Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Steve Langasek]

On Fri, Mar 08, 2024 at 08:04:59PM -0800, Ryan Tandy wrote:
> > This does fix the build failure.  I was about to push such a change to the
> > repo and do a maintainer upload, since I've never been removed from the
> > uploaders field after all these years ;)  Do you want me to do this, or
> > would you be able to do it tonight?  Getting openldap to build is a priority
> > wrt rebootstrapping 32-bit archs for time_t.

> I wasn't sure where openldap was on the priority list for arm* since it's
> still BD-Uninstallable on the buildds.

Yes, it's BD-Uninstallable there because it needs manually built to address
build loops (ldap->cyrus-sasl2->krb5).

> Yes, I can upload it tonight, in a couple of hours from now. Is that OK?

Works for me!

> > WARNING!
> > Running as root!
> > There's a fair chance slapd will fail to start.
> > Check file permissions!
> > 
> > Starting slapd on TCP/IP port 9011...
> > Testing slapd searching...
> > Creating a dynamic entry...
> > ldapadd failed (255)!
> > ../../../tests/scripts/test046-dds: 93: kill: No such process
> > 
> > > > > > > test046-dds failed for mdb after 2 seconds
> > (exit 255)
> > make[4]: *** [Makefile:303: mdb-mod] Error 255

> I have not seen this failure. Ran it again just now and it passed. But I
> only run amd64... I wouldn't be able to dig into that tonight, even if I
> could reproduce it. Do you think I should disable the test proactively?

Rather than disabling the test in the package, I can do a
DEB_BUILD_OPTIONS=nocheck build to help get things bootstrapped in the
archive.  I'm interested to know if it also shows up on hppa, since that
would definitely point to it being time_t related.

-- 
Steve Langasek   Give me a lever long enough and a Free OS
Debian Developer   to set it on, and I can move the world.
Ubuntu Developer   https://www.debian.org/
slanga...@ubuntu.com vor...@debian.org


signature.asc
Description: PGP signature

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Ryan Tandy]

On Fri, Mar 08, 2024 at 07:33:59PM -0800, Steve Langasek wrote:

This bug is also reproducible on the armel and armhf release archs as a
result of the time_t transition, so bumping this to serious.


ACK.


This does fix the build failure.  I was about to push such a change to the
repo and do a maintainer upload, since I've never been removed from the
uploaders field after all these years ;)  Do you want me to do this, or
would you be able to do it tonight?  Getting openldap to build is a priority
wrt rebootstrapping 32-bit archs for time_t.


I wasn't sure where openldap was on the priority list for arm* since 
it's still BD-Uninstallable on the buildds.


Yes, I can upload it tonight, in a couple of hours from now. Is that OK?


BTW there's also a reproducible test failure that I'm seeing on armhf in
both Debian and Ubuntu which is new; I'm not sure if it's related to time_t,
or if it affects other archs?


Starting test046-dds for mdb...

running defines.sh
Running slapadd to build slapd database...
Running slapindex to index slapd database...

WARNING!
Running as root!
There's a fair chance slapd will fail to start.
Check file permissions!

Starting slapd on TCP/IP port 9011...
Testing slapd searching...
Creating a dynamic entry...
ldapadd failed (255)!
../../../tests/scripts/test046-dds: 93: kill: No such process


test046-dds failed for mdb after 2 seconds

(exit 255)
make[4]: *** [Makefile:303: mdb-mod] Error 255


I have not seen this failure. Ran it again just now and it passed. But I 
only run amd64... I wouldn't be able to dig into that tonight, even if I 
could reproduce it. Do you think I should disable the test proactively?


thanks,
Ryan

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Steve Langasek]

Control: severity -1 serious

Hi Ryan,

This bug is also reproducible on the armel and armhf release archs as a
result of the time_t transition, so bumping this to serious.

