra2: whether has any mutation
u32* c = *src;
u32 c0 = *c++;
ffi_type rt;
B ro;
bool parseRepr=false, subParseRepr=false, mut=false;
switch (c0) {
default:
thrM("FFI: Error parsing type");
case 'i': case 'u': case 'f':;
u32 n = readUInt(&c);
if (c0=='f') {
if (n==32) rt = ffi_type_float;
else if (n==64) rt = ffi_type_double;
else thrM("FFI: Bad float width");
ro = m_c32(n==32? sty_f32 : sty_f64);
} else {
u32 scty;
if (n== 8) { scty = c0=='i'? sty_i8 : sty_u8; rt = c0=='i'? ffi_type_sint8 : ffi_type_uint8; }
else if (n==16) { scty = c0=='i'? sty_i16 : sty_u16; rt = c0=='i'? ffi_type_sint16 : ffi_type_uint16; }
else if (n==32) { scty = c0=='i'? sty_i32 : sty_u32; rt = c0=='i'? ffi_type_sint32 : ffi_type_uint32; }
else if (n==64) { scty = c0=='i'? sty_i64 : sty_u64; rt = c0=='i'? ffi_type_sint64 : ffi_type_uint64; }
else thrM("FFI: Bad integer width");
ro = m_c32(scty);
}
parseRepr = !inPtr;
subParseRepr = inPtr;
break;
case 'a':
ro = m_c32(sty_a);
rt = ffi_type_uint64;
break;
case '*': case '&':;
if (c0=='&') mut = true;
BQNFFIEnt* rp; ro = m_bqnFFIType(&rp, cty_ptr, 1);
rp[0] = ffi_parseTypeStr(&c, true);
mut|= rp[0].extra2;
rp[0].extra = c0=='&';
rt = ffi_type_pointer;
parseRepr = rp[0].extra;
break;
}
if (parseRepr && *c==':') {
c++;
u8 t = *c++;
u32 n = readUInt(&c);
if (t=='i' | t=='c') if (n!=8 & n!=16 & n!=32) { badW: thrF("Bad width in :%c%i", t, n); }
if (t=='u') if (n!=1 & n!=8 & n!=16 & n!=32) goto badW;
if (t=='f') if (n!=64) goto badW;
B roP = ro;
BQNFFIEnt* rp; ro = m_bqnFFIType(&rp, cty_repr, 1);
rp[0] = (BQNFFIEnt){.o=roP, .t=rt, .extra=t, .extra2=63-CLZ(n)};
}
*src = c;
return (BQNFFIEnt){.t=rt, .o=ro, .extra=subParseRepr, .extra2=mut};
}
B vfyStr(B x, char* name, char* arg);
BQNFFIEnt ffi_parseType(B arg, bool forRes) { // doesn't consume; parse argument side & other global decorators; .extra=side, .extra2=contains mutation
vfyStr(arg, "FFI", "type");
usz ia = a(arg)->ia;
if (ia==0) {
if (!forRes) thrM("FFI: Argument type empty");
return (BQNFFIEnt){.t = ffi_type_void, .o=m_c32(sty_void)};
}
MAKE_MUT(tmp, ia+1); mut_init(tmp, el_c32); MUTG_INIT(tmp);
mut_copyG(tmp, 0, arg, 0, ia);
mut_setG(tmp, ia, m_c32(0));
u32* xp = tmp->ac32;
u32* xpN = xp + ia;
u8 side;
if (*xp == U'𝕩') side = 2;
else if (*xp == U'𝕨') side = 1;
else side = 0;
if (forRes && side) thrM("FFI: Argument side cannot be specified for the result");
if (side) xp++;
BQNFFIEnt t = ffi_parseTypeStr(&xp, false);
// print(arg); printf(": "); printFFIType(stdout, t.o); printf("\n");
if (xp!=xpN) thrM("FFI: Bad type descriptor");
t.extra = side;
// keep extra2 as "contains mutation"
mut_pfree(tmp, 0);
return t;
}
TAlloc* ffiTmpObj;
u8* ffiTmpS;
u8* ffiTmpC;
u8* ffiTmpE;
static NOINLINE usz ffiTmpX(usz n) {
usz psz = ffiTmpC-ffiTmpS;
TAlloc* na = mm_alloc(sizeof(TAlloc)+psz+n, t_temp);
