Bug 29399 - openafs incompatible with backported 5.13 kernel (dkms-libafs fails to build the module)
Summary: openafs incompatible with backported 5.13 kernel (dkms-libafs fails to build ...
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: 8
Hardware: x86_64 Linux
Priority: Normal normal
Target Milestone: ---
Assignee: QA Team
QA Contact:
URL:
Whiteboard: MGA8-64-OK
Keywords: advisory, validated_update
Depends on:
Blocks:
 
Reported: 2021-08-23 20:32 CEST by Damien Neyret
Modified: 2021-11-09 18:51 CET (History)
4 users (show)

See Also:
Source RPM: openafs-1.9.0-1.mga8.src.rpm
CVE:
Status comment:


Attachments

Description Damien Neyret 2021-08-23 20:32:53 CEST
Description of problem:

dkms can't build the libafs module with kernel 5.13.12-desktop-2.mga8. When I try to do it by hand (make in /var/lib/dkms/libafs/1.9.0-1.mga8/build) I get the error:
/var/lib/dkms/libafs/1.9.0-1.mga8/build/src/libafs/MODLOAD-5.13.12-desktop-2.mga8-MP/afs_syscall.c:127:26: erreur: « TIF_IA32 » non déclaré (première utilisation dans cette fonction)
  127 |     if (test_thread_flag(TIF_IA32))
      |                          ^~~~~~~~
./include/linux/thread_info.h:131:45: note: dans la définition de la macro « test_thread_flag »
  131 |  test_ti_thread_flag(current_thread_info(), flag)

According to the openafs.org site a new openafs version named 1.8.8 (newer than 1.9.0 then...) should be compatible with kernels 5.11 and later.


Version-Release number of selected component (if applicable):
Mageia 8, openafs version 1.9.0-1, kernel 5.13.12


How reproducible:

dkms build -m libafs -v 1.9.0-1.mga8 -k 5.13.12-desktop-2.mga8

Thanks for your help !
Comment 1 David Walser 2021-08-24 00:17:57 CEST
Ouch, yeah they don't list a newer development 1.9.x release, but they do have the 1.8.7 emergency bugfix release and 1.8.8 update for kernel 5.13.

There is a 1.9.1 on the download site, which is newer than 1.8.7, so it probably includes the emergency bugfix, but not the kernel 5.13 support (it supports 5.12).

Version: 8 => Cauldron
Whiteboard: (none) => MGA8TOO
Source RPM: dkms-libafs-1.9.0-1.mga8 => openafs-1.9.0-1.mga8.src.rpm
Summary: dkms-libafs fails to build the module => openafs incompatible with backported 5.13 kernel (dkms-libafs fails to build the module)
Assignee: bugsquad => tmb

Comment 2 Thomas Backlund 2021-08-24 10:55:06 CEST

Should be fixed with openafs-1.9.0-1.1.mga8 just uploaded to updates testing:


SRPM:
openafs-1.9.0-1.1.mga8.src.rpm


i586:
dkms-libafs-1.9.0-1.1.mga8.noarch.rpm
libopenafs2-1.9.0-1.1.mga8.i586.rpm
libopenafs-devel-1.9.0-1.1.mga8.i586.rpm
libopenafs-static-devel-1.9.0-1.1.mga8.i586.rpm
openafs-1.9.0-1.1.mga8.i586.rpm
openafs-client-1.9.0-1.1.mga8.i586.rpm
openafs-doc-1.9.0-1.1.mga8.noarch.rpm
openafs-server-1.9.0-1.1.mga8.i586.rpm


x86_64:
dkms-libafs-1.9.0-1.1.mga8.noarch.rpm
lib64openafs2-1.9.0-1.1.mga8.x86_64.rpm
lib64openafs-devel-1.9.0-1.1.mga8.x86_64.rpm
lib64openafs-static-devel-1.9.0-1.1.mga8.x86_64.rpm
openafs-1.9.0-1.1.mga8.x86_64.rpm
openafs-client-1.9.0-1.1.mga8.x86_64.rpm
openafs-doc-1.9.0-1.1.mga8.noarch.rpm
openafs-server-1.9.0-1.1.mga8.x86_64.rpm

Whiteboard: MGA8TOO => (none)
Assignee: tmb => qa-bugs
Version: Cauldron => 8

Comment 3 Damien Neyret 2021-08-27 14:39:48 CEST
Hi !

