batv-milter-discuss — batv-milter project discussion

Re: [batv-milter-discuss] batv-milter, sendmail, and mail via mailx signed but relayed mail not signed

[Lars N. <Lar...@mc...>]

On Wednesday 24 June 2009 10:48:39 Lars Nordin wrote:
> On Tuesday 23 June 2009 18:16:24 SM wrote:
> > Did you start batv-filter with -m MSP-v4?
>
> Thanks for the reply.
>
> I just tried adding what you suggested but it did not work.
>
> FYI, here is the full argument list of how batv-filter is running:
> /usr/bin/batv-filter -u batv-filter -P /var/run/batv-filter/batv-filter.pid
> -p /var/run/batv-filter/batv-filter.sock -l -S -k /etc/mail/batv.key -m
> MSP-v4

Answering my own question here, I ended up creating /etc/mail/batv.localhosts 
(Ubuntu scheme) for the "-i" arg for batv-filter. I put the IP network of my 
PC where I'm running Thunderbird from.
Now messages sent from my PC thru sendmail are BATV signed.

Re: [batv-milter-discuss] batv-milter, sendmail, and mail via mailx signed but relayed mail not signed

[SM <sm...@re...>]

At 07:48 24-06-2009, Lars Nordin wrote:
>On Tuesday 23 June 2009 18:16:24 SM wrote:
>I just tried adding what you suggested but it did not work.
>
>FYI, here is the full argument list of how batv-filter is running:
>/usr/bin/batv-filter -u batv-filter -P 
>/var/run/batv-filter/batv-filter.pid -p
>/var/run/batv-filter/batv-filter.sock -l -S -k /etc/mail/batv.key -m MSP-v4

Can you post the mail log extract of the message submission?

Regards,
-sm 

Re: [batv-milter-discuss] batv-milter, sendmail, and mail via mailx signed but relayed mail not signed

[Lars N. <Lar...@mc...>]

On Tuesday 23 June 2009 18:16:24 SM wrote:
>
> Did you start batv-filter with -m MSP-v4?

Thanks for the reply.

I just tried adding what you suggested but it did not work.

FYI, here is the full argument list of how batv-filter is running:
/usr/bin/batv-filter -u batv-filter -P /var/run/batv-filter/batv-filter.pid -p 
/var/run/batv-filter/batv-filter.sock -l -S -k /etc/mail/batv.key -m MSP-v4

Re: [batv-milter-discuss] batv-milter, sendmail, and mail via mailx signed but relayed mail not signed

[SM <sm...@re...>]

At 11:56 23-06-2009, Lars Nordin wrote:
>I am having problems with BATV milter and sendmail not signing relayed mail.
>There is some thing that I have not configured correctly and it probably with
>sendmail.
>
>Here is what I am running:
>- Sendmail Version 8.14.3
>- BATV milter v0.5.0 (I compiled myself)
>- Ubuntu Linux v9.04
>
>I can send mail via login (i.e. mailx) and the messages are signed but when I
>send messages thru submission (port 587), those messages are not signed.

Did you start batv-filter with -m MSP-v4?

Regards,
-sm 

[batv-milter-discuss] batv-milter, sendmail, and mail via mailx signed but relayed mail not signed

[Lars N. <Lar...@mc...>]

I am having problems with BATV milter and sendmail not signing relayed mail. 
There is some thing that I have not configured correctly and it probably with 
sendmail.

Here is what I am running:
- Sendmail Version 8.14.3
- BATV milter v0.5.0 (I compiled myself)
- Ubuntu Linux v9.04

I can send mail via login (i.e. mailx) and the messages are signed but when I 
send messages thru submission (port 587), those messages are not signed. 

Any help would be appreciated - it has been quite a while since I have 
administered sendmail.


