In linux's 'struct inode', i_nlink is now read-only for some reason. They provide an inline function to set it: set_nlink(struct inode *, unsigned)
Attaching a patch...
Okay i'm having trouble with uploading files... probably due to my KDE update. Here's the patch:
--- openafs-1.6.0/src/afs/LINUX/osi_vfsops.c 2012-01-07 17:50:01.437528708 -0800
+++ openafs-1.6.0/src/afs/LINUX/osi_vfsops.c 2012-01-07 17:50:39.531045712 -0800
@@ -24,6 +24,10 @@
+# include <linux/version.h>
struct vcache *afs_globalVp = 0;
struct vfs *afs_globalVFS = 0;
struct vfsmount *afs_cacheMnt;
@@ -439,7 +443,11 @@
vattr2inode(struct inode *ip, struct vattr *vp)
ip->i_ino = vp->va_nodeid;
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,2,0)
+ set_nlink(ip, vp->va_nlink);
ip->i_nlink = vp->va_nlink;
ip->i_blocks = vp->va_blocks;
ip->i_blkbits = AFS_BLKBITS;
Here's the commit that broke openafs-kernel:
It doesn't seem like a big deal...
please try the latest 1.6.1 pre1 ebuild in portage and report.
Failed for me using gentoo-sources-3.2.0-r1, openafs-1.6.1_pre1, and heimdal-1.5.1-r1. If this was supposed to work against a 3.2 kernel, what test case *was* found to work? I would guess it was probably tested against mit-krb but not heimdal. (Just guessing.) More at Bug #398253.
I can't do it right now; 1.6.1_pre1 has the same fix according to git.openafs.org so I'm satisfied.
The heimdal stuff isn't relevant to this bug, and probably linux-3.2 isn't relevant to bug #398253.