rts/linker: Introduce a notion of strong symbols

4 files changed, 51 insertions, 26 deletions

diff --git a/rts/Linker.c b/rts/Linker.c
index bfbf8338af..9c1852d91c 100644
--- a/rts/Linker.c
+++ b/rts/Linker.c
@@ -266,7 +266,7 @@ int ghciInsertSymbolTable(
    StrHashTable *table,
    const SymbolName* key,
    SymbolAddr* data,
-   HsBool weak,
+   SymStrength strength,
    ObjectCode *owner)
 {
    RtsSymbolInfo *pinfo = lookupStrHashTable(table, key);
@@ -275,18 +275,27 @@ int ghciInsertSymbolTable(
       pinfo = stgMallocBytes(sizeof (*pinfo), "ghciInsertToSymbolTable");
       pinfo->value = data;
       pinfo->owner = owner;
-      pinfo->weak = weak;
+      pinfo->strength = strength;
       insertStrHashTable(table, key, pinfo);
       return 1;
    }
-   else if (weak && data && pinfo->weak && !pinfo->value)
+   else if (pinfo->strength == STRENGTH_STRONG)
+   {
+       /* The existing symbol is strong meaning we must never override it */
+       IF_DEBUG(linker, debugBelch("%s is already defined as a strong symbol; ignoring redefinition...", key));
+       return 1;
+   }
+   else if (strength == STRENGTH_WEAK &&
+            data &&
+            pinfo->strength == STRENGTH_WEAK &&
+            !pinfo->value)
    {
        /* The existing symbol is weak with a zero value; replace it with the new symbol. */
        pinfo->value = data;
        pinfo->owner = owner;
        return 1;
    }
-   else if (weak)
+   else if (strength == STRENGTH_WEAK)
    {
        return 1; /* weak symbol, because the symbol is weak, data = 0 and we
                  already know of another copy throw this one away.
@@ -296,12 +305,12 @@ int ghciInsertSymbolTable(
                  This also preserves the semantics of linking against
                  the first symbol we find. */
    }
-   else if (pinfo->weak && !weak) /* weak symbol is in the table */
+   else if (pinfo->strength == STRENGTH_WEAK && strength != STRENGTH_WEAK) /* weak symbol is in the table */
    {
       /* override the weak definition with the non-weak one */
       pinfo->value = data;
       pinfo->owner = owner;
-      pinfo->weak = HS_BOOL_FALSE;
+      pinfo->strength = strength;
       return 1;
    }
    else if (  pinfo->owner
@@ -322,7 +331,7 @@ int ghciInsertSymbolTable(
        if (owner && (owner->status == OBJECT_NEEDED || owner->status == OBJECT_RESOLVED)) {
            pinfo->value = data;
            pinfo->owner = owner;
-           pinfo->weak  = weak;
+           pinfo->strength = strength;
        }
 
        return 1;
@@ -376,10 +385,11 @@ HsBool ghciLookupSymbolInfo(StrHashTable *table,
         *result = NULL;
         return HS_BOOL_FALSE;
     }
-    if (pinfo->weak)
+    if (pinfo->strength == STRENGTH_WEAK) {
         IF_DEBUG(linker, debugBelch("lookupSymbolInfo: promoting %s\n", key));
-    /* Once it's looked up, it can no longer be overridden */
-    pinfo->weak = HS_BOOL_FALSE;
+        /* Once it's looked up, it can no longer be overridden */
+        pinfo->strength = STRENGTH_NORMAL;
+    }
 
     *result = pinfo;
     return HS_BOOL_TRUE;
@@ -412,7 +422,6 @@ void initLinker (void)
 void
 initLinker_ (int retain_cafs)
 {
-    RtsSymbolVal *sym;
 #if defined(OBJFORMAT_ELF) || defined(OBJFORMAT_MACHO)
     int compileResult;
 #endif
@@ -441,10 +450,10 @@ initLinker_ (int retain_cafs)
     symhash = allocStrHashTable();
 
     /* populate the symbol table with stuff from the RTS */
-    for (sym = rtsSyms; sym->lbl != NULL; sym++) {
+    for (const RtsSymbolVal *sym = rtsSyms; sym->lbl != NULL; sym++) {
         if (! ghciInsertSymbolTable(WSTR("(GHCi built-in symbols)"),
                                     symhash, sym->lbl, sym->addr,
-                                    sym->weak, NULL)) {
+                                    sym->strength, NULL)) {
             barf("ghciInsertSymbolTable failed");
         }
         IF_DEBUG(linker, debugBelch("initLinker: inserting rts symbol %s, %p\n", sym->lbl, sym->addr));
diff --git a/rts/LinkerInternals.h b/rts/LinkerInternals.h
index 167a14202d..7058ad355b 100644
--- a/rts/LinkerInternals.h
+++ b/rts/LinkerInternals.h
@@ -345,6 +345,12 @@ extern Mutex dl_mutex;
 /* Type of the initializer */
 typedef void (*init_t) (int argc, char **argv, char **env);
 
+typedef enum _SymStrength {
+    STRENGTH_NORMAL,
+    STRENGTH_WEAK,
+    STRENGTH_STRONG,
+} SymStrength;
+
 /* SymbolInfo tracks a symbol's address, the object code from which
    it originated, and whether or not it's weak.
 
