fmtcheck: use symbolic label names by default

Actual program counters may differ between different architecture, so avoid
using them in the testsuite.

* mh/fmtcheck.c: New option -pc requires printing program counters.  Unless
this option is used, -disass prints only labels where needed.
* mh/mh.h (mh_format_dump_disass): Change signature.
* mh/mh_format.c (mh_format_dump_disass): Take additional flag indicating
whether to print program counters.  Compute and print program labels, unless
this flag is non-zero.
* mh/tests/fmtcomp.at: Don't expect program counters in disassembled code.

4 files changed, 286 insertions, 123 deletions

diff --git a/mh/fmtcheck.c b/mh/fmtcheck.c
index 8c8509744..b0d7d37e2 100644
--- a/mh/fmtcheck.c
+++ b/mh/fmtcheck.c
@@ -31,6 +31,7 @@ static int debug_option;
 static char *input_file;
 static size_t width;
 static size_t msgno;
+static int pc_option;
 
 void
 opt_formfile (struct mu_parseopt *po, struct mu_option *opt, char const *arg)
@@ -58,12 +59,15 @@ static struct mu_option options[] = {
   { "format",  0, N_("FORMAT"), MU_OPTION_DEFAULT,
     N_("use this format string"),
     mu_c_string, NULL, opt_format },
-  { "dump",    0, NULL,     MU_OPTION_HIDDEN,
+  { "dump",    0, NULL,     MU_OPTION_DEFAULT,
     N_("dump the listing of compiled format code"),
     mu_c_bool,   &dump_option },
-  { "disassemble",    0, NULL,     MU_OPTION_HIDDEN,
+  { "disassemble",    0, NULL,     MU_OPTION_DEFAULT,
     N_("dump disassembled format code"),
     mu_c_bool,   &disass_option },
+  { "pc",      0, NULL,     MU_OPTION_DEFAULT,
+    N_("print program counter along with disassembled code (implies --disassemble)"),
+    mu_c_bool,   &pc_option },
   { "debug",   0, NULL,     MU_OPTION_DEFAULT,
     N_("enable parser debugging output"),
     mu_c_bool,   &debug_option },
@@ -94,6 +98,8 @@ int
 main (int argc, char **argv)
 {
   mh_getopt (&argc, &argv, options, 0, args_doc, prog_doc, NULL);
+  if (pc_option)
+    disass_option = 1;
   switch (argc)
     {
     case 0:
@@ -123,7 +129,7 @@ main (int argc, char **argv)
   if (dump_option)
     mh_format_dump_code (format);
   if (disass_option)
-    mh_format_dump_disass (format);
+    mh_format_dump_disass (format, pc_option);
 
   if (input_file)
     run ();
diff --git a/mh/mh.h b/mh/mh.h
index 312f9b0bc..c1d8ee85a 100644
--- a/mh/mh.h
+++ b/mh/mh.h
@@ -182,7 +182,7 @@ void mh_fvm_run (mh_fvm_t fvm, mu_message_t msg, size_t msgno);
 int mh_format_str (mh_format_t fmt, char *str, size_t width, char **pret);
 
 void mh_format_dump_code (mh_format_t fmt);
-void mh_format_dump_disass (mh_format_t fmt);
+void mh_format_dump_disass (mh_format_t fmt, int addr);
 
 #define MH_FMT_PARSE_DEFAULT 0
 #define MH_FMT_PARSE_TREE 0x01
diff --git a/mh/mh_format.c b/mh/mh_format.c
index 86eeb4ff3..0f1d4b728 100644
--- a/mh/mh_format.c
+++ b/mh/mh_format.c
@@ -1896,8 +1896,128 @@ _get_builtin_name (mh_builtin_fp ptr)
   return NULL;
 }
 