memcpy(na->data, ffiTmpS, psz);
ffiTmpS = na->data;
ffiTmpC = ffiTmpS + psz;
ffiTmpE = (u8*)na + mm_size((Value*) na);
if (ffiTmpObj) mm_free((Value*)ffiTmpObj);
ffiTmpObj = na;
return psz;
}
static NOINLINE usz ffiTmpAA(usz n) {
u64 align = _Alignof(max_align_t);
n = (n+align-1) & ~(align-1);
if (ffiTmpC+n > ffiTmpE) return ffiTmpX(n);
usz r = ffiTmpC-ffiTmpS;
ffiTmpC+= n;
return r;
}
static B ffiObjs;
static B toW(u8 tre, u8 wre, B x) {
switch(wre) { default: UD;
case 0: return taga(toBitArr(x)); break;
case 3: return tre=='c'? toC8Any(x) : toI8Any(x); break;
case 4: return tre=='c'? toC16Any(x) : toI16Any(x); break;
case 5: return tre=='c'? toC32Any(x) : toI32Any(x); break;
case 6: return toF64Any(x); break;
}
}
usz genObj(BQNFFIEnt ent, B c, bool anyMut) {
// printFFIType(stdout,ent.o); printf(" = "); print(c); printf("\n");
usz pos;
if (isC32(ent.o)) { // scalar
u32 t = o2cu(ent.o);
pos = ffiTmpAA(t==0? sizeof(BQNV) : 8);
void* ptr = ffiTmpS+pos;
f64 f = c.f;
switch(t) { default: thrF("FFI: Unimplemented scalar type %S", sty_names[t]);
case sty_a: *(BQNV*)ptr = makeX(inc(c)); break;
case sty_u8: *( u8*)ptr = f; break;
case sty_i8: *( i8*)ptr = f; break;
case sty_u16: *(u16*)ptr = f; break;
case sty_i16: *(i16*)ptr = f; break;
case sty_u32: *(u32*)ptr = f; break;
case sty_i32: *(i32*)ptr = f; break;
case sty_f32: *(float* )ptr = f; break;
case sty_f64: *(double*)ptr = f; break;
}
} else {
BQNFFIType* t = c(BQNFFIType, ent.o);
if (t->ty==cty_ptr) { // *any / &any
pos = ffiTmpAA(sizeof(void*));
B e = t->a[0].o;
if (!isC32(e)) thrM("FFI: Nested pointers unimplemented");
inc(c);
B cG;
if (!isArr(c)) thrF("FFI: Expected array corresponding to *%S", sty_names[o2cu(e)]);
usz ia = a(c)->ia;
bool mut = t->a[0].extra;
switch(o2cu(e)) { default: thrF("FFI: Unimplemented pointer to %S", sty_names[o2cu(e)]);
case sty_i8: cG = mut? taga(cpyI8Arr (c)) : toI8Any (c); break;
case sty_i16: cG = mut? taga(cpyI16Arr(c)) : toI16Any(c); break;
case sty_i32: cG = mut? taga(cpyI32Arr(c)) : toI32Any(c); break;
case sty_f64: cG = mut? taga(cpyF64Arr(c)) : toF64Any(c); break;
case sty_u8: { B t=toI16Any(c); i16* tp=i16any_ptr(t); i8* gp; cG= m_i8arrv(&gp, ia); u8* np=(u8* )gp; for (usz i=0; i<ia; i++) np[i]=tp[i]; dec(t); break; }
case sty_u16: { B t=toI32Any(c); i32* tp=i32any_ptr(t); i16* gp; cG=m_i16arrv(&gp, ia); u16* np=(u16* )gp; for (usz i=0; i<ia; i++) np[i]=tp[i]; dec(t); break; }
case sty_u32: { B t=toF64Any(c); f64* tp=f64any_ptr(t); i32* gp; cG=m_i32arrv(&gp, ia); u32* np=(u32* )gp; for (usz i=0; i<ia; i++) np[i]=tp[i]; dec(t); break; }
case sty_f32: { B t=toF64Any(c); f64* tp=f64any_ptr(t); i32* gp; cG=m_i32arrv(&gp, ia); float* np=(float*)gp; for (usz i=0; i<ia; i++) np[i]=tp[i]; dec(t); break; }
}
*(void**)(ffiTmpS+pos) = tyany_ptr(cG);
ffiObjs = vec_addN(ffiObjs, cG);
