Re: [Jim Blandy <jimb@redhat.com>] RFA: Add tests for lookup_symbol_aux bug

[Jim Blandy <jimb at redhat dot com>]

Michael Elizabeth Chastain <mec@shout.net> writes:
> Jim Blandy submits:
> 
>   2002-10-04  Jim Blandy  <jimb@redhat.com>
> 
> 	* gdb.c++/psmang.exp, gdb.c++/psmang1.cc, gdb.c++/psmang2.cc: New
> 	test.
> 
> This patch is approved, provided that you add one more comment.
> (It seems kinda strange to exhort Jim Blandy, of all people, to write
> more comments!)

That particular .exp file is 87% comments and blank lines.  :)

> My request is: in psmang.exp, mention whether you were testing with
> dwarf-2 or stabs+ debugging format when you saw the bug happening.
> This would be important if someone were trying to reproduce the bug.
> (I saw the bug happening with both formats).

Okay, I've committed this patch:

2002-12-21  Jim Blandy  <jimb@redhat.com>

	* gdb.c++/psmang.exp: Doc fix.

Index: gdb/testsuite/gdb.c++/psmang.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.c++/psmang.exp,v
retrieving revision 1.1
diff -u -r1.1 psmang.exp
--- gdb/testsuite/gdb.c++/psmang.exp	21 Dec 2002 22:56:56 -0000	1.1
+++ gdb/testsuite/gdb.c++/psmang.exp	22 Dec 2002 02:54:00 -0000
@@ -48,6 +48,8 @@
 #   Breakpoint 1 at 0x804841b: file psmang1.cc, line 13.
 #   (gdb) 
 #
+# We observed this bug first using Stabs, and then using Dwarf 2.
+#
 # The problem was in lookup_symbol_aux: when looking up s::method1, it
 # would fail to find it in any symtabs, find the minsym with the
 # corresponding mangled name (say, `_ZN1S7method1Ev'), pass the
@@ -117,6 +119,14 @@
 #
 # Note that #including any header file at all into both compilation
 # units --- say, <stdio.h> --- could create this sort of dependency.
+#
+# This is the aspect of the test which the debug format is most likely
+# to affect, I think.  The different formats create different kinds of
+# inter-CU dependencies, which could mask the bug.  It might be
+# possible for the test to check that at least one of the partial
+# symtabs remains unread, and fail otherwise --- the failure
+# indicating that the test itself isn't going to catch the bug it was
+# meant to, not that GDB is misbehaving.
 #
 # Third twist: given the way lookup_block_symbol is written, it's
 # possible to find the symbol even when it gets passed a mangled name