+/* Label array is used when disassembling the code, in order to create.
+   meaningful label names.  The array elements starting from index 1 keep
+   the code addesses to which branch instructions point, in ascending order.
+   Element 0 keeps the number of addresses stored.  Thus, the label
+   array LAB with contents { 3, 2, 5, 7 } declares three labels: "L1" on
+   address 2, "L2", on address 5, and "L3" on address 7.
+*/
+
+/* Find in LAB the index of the label corresponding to the given PC.  Return
+   0 if no label found. */
+size_t
+find_label (size_t *lab, size_t pc)
+{
+  if (lab)
+    {
+      size_t i;
+      for (i = 1; i <= lab[0]; i++)
+        {
+          if (lab[i] == pc)
+            return i;
+        }
+    }
+  return 0;
+}
+
+static int
+comp_pc (const void *a, const void *b)
+{
+  size_t pca = *(size_t*)a;
+  size_t pcb = *(size_t*)b;
+  if (pca < pcb)
+    return -1;
+  else if (pca > pcb)
+    return 1;
+  return 0;
+}
+
+/* Extract a label array from a compiled format FMT. */
+static size_t *
+extract_labels (mh_format_t fmt)
+{
+  size_t *lab;
+  size_t pc;
+  long n;
+  
+  lab = mu_calloc (fmt->progcnt, sizeof (lab[0]));
+  lab[0] = 0;
+  for (pc = 1; pc < fmt->progcnt; )
+    {
+      mh_opcode_t opcode = MHI_OPCODE (fmt->prog[pc++]);
+      if (opcode == mhop_stop)
+        break;
+      switch (opcode)
+        {
+        case mhop_branch:
+        case mhop_brzn:
+        case mhop_brzs:
+          n = MHI_NUM (fmt->prog[pc++]);
+          if (!find_label (lab, pc + n - 1))
+            lab[++lab[0]] = pc + n - 1;
+          break;
+          
+        case mhop_setn:
+          pc += 2;
+          break;
+
+        case mhop_sets:
+        case mhop_ldcomp:
+          pc += 2 + MHI_NUM (fmt->prog[pc + 1]);
+          break;
+
+        case mhop_movn:
+        case mhop_movs:
+          pc += 2;
+          break;
+
+        case mhop_ldbody:
+        case mhop_call:
+        case mhop_fmtspec:
+          pc++;
+          break;
+
+        case mhop_printlit:
+          pc += 1 + MHI_NUM (fmt->prog[pc]);
+          break;
+
+        case mhop_atoi:
+        case mhop_itoa:
+        case mhop_printn:
+        case mhop_prints:
+          break;
+
+        default:
+          abort ();
+        }
+    }
+  if (lab[0] > 0)
+    qsort (lab + 1, lab[0], sizeof lab[0], comp_pc);
+  return lab;
+}
+
+/* Print to *PBUF (of size *PSZ) the label corresponding to the address PC.
+   If there's no label having this address (in particular, if LAB==NULL),
+   format the address itself to *PBUF.
+   Reallocate *PBUF, updating *PSZ, if necessary.
+*/
 void
-mh_format_dump_disass (mh_format_t fmt)
+format_label (size_t *lab, size_t pc, char **pbuf, size_t *psz)
+{
+  size_t ln = find_label (lab, pc);
+  if (ln)
+    mu_asnprintf (pbuf, psz, "L%ld", (long) ln);
+  else
+    mu_asnprintf (pbuf, psz, "%ld", (long) pc);
+}
+
+/* Dump disassembled code of FMT to stdout.  If ADDR is 0, print label names
+   where necessary, otherwise, prefix each line of output with its program
+   counter in decimal.
+*/
+void
+mh_format_dump_disass (mh_format_t fmt, int addr)
 {
   mh_instr_t *prog = fmt->prog;
   size_t pc = 1;
@@ -1908,14 +2028,46 @@ mh_format_dump_disass (mh_format_t fmt)
     [R_ACC] = "acc"
   };
   static char c_trans[] = "\\\\\"\"a\ab\bf\fn\nr\rt\tv\v";
+  size_t *lab;
+  size_t lc;
+  char *lbuf = NULL;
+  size_t lsz = 0;
   
   if (!prog)
     return;
+
+  if (!addr)
+    lab = extract_labels (fmt);
+  else
+    lab = NULL;
+  lc = lab ? 1 : 0;
+
   while (!stop)
     {
       mh_opcode_t opcode;