strlits

7 files changed, 85 insertions, 20 deletions

diff --git a/ir.c b/ir.c
index cb4cc27..2f1c622 100644
--- a/ir.c
+++ b/ir.c
@@ -4,12 +4,12 @@ uchar type2cls[NTYPETAG];
 uchar cls2siz[KF8+1];
 const uchar siz2intcls[] = { [1] = KI4, [2] = KI4, [4] = KI4, [8] = KI8 };
 
-static vec_of(struct irdat) dats;
 struct instr instrtab[1<<14];
 static int ninstr;
 static int instrfreelist;
 struct calltab calltab;
 struct phitab phitab;
+struct dattab dattab;
 
 void
 irinit(struct function *fn)
@@ -59,14 +59,6 @@ addcon(const struct xcon *con)
    }
 }
 
-/* union ref
-adddat(struct function *fn, const struct irdat *dat)
-{
-   assert(dats.n < 1u<<29);
-   vpush(&dats, *dat);
-   return mkref(RDAT, dats.n - 1); 
-} */
-
 static void
 targwrite2(uchar *d, ushort x)
 {
@@ -163,6 +155,19 @@ mksymref(struct function *fn, const char *s)
    return mkref(RXCON, addcon(&con));
 }
 
+union ref
+mkdatref(struct function *fn, uint siz, uint align, const void *bytes, uint n)
+{
+   struct irdat dat = { align, 0, siz };
+   if (siz <= 8) memcpy(dat.sdat, bytes, n < siz ? n : siz);
+   else {
+      while (((uchar *)bytes)[n-1] == 0) --n; /* nip trailing zeroes */
+      vpushn(&dat.dat, bytes, n);
+   }
+   vpush(&dattab, dat);
+   return mkref(RDAT, dattab.n - 1); 
+}
+
 struct instr
 mkalloca(uint siz, uint align)
 {
@@ -220,6 +225,7 @@ insertinstr(struct block *blk, int idx, struct instr ins)
    else {
       assert(idx >= 0 && idx < blk->ins.n);
       vpush_((void **)&blk->ins.p, &blk->ins._cap, &blk->ins.n, sizeof *blk->ins.p);
+      vresize(&blk->ins, blk->ins.n);
       for (int i = blk->ins.n++; i > idx; --i)
          blk->ins.p[i] = blk->ins.p[i - 1];
       blk->ins.p[idx] = new;
diff --git a/ir.h b/ir.h
index a08bbce..4cce899 100644
--- a/ir.h
+++ b/ir.h
@@ -65,12 +65,13 @@ struct phi {
 
 enum refkind {
    RNONE,
-   RTMP,  /* reference to another instruction's result */
-   RPARAM,  /* function param */
-   RICON, /* small integer constants */
-   RXCON, /* other constants (incl. external symbols) */
+   RTMP,   /* reference to another instruction's result */
+   RPARAM, /* function param */
+   RICON,  /* small integer constants */
+   RXCON,  /* other constants (incl. external symbols) */
+   RDAT,   /* reference to irdat */
    RMORE,  /* reference to extra data for Ocall and Ophi */
-   RREG,  /* machine register */
+   RREG,   /* machine register */
 };
 
 union ref {
@@ -163,6 +164,7 @@ extern const uchar siz2intcls[];
 extern struct instr instrtab[];
 extern struct calltab {vec_of(struct call);} calltab;
 extern struct phitab {vec_of(struct phi);} phitab;
+extern struct dattab {vec_of(struct irdat);} dattab;
 #define NOREF ((union ref) {0})
 #define mkref(t, x) ((union ref) {{ (t), (x) }})
 #define mkzerocon() ((union ref) {{ RICON, 0 }})
@@ -173,6 +175,7 @@ union irtype mkirtype(union type);
 union ref mkintcon(struct function *, enum irclass, vlong);
 union ref mkfltcon(struct function *, enum irclass, double);
 union ref mksymref(struct function *, const char *);
+union ref mkdatref(struct function *, uint siz, uint align, const void *, uint n);
 struct instr mkalloca(uint siz, uint align);
 void conputdat(struct irdat *, uint off, enum typetag t, const void *dat);
 union ref mkcallarg(struct function *, bool sret, uint narg, int vararg, union ref *, union irtype *);
diff --git a/irdump.c b/irdump.c
index aa815e4..1e5302e 100644
--- a/irdump.c
+++ b/irdump.c
@@ -1,6 +1,49 @@
 #include "ir.h"
 
