Bug#771298: openafs-modules-dkms: fails to build against v3.18 based kernels
I did just run into a openafs-modules-dkms build failure with current sid ( linux 3.16.7-ckt4-1 ) tried both 1.6.10-3 and 1.6.11~pre1-1 I get this error: /var/lib/dkms/openafs/1.6.10/build/src/libafs/MODLOAD-3.16.0-4-amd64- SP/afs_daemons.c:403:24: error: 'struct dentry' has no member named 'd_alias' After downgrading the linux headers to 3.16.7-ckt2-1 it was possible again to build them. tried both 1.6.10-3 and 1.6.11~pre1-1 and both built fine. With all the patches being backported to linux 3.16 i wouldn't be surprised if all this is related (so i sent it here) for now. -- Wolfgang Scheicher Institut fur ComputerGraphik und WissensVisualisierung Inffeldgasse 16c -- TU Graz -- A-8010 Graz -- Austria -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#771298: openafs-modules-dkms: fails to build against v3.18 based kernels
On Thu, 22 Jan 2015, Wolfgang Scheicher wrote: I did just run into a openafs-modules-dkms build failure with current sid ( linux 3.16.7-ckt4-1 ) tried both 1.6.10-3 and 1.6.11~pre1-1 I get this error: /var/lib/dkms/openafs/1.6.10/build/src/libafs/MODLOAD-3.16.0-4-amd64- SP/afs_daemons.c:403:24: error: 'struct dentry' has no member named 'd_alias' This particular issue is fixed by http://gerrit.openafs.org/11642 . The upcoming 1.6.11pre2 should build against 3.18 kernels without modification, but maybe I should pull in individual patches instead of waiting for 1.6.11pre2 to appear. -Ben -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#771298: openafs-modules-dkms: fails to build against v3.18 based kernels
On Sat, Nov 29, 2014 at 02:43:45PM -0500, Benjamin Kaduk wrote: Thanks for considering these patches for Debian. Hi Andy, I'm happy to bring support for openafs on the 3.18 kernel to Debian; it hadn't been a priority yet since Debian is running a somewhat older kernel. However, I'm more likely to do so by cherry-picking upstream's patches than by importing Ubuntu's patches. There's actually a 1.6.11pre1 upstream release candidate in the process of being announced (e.g., https://github.com/openafs/openafs/releases/tag/openafs-stable-1_6_11pre1), which should work on at least 3.18rc5 (I don't have data about rc6). Would you prefer to have 1.6.11pre1 in debian experimental or just the 3.18- (and 3.17-)support patches pulled into debian unstable? There probably would not be substantial harm from puttin 1.6.11pre1 itself into unstable, but my general preference is to put prereleases into experimental. For Ubuntu I think what we have will tide us over until the upstream release is ready (and in Debian) so I don't think there is any need to pull a pre-release if that is not your normal practice. It is likely to make it long before we are set in stone here anyhow. Thank you for considering these changes. -apw -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#771298: openafs-modules-dkms: fails to build against v3.18 based kernels
On Fri, 28 Nov 2014, Andy Whitcroft wrote: Package: openafs Version: 1.6.10-3 Severity: important Tags: patch User: ubuntu-de...@lists.ubuntu.com Usertags: origin-ubuntu vivid ubuntu-patch Dear Maintainer, In Ubuntu we are about to switch up to a v3.18 based kernel. It seems there are a trio of incompatibilities exposed there for the DKMS module. They all seem pretty safely fixed, and we have just uploaded that to Ubuntu for testing. In Ubuntu, the attached patch was applied to achieve the following: * Upstream v3.18 porting (LP: #1397275): - follow removal of key_type match op - follow removal of ctl_table typedef - d_invalidate now returns void Thanks for considering these patches for Debian. Hi Andy, I'm happy to bring support for openafs on the 3.18 kernel to Debian; it hadn't been a priority yet since Debian is running a somewhat older kernel. However, I'm more likely to do so by cherry-picking upstream's patches than by importing Ubuntu's patches. There's actually a 1.6.11pre1 upstream release candidate in the process of being announced (e.g., https://github.com/openafs/openafs/releases/tag/openafs-stable-1_6_11pre1), which should work on at least 3.18rc5 (I don't have data about rc6). Would you prefer to have 1.6.11pre1 in debian experimental or just the 3.18- (and 3.17-)support patches pulled into debian unstable? There probably would not be substantial harm from puttin 1.6.11pre1 itself into unstable, but my general preference is to put prereleases into experimental. Thanks, Ben -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#771298: openafs-modules-dkms: fails to build against v3.18 based kernels
Package: openafs
Version: 1.6.10-3
Severity: important
Tags: patch
User: ubuntu-de...@lists.ubuntu.com
Usertags: origin-ubuntu vivid ubuntu-patch
Dear Maintainer,
In Ubuntu we are about to switch up to a v3.18 based kernel. It seems
there are a trio of incompatibilities exposed there for the DKMS module.
They all seem pretty safely fixed, and we have just uploaded that to
Ubuntu for testing.
In Ubuntu, the attached patch was applied to achieve the following:
* Upstream v3.18 porting (LP: #1397275):
- follow removal of key_type match op
- follow removal of ctl_table typedef
- d_invalidate now returns void
Thanks for considering these patches for Debian.
