From 7d2a1280cd4321d2a3b2fff0b2413085347a7b4d Mon Sep 17 00:00:00 2001 From: Nicholas Noll Date: Thu, 21 May 2020 18:53:23 -0700 Subject: feat: prototype of ast stmt and decl implementations --- sys/cmd/cc/cc.h | 259 +++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 183 insertions(+), 76 deletions(-) (limited to 'sys/cmd/cc/cc.h') diff --git a/sys/cmd/cc/cc.h b/sys/cmd/cc/cc.h index 144eb4a..7c9d679 100644 --- a/sys/cmd/cc/cc.h +++ b/sys/cmd/cc/cc.h @@ -15,14 +15,18 @@ typedef struct Lexer Lexer; typedef struct Sym Sym; typedef struct Type Type; +typedef struct Scope Scope; +typedef struct Parser Parser; + +typedef struct Node Node; typedef struct Ptr Ptr; typedef struct Name Name; +typedef struct Dtor Dtor; typedef struct Decl Decl; typedef struct Stmt Stmt; typedef struct Expr Expr; -/* maps */ typedef struct SymTab SymTab; typedef struct StrTab StrTab; @@ -34,21 +38,26 @@ typedef struct Compiler Compiler; KEYWORD(Kregister,"register") \ KEYWORD(Kstatic,"static") \ KEYWORD(Kextern,"extern") \ + KEYWORD(Ktls,"thread_local") \ KEYWORD(Ktypedef,"typedef") \ KEYWORD(Kconst,"const") \ KEYWORD(Kvolatile,"volatile") \ KEYWORD(Krestrict,"restrict") \ + KEYWORD(Katomic,"_Atomic") \ KEYWORD(Kinline,"inline") \ + KEYWORD(Knoret,"_Noreturn") \ + KEYWORD(Ksigned,"signed") \ + KEYWORD(Kunsigned,"unsigned") \ KEYWORD(Kvoid,"void") \ + KEYWORD(Kbool,"_Bool") \ KEYWORD(Kchar,"char") \ - KEYWORD(Kint,"int") \ KEYWORD(Kfloat,"float") \ KEYWORD(Kdouble,"double") \ - KEYWORD(Ksigned,"signed") \ - KEYWORD(Kunsigned,"unsigned") \ + KEYWORD(Kcomplex,"complex") \ + KEYWORD(Kimaginary,"imaginary") \ + KEYWORD(Kint,"int") \ KEYWORD(Kshort,"short") \ KEYWORD(Klong,"long") \ - KEYWORD(Kvlong,"vlong") \ KEYWORD(Kstruct,"struct") \ KEYWORD(Kunion,"union") \ KEYWORD(Kenum,"enum") \ @@ -201,7 +210,7 @@ extern byte *tokens[NUM_TOKENS]; struct Token { uint32 kind; - Range pos; + Pos pos; union { byte *s; double f; @@ -215,9 +224,12 @@ struct Token enum { - Svar = 1 << 0, - Sfunc = 1 << 1, - Smacro = 1 << 2, + Sobj = iota(0), + Stype = iota(2), + Stag = iota(3), + Senum = iota(4), + Sstmt = iota(5), + Smacro = iota(6), }; struct Sym @@ -226,7 +238,9 @@ struct Sym string name; union { string macro; - /*Func *func;*/ + Decl *obj; + Type *tag; + Stmt *blk; }; }; @@ -244,6 +258,7 @@ struct SymTab Sym *define(SymTab *tab, string ident, int kind); Sym *lookup(SymTab *tab, string ident); error forget(SymTab *tab, string ident); +void forgetall(SymTab *tab); struct Lexer { @@ -276,6 +291,7 @@ void puttok(Token); // parsing & type resolution // tokens -> ast +/* parent data */ struct Node { Range pos; @@ -287,9 +303,9 @@ enum { Nbad, /* labels */ - Slabel, Scase, + Sempty, Slabel, Scase, Sblock, - Sexpr, + Sexpr, Sdecl, Sselect, /* loops */ Sfor, Swhile, Sdo, @@ -317,8 +333,17 @@ enum /* lists */ Xcomma, + Dfunc, + Dtype, Dvar, + Dvars, + + /* names (shouldn't interact with base AST node enumeration */ + Nident = iota(0), + Nparen = iota(1), + Nindex = iota(2), + Ncall = iota(3), }; /* statements */ @@ -331,7 +356,7 @@ struct Stmt string ident; Expr *x; }; - Stmt *stmt; + Node *stmt; } lbl; struct { long n; @@ -339,13 +364,10 @@ struct Stmt } blk; Expr *x; struct { - union { - Expr *x; - Decl *d; - } init; + Node *init; Expr *cond; Expr *step; - Stmt *body; + Node *body; } loop; union{ string lbl; @@ -353,9 +375,9 @@ struct Stmt } jmp; struct { Expr *cond; - Stmt *body; - Stmt *orelse; - } sel; + Node *body; + Node *orelse; + } br; }; }; @@ -363,6 +385,7 @@ struct Stmt struct Expr { struct Node; + uint64 type; union { struct { Expr *l; @@ -403,83 +426,162 @@ struct Name { struct Node; union { - string ident; - struct Dtor *paren; - struct { - uint32 