The patches are all in /afs/andrew.cmu.edu/usr18/cg2v/www/openafs-patches, aka http://www.contrib.andrew.cmu.edu/~cg2v/openafs-patches
Descriptions follow:
linux-ioctl-update: The linux ioctl inode op is evolving, mostly to eliminate it's use of the 'big kernel lock'. A new operation, unlocked_ioctl was added to file_operations. if unlocked_ioctl is present, it is used to handle ioctl operations instead of ioctl.
In addition, another new operation was added, compat_ioctl, which is used when a 32 bit process calls ioctl on a 64 bit kernel (instead of the old way where the 32 bit handlers were registered on a globally on per-ioctl-number basis, rather than being part of the driver or filesystem.)
The nifty thing about this new feature is that it's self identifying!. no new configure tests are needed. details at <http://lwn.net/Articles/119652/>
In addition, this patch introduces the use of a new 'feature' define, NEED_IOCTL32, in place of the expression 'defined(AFS_LINUX_64BIT_KERNEL) && !defined(AFS_ALPHA_LINUX20_ENV) && !defi
ned(AFS_IA64_LINUX20_ENV)' One of the things I'd like people to comment on is whether this sort of thing is a good idea, and if it should be done in more places.
rx_clock-rollover: this will fix RT ticket 17990, which is the cause of the mysterious bosserver deaths that have been reported recently. It changes the way that the rx_clock module uses the interval timer. more details are in the ticket.
tbutc: tbutc is currently build against a mix of pthread-based and lwp-based libraries. While this does not appear to be causing problems, it's not a good idea. This patch replaces the lwp libraries with pthread libraries. I need someone who actually uses tbutc to build with this patch and make sure tbutc still works....
p7sKE14RLdZro.p7s
Description: S/MIME cryptographic signature
