Apache 2.2.20

8 views
Skip to first unread message

Dave Saville

unread,
Sep 7, 2011, 4:58:54 AM9/7/11
to apa...@googlegroups.com
Just tried Paul's latest and the first thing I see is there is now some
sort of logging going on to the VIO window it starts in. Wonder if Paul
left some debug code in there?

Odd thing is I start apache with:

bin\httpd.exe -d . 2>&1 >> d:\apps\apache2\logs\stderr-out

So I don't see why there is anything in the window anyway.

--
Kind regards

Dave Saville

Paul Smedley

unread,
Sep 7, 2011, 5:28:26 AM9/7/11
to apa...@googlegroups.com
Hi Dave,

On 07/09/11 18:28, Dave Saville wrote:
> Just tried Paul's latest and the first thing I see is there is now some
> sort of logging going on to the VIO window it starts in. Wonder if Paul
> left some debug code in there?

Not that I'm aware of but anything's possible....

Dave Saville

unread,
Sep 10, 2011, 4:44:18 AM9/10/11
to apa...@googlegroups.com

Seems to be dumping the environment to the VIO window for every access.

Paul Smedley

unread,
Sep 10, 2011, 5:04:12 AM9/10/11
to apa...@googlegroups.com
Hi Dave,

Not happening here.....

Dave Saville

unread,
Sep 10, 2011, 7:28:28 AM9/10/11
to apa...@googlegroups.com
On Sat, 10 Sep 2011 18:34:12 +0930 Paul Smedley wrote:
>
>Hi Dave,
>
>On 10/09/11 18:14, Dave Saville wrote:
>>On Wed, 07 Sep 2011 18:58:26 +0930 Paul Smedley wrote:
>>>
>>>Hi Dave,
>>>
>>>On 07/09/11 18:28, Dave Saville wrote:
>>>>Just tried Paul's latest and the first thing I see is there is now some
>>>>sort of logging going on to the VIO window it starts in. Wonder if Paul
>>>>left some debug code in there?
>>>Not that I'm aware of but anything's possible....
>>>
>>
>>Seems to be dumping the environment to the VIO window for every access.
>Not happening here.....
>

Well *something* is different. 2.2.17 does not do it. The only
difference is that the script that does it is accessed on port 40 and
other scripts on 80 work fine. The config has not changed. Making the
window bigger I get

Attempting to exec H:/home/DB/web2/admin/cgi-bin/cars.pl as CGI child
(argv0 = H:/home/DB/web2/admin/cgi-bin/cars.pl)
Environment:

Dave Saville

unread,
Sep 11, 2011, 5:06:13 AM9/11/11
to apa...@googlegroups.com
Cracked it by looking at the mod_cgi source. Must have been compiled
with DEBUG_CGI true.

And the reason I could not redirect the stream is it is neither stdout
nor stderr

#ifdef OS2
/* Under OS/2 need to use device con. */
FILE *dbg = fopen("con", "w");

fprintf(dbg, "Attempting to exec %s as CGI child (argv0 = %s)\n",


Hey ho.

Paul Smedley

unread,
Sep 11, 2011, 5:51:54 AM9/11/11
to apa...@googlegroups.com
Hi Dave,

On 11/09/11 18:36, Dave Saville wrote:
> Cracked it by looking at the mod_cgi source. Must have been compiled
> with DEBUG_CGI true.
>
> And the reason I could not redirect the stream is it is neither stdout
> nor stderr
>
> #ifdef OS2
> /* Under OS/2 need to use device con. */
> FILE *dbg = fopen("con", "w");
>
> fprintf(dbg, "Attempting to exec %s as CGI child (argv0 = %s)\n",

Yep you're correct - sorry for the inconvenience :(

Try
http://sourceforge.net/projects/ecsports/files/Apache2%20Web%20Server/httpd-2.2.20-os2-20110911.zip/download


Cheers,

Paul

Dave Saville

unread,
Sep 11, 2011, 7:13:57 AM9/11/11
to apa...@googlegroups.com

Thanks Paul - that's fixed it.

Dave Saville

unread,
Sep 12, 2011, 5:40:23 AM9/12/11
to apa...@googlegroups.com
Oh dear

The authentication seems broken - Flags all attempts as password
mismatch. Restored 2.2.17 and all is well again.

Anyone got authentication working on this release?

Lewis G Rosenthal

unread,
Sep 12, 2011, 11:12:47 AM9/12/11
to apa...@googlegroups.com
Hi, Dave!