I tried to install the new openafs version 1.9.0-1.1.mga8. The dkms module compiles correctly with kernel version 5.13.12-desktop-2.mga8, and that module loads correctly in the system:

[neyret@localhost neyret]# service openafs-client status
Redirecting to /bin/systemctl status openafs-client.service
● openafs-client.service - OpenAFS Client Service
     Loaded: loaded (/usr/lib/systemd/system/openafs-client.service; disabled; vendor preset: >
     Active: active (running) since Fri 2021-08-27 14:19:41 CEST; 8s ago
    Process: 11982 ExecStartPre=/bin/sed -n w/etc/openafs/CellServDB /etc/openafs/CellServDB.l>
    Process: 11988 ExecStartPre=/bin/chmod 0644 /etc/openafs/CellServDB (code=exited, status=0>
    Process: 11989 ExecStartPre=/sbin/modprobe libafs (code=exited, status=0/SUCCESS)
    Process: 11993 ExecStart=/sbin/afsd $AFSD_ARGS (code=exited, status=0/SUCCESS)
      Tasks: 5 (limit: 9511)
     Memory: 17.0M
        CPU: 245ms
     CGroup: /system.slice/openafs-client.service
             ├─12001 /sbin/afsd -dynroot -fakestat -afsdb
             ├─12009 /sbin/afsd -dynroot -fakestat -afsdb
             ├─12011 /sbin/afsd -dynroot -fakestat -afsdb
             ├─12013 /sbin/afsd -dynroot -fakestat -afsdb
             └─12014 /sbin/afsd -dynroot -fakestat -afsdb

août 27 14:19:40 localhost systemd[1]: Starting OpenAFS Client Service...
août 27 14:19:41 localhost afsd[12015]: afsd: All AFS daemons started.
août 27 14:19:41 localhost afsd[11993]: afsd: All AFS daemons started.
août 27 14:19:41 localhost systemd[1]: Started OpenAFS Client Service.

But when I try to access to a remote file system it fails (it does the same with other servers):
/home/neyret 2> ll /afs/grand.central.org
ls: impossible d'accéder à '/afs/grand.central.org': Connexion terminée par expiration du délai d'attente

Here are the messages in dmesg:
[ 3154.892362] afs: Lost contact with file server 18.9.48.14 in cell grand.central.org (code -1) (all multi-homed ip addresses down for the server)
[ 3154.892372] RXAFS_GetCapabilities failed with code -1

Any idea about this problem ?

Thanks and best regards
Comment 4 Len Lawrence 2021-09-19 16:50:51 CEST
@Damien
The status message indicates that the service is disabled.
In the past when we have rebooted the system seemed to enable it automatically.
Maybe you could try
# systemctl enable openafs-client
# systemctl start openafs-client
and check status after that if there are no obvious errors.

CC: (none) => tarazed25

Comment 5 Thomas Andrews 2021-10-16 18:36:57 CEST
Probook 6550b, i3, Intel graphics, Intel wifi, 64-bit Plasma system. Openafs was not previously installed on this system. I know nothing about openafs, so pretty much stumbling blind here.

Installed backported kernel-desktop 5.14.10, including devel packages, disabled backports, and rebooted. Used qarepo to download the packages from Comment 2, then installed them and dependencies using MCC. No installation issues, so a successful build of the module was assumed. Rebooted once more, to be sure all was fully in place.

# systemctl enable openafs-client
Created symlink /etc/systemd/system/multi-user.target.wants/openafs-client.service → /usr/lib/systemd/system/openafs-client.service.
Created symlink /etc/systemd/system/remote-fs.target.wants/openafs-client.service → /usr/lib/systemd/system/openafs-client.service.
[root@localhost ~]# systemctl start openafs-client
[root@localhost ~]# systemctl status openafs-client
● openafs-client.service - OpenAFS Client Service
     Loaded: loaded (/usr/lib/systemd/system/openafs-client.service; enabled; vendor preset: disabled)
     Active: active (running) since Sat 2021-10-16 12:15:07 EDT; 29s ago
    Process: 7691 ExecStartPre=/bin/sed -n w/etc/openafs/CellServDB /etc/openafs/CellServDB.local /etc/openafs/CellServDB.>
    Process: 7692 ExecStartPre=/bin/chmod 0644 /etc/openafs/CellServDB (code=exited, status=0/SUCCESS)
    Process: 7693 ExecStartPre=/sbin/modprobe libafs (code=exited, status=0/SUCCESS)
    Process: 7696 ExecStart=/sbin/afsd $AFSD_ARGS (code=exited, status=0/SUCCESS)
      Tasks: 5 (limit: 9272)
     Memory: 7.1M
        CPU: 227ms
     CGroup: /system.slice/openafs-client.service
             ├─7704 /sbin/afsd -dynroot -fakestat -afsdb
             ├─7712 /sbin/afsd -dynroot -fakestat -afsdb
             ├─7713 /sbin/afsd -dynroot -fakestat -afsdb
             ├─7715 /sbin/afsd -dynroot -fakestat -afsdb
             └─7716 /sbin/afsd -dynroot -fakestat -afsdb

Oct 16 12:15:06 localhost systemd[1]: Starting OpenAFS Client Service...
Oct 16 12:15:07 localhost afsd[7717]: afsd: All AFS daemons started.
Oct 16 12:15:07 localhost afsd[7696]: afsd: All AFS daemons started.
Oct 16 12:15:07 localhost systemd[1]: Started OpenAFS Client Service.

