UPDATE: netatalk-2.0.3 take 2

[Arnaud Bergeron]

Here is a updated tarball of my netatalk-2.0.3 port.

It is a tarball rather than a diff because I have trouble diffing my
port against what is in cvs.

I did some changes according to private comments from Ian McWilliam.
Primarly patches.

Tested on i386, would need testing especially on 64bit archs (amd64 , alpha).

Arnaud

[Bryan Irvine]

On 12/31/05, Arnaud Bergeron <[hidden email]> wrote:
> Here is a updated tarball of my netatalk-2.0.3 port.
>
> It is a tarball rather than a diff because I have trouble diffing my
> port against what is in cvs.
>
> I did some changes according to private comments from Ian McWilliam.
> Primarly patches.
>
> Tested on i386, would need testing especially on 64bit archs (amd64 , alpha).

Here's the output I get on a i386 machine.

# make
===>  netatalk-2.0.3 depends on: libtool-1.5.18p2 - found
===>  netatalk-2.0.3 depends on: db.4 (db-*) - found
===>  Checking files for netatalk-2.0.3
`/usr/ports/distfiles/netatalk-2.0.3.tar.gz' is up to date.
>> Checksum OK for netatalk-2.0.3.tar.gz. (sha1)
===>  Extracting for netatalk-2.0.3
===>  Patching for netatalk-2.0.3
===>   Ignoring patchfile patch-bin_megatron_nad_c.orig
===>   Ignoring patchfile patch-bin_psorder_psorder_c.orig
Ignoring previously applied (or reversed) patch.
1 out of 1 hunks ignored--saving rejects to etc/afpd/afp_asp.c.rej
***>   patch-etc_afpd_afp_asp_c did not apply cleanly
===>   Ignoring patchfile patch-etc_afpd_directory_c.orig
===>   Ignoring patchfile patch-etc_afpd_mangle_c.orig
===>   Ignoring patchfile patch-etc_afpd_volume_c.orig
===>   Ignoring patchfile patch-etc_cnid_dbd_cnid_metad_c.orig
===>   Ignoring patchfile patch-etc_papd_main_c.orig
===>   Ignoring patchfile patch-etc_papd_ppd_c.orig
*** Error code 1

Stop in /root/netatalk (line 1705 of /usr/ports/infrastructure/mk/bsd.port.mk).

[Arnaud Bergeron]

On 1/9/06, Bryan Irvine <[hidden email]> wrote:

> On 12/31/05, Arnaud Bergeron <[hidden email]> wrote:
> > Here is a updated tarball of my netatalk-2.0.3 port.
> >
> > It is a tarball rather than a diff because I have trouble diffing my
> > port against what is in cvs.
> >
> > I did some changes according to private comments from Ian McWilliam.
> > Primarly patches.
> >
> > Tested on i386, would need testing especially on 64bit archs (amd64 , alpha).
>
> Here's the output I get on a i386 machine.
>
> # make
> ===>  netatalk-2.0.3 depends on: libtool-1.5.18p2 - found
> ===>  netatalk-2.0.3 depends on: db.4 (db-*) - found
> ===>  Checking files for netatalk-2.0.3
> `/usr/ports/distfiles/netatalk-2.0.3.tar.gz' is up to date.
> >> Checksum OK for netatalk-2.0.3.tar.gz. (sha1)
> ===>  Extracting for netatalk-2.0.3
> ===>  Patching for netatalk-2.0.3
> ===>   Ignoring patchfile patch-bin_megatron_nad_c.orig
> ===>   Ignoring patchfile patch-bin_psorder_psorder_c.orig
> Ignoring previously applied (or reversed) patch.
> 1 out of 1 hunks ignored--saving rejects to etc/afpd/afp_asp.c.rej
> ***>   patch-etc_afpd_afp_asp_c did not apply cleanly
> ===>   Ignoring patchfile patch-etc_afpd_directory_c.orig
> ===>   Ignoring patchfile patch-etc_afpd_mangle_c.orig
> ===>   Ignoring patchfile patch-etc_afpd_volume_c.orig
> ===>   Ignoring patchfile patch-etc_cnid_dbd_cnid_metad_c.orig
> ===>   Ignoring patchfile patch-etc_papd_main_c.orig
> ===>   Ignoring patchfile patch-etc_papd_ppd_c.orig
> *** Error code 1
>
> Stop in /root/netatalk (line 1705 of /usr/ports/infrastructure/mk/bsd.port.mk).
>

Sorry about that, I just realized I created a patch for a file I

edited by mistake.  I wonder why my test build did alright but I got
the error too,  The fix is simple: remove
patch-etc_afpd_afp_asp_c.patch from the patch folder

For the others, yet another tarball with this problem fixed (and, I
hope, no more to fix).  As always, please test on 64 bit archs.

