Cleanup (reactorlabs#1205)

* update gnur

* Cleanup

Remove out-of-sync structs and use gnur's
Use Rf_ prefix for all R internals we call
Try to deal with the #define eval Rf_eval idiom

* some includes and formatting

* delete unused files

* small refactoring

* try sanitizer

* missing include

* fix perf regression

Author: JanJecmen

.vscode/c_cpp_properties.json

@@ -9,7 +9,8 @@
                 "${workspaceFolder}/external/llvm-12.0.0.src/include/**"
             ],
             "defines": [
-                "ENABLE_SLOWASSERT"
+                "ENABLE_SLOWASSERT",
+                "R_NO_REMAP"
             ],
             "cStandard": "c11",
             "cppStandard": "c++14",

external/custom-r

@@ -2,136 +2,31 @@
 #define RIR_FUNTAB_H
 
 #include "BuiltinIds.h"
-
 #include "R/r.h"
 
 #include <cassert>
 
-typedef struct sxpinfo_struct_rjit {
-    unsigned int type : 5; /* ==> (FUNSXP == 99) %% 2^5 == 3 == CLOSXP
-                        * -> warning: `type' is narrower than values
-                        *              of its type
-                        * when SEXPTYPE was an enum */
-    unsigned int obj : 1;
-    unsigned int named : 2;
-    unsigned int gp : 16;
-    unsigned int mark : 1;
-    unsigned int debug : 1;
-    unsigned int trace : 1; /* functions and memory tracing */
-    unsigned int spare : 1; /* currently unused */
-    unsigned int gcgen : 1; /* old generation number */
-    unsigned int gccls : 3; /* node class */
-} sxpifo_struct_rjit;       /*		    Tot: 32 */
-
-typedef struct cons_rjit {
-    SEXP car;
-    SEXP cdr;
-    SEXP tag;
-} cons_rjit;
-
-typedef struct sexprec_rjit {
-    struct sxpinfo_struct_rjit sxpinfo;
-    SEXP attrib;
-    SEXP gengc_next_node, gengc_prev_node;
-    union {
-        struct cons_rjit cons;
-        int i;
-    } u;
-} sexprec_rjit;
-
-typedef SEXP (*CCODE)(SEXP, SEXP, SEXP, SEXP);
-
-/* Information for Deparsing Expressions */
-typedef enum {
-    PP_INVALID = 0,
-    PP_ASSIGN = 1,
-    PP_ASSIGN2 = 2,
-    PP_BINARY = 3,
-    PP_BINARY2 = 4,
-    PP_BREAK = 5,
-    PP_CURLY = 6,
-    PP_FOR = 7,
-    PP_FUNCALL = 8,
-    PP_FUNCTION = 9,
-    PP_IF = 10,
-    PP_NEXT = 11,
-    PP_PAREN = 12,
-    PP_RETURN = 13,
-    PP_SUBASS = 14,
-    PP_SUBSET = 15,
-    PP_WHILE = 16,
-    PP_UNARY = 17,
-    PP_DOLLAR = 18,
-    PP_FOREIGN = 19,
-    PP_REPEAT = 20
-} PPkind;
-
-typedef enum {
-    PREC_FN = 0,
-    PREC_LEFT = 1,
-    PREC_EQ = 2,
-    PREC_RIGHT = 3,
-    PREC_TILDE = 4,
-    PREC_OR = 5,
-    PREC_AND = 6,
-    PREC_NOT = 7,
-    PREC_COMPARE = 8,
-    PREC_SUM = 9,
-    PREC_PROD = 10,
-    PREC_PERCENT = 11,
-    PREC_COLON = 12,
-    PREC_SIGN = 13,
-    PREC_POWER = 14,
-    PREC_DOLLAR = 15,
-    PREC_NS = 16,
-    PREC_SUBSET = 17
-} PPprec;
+static inline int getBuiltinNr(SEXP f) { return f->u.primsxp.offset; }
 
-typedef struct {
-    PPkind kind;             /* deparse kind */
-    PPprec precedence;       /* operator precedence */
-    unsigned int rightassoc; /* right associative? */
-} PPinfo;
-
-/* The type definitions for the table of built-in functions. */
-/* This table can be found in ../main/names.c */
-typedef struct {
-    char* name;  /* print name */
-    CCODE cfun;  /* c-code address */
-    int code;    /* offset within c-code */
-    int eval;    /* evaluate args? */
-    int arity;   /* function arity */
-    PPinfo gram; /* pretty-print info */
-} FUNTAB;
-
-extern FUNTAB R_FunTab[];
-
-/** Returns us the CCODE object from R_FunTab based on name.
-
-  TODO This exists in gnu-r (names.c), when integrated inside, we want to make
-  use of it.
- */
 static inline CCODE getBuiltin(SEXP f) {
-    int i = ((sexprec_rjit*)f)->u.i;
-    return R_FunTab[i].cfun;
+    return R_FunTab[getBuiltinNr(f)].cfun;
 }
-static inline int getBuiltinNr(SEXP f) { return ((sexprec_rjit*)f)->u.i; }
+
 static inline const char* getBuiltinName(int i) { return R_FunTab[i].name; }
-static inline int getBuiltinArity(SEXP f) { return R_FunTab[getBuiltinNr(f)].arity; }
-static inline int getFlag(SEXP f) {
-    int i = ((sexprec_rjit*)f)->u.i;
-    return (((R_FunTab[i].eval) / 100) % 10);
+static inline const char* getBuiltinName(SEXP f) {
+    return getBuiltinName(getBuiltinNr(f));
 }
-static inline int getFlag(int i) { return (((R_FunTab[i].eval) / 100) % 10); }
 