On Fri, Mar 08, 2024 at 06:04:45PM -0800, Ryan Tandy wrote:
> Would you be willing to test build the attached patch on hppa? I've tested
> it on amd64 with -Werror=implicit-function-declaration appended.


> From fa0b704371762bdc479a5d8dc6a0a6df4ec3a52e Mon Sep 17 00:00:00 2001
> From: Ryan Tandy 
> Date: Fri, 8 Mar 2024 17:58:15 -0800
> Subject: [PATCH] Fix implicit declaration of kadm5_s_init_with_password_ctx
> 
> ---
>  debian/patches/series|  1 +
>  debian/patches/smbk5pwd-implicit-declaration | 12 
>  2 files changed, 13 insertions(+)
>  create mode 100644 debian/patches/smbk5pwd-implicit-declaration
> 
> diff --git a/debian/patches/series b/debian/patches/series
> index a8d57cb99c..7381d0a06b 100644
> --- a/debian/patches/series
> +++ b/debian/patches/series
> @@ -13,3 +13,4 @@ add-tlscacert-option-to-ldap-conf
>  fix-build-top-mk
>  switch-to-lt_dlopenadvise-to-get-RTLD_GLOBAL-set.diff
>  set-maintainer-name
> +smbk5pwd-implicit-declaration
> diff --git a/debian/patches/smbk5pwd-implicit-declaration 
> b/debian/patches/smbk5pwd-implicit-declaration
> new file mode 100644
> index 00..a704086286
> --- /dev/null
> +++ b/debian/patches/smbk5pwd-implicit-declaration
> @@ -0,0 +1,12 @@
> +Description: Fix implicit declaration of kadm5_s_init_with_password_ctx
> +Bug-Debian: https://bugs.debian.org/1065633
> +--- a/contrib/slapd-modules/smbk5pwd/smbk5pwd.c
>  b/contrib/slapd-modules/smbk5pwd/smbk5pwd.c
> +@@ -45,6 +45,7 @@
> + #include 
> + #include 
> + #include 
> ++#include 
> + 
> + #ifndef HDB_INTERFACE_VERSION
> + #define HDB_MASTER_KEY_SET  master_key_set
> -- 
> 2.39.2

This does fix the build failure.  I was about to push such a change to the
repo and do a maintainer upload, since I've never been removed from the
uploaders field after all these years ;)  Do you want me to do this, or
would you be able to do it tonight?  Getting openldap to build is a priority
wrt rebootstrapping 32-bit archs for time_t.

BTW there's also a reproducible test failure that I'm seeing on armhf in
both Debian and Ubuntu which is new; I'm not sure if it's related to time_t,
or if it affects other archs?

> Starting test046-dds for mdb...
running defines.sh
Running slapadd to build slapd database...
Running slapindex to index slapd database...

WARNING!
Running as root!
There's a fair chance slapd will fail to start.
Check file permissions!

Starting slapd on TCP/IP port 9011...
Testing slapd searching...
Creating a dynamic entry...
ldapadd failed (255)!
../../../tests/scripts/test046-dds: 93: kill: No such process  

> test046-dds failed for mdb after 2 seconds
(exit 255)
make[4]: *** [Makefile:303: mdb-mod] Error 255


In Ubuntu we disabled the tests on armhf to not hold up the bootstrap.

Thanks,
-- 
Steve Langasek   Give me a lever long enough and a Free OS
Debian Developer   to set it on, and I can move the world.
Ubuntu Developer   https://www.debian.org/
slanga...@ubuntu.com vor...@debian.org


signature.asc
Description: PGP signature

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Ryan Tandy]

Would you be willing to test build the attached patch on hppa? I've 
tested it on amd64 with -Werror=implicit-function-declaration appended.


thanks,
Ryan
>From fa0b704371762bdc479a5d8dc6a0a6df4ec3a52e Mon Sep 17 00:00:00 2001
From: Ryan Tandy 
Date: Fri, 8 Mar 2024 17:58:15 -0800
Subject: [PATCH] Fix implicit declaration of kadm5_s_init_with_password_ctx