} else if (t->ty==cty_repr) { // any:any
B o2 = t->a[0].o;
u8 tre = t->a[0].extra;
u8 wre = t->a[0].extra2;
if (isC32(o2)) { // scalar:any
pos = ffiTmpAA(8);
u8 et = o2cu(o2);
u8 etw = sty_w[et]*8;
// TODO if etw==1<<wre, array isn't necessary
if (etw < 1<<wre) thrF("FFI: Representation wider than the value for %S:%c%i", sty_names[et], tre, 1<<wre);
if (!isArr(c)) thrF("FFI: Expected array corresponding to %S:%c%i", sty_names[et], tre, 1<<wre);
if (a(c)->ia != etw>>wre) thrM("FFI: Bad input array length");
B cG = toW(tre, wre, inc(c));
memcpy(ffiTmpS+pos, tyany_ptr(cG), 8); // may over-read, ¯\_(ツ)_/¯
dec(cG);
} else { // *scalar:any / &scalar:any
pos = ffiTmpAA(sizeof(void*));
if (!isArr(c)) thrM("FFI: Expected array corresponding to a pointer");
BQNFFIType* t2 = c(BQNFFIType, o2);
B ore = t2->a[0].o;
assert(t2->ty==cty_ptr && isC32(ore)); // we shouldn't be generating anything else
inc(c);
B cG;
bool mut = t->a[0].extra;
if (mut) {
Arr* cGp;
switch(wre) { default: UD;
case 0: cGp = (Arr*) cpyBitArr(c); break;
case 3: cGp = tre=='c'? (Arr*) cpyC8Arr(c) : (Arr*) cpyI8Arr(c); break;
case 4: cGp = tre=='c'? (Arr*)cpyC16Arr(c) : (Arr*)cpyI16Arr(c); break;
case 5: cGp = tre=='c'? (Arr*)cpyC32Arr(c) : (Arr*)cpyI32Arr(c); break;
case 6: cGp = (Arr*) cpyF64Arr(c); break;
}
cG = taga(cGp);
} else cG = toW(tre, wre, c);
*(void**)(ffiTmpS+pos) = tyany_ptr(cG);
ffiObjs = vec_addN(ffiObjs, cG);
}
} else thrM("FFI: Unimplemented type");
}
return pos;
}
B buildObj(BQNFFIEnt ent, B c, bool anyMut, B* objs, usz* objPos) {
if (isC32(ent.o)) return m_f64(0); // scalar
BQNFFIType* t = c(BQNFFIType, ent.o);
if (t->ty==cty_ptr) { // *any / &any
B e = t->a[0].o;
B f = objs[(*objPos)++];
bool mut = t->a[0].extra;
if (mut) {
usz ia = a(c)->ia;
switch(o2cu(e)) { default: UD;
case sty_i8: case sty_i16: case sty_i32: case sty_f64: return inc(f);
case sty_u8: { u8* tp=tyarr_ptr(f); i16* rp; B r=m_i16arrv(&rp, ia); for (usz i=0; i<ia; i++) rp[i]=tp[i]; return r; }
case sty_u16: { u16* tp=tyarr_ptr(f); i32* rp; B r=m_i32arrv(&rp, ia); for (usz i=0; i<ia; i++) rp[i]=tp[i]; return r; }
case sty_u32: { u32* tp=tyarr_ptr(f); f64* rp; B r=m_f64arrv(&rp, ia); for (usz i=0; i<ia; i++) rp[i]=tp[i]; return r; }
case sty_f32: { u32* tp=tyarr_ptr(f); f64* rp; B r=m_f64arrv(&rp, ia); for (usz i=0; i<ia; i++) rp[i]=tp[i]; return r; }
}
} else return m_f64(0);
} else if (t->ty==cty_repr) { // any:any
B o2 = t->a[0].o;
if (isC32(o2)) return m_f64(0); // scalar:any
// *scalar:any / &scalar:any
B f = objs[(*objPos)++];
bool mut = t->a[0].extra;
if (!mut) return m_f64(0);
return inc(f);
} else thrM("FFI: Unimplemented type");
}
B libffiFn_c2(B t, B w, B x) {
BoundFn* c = c(BoundFn,t);
B argObj = c(HArr,c->obj)->a[0];
B cifObj = c(HArr,c->obj)->a[1];
ffi_cif* cif = (void*) c(TAlloc,cifObj)->data;