@@ -360,7 +366,7 @@ typedef void (*init_t) (int argc, char **argv, char **env);
 typedef struct _RtsSymbolInfo {
     SymbolAddr* value;
     ObjectCode *owner;
-    HsBool weak;
+    SymStrength strength;
 } RtsSymbolInfo;
 
 void exitLinker( void );
@@ -389,7 +395,7 @@ int ghciInsertSymbolTable(
     StrHashTable *table,
     const SymbolName* key,
     SymbolAddr* data,
-    HsBool weak,
+    SymStrength weak,
     ObjectCode *owner);
 
 /* Lock-free version of lookupSymbol. When 'dependent' is not NULL, adds it as a
diff --git a/rts/RtsSymbols.c b/rts/RtsSymbols.c
index e8c60d2c07..0a821de64b 100644
--- a/rts/RtsSymbols.c
+++ b/rts/RtsSymbols.c
@@ -81,6 +81,16 @@
 #endif
 
 /*
+ * Note [Strong symbols]
+ * ~~~~~~~~~~~~~~~~~~~~~
+ *
+ * The notion of a *weak* symbol is fairly common in linking: a symbol is weak
+ * if it is declared but not defined, allowing it to be defined by an object
+ * which is loaded later. GHC generalizes this notion, allowing symbol
+ * definitions to be declared as *strong*. A strong symbol is one which will
+ * silently supercede definitions of the same name by later objects.
+ */
+/*
  * Note [Symbols for MinGW's printf]
  * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  *
@@ -150,9 +160,9 @@
       /* ^^ Need to figure out why this is needed.  */   \
       /* See Note [_iob_func symbol] */                  \
       RTS_WIN64_ONLY(SymI_HasProto_redirect(             \
-         __imp___acrt_iob_func, __rts_iob_func, true))   \
+         __imp___acrt_iob_func, __rts_iob_func, STRENGTH_WEAK))   \
       RTS_WIN32_ONLY(SymI_HasProto_redirect(             \
-         __imp____acrt_iob_func, __rts_iob_func, true))  \
+         __imp____acrt_iob_func, __rts_iob_func, STRENGTH_WEAK))  \
       SymI_HasProto(__mingw_vsnwprintf)                  \
       /* ^^ Need to figure out why this is needed.  */   \
       SymI_HasProto(__mingw_vfprintf)                    \
@@ -1059,7 +1069,7 @@
 #define SymE_HasProto(vvv)    SymI_HasProto(vvv)
 #endif
 #define SymI_HasProto(vvv) /**/
-#define SymI_HasProto_redirect(vvv,xxx,weak) /**/
+#define SymI_HasProto_redirect(vvv,xxx,strength) /**/
 #define SymI_HasProto_deprecated(vvv) /**/
 RTS_SYMBOLS
 RTS_RET_SYMBOLS
@@ -1080,11 +1090,11 @@ RTS_LIBFFI_SYMBOLS
 #undef SymE_NeedsDataProto
 
 #define SymI_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
-                    (void*)(&(vvv)), false },
+                    (void*)(&(vvv)), STRENGTH_NORMAL },
 #define SymI_HasDataProto(vvv) \
                     SymI_HasProto(vvv)
 #define SymE_HasProto(vvv) { MAYBE_LEADING_UNDERSCORE_STR(#vvv), \
-            (void*)DLL_IMPORT_DATA_REF(vvv), false },
+            (void*)DLL_IMPORT_DATA_REF(vvv), STRENGTH_NORMAL },
 #define SymE_HasDataProto(vvv) \
                     SymE_HasProto(vvv)
 
@@ -1095,9 +1105,9 @@ RTS_LIBFFI_SYMBOLS
 
 // SymI_HasProto_redirect allows us to redirect references to one symbol to
 // another symbol.  See newCAF/newRetainedCAF/newGCdCAF for an example.
-#define SymI_HasProto_redirect(vvv,xxx,weak) \
+#define SymI_HasProto_redirect(vvv,xxx,strength) \
     { MAYBE_LEADING_UNDERSCORE_STR(#vvv),    \
-      (void*)(&(xxx)), weak },
+      (void*)(&(xxx)), strength },
 
 // SymI_HasProto_deprecated allows us to redirect references from their deprecated
 // names to the undeprecated ones. e.g. access -> _access.
@@ -1107,7 +1117,7 @@ RTS_LIBFFI_SYMBOLS
 // define them, since on Windows these functions shouldn't be in the top level
 // namespace, but we have them for POSIX compatibility.
 #define SymI_HasProto_deprecated(vvv)   \
-   { #vvv, (void*)0xBAADF00D, true },
+   { #vvv, (void*)0xBAADF00D, STRENGTH_WEAK },
 
 RtsSymbolVal rtsSyms[] = {
       RTS_SYMBOLS
@@ -1124,7 +1134,7 @@ RtsSymbolVal rtsSyms[] = {
       // dyld stub code contains references to this,
       // but it should never be called because we treat
       // lazy pointers as nonlazy.
-      { "dyld_stub_binding_helper", (void*)0xDEADBEEF, false },
+      { "dyld_stub_binding_helper", (void*)0xDEADBEEF, STRENGTH_NORMAL },
 #endif
-      { 0, 0, false } /* sentinel */
+      { 0, 0, STRENGTH_NORMAL } /* sentinel */
 };
diff --git a/rts/RtsSymbols.h b/rts/RtsSymbols.h
index b17c56e0dd..0efc16be0f 100644
--- a/rts/RtsSymbols.h
+++ b/rts/RtsSymbols.h
@@ -21,7 +21,7 @@
 typedef struct _RtsSymbolVal {
     const SymbolName* lbl;
     SymbolAddr* addr;
-    bool weak;
+    SymStrength strength;
 } RtsSymbolVal;
 
 extern RtsSymbolVal rtsSyms[];