Changeset 2431 for trunk

Timestamp:

Nov 13, 2005, 3:27:34 AM (20 years ago)

Author:

bird

Message:

Fixed problems with symbol truncation.
Kudos to Yuri for tracking down the places which required attention.

Location:

trunk/src/emx

Files:

• ChangeLog.LIBC (modified) (1 diff)
• src/emxomf/emxomf.c (modified) (14 diffs)

trunk/src/emx/ChangeLog.LIBC

@@ -4 +4 @@
 
 2005-11-12: knut st. osmundsen <bird-gccos2-spam@anduin.net>
+    - emxomf:
+        o Fixed problems with symbol truncation. 
+          Kudos to Yuri for tracking down the places which required attention.
     - libc:
         o Fixed bug in the sigsuspend (includes pause ++) handling,

trunk/src/emx/src/emxomf/emxomf.c

@@ -35 +35 @@
 #include <sys/time.h>
 #include <ar.h>
+#include <assert.h>
 
 /* Insert private header files. */
@@ -51 +52 @@
 #define SF_FAR16        0x01    /* 16:16 pointer */
 
+/* Length of the weak marker. */
+
+#define SYMBOL_WEAK_LENGTH  (3 + 11 + 6)
+
+/* Length of the OMF symbol hash */
+
+#define SYMBOL_HASH_LENGTH  (2 + 6)
+
+/* Max OMF symbol length */
+
+#define SYMBOL_MAX_LENGTH   (255 - SYMBOL_HASH_LENGTH - SYMBOL_WEAK_LENGTH)
 /* This structure holds additional data for symbols. */
 
@@ -240 +252 @@
 /* Remove underscores from all symbol names */
 static int opt_rmunder = FALSE;
-
-/* Add timestamp to symbol to make it more unique. */
-static int opt_weakts = TRUE;
 
 /* This is the page size for OMF libraries.  It is set by the -p
@@ -318 +327 @@
 static char base_dir[256];
 
-/* This variable holds the simple weak marker for the current module. See get_mod_name(). */
-static char weak_marker_simple[128];
-static int  weak_marker_simple_size;
-
-/* This variable holds the timestamped weak marker for the current module. See get_mod_name(). */
-static char weak_marker_ts[148];
-static int  weak_marker_ts_size;
+/* This variable holds the timestamped weak marker for the current module. */
+static char weak_marker[SYMBOL_WEAK_LENGTH + 1];
 
 /* This growing array holds the file name table for generating line
@@ -623 +627 @@
   /* Init weak markers */
 
-  weak_marker_simple[0] = '\0';
-  weak_marker_simple_size = 0;
-  weak_marker_ts[0] = '\0';
-  weak_marker_ts_size = 0;
+  weak_marker[0] = '\0';
 }
 
@@ -763 +764 @@
 
 
+/* Calculates the hash for a string using the original djb2 aglorithm. */
+  
+static unsigned hash_string(const char *pch, size_t cch)
+{
+    unsigned uHash;
+    for (uHash = 5381; cch > 0; pch++, cch--)
+        uHash += (uHash << 5) + *pch;
+    return uHash;
+}
+
+/* Formats a 64-bit number with a fixed with. The purpose is to encode
+   a unique number as tiny as possible while keeping within what any 
+   linker should accept. The width is fixed, and the string is clipped 
+   or padded with zeros to satisfy that. The return value is psz + cchWidth. */
+
+static char *format_u64(uint64_t u64, char *psz, unsigned uRadix, int cchWidth)
+{
+    static const char       s_achDigits[] = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
+    static const unsigned   s_cchDigits = sizeof(s_achDigits) / sizeof(s_achDigits[0]) - 1;
+    assert(uRadix <= s_cchDigits);
+    while (cchWidth-- > 0)
+    {
+        unsigned i = u64 % uRadix;
+        *psz++ = s_achDigits[i];
+        u64 /= uRadix;
+    }
+    assert(!u64);
+    *psz = '\0';
+    return psz;
+}
+
+
+
+
+
 /* Put the string pszName into the current OMF record.
    In the OMF record, the string is preceded by a length byte.  The
@@ -772 +808 @@
 static void put_nstr(const char *pszName, size_t cch)
 {
-    if (cch > 255)
+    if (    cch > SYMBOL_MAX_LENGTH
+        &&  !strstr(pszName + SYMBOL_MAX_LENGTH - SYMBOL_WEAK_LENGTH, "$w$"))
     {
         /* Hash the symbol to help making it unique.
@@ -778 +815 @@
          *      the weakld is parsing it!
          */
-        char        szHash[16];
-        int         cch2;
-        const char *psz;
-        unsigned    uHash;
-        for (psz = pszName, uHash = 5381, cch2 = cch; cch > 0; psz++, cch--) /* hash alg: original djb2. */
-            uHash += (uHash << 5) + *psz;
-        cch2 = sprintf(szHash, "!_%X", 0x7fffffff & uHash);
-
-        if (!fits(1+255))
+        unsigned    uHash = hash_string(pszName, cch);
+        char        szHash[SYMBOL_HASH_LENGTH + 1];
+        char       *psz;
+        szHash[0] = '!';
+        szHash[1] = '_';
+        psz = format_u64((uint32_t)uHash, &szHash[2], 62, 6);
+        assert(psz - &szHash[0] == SYMBOL_HASH_LENGTH);
+
+        if (!fits(1 + SYMBOL_MAX_LENGTH + SYMBOL_HASH_LENGTH))
             doesn_fit();
-        out_data[out_idx++] = 255;
-        memcpy(out_data + out_idx, pszName, 255);
-        memcpy(out_data + out_idx + 255 - cch2, szHash, cch2);
-        out_idx += 255;
-
-        warning ("Symbol length truncated to 255 characters (from %d) for `%s'", cch, pszName);
+        out_data[out_idx++] = SYMBOL_MAX_LENGTH + SYMBOL_HASH_LENGTH;
+        memcpy(out_data + out_idx, pszName, SYMBOL_MAX_LENGTH);
+        out_idx += SYMBOL_MAX_LENGTH;
+        memcpy(out_data + out_idx, szHash, SYMBOL_HASH_LENGTH);
+        out_idx += SYMBOL_HASH_LENGTH;
+
+        warning ("Truncated symbol '%.*s' to '%.*s%s'", cch, pszName, SYMBOL_MAX_LENGTH, pszName, szHash);
     }
     else
@@ -808 +846 @@
    See put_nstr() for full details on string handling. */
 