Arnaud

[Bryan Irvine]

in /root/netatalk (line 1705 of /usr/ports/infrastructure/mk/bsd.port.mk).
> >
> Sorry about that, I just realized I created a patch for a file I
> edited by mistake.  I wonder why my test build did alright but I got
> the error too,  The fix is simple: remove
> patch-etc_afpd_afp_asp_c.patch from the patch folder

This one *seemed* to compile fine.  AFter install it is missing the
afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
copying over a running config from another server to to check for
config issues.  hrmmm

I'll keep playing with it.  It's nice so far though.

[Morris, Roy]

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]]On Behalf
> Of Bryan Irvine
> Sent: Tuesday, January 10, 2006 1:57 PM
> To: Arnaud Bergeron
> Cc: [hidden email]
> Subject: Re: UPDATE: netatalk-2.0.3 take 2
>
>
> in /root/netatalk (line 1705 of
> /usr/ports/infrastructure/mk/bsd.port.mk).
> > >
> > Sorry about that, I just realized I created a patch for a file I
> > edited by mistake.  I wonder why my test build did alright but I got
> > the error too,  The fix is simple: remove
> > patch-etc_afpd_afp_asp_c.patch from the patch folder
>
> This one *seemed* to compile fine.  AFter install it is missing the
> afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
> copying over a running config from another server to to check for
> config issues.  hrmmm
>
> I'll keep playing with it.  It's nice so far though.
>
>

I installed the previous version, I am not sure if you noticed the
/etc/netatalk/rc.atalk was missing.

[Bryan Irvine]

On 1/10/06, Bryan Irvine <[hidden email]> wrote:

> in /root/netatalk (line 1705 of /usr/ports/infrastructure/mk/bsd.port.mk).
> > >
> > Sorry about that, I just realized I created a patch for a file I
> > edited by mistake.  I wonder why my test build did alright but I got
> > the error too,  The fix is simple: remove
> > patch-etc_afpd_afp_asp_c.patch from the patch folder
>
> This one *seemed* to compile fine.  AFter install it is missing the
> afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
> copying over a running config from another server to to check for
> config issues.  hrmmm
>

After a lot of tinkering the uams_guest.so doesnt' seem to work with
the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
strange.

[Bryan Irvine]

> > This one *seemed* to compile fine.  AFter install it is missing the
> > afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
> > copying over a running config from another server to to check for
> > config issues.  hrmmm
> >
>
> After a lot of tinkering the uams_guest.so doesnt' seem to work with
> the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
> strange.
>

Downloaded and compiled from source and guest access works.  Perhaps
it's just a build flag.

--Bryan

[Bryan Irvine]

On 1/9/06, Arnaud Bergeron <[hidden email]> wrote:

> On 1/9/06, Bryan Irvine <[hidden email]> wrote:
> > On 12/31/05, Arnaud Bergeron <[hidden email]> wrote:
> > > Here is a updated tarball of my netatalk-2.0.3 port.
> > >
> > > It is a tarball rather than a diff because I have trouble diffing my
> > > port against what is in cvs.
> > >
> > > I did some changes according to private comments from Ian McWilliam.
> > > Primarly patches.
> > >
> > > Tested on i386, would need testing especially on 64bit archs (amd64 , alpha).
> >
> > Here's the output I get on a i386 machine.
> >
> > # make
> > ===>  netatalk-2.0.3 depends on: libtool-1.5.18p2 - found
> > ===>  netatalk-2.0.3 depends on: db.4 (db-*) - found
> > ===>  Checking files for netatalk-2.0.3
> > `/usr/ports/distfiles/netatalk-2.0.3.tar.gz' is up to date.
> > >> Checksum OK for netatalk-2.0.3.tar.gz. (sha1)
> > ===>  Extracting for netatalk-2.0.3
> > ===>  Patching for netatalk-2.0.3
> > ===>   Ignoring patchfile patch-bin_megatron_nad_c.orig
> > ===>   Ignoring patchfile patch-bin_psorder_psorder_c.orig
> > Ignoring previously applied (or reversed) patch.
> > 1 out of 1 hunks ignored--saving rejects to etc/afpd/afp_asp.c.rej
> > ***>   patch-etc_afpd_afp_asp_c did not apply cleanly
> > ===>   Ignoring patchfile patch-etc_afpd_directory_c.orig
> > ===>   Ignoring patchfile patch-etc_afpd_mangle_c.orig
> > ===>   Ignoring patchfile patch-etc_afpd_volume_c.orig
> > ===>   Ignoring patchfile patch-etc_cnid_dbd_cnid_metad_c.orig
> > ===>   Ignoring patchfile patch-etc_papd_main_c.orig
> > ===>   Ignoring patchfile patch-etc_papd_ppd_c.orig
> > *** Error code 1
> >
> > Stop in /root/netatalk (line 1705 of /usr/ports/infrastructure/mk/bsd.port.mk).
> >
> Sorry about that, I just realized I created a patch for a file I
> edited by mistake.  I wonder why my test build did alright but I got
> the error too,  The fix is simple: remove
> patch-etc_afpd_afp_asp_c.patch from the patch folder
>
> For the others, yet another tarball with this problem fixed (and, I
> hope, no more to fix).  As always, please test on 64 bit archs.
>
> Arnaud
>
>