---
 debian/patches/series|  1 +
 debian/patches/smbk5pwd-implicit-declaration | 12 
 2 files changed, 13 insertions(+)
 create mode 100644 debian/patches/smbk5pwd-implicit-declaration

diff --git a/debian/patches/series b/debian/patches/series
index a8d57cb99c..7381d0a06b 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -13,3 +13,4 @@ add-tlscacert-option-to-ldap-conf
 fix-build-top-mk
 switch-to-lt_dlopenadvise-to-get-RTLD_GLOBAL-set.diff
 set-maintainer-name
+smbk5pwd-implicit-declaration
diff --git a/debian/patches/smbk5pwd-implicit-declaration b/debian/patches/smbk5pwd-implicit-declaration
new file mode 100644
index 00..a704086286
--- /dev/null
+++ b/debian/patches/smbk5pwd-implicit-declaration
@@ -0,0 +1,12 @@
+Description: Fix implicit declaration of kadm5_s_init_with_password_ctx
+Bug-Debian: https://bugs.debian.org/1065633
+--- a/contrib/slapd-modules/smbk5pwd/smbk5pwd.c
 b/contrib/slapd-modules/smbk5pwd/smbk5pwd.c
+@@ -45,6 +45,7 @@
+ #include 
+ #include 
+ #include 
++#include 
+ 
+ #ifndef HDB_INTERFACE_VERSION
+ #define	HDB_MASTER_KEY_SET	master_key_set
-- 
2.39.2

Bug#1065633: [SPAM] Re: Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[John David Anglin]

On 2024-03-07 7:21 p.m., Ryan Tandy wrote:

On Thu, Mar 07, 2024 at 01:29:54PM -0800, Ryan Tandy wrote:
The binNMUs succeeded on several release arches already. I'm not sure why hppa would be different. I see 
-Werror=implicit-function-declaration in its compiler commands, but I don't know where it's coming from.


I remembered later maybe seeing something about this around time64 build flags, 
and sure enough, in dpkg 1.22.5:

https://git.dpkg.org/cgit/dpkg/dpkg.git/commit/?id=ef90821fe45b99fa8c8c4279b9a74c30f59f491d

So actually it looks like hppa was only the first to encounter it, and armel/armhf are likely to fail the same way (right now they are still 
BD-Uninstallable), in which case this bug will be RC.

Yes, see:
https://wiki.debian.org/BrainDumpT64

We have been doing a lot of manual builds, so it's possible we may have messed 
up.

--
John David Anglin  dave.ang...@bell.net

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Ryan Tandy]

On Thu, Mar 07, 2024 at 01:29:54PM -0800, Ryan Tandy wrote:
The binNMUs succeeded on several release arches already. I'm not sure 
why hppa would be different. I see 
-Werror=implicit-function-declaration in its compiler commands, but I 
don't know where it's coming from.


I remembered later maybe seeing something about this around time64 build 
flags, and sure enough, in dpkg 1.22.5:


https://git.dpkg.org/cgit/dpkg/dpkg.git/commit/?id=ef90821fe45b99fa8c8c4279b9a74c30f59f491d

So actually it looks like hppa was only the first to encounter it, and 
armel/armhf are likely to fail the same way (right now they are still 
BD-Uninstallable), in which case this bug will be RC.

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Ryan Tandy]

The implicit declaration occurs on all arches, but it's a warning, not 
an error. For example on amd64:


smbk5pwd.c:917:23: warning: implicit declaration of function 
‘kadm5_s_init_with_password_ctx’; did you mean ‘kadm5_init_with_password_ctx’? 
[-Wimplicit-function-declaration]
  917 | ret = kadm5_s_init_with_password_ctx( context,
  |   ^~
  |   kadm5_init_with_password_ctx


The binNMUs succeeded on several release arches already. I'm not sure 
why hppa would be different. I see -Werror=implicit-function-declaration 
in its compiler commands, but I don't know where it's coming from.