No obvious errors there. The server was a different story.

[root@localhost ~]# systemctl enable openafs-server
Created symlink /etc/systemd/system/multi-user.target.wants/openafs-server.service → /usr/lib/systemd/system/openafs-server.service.
[root@localhost ~]# systemctl start openafs-server
[root@localhost ~]# systemctl status openafs-server
● openafs-server.service - OpenAFS Server Service
     Loaded: loaded (/usr/lib/systemd/system/openafs-server.service; enabled; vendor preset: disabled)
     Active: failed (Result: exit-code) since Sat 2021-10-16 12:24:53 EDT; 21s ago
    Process: 20898 ExecStart=/usr/sbin/bosserver -nofork $BOSSERVER_ARGS (code=exited, status=1/FAILURE)
   Main PID: 20898 (code=exited, status=1/FAILURE)
        CPU: 5ms

Oct 16 12:24:53 localhost systemd[1]: Started OpenAFS Server Service.
Oct 16 12:24:53 localhost bosserver[20898]: No non-loopback addresses found for host localhost
Oct 16 12:24:53 localhost systemd[1]: openafs-server.service: Main process exited, code=exited, status=1/FAILURE
Oct 16 12:24:53 localhost systemd[1]: openafs-server.service: Failed with result 'exit-code'.

This looks like I stubbed my toe on a chair in the dark to me - user error. I fully suspect that if someone who knew something about it were to try, it would be OK. 

Since this is an update candidate and not a backport, it needs to work with the current 5.10 series kernel, as well. If someone with a bit of knowledge could try that out and find it to be OK, it isn't much of a stretch to extend that OK to the backported kernel, too.

CC: (none) => andrewsfarm

Comment 6 Thomas Andrews 2021-10-26 21:02:37 CEST
I guess we are stuck with me, then. 

Installed openafs in a VirtualBox MGA8 Plasma guest, rebooted, then updated the packages using qarepo. No installation issues. Rebooted once more, then issued the same commands as in Comment 5, with the same results, indicating that at least the module was successfully built.

That's as far as I can take it. Calling it OK and validating.

Whiteboard: (none) => MGA8-64-OK
CC: (none) => sysadmin-bugs
Keywords: (none) => validated_update

Dave Hodgins 2021-10-26 23:20:30 CEST

Keywords: (none) => advisory
CC: (none) => davidwhodgins

Comment 7 Mageia Robot 2021-10-27 14:14:40 CEST
An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGAA-2021-0206.html

Resolution: (none) => FIXED
Status: NEW => RESOLVED

Comment 8 Damien Neyret 2021-11-09 13:48:53 CET
I didn't try again since a long time if openafs was working again. I tried with the openafs version 1.9.0-1.1.mga8, with kernel 5.14.14-desktop-1.mga8. There is no problem with the module compilation with dkms (gcc version 10.3.0-2.mga8).
The module is well loaded when I do service openafs-client start, here are the dmesg messages:

[ 4609.169910] Key type afs_pag registered
[ 4609.437610] afs: Binding rx to 0.0.0.0:7001
[ 4609.437835] enabling dynamically allocated vcaches
[ 4609.437837] Starting AFS cache scan...
[ 4610.553500] found 1 non-empty cache files (0%).
[ 4610.607498] audit: type=1130 audit(1636460828.058:175): pid=1 uid=0 auid=4294967295 ses=4294967295 msg='unit=openafs-client comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'

But when I try to access any afs site the access becomes stuck for a while and then failed:
/home/neyret 0> ll /afs/grand.central.org
ls: impossible d'ouvrir le répertoire '/afs/grand.central.org': Connexion terminée par expiration du délai d'attente

Here are the dmesg messages:
[ 4689.793703] afs: Lost contact with file server 128.2.172.19 in cell grand.central.org (code -1) (all multi-homed ip addresses down for the server)
[ 4689.793712] RXAFS_GetCapabilities failed with code -1

So in short the module compiles now well, but openafs accessses still do not work properly.

Thanks for your help

Resolution: FIXED => (none)
Status: RESOLVED => REOPENED

Comment 9 Thomas Backlund 2021-11-09 17:56:31 CET
Closing this again as the update has been pushed.

Please try with openafs-1.9.1-1.mga8 that just got built in Core Updates Testing and report how/if it works in:

https://bugs.mageia.org/show_bug.cgi?id=29639

Status: REOPENED => RESOLVED
Resolution: (none) => FIXED

Comment 10 Damien Neyret 2021-11-09 18:51:52 CET
Ok thanks, it seems it works with this new version.

Note You need to log in before you can comment on or make changes to this bug.