void* sym = c->w_c2;
usz argn = cif->nargs;
ffiTmpS = ffiTmpC = ffiTmpE = NULL;
ffiTmpObj = NULL;
ffiObjs = emptyHVec();
ffiTmpX(128);
ffiTmpAA(argn * sizeof(u32));
#define argOffs ((u32*)ffiTmpS)
SGetU(x)
BQNFFIEnt* ents = c(BQNFFIType,argObj)->a;
usz mutArgs = 0;
for (usz i = 0; i < argn; i++) {
bool mut = ents[i+1].extra2;
argOffs[i] = genObj(ents[i+1], GetU(x,i), mut);
mutArgs+= mut;
}
u32 resPos;
bool simpleRes = isC32(ents[0].o);
if (simpleRes) {
resPos = ffiTmpAA(o2cu(ents[0].o)==0? sizeof(BQNV) : sizeof(ffi_arg)>8? sizeof(ffi_arg) : 8);
} else thrM("FFI: Unimplemented result type");
void** argPtrs = (void**) (ffiTmpS + ffiTmpAA(argn*sizeof(void*))); // must be the very last alloc to be able to store pointers
for (usz i = 0; i < argn; i++) argPtrs[i] = ffiTmpS + argOffs[i];
void* res = ffiTmpS + resPos;
// for (usz i = 0; i < ffiTmpC-ffiTmpS; i++) { // simple hexdump of ffiTmp
// if (!(i&15)) printf("%s%p ", i?"\n":"", (void*)(ffiTmpS+i));
// else if (!(i&3)) printf(" ");
// printf("%x%x ", ffiTmpS[i]>>4, ffiTmpS[i]&15);
// }
// printf("\n");
ffi_call(cif, sym, res, argPtrs);
B r;
bool resVoid = false;
if (simpleRes) {
switch(o2cu(ents[0].o)) { default: thrM("FFI: Unimplemented type");
case sty_void: r = m_c32(0); resVoid = true; break;
case sty_a: r = getB(*(BQNV*)res); break;
case sty_i8: r = m_i32(sizeof(ffi_arg )>sizeof( i8)? ( i8)*(ffi_arg *)res : *( i8*)res); break;
case sty_i16: r = m_i32(sizeof(ffi_arg )>sizeof(i16)? (i16)*(ffi_arg *)res : *(i16*)res); break;
case sty_i32: r = m_i32(sizeof(ffi_arg )>sizeof(i32)? (i32)*(ffi_arg *)res : *(i32*)res); break;
case sty_u8: r = m_i32(sizeof(ffi_sarg)>sizeof( u8)? ( u8)*(ffi_sarg*)res : *( u8*)res); break;
case sty_u16: r = m_i32(sizeof(ffi_sarg)>sizeof(u16)? (u16)*(ffi_sarg*)res : *(u16*)res); break;
case sty_u32: r = m_f64(sizeof(ffi_sarg)>sizeof(u32)? (u32)*(ffi_sarg*)res : *(u32*)res); break;
case sty_f32: r = m_f64(*(float* )res); break;
case sty_f64: r = m_f64(*(double*)res); break;
}
} else UD;
mm_free((Value*)ffiTmpObj);
if (mutArgs) {
usz objPos = 0;
M_HARR(ra, mutArgs+(resVoid? 0 : 1));
if (!resVoid) HARR_ADDA(ra, r);
for (usz i = 0; i < argn; i++) {
bool mut = ents[i+1].extra2;
B c = buildObj(ents[i+1], GetU(x,i), mut, harr_ptr(ffiObjs), &objPos);
if (mut) HARR_ADDA(ra, c);
}
r = HARR_FV(ra);
}
dec(w); dec(x); dec(ffiObjs);
return r;
}
B libffiFn_c1(B t, B x) { return libffiFn_c2(t, bi_N, x); }
#endif
B loadffi_c2(B t, B w, B x) {
usz xia = a(x)->ia;
if (xia<2) thrM("FFI: Function specification must have at least two items");
usz argn = xia-2;
SGetU(x)
B name = GetU(x,1);
vfyStr(name, "FFI", "type");
BQNFFIEnt tRes = ffi_parseType(GetU(x,0), true);
BQNFFIEnt* args; B argObj = m_bqnFFIType(&args, 255, argn+1);
args[0] = tRes;
for (usz i = 0; i < argn; i++) args[i+1] = ffi_parseType(GetU(x,i+2), false);