As for the implicit declaration, I don't remember why I didn't already 
submit a patch for it... At a quick glance it looks like the prototype 
is now in the private header . I don't know if 
there's a reason we can't/shouldn't include that one... The symbol may 
be private, but we're already using it regardless :)

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function 'kadm5_s_init_with_password_ctx'

[Quanah Gibson-Mount]

--On Thursday, March 7, 2024 5:41 PM + John David Anglin 
 wrote:



Source: openldap
Version: 2.5.13+dfsg-5+b4
Severity: normal
Tags: ftbfs

Dear Maintainer,

See:
https://buildd.debian.org/status/fetch.php?pkg=openldap=hppa=2.5
.13%2Bdfsg-5%2Bb4=1709830559=0

smbk5pwd.c: In function 'smbk5pwd_modules_init':
smbk5pwd.c:917:23: error: implicit declaration of function
'kadm5_s_init_with_password_ctx'; did you mean
'kadm5_init_with_password_ctx'?
[-Werror=implicit-function-declaration]   917 | ret =
kadm5_s_init_with_password_ctx( context,   |
^~


Maybe you're missing a header?  This is a standard Heimdal function:



Or, Debian failed to properly package the Heimdal includes.  But this 
doesn't seem to be the right place to report the issue.


--Quanah

Bug#1065633: openldap: FTBFS on hppa - implicit declaration of function ‘kadm5_s_init_with_password_ctx’

[John David Anglin]

Source: openldap
Version: 2.5.13+dfsg-5+b4
Severity: normal
Tags: ftbfs

Dear Maintainer,

See:
https://buildd.debian.org/status/fetch.php?pkg=openldap=hppa=2.5.13%2Bdfsg-5%2Bb4=1709830559=0

smbk5pwd.c: In function ‘smbk5pwd_modules_init’:
smbk5pwd.c:917:23: error: implicit declaration of function 
‘kadm5_s_init_with_password_ctx’; did you mean ‘kadm5_init_with_password_ctx’? 
[-Werror=implicit-function-declaration]
  917 | ret = kadm5_s_init_with_password_ctx( context,
  |   ^~
  |   kadm5_init_with_password_ctx
smbk5pwd.c:924:25: warning: ‘krb5_get_error_string’ is deprecated 
[-Wdeprecated-declarations]
  924 | err_str = krb5_get_error_string( context );
  | ^~~
In file included from /usr/include/heimdal/krb5.h:967,
 from smbk5pwd.c:45:
/usr/include/heimdal/krb5-protos.h:4188:1: note: declared here
 4188 | krb5_get_error_string (krb5_context /*context*/)
  | ^
smbk5pwd.c:926:33: warning: ‘krb5_get_err_text’ is deprecated 
[-Wdeprecated-declarations]
  926 | err_msg = (char *)krb5_get_err_text( 
context, ret );
  | ^~~
/usr/include/heimdal/krb5-protos.h:4152:1: note: declared here
 4152 | krb5_get_err_text (
  | ^
smbk5pwd.c:931:33: warning: ‘krb5_free_error_string’ is deprecated 
[-Wdeprecated-declarations]
  931 | krb5_free_error_string( context, 
err_str );
  | ^~
/usr/include/heimdal/krb5-protos.h:3721:1: note: declared here
 3721 | krb5_free_error_string (
  | ^~
cc1: some warnings being treated as errors
make[2]: *** [Makefile:54: smbk5pwd.lo] Error 1

Regards,
Dave Anglin

-- System Information:
Debian Release: trixie/sid
  APT prefers buildd-unstable
  APT policy: (500, 'buildd-unstable'), (500, 'unstable')
Architecture: hppa (parisc64)

Kernel: Linux 6.1.80+ (SMP w/4 CPU threads)
Locale: LANG=C, LC_CTYPE=C.UTF-8 (charmap=UTF-8), LANGUAGE not set
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)