Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm
Sender: cygwin-developers-owner AT sourceware DOT cygnus DOT com
Delivered-To: mailing list cygwin-developers AT sourceware DOT cygnus DOT com
Message-Id: <199904292008.PAA23228@modi.xraylith.wisc.edu>
X-Authentication-Warning: modi.xraylith.wisc.edu: localhost.xraylith.wisc.edu [127.0.0.1] didn't use HELO protocol
To: cygwin-developers AT sourceware DOT cygnus DOT com
Subject: (patch) reg_key copy construction bug
Date: Thu, 29 Apr 1999 15:08:36 -0500
From: Mumit Khan <khan AT xraylith DOT wisc DOT EDU>

The following fix is hopefully obvious. Think of what happens when you 
pass a copy of a (open) reg_key to a function -- it calls the destructor 
on the way back, and so the does the original and RegClose crashes trying.

The current implementation of reg_key is dangerous in that it blindly
allows a copy construction;  it should really disable copying by using
a private copy. However that'll break the subkey generation, so it'll 
take a bit of rework.

The current code should always crash in mount_info::from_registry ()
when Cygwin1.dll starts up.

Against ss-1999-04-27.

Thu Apr 29 13:55:57 1999  Mumit Khan  <khan AT xraylith DOT wisc DOT edu>

	* shared.h (read_mounts): Change prototype to accept a reference
	to reg_key, not a copy.
	* path.cc (read_mounts): Likewise.

--- shared.h.~1	Thu Apr 29 13:51:51 1999
+++ shared.h	Thu Apr 29 13:52:13 1999
@@ -335,7 +335,7 @@ public:
 private:
 
   void sort ();
-  void read_mounts (reg_key r);
+  void read_mounts (reg_key& r);
   void read_v1_mounts (reg_key r, const int which);
   void mount_slash ();
   void to_registry ();
--- path.cc.~1	Thu Apr 29 13:55:20 1999
+++ path.cc	Thu Apr 29 13:52:19 1999
@@ -1158,7 +1158,7 @@ mount_info::binary_win32_path_p (const c
    key. */
 
 void
-mount_info::read_mounts (reg_key r)
+mount_info::read_mounts (reg_key& r)
 {
   HKEY key = r.get_key ();
   LONG err;

Regards,
Mumit