char* ws = toCStr(w);
void* dl = dlopen(ws, RTLD_NOW);
freeCStr(ws);
dec(w);
if (dl==NULL) thrF("Failed to load: %S", dlerror());
char* nameStr = toCStr(name);
void* sym = dlsym(dl, nameStr);
freeCStr(nameStr);
dec(x);
if (sym==NULL) thrF("Failed to find symbol: %S", dlerror());
#if FFI==1
return m_ffiFn(foreignFnDesc, bi_N, directFn_c1, directFn_c2, sym, sym);
#else
TAlloc* argsRaw = ARBOBJ(argn*sizeof(ffi_type*));
ffi_type** argsRawArr = (ffi_type**)argsRaw->data;
for (usz i = 0; i < argn; i++) argsRawArr[i] = &args[i+1].t;
// for (usz i = 0; i < argn; i++) {
// ffi_type c = *argsRawArr[i];
// printf("%zu %d %d %p\n", c.size, c.alignment, c.type, c.elements);
// }
TAlloc* cif = ARBOBJ(sizeof(ffi_cif));
ffi_status s = ffi_prep_cif((ffi_cif*)cif->data, FFI_DEFAULT_ABI, argn, &args[0].t, argsRawArr);
if (s!=FFI_OK) thrM("FFI: Error preparing call interface");
// mm_free(argsRaw)
return m_ffiFn(foreignFnDesc, m_hVec3(argObj, tag(cif, OBJ_TAG), tag(argsRaw, OBJ_TAG)), libffiFn_c1, libffiFn_c2, NULL, sym);
#endif
}
DEF_FREE(ffiType) {
usz ia = ((BQNFFIType*)x)->ia;
for (usz i = 0; i < ia; i++) dec(((BQNFFIType*)x)->a[i].o);
}
void ffiType_visit(Value* x) {
usz ia = ((BQNFFIType*)x)->ia;
for (usz i = 0; i < ia; i++) mm_visit(((BQNFFIType*)x)->a[i].o);
}
void ffiType_print(FILE* f, B x) {
BQNFFIType* t = c(BQNFFIType,x);
fprintf(f, "cty_%d⟨", t->ty);
for (usz i=0, ia=t->ia; i<ia; i++) {
if (i) fprintf(f, ", ");
printFFIType(f, t->a[i].o);
}
fprintf(f, "");
}
void ffi_init() {
boundFnDesc = registerNFn(m_str8l("(foreign function)"), boundFn_c1, boundFn_c2);
foreignFnDesc = registerNFn(m_str8l("(foreign function)"), directFn_c1, directFn_c2);
TIi(t_ffiType,freeO) = ffiType_freeO;
TIi(t_ffiType,freeF) = ffiType_freeF;
TIi(t_ffiType,visit) = ffiType_visit;
TIi(t_ffiType,print) = ffiType_print;
}
#else
void ffi_init() { }
B loadffi_c2(B t, B w, B x) { thrM("CBQN was compiled without FFI"); }
#endif

94
src/ffi.h Normal file
View File

@ -0,0 +1,94 @@
#include<stddef.h>
#include<stdint.h>
#include<stdbool.h>
typedef uint64_t BQNV;
// to define a native function, define one with one of the following signatures:
// BQNV yourName(BQNV x);
// BQNV yourName(BQNV w, BQNV x);
// and load it in BQN with something like `"path/to/compiled/file.so" •LoadNativeFnOrWhatever arity‿"yourName"`
#ifdef __cplusplus
extern "C" {
#endif
void bqn_free(BQNV v);
double bqn_toF64 (BQNV v); // includes bqn_free(v)
uint32_t bqn_toChar(BQNV v); // includes bqn_free(v)
double bqn_readF64 (BQNV v); // doesn't include bqn_free(v)
uint32_t bqn_readChar(BQNV v); // doesn't include bqn_free(v)
// invoke BQN function
BQNV bqn_call1(BQNV f, BQNV x);
BQNV bqn_call2(BQNV f, BQNV w, BQNV x);
// evaluate BQN code in a fresh environment
BQNV bqn_eval(BQNV src);
