[dmd-internals] segv regression found by dmd test suite

[Brad Roberts]

This patch fixes the one error.  Still running the test suite to see if anything
dislikes it:

===================================================================
--- mangle.c    (revision 792)
+++ mangle.c    (working copy)
@@ -222,15 +222,18 @@
     printf("\n");
 #endif
     id = ident ? ident->toChars() : toChars();
-    Dsymbol *par = isnested || isTemplateMixin() ? parent : tempdecl->parent;
     if (!tempdecl)
         error("is not defined");
-    else if (par)
+    else
     {
-        char *p = par->mangle();
-        if (p[0] == '_' && p[1] == 'D')
-            p += 2;
-        buf.writestring(p);
+        Dsymbol *par = isnested || isTemplateMixin() ? parent : tempdecl->parent;
+        if (par)
+        {
+            char *p = par->mangle();
+            if (p[0] == '_' && p[1] == 'D')
+                p += 2;
+            buf.writestring(p);
+        }
     }
     buf.printf("%zu%s", strlen(id), id);
     id = buf.toChars();

On 12/7/2010 8:30 PM, Brad Roberts wrote:
> It looks like dmd change 777 introduced a change that causes one of the failure
> test cases to segv.  Yeah, the auto-tester has been failing on linux for a
> rather long time.  Just got around to looking at it.
> 
> http://www.dsource.org/projects/dmd/changeset/777
> 
> $ gdb --args ../src/dmd -v -Ifail_compilation   -odtest_results/fail_compilation
> -oftest_results/fail_compilation/fail278_0.o -c fail_compilation/fail278.d
> (gdb) run
> binary    /home/braddr/sandbox/d/source/trunk/dmd/src/dmd
> version   v2.051
> config    /home/braddr/sandbox/d/source/trunk/dmd/src/dmd.conf
> parse     fail278
> importall fail278
> import    object        (~/sandbox/d/source/trunk/druntime/import/object.di)
> semantic  fail278
> 
> Program received signal SIGSEGV, Segmentation fault.
> 0x0812baf9 in TemplateInstance::mangle (this=0x82496f8) at mangle.c:225
> 225         Dsymbol *par = isnested || isTemplateMixin() ? parent :
> tempdecl->parent;
> 
> (gdb) print tempdecl
> $1 = (TemplateDeclaration *) 0x0
> 
> (gdb) bt
> #0  0x0812baf9 in TemplateInstance::mangle (this=0x82496f8) at mangle.c:225
> #1  0x08186925 in TemplateInstance::genIdent (this=0x8249650) at template.c:4684
> #2  0x08187b58 in TemplateMixin::semantic (this=0x8249650, sc=0x8249838) at
> template.c:5190
> #3  0x0818537c in TemplateInstance::semantic (this=0x8249460, sc=0x82493d0,
> fargs=0x0) at template.c:4016
> #4  0x08184a94 in TemplateInstance::semantic (this=0x8249460, sc=0x82493d0) at
> template.c:3687
> #5  0x080db0f3 in ScopeExp::semantic (this=0x8249500, sc=0x82493d0) at
> expression.c:3587
> #6  0x080b2f4c in VarDeclaration::semantic (this=0x8248ef0, sc=0x82493d0) at
> declaration.c:1243
> #7  0x0804b1ee in AttribDeclaration::semanticNewSc (this=0x8244538,
> sc=0x8249170, stc=4, linkage=LINKd, protection=PROTpublic, explicitProtection=0,
> structalign=8) at attrib.c:126
> #8  0x0804bd14 in StorageClassDeclaration::semantic (this=0x8244538,
> sc=0x8249170) at attrib.c:414
> #9  0x0818537c in TemplateInstance::semantic (this=0x8213ef8, sc=0x82141f0,
> fargs=0x0) at template.c:4016
> #10 0x08184a94 in TemplateInstance::semantic (this=0x8213ef8, sc=0x82141f0) at
> template.c:3687
> #11 0x0813f569 in TypeInstance::resolve (this=0x8214030, loc=..., sc=0x82141f0,
> pe=0xffffcfa8, pt=0xffffcfac, ps=0xffffcfa4) at mtype.c:5902
> #12 0x0813f7d8 in TypeInstance::toDsymbol (this=0x8214030, sc=0x82141f0) at
> mtype.c:5968
> #13 0x080b026d in AliasDeclaration::semantic (this=0x82140a8, sc=0x82141f0) at
> declaration.c:463
> #14 0x08130b29 in Module::semantic (this=0x8213208) at module.c:787
> #15 0x0812e0d9 in main (argc=13, argv=0x8209520) at mars.c:1171
> 
> _______________________________________________
> dmd-internals mailing list
> dmd-internals at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-internals