X-Authentication-Warning: delorie.com: mail set sender to djgpp-bounces using -f
X-Recipient: djgpp AT delorie DOT com
X-Original-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20120113;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        bh=VyKqqo+J4yw6/7YCeakOZ1VfGTBUlKlviKrROCcfM/U=;
        b=V9W8wG3DWw9zxB4blBqVJIU44GOjFgBbRfzFCEj6uOP6QLL+ZoyWFOdCdcESkQgV9k
         sWJwWLUKt2eh4uxR5f3SbV24hOT/xj04KhdKt00SmUknpmuu0O/XB9n81NDDnjbl9PFn
         qey9KOt8W135I54LzBKth2dNhc9Vij5ai2J73gnGuMYQNohZsXUhgj/ZLGsF27789zs+
         KiLx/XcwNY7DhUJGTeHVSuAcrLwaVrIYMNEPT9++PIkKm+Boy4mhvr8tqURsXfMz++DJ
         mXCvRxqwPBtHVXxCF9kqM+anBN3+4EQclU8Vqb4cYOfNFRxTUP4mC8NGv537sT1dDYUf
         genw==
MIME-Version: 1.0
X-Received: by 10.107.161.197 with SMTP id k188mr7837742ioe.190.1440170753359;
 Fri, 21 Aug 2015 08:25:53 -0700 (PDT)
In-Reply-To: <CAA2C=vA6taarO7vJhQkrqSbozn=KeKhDwh7w1E3bysAC3QJk+g@mail.gmail.com>
References: <CAA2C=vA6taarO7vJhQkrqSbozn=KeKhDwh7w1E3bysAC3QJk+g AT mail DOT gmail DOT com>
Date: Fri, 21 Aug 2015 18:25:53 +0300
Message-ID: <CAA2C=vCr3P9qnLZoW+YmaRCpQUf5XNmKYZh+eULi1mnb1rdnNw@mail.gmail.com>
Subject: Re: dlopen() failure with ELOOP with legitimate dxe
From: "Ozkan Sezer (sezeroz AT gmail DOT com) [via djgpp AT delorie DOT com]" <djgpp AT delorie DOT com>
To: djgpp <djgpp AT delorie DOT com>
Content-Type: text/plain; charset=UTF-8
Reply-To: djgpp AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On 8/21/15, Ozkan Sezer <sezeroz AT gmail DOT com> wrote:
> There is a bug in dlopen() preventing dlopen()ing a dxe.
> Here is the scenario, what we experienced with q2dos:
>
> ref_soft.dxe -> has no dependency dxes
> ref_gl.dxe -> depends on gl.dxe (present)
> gl.dxe -> depends on glide3x.dxe (not present)
>
> Start the exe, dlopen()s ref_soft.dxe -> OK, runs just fine.
>
> Do a dlclose() ref_soft.dxe, and attempt to dlopen() ref_gl.dxe:
> fails, no surprise, (dlerror() misleadingly says ENOENT, but ok)
>
> The failure triggers dlopen()ing of ref_soft.dxe again: it fails
> with ELOOP which is NOT expected because the dxe is there and it
> is legitimate and it loaded and ran successfully the first time.
>
> As far as I can see, the "Load the dependencies" loop around line
> 237 of dlopen.c does not reset stk_top upon failure which should
> be the reason for failure of further dlopen() attempts?
>

Applied the following change  to dlopen.c and the failure is gone:

Index: dlopen.c
===================================================================
RCS file: /cvs/djgpp/djgpp/src/libc/dxe/dlopen.c,v
retrieving revision 1.8
diff -u -p -r1.8 dlopen.c
--- dlopen.c	27 Jun 2015 12:21:06 -0000	1.8
+++ dlopen.c	21 Aug 2015 15:16:41 -0000
@@ -243,8 +243,10 @@ found:
     tmp.next = stk_top;
     stk_top = &tmp;

-    if (dlopen(scan, RTLD_GLOBAL) == NULL)
+    if (dlopen(scan, RTLD_GLOBAL) == NULL) {
+      stk_top = tmp.next;
       goto unrecoverable;
+    }

     stk_top = tmp.next;


Comments?  OK to apply?

--
O.S.