BQNV bqn_evalCStr(char* str); // evaluates the null-terminated UTF8-encoded str; equal to `BQNV s = bqn_makeUTF8Str(str, strlen(str)); result = bqn_eval(s); bqn_free(s);`
// read array data
size_t bqn_bound(BQNV a); // aka product of shape, ×´≢a
size_t bqn_rank(BQNV a);
void bqn_shape(BQNV a, size_t* buf); // writes bqn_rank(a) items in buf
BQNV bqn_pick(BQNV a, size_t pos);
// read all elements of `a` into the specified buffer
void bqn_readI8Arr (BQNV a, int8_t* buf);
void bqn_readI16Arr(BQNV a, int16_t* buf);
void bqn_readI32Arr(BQNV a, int32_t* buf);
void bqn_readF64Arr(BQNV a, double* buf);
void bqn_readC8Arr (BQNV a, uint8_t* buf);
void bqn_readC16Arr(BQNV a, uint16_t* buf);
void bqn_readC32Arr(BQNV a, uint32_t* buf);
void bqn_readObjArr(BQNV a, BQNV* buf);
// create objects
BQNV bqn_makeF64(double d);
BQNV bqn_makeChar(uint32_t c);
BQNV bqn_makeI8Arr (size_t rank, size_t* shape, int8_t* data);
BQNV bqn_makeI16Arr(size_t rank, size_t* shape, int16_t* data);
BQNV bqn_makeI32Arr(size_t rank, size_t* shape, int32_t* data);
BQNV bqn_makeF64Arr(size_t rank, size_t* shape, double* data);
BQNV bqn_makeC8Arr (size_t rank, size_t* shape, uint8_t* data);
BQNV bqn_makeC16Arr(size_t rank, size_t* shape, uint16_t* data);
BQNV bqn_makeC32Arr(size_t rank, size_t* shape, uint32_t* data);
BQNV bqn_makeObjArr(size_t rank, size_t* shape, BQNV* data); // frees the taken elements of data
BQNV bqn_makeI8Vec (size_t len, int8_t* data);
BQNV bqn_makeI16Vec(size_t len, int16_t* data);
BQNV bqn_makeI32Vec(size_t len, int32_t* data);
BQNV bqn_makeF64Vec(size_t len, double* data);
BQNV bqn_makeC8Vec (size_t len, uint8_t* data);
BQNV bqn_makeC16Vec(size_t len, uint16_t* data);
BQNV bqn_makeC32Vec(size_t len, uint32_t* data);
BQNV bqn_makeObjVec(size_t len, BQNV* data); // frees the taken elements of data
BQNV bqn_makeUTF8Str(size_t len, char* str);
typedef BQNV (*bqn_boundFn1)(BQNV obj, BQNV x);
typedef BQNV (*bqn_boundFn2)(BQNV obj, BQNV w, BQNV x);
// when called, 1st arg to `f` will be `obj`
BQNV bqn_makeBoundFn1(bqn_boundFn1 f, BQNV obj);
BQNV bqn_makeBoundFn2(bqn_boundFn2 f, BQNV obj);
// direct (zero copy) array item access
typedef enum { elt_i8, elt_i16, elt_i32, elt_f64, elt_c8, elt_c16, elt_c32, elt_unk } BQNElType;
BQNElType bqn_directType(BQNV a);
// can only use the functions below if bqn_elType returns the corresponding type
// a valid implementation of bqn_elType would be to always return elt_unk, thus disallowing the use of direct access entirely
int8_t* bqn_directI8 (BQNV a);
int16_t* bqn_directI16(BQNV a);
int32_t* bqn_directI32(BQNV a);
double* bqn_directF64(BQNV a);
uint8_t* bqn_directC8 (BQNV a);
uint16_t* bqn_directC16(BQNV a);
uint32_t* bqn_directC32(BQNV a);
#ifdef __cplusplus
}
#endif

11
src/h.h
View File

@ -39,6 +39,9 @@
#ifndef RANDSEED