kind; - Name *base; - union { - Expr *index; - /* TODO: - * func params - * variadic arrays - * compound set notation - */ - }; - } suffix; + string ident; + struct Dtor *paren; + }; + union { + Expr *i; + Expr *p; }; }; struct Dtor { - Range pos; Ptr ptr; Name name; }; // specifiers +/* + * the design is the following: + * type info is held w/in a 64 bit integer. + * the bottom 32 bits are associated to specializations + * the top 32 bits index into a type-info array held by the compiler. + */ enum { /* memory */ - Mauto = iota(0), - Mtype = iota(1), - Mstatic = iota(2), - Mreg = iota(3), - Mtls = iota(4), + Mauto = iota(Kauto), + Mstatic = iota(Kstatic), + Mreg = iota(Kregister), + Mtls = iota(Ktls), + Mtype = iota(Ktypedef), + + MaskMem = Mauto | Mstatic | Mreg | Mtls | Mtype, /* qualifiers */ - Qconst = iota(5), - Qrestr = iota(6), - Qvoltl = iota(7), - Qatom = iota(8), + Qconst = iota(Kconst), + Qrestr = iota(Krestrict), + Qvoltl = iota(Kvolatile), + Qatom = iota(Katomic), + + MaskQul = Qconst | Qrestr | Qvoltl | Qatom, - Finlne = iota(9), - Fnoret = iota(10), + Finlne = iota(Kinline), + Fnoret = iota(Knoret), + + MaskFcn = Finlne | Fnoret, /* types */ - Tlong = iota(12), - Tvlong = iota(13), - Tsign = iota(14), - Tunsign = iota(15), - - Tvoid = iota(16), - Tchar = iota(17), - Tshort = iota(18), - Tfloat = iota(19), - Tdouble = iota(20), - Tcmplx = iota(21), - Timag = iota(22), - - Tatom = iota(23), - - Taggr = iota(24), - Tenum = iota(25), - Tname = iota(26), - - /* alignment */ - Alnas = iota(27), + Tsign = iota(Ksigned), + Tunsign = iota(Kunsigned), + + MaskSgn = Tsign | Tunsign, + + Tvoid = iota(Kvoid), + + Tfloat = iota(Kfloat), + Tdouble = iota(Kdouble), + Tcmplx = iota(Kcomplex), + Timag = iota(Kimaginary), + + MaskFlt = Tfloat | Tdouble | Tcmplx | Timag, + + Tchar = iota(Kchar), + Tbool = iota(Kbool), + + Tshort = iota(Kshort), + Tint = iota(Kint), + Tlong = iota(Klong), + Tvlong = iota(Klong+1), + + MaskInt = Tshort | Tint | Tlong | Tvlong, + MaskTyp = Tvoid | Tbool | Tchar | Tint | Tfloat | Timag | Tcmplx, + /* + * NOTE IMPORTANT: vlong takes over the struct bit place + * DON'T MOVE KEYWORDS WITHOUT REORGANIZING + */ + Tstruct = iota(Kstruct+1), + Tunion = iota(Kunion+1), + Tenum = iota(Kenum+1), + Tname = iota(Kenum+2), + + Sbad = -1, +}; + +struct Decls +{ + union { + struct Dtor; + Dtor dtor; + }; + Expr *init; + struct Decls *link; }; struct Decl { struct Node; + uint64 spec; union { + struct { + struct Dtor; + Stmt *body; + } func; + struct { + struct Dtor; + Expr *init; + struct Decls *link; + } var; + struct Dtor type; + }; +}; +/* types */ +struct Type +{ + Sym *sym; + string ident; + uintptr size; + uint16 align; + union { + struct { + int len; + Dtor *d; + Expr *x; + } su; + struct { + int len; + string *s; + Expr *x; + } en; }; }; +struct Scope +{ + SymTab tags; + SymTab objs; +}; + +struct Parser +{ + Token tok[2]; + struct { + int cap; + int len; + Decl **decls; + } ast; + + Scope *sp; + Scope spstk[20]; +}; + +/* ast.c functions */ +void parse(Parser *, Lexer *); + // ----------------------------------------------------------------------- -// compiler +// global compiler enum { @@ -525,9 +627,18 @@ string internview(byte* beg, byte *end); /* main data */ struct Compiler { + /* storage */ mem·Arena *heap; StrTab strs; + string outfile; + + struct { + int cap; + int len; + Type *info; + } type; + /* i/o data */ struct { int cap; int len; @@ -542,15 +653,11 @@ struct Compiler string *path; } omit; + /* partitioned data for stages */ Lexer lxr; - - string outfile; - struct { - int cap; - int len; - Decl *decls; - } ast; + Parser psr; }; + extern Compiler C; /* cc.c compiler functions */ -- cgit v1.2.1