2004-10-21 15:01 shadow * src/afs/LINUX/osi_vnodeops.c: DELTA fstracing on a linux machine and trying to fs flush a file in the cache would tr igger a hang on a machine with SMP kernel. fix it. STABLE12-fstrace-avoid-double-glock-in-dentry-code-during-fstrace-2 0040719 AUTHOR shadow@dementia.org FIXES 2004-10-20 07:41 shadow * src/viced/callback.c: DELTA since it looks like CheckHost is using a gc'd client conn, this more or less has to be it. and don't deadlock. STABLE12-viced-callback-lock-host-during-multi-breakcallback-200410 11 AUTHOR shadow@dementia.org FIXES 2004-10-18 06:38 shadow * src/afs/DARWIN/osi_vnodeops.c: DELTA STABLE12-darwin-updates-20040623 AUTHOR shadow@dementia.org FIXES this should be switched to use the freebsd style getnewvnode perhaps it will be 2004-10-18 06:36 shadow * src/viced/host.c: DELTA STABLE12-findclient-print-stolen-clients-20040805 AUTHOR shadow@dementia.org FIXES when a client is "stolen" due to a race, print it. 2004-10-18 06:26 shadow * src/vol/vol-salvage.c: DELTA STABLE12-vol-fix-nuke-iteration-and-nuke-globals-20040928 AUTHOR shadow@dementia.org allInodes global dies; don't use ti as an iterator right after freeing it. 2004-10-18 06:09 shadow * src/afs/LINUX/osi_vnodeops.c: DELTA STABLE12-linux-pagelaunder-avoid-deadlock-20041012 AUTHOR jbuehler@hekimian.com FIXES 15239 mmap'd files can deadlock on linux; fix by not letting kernel free pages out of our space predictively 2004-10-18 06:04 shadow * src/afs/: afs_call.c, afs_osi.c, DARWIN/osi_prototypes.h, DARWIN/osi_sleep.c: DELTA STABLE12-darwin-afsdb-handler-signal-mask-20040728 AUTHOR hollandp@umich.edu FIXES 6027 mask signals on darwin so afsdb handler doesn't die 2004-10-18 05:58 shadow * src/afs/LINUX/osi_vnodeops.c: DELTA fstracing on a linux machine and trying to fs flush a file in the cache would tr igger a hang on a machine with SMP kernel. fix it. STABLE12-fstrace-avoid-double-glock-in-dentry-code-during-fstrace-2 0040719 AUTHOR shadow@dementia.org FIXES 2004-10-18 05:55 shadow * src/afs/afs_vcache.c: DELTA STABLE12-banish-fstrace-trydcachechildren-entirely-20040818 AUTHOR shadow@dementia.org we really don't need this... 2004-10-18 05:54 shadow * src/afs/LINUX/osi_alloc.c: DELTA STABLE12-linux-free-dont-drop-alloc-sem-20040714 AUTHOR chas@cmf.nrl.navy.mil FIXES 5831 dropping semaphore not needed, so don't 2004-10-18 05:52 shadow * src/: lwp/iomgr.c, rx/rx_lwp.c, vol/fssync.c, rx/rx_globals.c: DELTA STABLE12-rx-lwp-fdsetsize-20040708 AUTHOR jhutz@cmu.edu FIXES 5615 limit our fd set size so we don't "lose" fds. 2004-10-18 05:45 shadow * src/: afs/afs_vcache.c, afs/DARWIN/osi_vm.c, afs/DARWIN/osi_vnodeops.c, afs/VNOPS/afs_vnop_remove.c, config/param.ppc_darwin_12.h, config/param.ppc_darwin_12_usr.h, config/param.ppc_darwin_13.h, config/param.ppc_darwin_13_usr.h, config/param.ppc_darwin_14.h, config/param.ppc_darwin_14_usr.h, config/param.ppc_darwin_70.h, config/param.ppc_darwin_70_usr.h: DELTA STABLE12-darwin-updates-20040623 AUTHOR shadow@dementia.org FIXES this should be switched to use the freebsd style getnewvnode perhaps it will be 2004-10-18 05:23 shadow * src/volser/vsprocs.c: DELTA STABLE12-vos-new-repsite-cleanup-20040601 AUTHOR shadow@dementia.org FIXES 4730 if for some reason we got into a situation where all ROs were marked NEW_REPSITE in the vldb, vos release without -f deals very poorly. force a -f, and also, ne ver print "Old Release" for the RW, to avoid confusion 2004-10-18 05:20 shadow * src/rx/rx.c: DELTA STABLE12-rx-debug-20040529 AUTHOR jaltman@mit.edu Fix MakeDebugCall to properly check the return code of select() and recvfrom(). Do not copy data buffer unless recvfrom() succeeds. 