#define RANDSEED 0 // random seed used to make •rand (0 for using time)
#endif
#ifndef FFI
#define FFI 2
#endif
// #define HEAP_VERIFY // enable usage of heapVerify()
// #define ALLOC_STAT // store basic allocation statistics
@ -212,11 +215,11 @@ typedef union B {
/*22*/ F(harr ) F(fillarr ) F(i8arr ) F(i16arr ) F(i32arr ) F(c8arr ) F(c16arr ) F(c32arr ) F(f64arr ) \
/*31*/ F(bitarr) \
\
/*32*/ F(comp) F(block) F(body) F(scope) F(scopeExt) F(blBlocks) \
/*38*/ F(ns) F(nsDesc) F(fldAlias) F(vfyObj) F(hashmap) F(temp) F(nfn) F(nfnDesc) \
/*46*/ F(freed) F(harrPartial) F(customObj) F(mmapH) \
/*32*/ F(comp) F(block) F(body) F(scope) F(scopeExt) F(blBlocks) F(arbObj) F(ffiType) \
/*40*/ F(ns) F(nsDesc) F(fldAlias) F(vfyObj) F(hashmap) F(temp) F(nfn) F(nfnDesc) \
/*48*/ F(freed) F(harrPartial) F(customObj) F(mmapH) \
\
/*49*/ IF_WRAP(F(funWrap) F(md1Wrap) F(md2Wrap))
/*51*/ IF_WRAP(F(funWrap) F(md1Wrap) F(md2Wrap))
enum Type {
#define F(X) t_##X,

View File

@ -5,7 +5,7 @@
#include "ns.h"
#include "builtins.h"
#define FOR_INIT(F) F(base) F(harr) F(mutF) F(fillarr) F(tyarr) F(hash) F(sfns) F(fns) F(arith) F(md1) F(md2) F(derv) F(comp) F(rtWrap) F(ns) F(nfn) F(sysfn) F(inverse) F(load) F(sysfnPost) F(dervPost) F(mmap)
#define FOR_INIT(F) F(base) F(harr) F(mutF) F(fillarr) F(tyarr) F(hash) F(sfns) F(fns) F(arith) F(md1) F(md2) F(derv) F(comp) F(rtWrap) F(ns) F(nfn) F(sysfn) F(inverse) F(load) F(sysfnPost) F(dervPost) F(ffi) F(mmap)
#define F(X) void X##_init(void);
FOR_INIT(F)
#undef F
@ -652,6 +652,7 @@ void base_init() { // very first init function
TIi(t_customObj,freeO) = customObj_freeO;
TIi(t_customObj,freeF) = customObj_freeF;
TIi(t_customObj,visit) = customObj_visit;
TIi(t_arbObj,visit) = noop_visit;
assert((MD1_TAG>>1) == (MD2_TAG>>1)); // just to be sure it isn't changed incorrectly, `isMd` depends on this

View File

@ -29,5 +29,6 @@
#include "../ns.c"
#include "../nfns.c"
#include "../rtwrap.c"
#include "../ffi.c"
#include "../jit/nvm.c"
#include "../main.c"

View File

@ -29,7 +29,10 @@ typedef struct TStack {
#define TSFREE(N) mm_free((Value*)N##_o);
#define TSUPD(N,AM) { N##_o = ts_e(N##_o, N##_e, AM); N = (void*)N##_o->data; }
#define TSADD(N,X) { if (N##_o->size==N##_o->cap) TSUPD(N, 1); N[N##_o->size++] = X; }
#define TSADDA(N,P,AM) { u64 n=AM; if(N##_o->size+n>N##_o->cap) TSUPD(N, n); memcpy(N+N##_o->size,P,n*N##_e); N##_o->size+= n; }
#define TSADDA(N,P,AM) { u64 n_=(AM); if(N##_o->size+n_>N##_o->cap) TSUPD(N, n_); memcpy(N+N##_o->size,P,n_*N##_e); N##_o->size+= n_; }
#define TSADDAU(N,AM) { u64 n_=(AM); if(N##_o->size+n_>N##_o->cap) TSUPD(N, n_); N##_o->size+= n_; }
#define TSFREEP(N) mm_free((void*)RFLD(N, TStack, data));
#define TSSIZE(N) (N##_o->size)
TStack* ts_e(TStack* o, u32 elsz, u64 am);
#define ARBOBJ(SZ) (TAlloc*)mm_alloc(sizeof(TAlloc)+(SZ), t_arbObj)