X-Recipient: archive-cygwin AT delorie DOT com
X-SWARE-Spam-Status: No, hits=-1.8 required=5.0	tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,T_TO_NO_BRKTS_FREEMAIL
X-Spam-Check-By: sourceware.org
Message-ID: <4C8E7CD2.6020006@gmail.com>
Date: Mon, 13 Sep 2010 20:34:42 +0100
From: Dave Korn <dave DOT korn DOT cygwin AT gmail DOT com>
User-Agent: Thunderbird 2.0.0.17 (Windows/20080914)
MIME-Version: 1.0
To: cygwin AT cygwin DOT com
Subject: Re: Cygwin instabilities
References: <AANLkTim+hU+Cg6_J83B_h9VHsL8wiC4Xh7TbcwzjSjkW AT mail DOT gmail DOT com> <4C8E3A4B DOT 8030909 AT cygwin DOT com> <AANLkTimX-AztuJvS-RG+bZqWCVN7tEdQXs6Ogw_fCZC5 AT mail DOT gmail DOT com> <20100913173648 DOT 15137e2d0sfh16o0 AT webmail DOT df DOT eu> <AANLkTi==Dok1uNspQumrPZmm3dEmrt8RN9cjLvEhiiC0 AT mail DOT gmail DOT com> <20100913163523 DOT GA25859 AT ednor DOT casa DOT cgf DOT cx>
In-Reply-To: <20100913163523.GA25859@ednor.casa.cgf.cx>
Content-Type: multipart/mixed; boundary="------------050809060001020500050601"
X-IsSubscribed: yes
Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Id: <cygwin.cygwin.com>
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sourceware.org/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sourceware.org/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com

--------------050809060001020500050601
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

On 13/09/2010 17:35, Christopher Faylor wrote:

> You seem to be approaching this problem as if people will say "Ah!
> Stability issues! Right.  Well, ok, here's what you need to know."
> 
> If we knew of stability issues they would be fixed.

  Well, I know of one, but haven't had time to fix it yet, so I keep this hack
in my local builds.  Can't run "make check -jN" without it, but even then it
sometimes locks up.

  The problem I've run into is that on a heavily loaded system, a pinfo struct
can get truncated into a redirector in between the time a syscall checks the
process_state (using ISSTATE or NOTSTATE) and the time it subsequently
attempts to access a pinfo member which it hoped to guard by that check.

  I have a bad feeling that the only way to totally resolve this is going to
be adding lots of locking or mutexing around pinfo calls, which is almost
bound to have performance implications :-(

    cheers,
      DaveK


--------------050809060001020500050601
Content-Type: text/x-c;
 name="pinfo-fix.diff"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="pinfo-fix.diff"

--- src.clean/winsup/cygwin/pinfo.h	2010-09-01 22:06:36.000000000 +0100
+++ src/winsup/cygwin/pinfo.h	2010-09-06 20:36:17.062500000 +0100
@@ -51,8 +51,6 @@ public:
 
   DWORD exitcode;	/* set when process exits */
 
-#define PINFO_REDIR_SIZE ((char *) &myself.procinfo->exitcode - (char *) myself.procinfo)
-
   /* > 0 if started by a cygwin process */
   DWORD cygstarted;
 
@@ -64,9 +62,6 @@ public:
     signals.  */
   DWORD dwProcessId;
 
-  /* Used to spawn a child for fork(), among other things. */
-  WCHAR progname[NT_MAX_PATH];
-
   /* User information.
      The information is derived from the GetUserName system call,
      with the name looked up in /etc/passwd and assigned a default value
@@ -121,6 +116,12 @@ public:
   HANDLE wr_proc_pipe;
   DWORD wr_proc_pipe_owner;
   friend class pinfo;
+
+  /* Used to spawn a child for fork(), among other things. */
+  WCHAR progname[NT_MAX_PATH];
+  /* Truncate it after execed process exits. */
+#define PINFO_REDIR_SIZE ((char *) &myself.procinfo->progname[0] - (char *) myself.procinfo)
+
 };
 
 DWORD WINAPI commune_process (void *);


--------------050809060001020500050601
Content-Type: text/plain; charset=us-ascii

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
--------------050809060001020500050601--