-static inline bool builtinUpdatesVisibility(int id) { return getFlag(id) < 2; }
-static inline bool builtinVisibility(int id) {
-    assert(builtinUpdatesVisibility(id));
-    return getFlag(id) != 1;
+static inline int getBuiltinArity(SEXP f) {
+    return R_FunTab[getBuiltinNr(f)].arity;
 }
 
+static inline int getFlag(int i) { return ((R_FunTab[i].eval) / 100) % 10; }
+static inline int getFlag(SEXP f) { return getFlag(getBuiltinNr(f)); }
+
 static inline SEXP getBuiltinFun(char const* name) {
-    assert(R_FunTab[rir::blt(name)].eval % 10 == 1 && "Only use for BUILTINSXP");
+    assert(R_FunTab[rir::blt(name)].eval % 10 == 1 &&
+           "Only use for BUILTINSXP");
     if (R_FunTab[rir::blt(name)].eval % 100 / 10 == 0)
         return Rf_install(name)->u.symsxp.value;
     else

rir/src/R/Funtab.h

@@ -3,39 +3,42 @@
 
 #include "common.h"
 
+#define R_NO_REMAP
+#define USE_RINTERNALS
 #include <R.h>
+#include <R_ext/Print.h>
 #include <Rinterface.h>
-#define USE_RINTERNALS
 #include <Rinternals.h>
 
-// r print statement
-#include <R_ext/Print.h>
-
-#undef error
-#undef TRUE
-#undef FALSE
-#undef length
-#undef eval
-#undef cons
-
+// Use the function versions (some of the names clash with LLVM)
 #undef isNull
-inline bool isNull(SEXP s) { return TYPEOF(s) == NILSXP; }
+#undef isSymbol
+#undef isLogical
+#undef isReal
 #undef isComplex
-inline bool isComplex(SEXP s) { return TYPEOF(s) == CPLXSXP; }
+#undef isExpression
+#undef isEnvironment
 #undef isString
-inline bool isString(SEXP s) { return TYPEOF(s) == STRSXP; }
-
 #undef isObject
-inline bool isObject(SEXP s) { return OBJECT(s) != 0; }
-
-#undef isVector
-
+inline bool Rf_isNull(SEXP s) { return TYPEOF(s) == NILSXP; }
+inline bool Rf_isSymbol(SEXP s) { return TYPEOF(s) == SYMSXP; }
+inline bool Rf_isLogical(SEXP s) { return TYPEOF(s) == LGLSXP; }
+inline bool Rf_isReal(SEXP s) { return TYPEOF(s) == REALSXP; }
+inline bool Rf_isComplex(SEXP s) { return TYPEOF(s) == CPLXSXP; }
+inline bool Rf_isExpression(SEXP s) { return TYPEOF(s) == EXPRSXP; }
+inline bool Rf_isEnvironment(SEXP s) { return TYPEOF(s) == ENVSXP; }
+inline bool Rf_isString(SEXP s) { return TYPEOF(s) == STRSXP; }
+inline bool Rf_isObject(SEXP s) { return OBJECT(s) != 0; }
+
+// Clash with LLVM
 #undef PI
 
+// Get rid of the macro version of this
+#undef R_CheckStack
+
+// Bypass PREXPR from GNU R which causes code objects to be converted to AST
 #undef PREXPR
 inline SEXP PREXPR(SEXP pr) {
-    // bypassing PREXPR from Gnur, which causes code objects to be converted to
-    // AST
     SLOWASSERT(TYPEOF(pr) == PROMSXP);
     auto res = pr->u.promsxp.expr;
     if (TYPEOF(res) == BCODESXP)
@@ -44,10 +47,11 @@ inline SEXP PREXPR(SEXP pr) {
 }
 
 extern "C" {
+extern FUNTAB R_FunTab[];
 extern SEXP R_TrueValue;
 extern SEXP R_FalseValue;
 extern SEXP R_LogicalNAValue;
-};
+}
 
 // Performance critical stuff copied from Rinlinedfun.h
 
@@ -119,5 +123,3 @@ extern int R_PPStackTop;
     } while (0)
 
 #endif
-
-#undef R_CheckStack

rir/src/R/RVector.cpp

@@ -3,7 +3,6 @@
  */
 
 #include "api.h"
-#include "R/Funtab.h"
 #include "R/Serialize.h"
 #include "bc/BC.h"
 #include "bc/Compiler.h"

rir/src/R/RVector.h

@@ -4,6 +4,7 @@
 #include "R/r.h"
 #include "compiler/log/debug.h"
 #include "runtime/Context.h"
+
 #include <stdint.h>
 
 #define REXPORT extern "C"