Here is my sendmail.mc file:

define(`_USE_ETC_MAIL_')dnl
include(`/usr/share/sendmail/cf/m4/cf.m4')dnl
VERSIONID(`$Id: sendmail.mc, v 8.14.3-6 2008-12-06 22:15:17 cowboy Exp $')
OSTYPE(`debian')dnl
DOMAIN(`debian-mta')dnl
undefine(`confHOST_STATUS_DIRECTORY')dnl        #DAEMON_HOSTSTATS=
FEATURE(`no_default_msa')dnl
DAEMON_OPTIONS(`Family=inet,  Name=MTA-v4, Port=smtp')dnl
DAEMON_OPTIONS(`Family=inet,  Name=MSP-v4, Port=submission, M=Ea')dnl
define(`confPRIVACY_FLAGS',dnl
`needmailhelo,needexpnhelo,needvrfyhelo,restrictqrun,restrictexpand,nobodyreturn,authwarnings')dnl
define(`confCONNECTION_RATE_THROTTLE', `15')dnl
define(`confCONNECTION_RATE_WINDOW_SIZE',`10m')dnl
FEATURE(`use_cw_file')dnl
FEATURE(`access_db', , `skip')dnl
FEATURE(`greet_pause', `1000')dnl 1 seconds
FEATURE(`delay_checks', `friend', `n')dnl
define(`confBAD_RCPT_THROTTLE',`3')dnl
FEATURE(`conncontrol', `nodelay', `terminate')dnl
FEATURE(`ratecontrol', `nodelay', `terminate')dnl
include(`/etc/mail/m4/dialup.m4')dnl
include(`/etc/mail/m4/provider.m4')dnl
include(`/etc/mail/tls/starttls.m4')dnl
define(`confAUTH_MECHANISMS', `DIGEST-MD5 CRAM-MD5 LOGIN PLAIN')dnl
TRUST_AUTH_MECH(`DIGEST-MD5 CRAM-MD5 LOGIN PLAIN')dnl
define(`confAUTH_OPTIONS', `A p y')dnl
INPUT_MAIL_FILTER(`batv-filter', `S=/var/run/batv-filter/batv-filter.sock')
FEATURE(`always_add_domain')dnl
MASQUERADE_AS(`sendmail1.lars.ctqa.net')dnl
FEATURE(`allmasquerade')dnl
FEATURE(`masquerade_envelope')dnl
MAILER_DEFINITIONS
MAILER(`local')dnl
MAILER(`smtp')dnl

[batv-milter-discuss] 0.5.0 build warnings under FreeBSD/amd64

[Hirohisa Y. <um...@gm...>]

Hi,
I'm new here.

A few things.

1. I got warning the same as we experienced in dkim-filter:
cc -O -I. -I../../include  -I/usr/include -DWITHOUT_SMSTRING -D_THREAD_SAFE -DXP_MT -c batv-filter.c
batv-filter.c: In function 'main':
batv-filter.c:3467: warning: comparison is always true due to limited range of data type
batv-filter.c:3467: warning: comparison is always true due to limited range of data type

2. There might be some people who does not want to use sm_strl*()
while their OS already have them.

I attach a patch for them.


Btw, is there any ongoing work for standardizing BATV now?
draft-levine-smtp-batv looks to be expired.


Regards,
-- 

	Hirohisa Yamaguchi
	  um...@gm...

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[SM <sm...@re...>]

Hi Todd,
At 12:29 28-02-2008, Todd Lyons wrote:
>I applied the patch.  It builds and it runs, but I get timeouts when
>sendmail is talking to the milter.

Apply the attached patch.

>Almost the whole patch was just whitespace changes except for two parts.

Sorry about that; it was a quick fix.

>The first part was just a s/strlcpy/sm_strlcpy/, so that makes perfect
>sense.  But this part uses a different variable.  Is that right?
>
> >@@ -958,8 +958,8 @@
> >               buf[1] = sig[1];
> >               buf[2] = sig[2];
> >               buf[3] = sig[3];
> >-              strlcat(buf, orcpt, sizeof buf);
> >-              strlcat(buf, key, sizeof buf);
> >+              sm_strlcat(buf, orcpt, sizeof buf);
> >+              sm_strlcat(buf, key, sizeof buf);

We are doing s/strlcat/sm_strlcat/ here.