-static void put_str(const char *pszName)
-{
-    int cch = strlen(pszName);
-    put_nstr(pszName, cch);
+static inline void put_str(const char *pszName)
+{
+    put_nstr(pszName, strlen(pszName));
 }
 
@@ -963 +1000 @@
 static void add_extdef (int *pstarted, const char *name, int type)
 {
-  if (*pstarted && !fits (strlen (name) + 3 + (type > 127)))
-    {
-      write_rec ();
-      *pstarted = FALSE;
+  if (*pstarted)
+    {
+      size_t cchEncodedName = strlen(name);
+      if (   cchEncodedName > SYMBOL_MAX_LENGTH 
+          && !strstr(name + SYMBOL_MAX_LENGTH - SYMBOL_WEAK_LENGTH, "$w$"))
+        cchEncodedName = SYMBOL_MAX_LENGTH + SYMBOL_HASH_LENGTH;
+      if (!fits(cchEncodedName + 3 + (type > 127)))
+        {
+           write_rec();
+           *pstarted = FALSE;
+        }
     }
   if (!*pstarted)
     {
-      init_rec (EXTDEF);
+      init_rec(EXTDEF);
       *pstarted = TRUE;
     }
-  put_sym (name);
-  put_idx (type);                       /* type index */
+  put_sym(name);
+  put_idx(type);                        /* type index */
 }
 
@@ -1096 +1140 @@
         case N_WEAKB:               /* 0x11  Weak bss symbol. */
         {
-            int cch = strlen(pszOrgName);
-            int cch2;
-
-            /* Init the markers if not done already. */
-            if (!weak_marker_simple[0])
+            int cch;
+
+            /* Init the weak marker if it hasn't already been done for this module. */
+            if (!weak_marker[0])
             {
-                static struct timeval   tv_prev;
-                static unsigned         iOpaque = 1;
-                struct timeval          tv = {0, 0};
-                const char             *p1;
-                char                   *p2, *p3;
-
-                /* simple */
-                memcpy(weak_marker_simple, "$w$", 4);
-                for (p1 = "\\/:", p2 = mod_name; *p1; p1++)
-                  if ((p3 = strrchr(p2, *p1)) != NULL)
-                    p2 = p3 + 1;
-                p1 = &weak_marker_simple[sizeof(weak_marker_simple) - 1];
-                for (p3 = &weak_marker_simple[3]; *p2 && p3 < p1; p2++)
-                  if (isalnum(*p2) || *p2 == '_' || *p2 == '@')
-                      *p3++ = *p2;
-                  else if (*p2 == '.')
-                      *p3++ = '_';
-                *p3 = '\0';
-                weak_marker_simple_size = p3 - &weak_marker_simple[0];
-
-                /* timestamped: $w$[1-7][1-4]#[-1-7]#[1-123] */
+                struct timeval  tv = {0, 0};
+                uint64_t        u64;
+                char           *psz;
+
+                /* prefix */
+                psz = (char *)memcpy(&weak_marker[0], "$w$", sizeof("$w$")) + sizeof("$w$") - 1;
+
+                /* use time-of-day + 4 random bits for the 11 char value */
                 gettimeofday(&tv, NULL);
-                memcpy(&weak_marker_ts[0], "$w$", 4);
-                ltoa(tv.tv_sec, &weak_marker_ts[3], 36);
-                cch2 = 3 + strlen(&weak_marker_ts[3]);
-                ltoa(tv.tv_usec, &weak_marker_ts[cch2], 36);
-                cch2 += strlen(&weak_marker_ts[cch2]);
-                weak_marker_ts[cch2++] = '#';
-                if (tv_prev.tv_usec == tv.tv_usec && tv_prev.tv_sec == tv.tv_sec)
-                {
-                    if (!iOpaque)
-                    {
-                        unsigned short seed[3];
-                        seed[0] = (unsigned short)tv.tv_usec;
-                        seed[1] = (unsigned short)tv.tv_sec;
-                        seed[2] = (unsigned short)tv.tv_sec >> 16;
-                        iOpaque = nrand48(seed);
-                    }
-                    iOpaque++;
-                    ltoa(tv.tv_usec, &weak_marker_ts[cch2], 36);
-                    cch2 += strlen(&weak_marker_ts[cch2]);
-                    weak_marker_ts[cch2++] = '#';
-                }
-                tv_prev = tv;
-                memcpy(&weak_marker_ts[cch2], &weak_marker_simple[3], weak_marker_simple_size - 3 + 1);
-                weak_marker_ts_size = cch2 + weak_marker_simple_size - 3;
+                u64 = (uint64_t)tv.tv_sec * 1000000 + tv.tv_usec;
+                u64 <<= 12;
+                u64 |= rand() & 0xfff;
+                psz = format_u64(u64, psz, 62, 11);
+
+                /* simply hash the current filename for the next 6 chars value. */
+                u64 = hash_string(mod_name, strlen(mod_name));
+                psz = format_u64((uint32_t)u64, psz, 62, 6);
+                assert((psz - &weak_marker[0]) == SYMBOL_WEAK_LENGTH);
             }
 