On 09/12/11 05:40 am, Dave Saville thus wrote :


> Oh dear
>
> The authentication seems broken - Flags all attempts as password
> mismatch. Restored 2.2.17 and all is well again.
>
> Anyone got authentication working on this release?
>

I was going to test the build a couple of nights ago, to see if I could
replicate the noise you were getting in the VIO window. It got late, and
I figured I'd wait. Yesterday morning, I saw Paul's post about the
updated build to address that, but again got involved with other things,
and didn't have an opportunity to dig in.

I'll install it on my ThinkPad and see what I get. I have a couple
password protected areas which utilize AuthType Basic. Is that the type
causing you difficulty?

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE, CWTS
Rosenthal& Rosenthal, LLC www.2rosenthals.com
Need a managed Wi-Fi hotspot? www.hautspot.com
visit my IT blog www.2rosenthals.net/wordpress
please do not add my address to any non-bcc mass mailings
-------------------------------------------------------------

Dave Saville

unread,
Sep 13, 2011, 4:07:51 AM9/13/11
to apa...@googlegroups.com
On Mon, 12 Sep 2011 11:12:47 -0400 Lewis G Rosenthal wrote:
>
>Hi, Dave!
>
>On 09/12/11 05:40 am, Dave Saville thus wrote :
>>Oh dear
>>
>>The authentication seems broken - Flags all attempts as password
>>mismatch. Restored 2.2.17 and all is well again.
>>
>>Anyone got authentication working on this release?
>> I was going to test the build a couple of nights ago, to see if I could replicate the noise you were getting in the VIO window. It got late, and I figured I'd wait.
>Yesterday morning, I saw Paul's post about the updated build to address that, but again got involved with other things, and didn't have an opportunity to dig in.
>
>I'll install it on my ThinkPad and see what I get. I have a couple password protected areas which utilize AuthType Basic. Is that the type causing you difficulty?
>

Hi Lewis, Yes Basic.

Dave Saville

unread,
Sep 18, 2011, 7:45:20 AM9/18/11
to apa...@googlegroups.com
On Mon, 12 Sep 2011 11:12:47 -0400 Lewis G Rosenthal wrote:
>
>Hi, Dave!
>
>On 09/12/11 05:40 am, Dave Saville thus wrote :
>>Oh dear
>>
>>The authentication seems broken - Flags all attempts as password
>>mismatch. Restored 2.2.17 and all is well again.
>>
>>Anyone got authentication working on this release?
>> I was going to test the build a couple of nights ago, to see if I could replicate the noise you were getting in the VIO window. It got late, and I figured I'd wait.
>Yesterday morning, I saw Paul's post about the updated build to address that, but again got involved with other things, and didn't have an opportunity to dig in.
>
>I'll install it on my ThinkPad and see what I get. I have a couple password protected areas which utilize AuthType Basic. Is that the type causing you difficulty?
>

Hi Lewis

Have you found time to test this? Have opened a mantis ticket #487

Piersante Sestini

unread,
Sep 18, 2011, 9:29:49 AM9/18/11
to Apache for OS/2


On Sep 12, 11:40 am, "Dave Saville" <d...@deezee.org> wrote:
>
> The authentication seems broken - Flags all attempts as password
> mismatch. Restored 2.2.17 and all is well again.
>
> Anyone got authentication working on this release?
>

The SWAT server for Samba administration uses basic authentication and
appears to work here.
Tested only from localhost, though.

Piersante

Lewis G Rosenthal

unread,
Sep 18, 2011, 10:09:36 AM9/18/11
to apa...@googlegroups.com
Not yet, Dave, but I should get to it today (it hasn't dropped off my list).

Cheers/2

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE

Rosenthal& Rosenthal, LLC www.2rosenthals.com
Need a managed Wi-Fi hotspot? www.hautspot.com
visit my IT blog www.2rosenthals.net/wordpress

-------------------------------------------------------------

Lewis G Rosenthal

unread,
Sep 20, 2011, 4:47:29 PM9/20/11
to apa...@googlegroups.com
Hi, Piersante...

On 09/18/11 09:29 am, Piersante Sestini thus wrote :

Basic auth is *not* working here. The 2.2.20 version of auth_bas.dll
does not seem to want to properly handle passwords generated from an
earlier Apache2 htpasswd.exe, however, the reverse does seem to work (I
can use htpasswd.exe to generate a new password file and then replace
the one on the server running Apache 2.2.15 and it will properly
decrypt. So, whatever happened seems to have only impacted the
decryption side of the situation, namely, the dll.

The log simply shows a failed password; no specifics.

Dave, apologies for taking so long to actually test this.

Anyone else?