Regards,
-sm  

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[Todd L. <tl...@iv...>]

On Thu, Feb 28, 2008 at 10:57:08AM -0800, SM wrote:

>> batv-filter.o: In function `mlfi_envrcpt':
>> batv-filter.c:(.text+0x1557): undefined reference to `strlcpy'
> That's not needed.  Replace strlcpy in the source with sm_strlcpy and 
> strlcat with sm_strlcat or apply the attached patch.

I applied the patch.  It builds and it runs, but I get timeouts when
sendmail is talking to the milter.

sendmail.mc:
INPUT_MAIL_FILTER(`batv',`S=inet:10038@localhost, F= T=C:1m;S:60s;R:2m')

/usr/bin/batv-filter -l -h -p inet:10038 -P /var/run/batv-milter.pid -a
/etc/mail/batv-allow -i /etc/mail/batv-sign-ips -k
/etc/mail/domainkeys/test.pem -n -d mrball.net

Results in this:
Feb 28 12:16:16 mail sm-mta[9869]: m1SKG4Cq009869: Milter (batv): timeout before data read, where=body
Feb 28 12:16:16 mail sm-mta[9869]: m1SKG4Cq009869: Milter (batv): to error state
Feb 28 12:16:16 mail sm-mta[9869]: m1SKG4Cq009869: Milter: data, reject=451 4.3.2 Please try again later
Feb 28 12:16:16 mail sm-mta[9869]: m1SKG4Cq009869: to=<to...@mr...>, delay=0 0:00:11, pri=35220, stat=Please try again later

Almost the whole patch was just whitespace changes except for two parts.
The first part was just a s/strlcpy/sm_strlcpy/, so that makes perfect
sense.  But this part uses a different variable.  Is that right?

>@@ -958,8 +958,8 @@
> 		buf[1] = sig[1];
> 		buf[2] = sig[2];
> 		buf[3] = sig[3];
>-		strlcat(buf, orcpt, sizeof buf);
>-		strlcat(buf, key, sizeof buf);
>+		sm_strlcat(buf, orcpt, sizeof buf);
>+		sm_strlcat(buf, key, sizeof buf);
> 
> 		SHA1_Init(&sha1);
> 		SHA1_Update(&sha1, buf, strlen(buf));

Thanks for the very quick patch, BTW!
-- 
Regards...		Todd
  We should not be building surveillance technology into standards.
  Law enforcement was not supposed to be easy.  Where it is easy, 
  it's called a police state.             -- Jeff Schiller on NANOG
Linux kernel 2.6.22-14-generic   load average: 0.00, 0.01, 0.00

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[SM <sm...@re...>]

Hi Murray,
At 11:46 28-02-2008, Murray S. Kucherawy wrote:
>libsm is not part of the batv distribution yet, so those symbols will also
>be unresolved.

strcpy and strcat could be replaced with the g_ variant meanwhile as 
a workaround.

Regards,
-sm 

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[Murray S. K. <ms...@se...>]

On Thu, 28 Feb 2008, Todd Lyons wrote:
> WDYT?

Someone opened a bug against batv-milter earlier this week about the same 
issue.

The solution for dk-milter and dkim-milter was to include libsm from the 
open source sendmail distribution which has our own implementations of 
those functions (i.e. sm_strlcpy() and sm_strlcat()).  I'm inclined to do 
that for batv-milter as well since that seems to be the simplest and most 
universal solution.

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[Murray S. K. <ms...@se...>]

On Thu, 28 Feb 2008, SM wrote:
> That's not needed.  Replace strlcpy in the source with sm_strlcpy and 
> strlcat with sm_strlcat or apply the attached patch.

libsm is not part of the batv distribution yet, so those symbols will also 
be unresolved.