 
             /* generate the weak symbol name. */
-            if (!opt_weakts && cch + weak_marker_simple_size < 255) /* total is one byte short of full length. */
-                snprintf(pachName, cchName, "%s%s", pszOrgName, weak_marker_simple);
-            else if (cch + weak_marker_ts_size < 255)
-                snprintf(pachName, cchName, "%s%s", pszOrgName, weak_marker_ts);
-            else if (cch + 24 < 255)
-                snprintf(pachName, cchName, "%s%.16s", pszOrgName, weak_marker_ts);
+            cch = strlen(pszOrgName);
+            if (cch <= SYMBOL_MAX_LENGTH)           /* Must cut exactly where add_nstr cuts. */
+                snprintf(pachName, cchName, "%s%s", pszOrgName, weak_marker);
             else
-            {   /* too long. */
-                int         cch3 = weak_marker_ts_size > 24 ? 24 : weak_marker_ts_size;
-                char        szHash[16];
-                const char *psz;
-                unsigned    uHash;
-                for (psz = pszOrgName, uHash = 5381; *psz; psz++)   /* hash alg: original djb2. */
-                    uHash += (uHash << 5) + *psz;
-
-                cch2 = sprintf(szHash, "!_%X", 0x7fffffff & uHash);
-                memcpy(pachName, pszOrgName, cch > 255 ? 255 : cch);
-                memcpy(&pachName[255 - cch3 - cch2], szHash, cch2);
-                memcpy(&pachName[255 - cch3], weak_marker_ts, cch3);
-                pachName[255] = '\0';
-                warning ("Symbol length truncated to 255 characters (from %d) for `%s'", cch, pszOrgName);
+            {   /* too long. truncate to: name+!_[hash]$w$[weakmarker] */
+                uint32_t u32Hash = hash_string(pszOrgName, cch);
+                memcpy(pachName, pszOrgName, SYMBOL_MAX_LENGTH);
+                pachName[SYMBOL_MAX_LENGTH + 0] = '!';
+                pachName[SYMBOL_MAX_LENGTH + 1] = '_';
+                format_u64(u32Hash, &pachName[SYMBOL_MAX_LENGTH + 2], 62, 6);
+                memcpy(&pachName[SYMBOL_MAX_LENGTH + SYMBOL_HASH_LENGTH], weak_marker, SYMBOL_WEAK_LENGTH + 1);
+                warning("Truncated symbol '%s' to '%s' (weak)", pszOrgName, pachName);
             }
 
@@ -3707 +3713 @@
   puts ("  -I <idmdll>        Name the identifier manipulation DLL");
   puts ("  -O <directory>     Extract files to <directory>");
-  puts ("  -t                 Add timestamping to the weak symbol mangling (def)");
-  puts ("  -j                 Do not add timestamping to the weak symbol mangling");
   puts ("  -z                 Remove underscores from all symbol names");
   puts ("  -P <dbgpackdll>    Name the dbgpack DLL (ilink)");
@@ -4152 +4156 @@
         *libreq_add = lrp;
         libreq_add = &lrp->next;
-        break;
-      case 'j':
-        opt_weakts = FALSE;
         break;
       case 'I':
@@ -4220 +4221 @@
         opt_rmunder = TRUE;
         break;
-      case 't':
-        opt_weakts = TRUE;
-        break;
       default:
         usage ();