[Arnaud Bergeron]

On 1/10/06, Bryan Irvine <[hidden email]> wrote:

> > > This one *seemed* to compile fine.  AFter install it is missing the
> > > afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
> > > copying over a running config from another server to to check for
> > > config issues.  hrmmm
> > >
> >
> > After a lot of tinkering the uams_guest.so doesnt' seem to work with
> > the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
> > strange.
> >
>
> Downloaded and compiled from source and guest access works.  Perhaps
> it's just a build flag.
>

Maybe it is also the patches I made.  Only some of them are really
useful to compile.  The rest try to remove all occurences of strcpy,
strcat and sprintf.  The "speed hack" logic used with those functions
is hard to get sometimes.

If you are convinced it is build flags, compare those you used with
those I use for the package.

Otherwise here is yet another updated tarball which installs afpd.conf
and rc.atalk as it should.

> --Bryan
>


--
"You would end up with a machine which won't work, but won't reboot,
either (whoa. Windows emulation on OpenBSD!)."
-Nick, http://marc.theaimsgroup.com/?l=openbsd-misc&m=113676535806990&w=2

[Ian McWilliam]

On 13 Jan 2006, at 1:15 PM, Arnaud Bergeron wrote:

> On 1/10/06, Bryan Irvine <[hidden email]> wrote:
>>>> This one *seemed* to compile fine.  AFter install it is missing the
>>>> afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
>>>> copying over a running config from another server to to check for
>>>> config issues.  hrmmm
>>>>
>>>
>>> After a lot of tinkering the uams_guest.so doesnt' seem to work with
>>> the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
>>> strange.
>>>
>>
>> Downloaded and compiled from source and guest access works.  Perhaps
>> it's just a build flag.
>>
>

Can you
tail -f /var/log/daemon
as you log in/out on the compiled from source version and copy me the  
output. I'm curious as to why this would act differently as the auth  
process for guest is to use the user nobody which has a non-existant  
home directory. This is where guest access is going to fail as afpd  
can't stat /nonexistant.

thanx

Jan 23 19:10:24 kallasjarvi afpd[8122]: login noauth
Jan 23 19:10:24 kallasjarvi afpd[8122]: login nobody (uid 32767, gid  
32767) AFP3.1
Jan 23 19:10:24 kallasjarvi afpd[8122]: afp_getsrvrparms(/
nonexistent): stat: No such file or directory
Jan 23 19:10:42 kallasjarvi afpd[8122]: logout nobody

Ian McWilliam

[Arnaud Bergeron]

On 1/23/06, Ian McWilliam <[hidden email]> wrote:

>
> On 13 Jan 2006, at 1:15 PM, Arnaud Bergeron wrote:
>
> > On 1/10/06, Bryan Irvine <[hidden email]> wrote:
> >>>> This one *seemed* to compile fine.  AFter install it is missing the
> >>>> afpd.conf file, and uamlist_guest.so doesnt' work.  I even tried
> >>>> copying over a running config from another server to to check for
> >>>> config issues.  hrmmm
> >>>>
> >>>
> >>> After a lot of tinkering the uams_guest.so doesnt' seem to work with
> >>> the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
> >>> strange.
> >>>
> >>
> >> Downloaded and compiled from source and guest access works.  Perhaps
> >> it's just a build flag.
> >>
> >
>
> Can you
> tail -f /var/log/daemon
> as you log in/out on the compiled from source version and copy me the
> output. I'm curious as to why this would act differently as the auth
> process for guest is to use the user nobody which has a non-existant
> home directory. This is where guest access is going to fail as afpd
> can't stat /nonexistant.
>
> thanx
>
> Jan 23 19:10:24 kallasjarvi afpd[8122]: login noauth
> Jan 23 19:10:24 kallasjarvi afpd[8122]: login nobody (uid 32767, gid
> 32767) AFP3.1
> Jan 23 19:10:24 kallasjarvi afpd[8122]: afp_getsrvrparms(/
> nonexistent): stat: No such file or directory
> Jan 23 19:10:42 kallasjarvi afpd[8122]: logout nobody
>
> Ian McWilliam
>

Guest access does work on my machine ... I just tested it (granted, on
3.8-release).

It does give the message about /nonexistent not existing (twice).  But
this seems to have no consequences.

Some details that could be helpful:

1. Are you using cnid_metad?
2. Is the cnid db inside the share?
3. What is your cnid scheme?

--
"You would end up with a machine which won't work, but won't reboot,
either (whoa. Windows emulation on OpenBSD!)."
-Nick, http://marc.theaimsgroup.com/?l=openbsd-misc&m=113676535806990&w=2

[Morris, Roy]

> On 1/23/06, Ian McWilliam <[hidden email]> wrote:
> >
> > On 13 Jan 2006, at 1:15 PM, Arnaud Bergeron wrote:
> >
> > > On 1/10/06, Bryan Irvine <[hidden email]> wrote:
> > >>>> This one *seemed* to compile fine.  AFter install it
> is missing the
> > >>>> afpd.conf file, and uamlist_guest.so doesnt' work.  I
> even tried
> > >>>> copying over a running config from another server to
> to check for
> > >>>> config issues.  hrmmm
> > >>>>
> > >>>
> > >>> After a lot of tinkering the uams_guest.so doesnt' seem
> to work with
> > >>> the old 1.6 port either.  Perhaps it's an OpenBSD thing?   Veery
> > >>> strange.
> > >>>
> > >>
> > >> Downloaded and compiled from source and guest access
> works.  Perhaps
> > >> it's just a build flag.
> > >>
> > >
> >
> > Can you
> > tail -f /var/log/daemon
> > as you log in/out on the compiled from source version and
> copy me the
> > output. I'm curious as to why this would act differently as the auth
> > process for guest is to use the user nobody which has a non-existant
> > home directory. This is where guest access is going to fail as afpd
> > can't stat /nonexistant.
> >
> > thanx
> >
> > Jan 23 19:10:24 kallasjarvi afpd[8122]: login noauth
> > Jan 23 19:10:24 kallasjarvi afpd[8122]: login nobody (uid 32767, gid
> > 32767) AFP3.1
> > Jan 23 19:10:24 kallasjarvi afpd[8122]: afp_getsrvrparms(/
> > nonexistent): stat: No such file or directory
> > Jan 23 19:10:42 kallasjarvi afpd[8122]: logout nobody
> >
> > Ian McWilliam
> >
>
> Guest access does work on my machine ... I just tested it (granted, on
> 3.8-release).
>
> It does give the message about /nonexistent not existing (twice).  But
> this seems to have no consequences.
>
> Some details that could be helpful:
>
> 1. Are you using cnid_metad?
> 2. Is the cnid db inside the share?
> 3. What is your cnid scheme?
>
> --
> "You would end up with a machine which won't work, but won't reboot,
> either (whoa. Windows emulation on OpenBSD!)."
> -Nick,

http://marc.theaimsgroup.com/?l=openbsd-misc&m=113676535806990&w=2

Check the netatalk archives, and provide them with your afpd -v.
Most people I suspect are using the default cdb.

mail:/home/rmorris$ afpd -v
afpd 2.0.3 - Apple Filing Protocol (AFP) daemon of Netatalk

afpd has been compiled with support for these features:

        AFP3.1 support: Yes
      Transport layers: TCP/IP DDP
         CNID backends: cdb dbd last

             afpd.conf: /etc/netatalk/afpd.conf
   AppleVolumes.system: /etc/netatalk/AppleVolumes.system
  AppleVolumes.default: /etc/netatalk/AppleVolumes.default
       UAM search path: /usr/local/lib/netatalk/uams/