--
Lewis
-------------------------------------------------------------
Lewis G Rosenthal, CNA, CLP, CLE, CWTS


Rosenthal& Rosenthal, LLC www.2rosenthals.com
Need a managed Wi-Fi hotspot? www.hautspot.com
visit my IT blog www.2rosenthals.net/wordpress

Dave Saville

unread,
Sep 21, 2011, 3:43:40 AM9/21/11
to apa...@googlegroups.com
On Tue, 20 Sep 2011 16:47:29 -0400 Lewis G Rosenthal wrote:
>Basic auth is *not* working here. The 2.2.20 version of auth_bas.dll does
>not seem to want to properly handle passwords generated from an earlier
>Apache2 htpasswd.exe, however, the reverse does seem to work (I can use
>htpasswd.exe to generate a new password file and then replace the one on
>the
>server running Apache 2.2.15 and it will properly decrypt. So, whatever
>happened seems to have only impacted the decryption side of the situation,
>namely,
>the dll.
>
>The log simply shows a failed password; no specifics.
>

You think this may have something to do with it? From the 2.2.18 changelog

*) htpasswd: Change the default algorithm for htpasswd to MD5 on all
platforms. Crypt with its 8 character limit is not useful anymore;
improve out of disk space handling (PR 30877); print a warning if
a password is truncated by crypt. [Stefan Fritsch]

Paul Smedley

unread,
Sep 21, 2011, 5:51:12 AM9/21/11
to apa...@googlegroups.com
Hi Dave,

I'd wager this is the root cause of the issue...

Cheers,

Paul

Lewis G Rosenthal

unread,
Sep 21, 2011, 11:57:25 AM9/21/11
to apa...@googlegroups.com
On 09/21/11 05:51 am, Paul Smedley thus wrote :
I've added my own note to the issue, as well.

In my case, I can workaround by re-generating new password files. Why,
though, have I come up empty searching even Apache's bugzilla for
mentions of this under other platforms?

Odd...

BTW, I *am* using MD5 in my password files, and these did *not* match up
to the new ones. I generated a new file, specifying MD5 on the command
line, too.

Cheers/2

Piersante Sestini

unread,
Sep 21, 2011, 2:13:17 PM9/21/11
to apa...@googlegroups.com
In effects I had to re-generate the password recently. As it is the only one that I use, it wasn't a problem for me, so  I forgot.

Piersante

2011/9/20 Lewis G Rosenthal <lgros...@2rosenthals.com>
Hi, Piersante...Basic auth is *not* working here. The 2.2.20 version of  auth_bas.dll does not seem to want to properly handle passwords generated from an earlier Apache2 htpasswd.exe, however, the reverse does seem to work (I can use htpasswd.exe to generate a new password file and then replace the one on the server running Apache 2.2.15 and it will properly decrypt. So, whatever happened seems to have only impacted the decryption side of the situation, namely, the dll.

Dave Saville

unread,
Sep 22, 2011, 4:40:55 AM9/22/11
to apa...@googlegroups.com

Neither could I find anything. Therefore I am forced to the conclusion it
must be an OS.2 problem. In my case the password file is using CRYPT (the
old default) and fails. I cannot regenerate the file as users are able to
change their passwords - I have over a thousand :-(

Steven Levine

unread,
Sep 22, 2011, 1:46:03 PM9/22/11
to apa...@googlegroups.com
In <001.d8d00100...@deezee.org>, on 09/22/11
at 09:40 AM, "Dave Saville" <da...@deezee.org> said:

Hi,

>>BTW, I *am* using MD5 in my password files, and these did *not* match up
>>to the new ones. I generated a new file, specifying MD5 on the command
>>line, too.

>Neither could I find anything. Therefore I am forced to the conclusion it
> must be an OS.2 problem. In my case the password file is using CRYPT
>(the old default) and fails. I cannot regenerate the file as users are
>able to change their passwords - I have over a thousand :-(

I have to suspect that problem is in

apr-trunk\crypto\apr_md5.c

The code is supposed to be backwards compatible. There is some code that
disables crypt for Windows and a couple of other platforms, but not OS/2.

I've not installed Paul's latest build yet, so I can't test here. What
happens if you generate a crypt hash with the old and the new
htpasswd.exe? If they differ when the new htpasswd is run with the -d
switch, I would say there's some sort of regression.

Steven

--
----------------------------------------------------------------------
"Steven Levine" <ste...@earthlink.net> eCS/Warp/DIY etc.
www.scoug.com www.ecomstation.com
----------------------------------------------------------------------

Reply all
Reply to author
Forward
0 new messages