Re: [batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[SM <sm...@re...>]

Hi Todd,
At 09:29 28-02-2008, Todd Lyons wrote:
>Hi Murray, while reading a post in NANOG, I stumbled across Sendmail's
>milter implemenation of batv, so I downloaded and tried to build it, but
>I get this:

[snip]

>batv-filter.o: In function `mlfi_envrcpt':
>batv-filter.c:(.text+0x1557): undefined reference to `strlcpy'


>In reading some mailing list posts, it seems like the GNU libc
>maintainers won't put them in libc as shown in:
>   http://en.wikipedia.org/wiki/Strlcpy
>   http://sources.redhat.com/ml/libc-alpha/2000-08/msg00053.html
>   http://sources.redhat.com/ml/libc-alpha/2002-01/msg00001.html
>
>What solutions do I have for getting this to work on my Linux machines?
>I see an inference that on GNU systems it should be provided by glib and
>are named g_strlcpy and g_strlcat.  It does appear to be that way on my
>system:

That's not needed.  Replace strlcpy in the source with sm_strlcpy and 
strlcat with sm_strlcat or apply the attached patch.

Regards,
-sm

[batv-milter-discuss] strlcpy not provided by GNU libc on Linux

[Todd L. <tl...@iv...>]

Hi Murray, while reading a post in NANOG, I stumbled across Sendmail's
milter implemenation of batv, so I downloaded and tried to build it, but
I get this:

make[1]: Entering directory `/root/src/batv-milter-0.2.0/obj.Linux.2.6.22-gentoo-r8.i686/batv-filter'
cc -O2 -I. -I../../include  -UNIS -DSTARTTLS  -D_REENTRANT -DXP_MT   -c -o batv-filter.o batv-filter.c
cc -o batv-filter -lpthread  batv-filter.o      -lmilter   -ldb -lresolv -lcrypt -lnsl -ldl -lssl -lcrypto -lssl -lcrypto  
batv-filter.o: In function `mlfi_envrcpt':
batv-filter.c:(.text+0x1557): undefined reference to `strlcpy'
batv-filter.c:(.text+0x17ca): undefined reference to `strlcpy'
batv-filter.c:(.text+0x19cc): undefined reference to `strlcat'
batv-filter.c:(.text+0x19e6): undefined reference to `strlcat'
batv-filter.o: In function `mlfi_envfrom':
batv-filter.c:(.text+0x2109): undefined reference to `strlcat'
batv-filter.c:(.text+0x2130): undefined reference to `strlcat'
collect2: ld returned 1 exit status
make[1]: *** [batv-filter] Error 1
make[1]: Leaving directory `/root/src/batv-milter-0.2.0/obj.Linux.2.6.22-gentoo-r8.i686/batv-filter'
make: *** [all] Error 2

In reading some mailing list posts, it seems like the GNU libc
maintainers won't put them in libc as shown in:
  http://en.wikipedia.org/wiki/Strlcpy
  http://sources.redhat.com/ml/libc-alpha/2000-08/msg00053.html
  http://sources.redhat.com/ml/libc-alpha/2002-01/msg00001.html

What solutions do I have for getting this to work on my Linux machines?
I see an inference that on GNU systems it should be provided by glib and
are named g_strlcpy and g_strlcat.  It does appear to be that way on my
system:
mail batv-milter-0.2.0 # nm -D /usr/lib/libglib-2.0.so | grep strlc
0004e320 T g_strlcat
0004df20 T g_strlcpy

And from /usr/include/glib-2.0/glib/gstrfuncs.h:
gsize                 g_strlcpy        (gchar        *dest,
                                        const gchar  *src,
                                        gsize         dest_size);
gsize                 g_strlcat        (gchar        *dest,
                                        const gchar  *src,
                                        gsize         dest_size);

So for a linux box which doesn't have strl* provided either manually by
downloading the source from ftp.openbsd.org or specifically by a patched
GNU libc, then it could set some #DEFINE.  And that setting could make
it read in a header that wraps g_strl* around strl*.  (I guess the
biggest issue here is _what_ to do to actually detect that the OS needs
to use one or the other.)

WDYT?
-- 
Regards...		Todd
The greatest shortcoming of the human race is our inability to
understand the exponential function.   --Albert Bartlett, physicist
Linux kernel 2.6.22-14-generic   load average: 0.12, 0.11, 0.09

Re: [batv-milter-discuss] Number of days valid..