 extern struct xcon conht[];
+static int nextdat;
+
+#define aisprint(c) in_range(c, ' ', '~')
+
+static void
+pridat(const struct irdat *dat)
+{
+   efmt("dat ^%d(align %d, size %d):\n\t", dat - dattab.p, dat->align, dat->siz);
+   assert(!dat->syms);
+   if (dat->siz <= 8) {
+      efmt("b ");
+      for (int i = 0; i < dat->siz; ++i)
+         efmt("%d,", dat->sdat[i]);
+   } else {
+      enum {
+         MINZERO = 4,
+         MAXLINE = 60,
+      };
+      int npri = 0;
+      int nzero = dat->siz - dat->dat.n;
+      int strbegin = 0, nstr = 0;
+      for (int i = 0; i < dat->dat.n + (nzero & -(nzero <= MINZERO)); ++i) {
+         int c = i < dat->dat.n ? dat->dat.p[i] : 0;
+         if (npri > MAXLINE) {
+            npri = 0;
+            efmt("\n\t");
+         }
+         if (aisprint(c)) {
+            if (!nstr++) strbegin = i;
+         } else {
+            if (nstr) {
+               npri += efmt("asc %'S,", dat->dat.p+strbegin, nstr);
+               nstr = 0;
+               efmt("b ");
+            }
+            npri += efmt("%d,", c);
+         }
+      }
+      if (nstr) npri += efmt("asc %'S,", dat->dat.p+strbegin, nstr);
+      if ((nzero -= MINZERO) > 0) efmt("z %d", nzero);
+   }
+   efmt("\n");
+}
 
 static const char *clsname[] = {
    "?", "i4", "i8", "ptr", "f4", "f8"
@@ -51,6 +94,9 @@ dumpref(enum op o, union ref ref)
       default: assert(0);
       }
       break;
+   case RDAT:
+      efmt("$^%d", ref.i);
+      break;
    case RMORE:
       if (o == Ocall || o == Ointrin) {
          struct call *call = &calltab.p[ref.idx];
@@ -74,7 +120,7 @@ dumpref(enum op o, union ref ref)
          struct phi *phi = &phitab.p[ref.idx];
          for (int i = 0; i < phi->n; ++i) {
             if (i > 0) efmt(", ");
-            efmt("[.L%d ", phi->blk[i]->id);
+            efmt("[@%d ", phi->blk[i]->id);
             dumpref(0, phi->ref[i]);
             efmt("]");
          }
@@ -120,7 +166,7 @@ dumpblk(struct function *fn, struct block *blk)
 {
    static const char *jnames[] = { 0, "b", "ret" };
    int i;
-   efmt("  .L%d:\n", blk->id);
+   efmt("  @%d:\n", blk->id);
    for (i = 0; i < blk->phi.n; ++i) {
       dumpinst(&instrtab[blk->phi.p[i]]);
    }
@@ -138,8 +184,8 @@ dumpblk(struct function *fn, struct block *blk)
       dumpref(0, blk->jmp.arg[i]);
    }
    if (i && blk->s1) efmt(", ");
-   if (blk->s1 && blk->s2) efmt(".L%d, .L%d", blk->s1->id, blk->s2->id);
-   else if (blk->s1) efmt(".L%d", blk->s1->id);
+   if (blk->s1 && blk->s2) efmt("@%d, @%d", blk->s1->id, blk->s2->id);
+   else if (blk->s1) efmt("@%d", blk->s1->id);
    efmt("\n");
 }
 
@@ -148,6 +194,9 @@ irdump(struct function *fn, const char *fname)
 {
    struct block *blk;
 
+   /* print datas that have never been printed before */
+   while (nextdat < dattab.n) pridat(&dattab.p[nextdat++]);
+
    efmt("function %s : %ty\n", fname, fn->fnty);
    if (fn->abiarg || fn->nabiret) {
       efmt("abi: (");
diff --git a/parse.c b/parse.c
index 4ed63c7..0dd984d 100644
--- a/parse.c
+++ b/parse.c
@@ -977,7 +977,6 @@ exprvalue(struct function *fn, const struct expr *ex)
 {
    return compileexpr(fn, ex, /*discard*/ 0);
 }
-
 static inline void
 expreffects(struct function *fn, const struct expr *ex)
 {
@@ -1025,6 +1024,8 @@ expraddr(struct function *fn, const struct expr *ex)
          assert(0);
       }
       break;
+   case ESTRLIT:
+      return mkdatref(fn, ex->s.n+1, /*align*/ 1, ex->s.p, ex->s.n);
    case EDEREF:
       return exprvalue(fn, ex->sub);
    case EGETF:
@@ -1352,7 +1353,8 @@ compileexpr(struct function *fn, const struct expr *ex, bool discard)
    sub = ex->sub;
 
    if (ex->ty.t != TYVOID && !isscalar(ex->ty))
-      /* fn & array designators evaluate to their address; so do aggregates for the purpose of code generation */
+      /* fn & array designators evaluate to their address;
+       * so do aggregates for the purpose of code generation */
       return expraddr(fn, ex);
    switch (ex->t) {
    case ENUMLIT:
diff --git a/test/hello.c b/test/hello.c
new file mode 100644
index 0000000..98a93a0
--- /dev/null
+++ b/test/hello.c
@@ -0,0 +1,5 @@
+int printf(const char *, ...);
+
+int main() {
+    printf("hello world\n");
+}
diff --git a/test.c b/test/test.c
index 40b59e6..40b59e6 100644
--- a/test.c
+++ b/test/test.c
diff --git a/test2.c b/test/test2.c
index 7d39164..7d39164 100644
--- a/test2.c
+++ b/test/test2.c