-apw
-- System Information:
Debian Release: jessie/sid
APT prefers vivid-updates
APT policy: (500, 'vivid-updates'), (500, 'vivid-security'), (500, 'vivid'),
(100, 'vivid-backports')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 3.18.0-4-generic (SMP w/4 CPU cores)
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
diff -Nru openafs-1.6.10/debian/changelog openafs-1.6.10/debian/changelog
diff -Nru openafs-1.6.10/debian/patches/series openafs-1.6.10/debian/patches/series
--- openafs-1.6.10/debian/patches/series 2014-11-06 20:06:47.0 +
+++ openafs-1.6.10/debian/patches/series 2014-11-27 21:21:32.0 +
@@ -1,3 +1,6 @@
0001-userok.c-Fix-fixed-size-on-stack-path-buffers.patch
0002-Tweak-AFSDIR_PATH_MAX-definition.patch
0003-Add-dummy-exit-command-for-afsd-to-do-nothing.patch
+ubuntu-v3.18-key_type-removes-match.patch
+ubuntu-v3.18-typedef-ctl_table-removed.patch
+ubuntu-v3.18-d_invalidate-returns-void.patch
diff -Nru openafs-1.6.10/debian/patches/ubuntu-v3.18-d_invalidate-returns-void.patch openafs-1.6.10/debian/patches/ubuntu-v3.18-d_invalidate-returns-void.patch
--- openafs-1.6.10/debian/patches/ubuntu-v3.18-d_invalidate-returns-void.patch 1970-01-01 01:00:00.0 +0100
+++ openafs-1.6.10/debian/patches/ubuntu-v3.18-d_invalidate-returns-void.patch 2014-11-27 21:42:33.0 +
@@ -0,0 +1,61 @@
+Description: v3.18 porting -- d_invalidate now returns void
+ In v3.18 they have changed d_invalidate to return void as it can no
+ longer fail. Comment out the recovery code as it is no longer needed.
+Author: Andy Whitcroft a...@ubuntu.com
+---
+
+--- openafs-1.6.10.orig/src/afs/LINUX/osi_vcache.c
openafs-1.6.10/src/afs/LINUX/osi_vcache.c
+@@ -39,11 +39,15 @@ restart:
+ dget_locked(dentry);
+
+ spin_unlock(dcache_lock);
++#if defined(D_INVALIDATE_NOT_VOID)
+ if (d_invalidate(dentry) == -EBUSY) {
+ dput(dentry);
+ /* perhaps lock and try to continue? (use cur as head?) */
+ goto inuse;
+ }
++#else
++ d_invalidate(dentry);
++#endif
+ dput(dentry);
+ spin_lock(dcache_lock);
+ goto restart;
+@@ -71,11 +75,15 @@ restart:
+ dget(dentry);
+
+ spin_unlock(inode-i_lock);
++#if defined(D_INVALIDATE_NOT_VOID)
+ if (d_invalidate(dentry) == -EBUSY) {
+ dput(dentry);
+ /* perhaps lock and try to continue? (use cur as head?) */
+ goto inuse;
+ }
++#else
++ d_invalidate(dentry);
++#endif
+ dput(dentry);
+ spin_lock(inode-i_lock);
+ goto restart;
+--- openafs-1.6.10.orig/src/cf/linux-test4.m4
openafs-1.6.10/src/cf/linux-test4.m4
+@@ -182,6 +182,18 @@ AC_DEFUN([LINUX_INODE_SETATTR_RETURN_TYP
+ ])
+
+
++AC_DEFUN([LINUX_D_INVALIDATE_RETURN_TYPE], [
++ AC_CHECK_LINUX_BUILD([for d_invalidate return type],
++ [ac_cv_linux_func_d_invalidater_returns_int],
++ [#include linux/dentry.h],
++ [struct dentry _dentry;
++ int i;
++ i = d_invalidate(_dentry);],
++ [D_INVALIDATE_NOT_VOID],
++ [define if your d_invalidate return return non-void],
++ [])
++])
++
+
+ AC_DEFUN([LINUX_AOP_WRITEBACK_CONTROL], [
+ AC_CHECK_LINUX_BUILD([whether aop.writepage takes a writeback_control],
diff -Nru openafs-1.6.10/debian/patches/ubuntu-v3.18-key_type-removes-match.patch openafs-1.6.10/debian/patches/ubuntu-v3.18-key_type-removes-match.patch
--- openafs-1.6.10/debian/patches/ubuntu-v3.18-key_type-removes-match.patch 1970-01-01 01:00:00.0 +0100
+++ openafs-1.6.10/debian/patches/ubuntu-v3.18-key_type-removes-match.patch 2014-11-27 21:42:13.0 +
@@ -0,0 +1,46 @@
+Description: v3.18 porting -- follow removal of key_type match op
+ In v3.18 they have removed the key_type match op. This now defaults to
+ a direct match. As this is what the afs match does we can simply stop
+ supplying our match.
+Author: Andy Whitcroft a...@ubuntu.com
+---
+
+Index: openafs-1.6.10/acinclude.m4
+===
+--- openafs-1.6.10.orig/acinclude.m4
openafs-1.6.10/acinclude.m4
+@@ -842,6 +842,7 @@ case $AFS_SYSNAME in *_linux* | *_umlinu
+ AC_CHECK_LINUX_STRUCT([inode_operations], [truncate], [fs.h])
+ AC_CHECK_LINUX_STRUCT([key_type], [preparse], [key-type.h])
+