[Murray S. K. <ms...@se...>]

On Wed, 29 Aug 2007, Andy Fiddaman wrote:
> if (sigexpire < SIGLIFETIME && daynum >= 1000 - SIGLIFETIME)
> 	daynum -= 1000;
> if (daynum > sigexpire)
> 	/* expired */

Ah yes, that's simpler.  I'd forgotten that I was using signed quantities 
there rather than unsigned ones.  Nicely done!

I'll put a new version out soon.

Re: [batv-milter-discuss] Number of days valid..

[Andy F. <ba...@fi...>]

On Wed, 29 Aug 2007, Murray S. Kucherawy wrote:
; The attached patch fixes that and one other problem related to processing the
; "-i" list.  I'll post v0.2.0 soon unless you find issues with these.

Just a query about the expiry check..

It's currently:

daynum = (now / 86400) % 1000;
if (!(sigexpire > daynum ||
     (daynum <= 1000 && sigexpire > daynum + 1000)))

but (daynum <= 1000) is always true and
(sigexpire > daynum + 1000) is always false.

so it doesn't appear to handle the wraparound case properly.

How about something like the following?

if (sigexpire < SIGLIFETIME && daynum >= 1000 - SIGLIFETIME)
	daynum -= 1000;
if (daynum > sigexpire)
	/* expired */

Andy

Re: [batv-milter-discuss] Number of days valid..

[Murray S. K. <ms...@se...>]

On Thu, 23 Aug 2007, Andy Fiddaman wrote:
> so based on this I would expect the days in the inserted key to be based
> on:
> 	(now / 86400) % 1000 + KEY_LIFETIME

Actually it should be:

 	(now / 86400 + KEY_LIFETIME) % 1000

...but basically you're right.

The attached patch fixes that and one other problem related to processing 
the "-i" list.  I'll post v0.2.0 soon unless you find issues with these.

Re: [batv-milter-discuss] Number of days valid..

[SM <sm...@re...>]

Hi Andy,
At 07:39 23-08-2007, Andy Fiddaman wrote:
>I've been looking at the batv milter implementation along with the
>draft-levine-batv-03 specification and I've got a question about how the
>key lifetime is implemented in the milter.
>
>In the draft, the three digit day number (DDD) is defined as
>         ; ... low three digits of
>         ; the number of days since 1970
>         ; when the address will expire
>
>so based on this I would expect the days in the inserted key to be based
>on:
>         (now / 86400) % 1000 + KEY_LIFETIME
>
>and the validator would then just compare DDD against the current day.
>
>The milter currently implements this in reverse - DDD is populated
>as the current day and the validation component checks that DDD < current
>day + KEY_LIFETIME.
>
>I appreciate that if the same implementation is used for both signing and
>verifying then the end result is the same but is my reading of the
>standard correct or is the milter correct as implemented?

DDD is expiry day.  We have expiry if

  DDD < (days from 1970 % 1000)

The source code does:

(void) time(&now);
daynum = (now / 86400) % 1000;

if (!(sigdaynum + SIGLIFETIME > daynum ||
                       (daynum <= 1000 &&
                        sigdaynum + SIGLIFETIME > daynum + 1000)))
                 {

I don't see why we have SIGLIFETIME in there as we only need that 
when computing the expiry days for signing.

Regards,
-sm 

[batv-milter-discuss] Number of days valid..

[Andy F. <ba...@fi...>]

I've been looking at the batv milter implementation along with the
draft-levine-batv-03 specification and I've got a question about how the
key lifetime is implemented in the milter.

In the draft, the three digit day number (DDD) is defined as
	; ... low three digits of
	; the number of days since 1970
	; when the address will expire

so based on this I would expect the days in the inserted key to be based
on:
	(now / 86400) % 1000 + KEY_LIFETIME

and the validator would then just compare DDD against the current day.

The milter currently implements this in reverse - DDD is populated
as the current day and the validation component checks that DDD < current
day + KEY_LIFETIME.

I appreciate that if the same implementation is used for both signing and
verifying then the end result is the same but is my reading of the
standard correct or is the milter correct as implemented?

Thanks,

Andy