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
Date: Thu, 1 Jul 1999 23:20:49 -0400
From: Chris Faylor <cgf AT cygnus DOT com>
To: Corinna Vinschen <corinna AT vinschen DOT de>
Cc: DJ Delorie <dj AT delorie DOT com>, cygwin-developers AT sourceware DOT cygnus DOT com
Subject: Re: Change in errno.cc
Message-ID: <19990701232049.A2754@cygnus.com>
References: <199907011359 DOT JAA05833 AT indy DOT delorie DOT com> <377BE2F2 DOT 4688C54A AT vinschen DOT de>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.3i
In-Reply-To: <377BE2F2.4688C54A@vinschen.de>; from Corinna Vinschen on Thu, Jul 01, 1999 at 11:51:46PM +0200

I just checked in a patch that did things slightly differently.

I changed stat_worker to tack on the .exe extension iff the previous
operation failed due to a ERROR_FILE_NOT_FOUND and the file ddid
not already contain an extension.

I also restored the file to the "non .exe" state as Corrinna did below
if GetFileAttributes 

Does this sound reasonable?

-chris

On Thu, Jul 01, 1999 at 11:51:46PM +0200, Corinna Vinschen wrote:
>DJ Delorie wrote:
>> 
>> ETXTBSY may only be returned if you're trying to run a program that's
>> being written to, or write to a program that's currently running.
>> 
>> If two programs open a data file for exclusive writing, one of them
>> should get EACCES.
>> 
>> Is this still the case after your patch is applied?
>
>Hmm, it seems I misinterpreted something. Forget the patch. I hope,
>the following is more devised.
>
>The problem is the output of `ls -l' e.g. in the root directory of the
>system drive:
>
>  ls: pagefile.sys: No such file or directory
>
>An strace shows the following:
>
>  stat_worker calls real_path.file_attributes() which calls
>  GetFileAttributesA(). This call fails with ERROR_ACCESS_DENIED.
>  This would give a correct error message but unfortunately,
>  stat_worker tries it again with an additional .exe suffix that
>  isn't erased later, if the corresponding GetFileAttributesA() call
>  returns -1. Whenever now fhandler_disk_file::fstat() is called,
>  it tries to read a non existant file. The resulting error code is
>  ERROR_FILE_NOT_FOUND :(
>
>The following patch only resets the filename to the input filename,
>if it was extended by the .exe suffix and the corresponding call to
>GetFileAttributesA() returns -1 and GetLastError() returns
>ERROR_FILE_NOT_FOUND.
>
>Regards,
>Corinna
>
>ChangeLog:
>==========
>
>Thu Jul 1 23:47:00  Corinna Vinschen  <corinna AT vinschen DOT de>
>
>	* syscalls.cc (stat_worker): Reset the filename to the input
>	filename, if it was extended by the .exe suffix and the
>	corresponding call to GetFileAttributesA() returns -1 and
>	GetLastError() returns ERROR_FILE_NOT_FOUND.
>Index: syscalls.cc
>===================================================================
>RCS file: /src/cvsroot/winsup-990627/syscalls.cc,v
>retrieving revision 1.1.1.1
>diff -u -p -r1.1.1.1 syscalls.cc
>--- syscalls.cc	1999/06/28 09:00:09	1.1.1.1
>+++ syscalls.cc	1999/07/01 21:38:14
>@@ -917,6 +917,8 @@ stat_worker (const char *caller, const c
>       debug_printf ("trying with .exe suffix");
>       strcat (win32_name, ".exe");
>       atts = GetFileAttributesA (win32_name);
>+      if (atts == -1 && GetLastError () == ERROR_FILE_NOT_FOUND)
>+        win32_name[strlen (win32_name) - 4] = '\0';
>     }
> 
>   debug_printf ("%d = GetFileAttributesA (%s)", atts, win32_name);


-- 
cgf AT cygnus DOT com
http://www.cygnus.com/