2004-10-18 05:12 shadow * src/rx/rx_packet.c: DELTA STABLE12-rx-protect-queue-during-debug-20040601 AUTHOR rtb@pclella.cern.ch FIXES 4811 "in rx_packet.c:rxi_ReceiveDebugPacket() the rx_idleServerQueue is scanned without taking a lock first. The queue_Remove macro zeroes the ->next field (luckily), but this can lead to a crash when Murphy strikes and the entry gets removed during the scan: queue_Scan will dereference 0x0 -> next. " 2004-10-18 05:11 shadow * src/afs/DARWIN/osi_vnodeops.c: DELTA STABLE12-darwin-pageout-limit-to-filesize-20040405 AUTHOR cg2v@andrew.cmu.edu FIXES 3870 previously we weren't limiting to file size, now we are 2004-10-18 05:08 shadow * acconfig.h, acinclude.m4, src/auth/cellconfig.c, src/cf/ressearch.m4: DELTA STABLE12-macos-build-update-20040310 AUTHOR shadow@dementia.org FIXES make the head build again. 2004-10-18 05:00 shadow * src/afs/DARWIN/osi_vnodeops.c: DELTA STABLE12-darwin-avoid-hang-on-rename-20040202 AUTHOR ssen@apple.com FIXES 2967 this version deals with afs being multiply mounted. return EXDEV on cross device rename attempts as upper layer doesn't deal for us 2004-10-18 04:59 shadow * src/kauth/kaprocs.c: DELTA STABLE12-kaanswertoolong-20040420 AUTHOR jaltman@mit.edu when determining whether or not the ticket len is too long for the receiver to accept we want to compare against the actual ticket length and not MAXKTCTICKETLEN which is what is used when sizeof(ktc_ticketAnswer) is used for comparison. 2004-10-18 04:55 shadow * src/afs/LINUX/osi_vfsops.c: DELTA STABLE12-linux-null-superblock-if-mount-failed-20040728 AUTHOR rainer.schoepf@proteosys.com FIXES 6002 The appended patch resets afs_globalVFS to NULL if /afs cannot be mounted, eg., if no cell server is accessible. There are two advantages: - it is possible to do "afsd -shutdown" to terminate all daemons - more importantly, if /afs cannot be mounted, the linux kernel will probably free the superblock, ie the memory pointed to afs_globalVFS. 2004-10-18 04:53 shadow * src/ubik/beacon.c: DELTA STABLE12-ubik-multihome-byte-order-problem-20040818 AUTHOR reuter@rzg.mpg.de "Playing around with multi-address database servers on little-endian machines I found out that there is one conversion of the ip-address too much" 2004-10-18 04:50 shadow * src/vol/: listinodes.c, namei_ops.c, namei_ops.h, ntops.c, nuke.c, vol-salvage.c: DELTA STABLE12-vol-fix-nuke-iteration-and-nuke-globals-20040928 AUTHOR shadow@dementia.org allInodes global dies; don't use ti as an iterator right after freeing it. 2004-10-18 04:32 shadow * src/ptserver/ptuser.c: DELTA STABLE12-pr-initialize-dont-leak-ref-20041009 AUTHOR shadow@dementia.org FIXES we were leaking an afsconf_Open ref. let's not. 2004-10-18 04:30 shadow * src/afs/SOLARIS/osi_vnodeops.c: DELTA STABLE12-solaris-fix-ulimit-write-size-bug-20041012 AUTHOR ericm@electric-cloud.com FIXES 15609 convert nonsensical comparison to one which makes sense in light of code and comments below it 2004-10-18 03:34 shadow * src/rx/rx.c: DELTA nWaiting ended up negative, because the dequeuer (GetCall) decremented. STABLE12-attachserverproc-dont-dequeue-if-not-in-queue-take-two-200 40818 AUTHOR shadow@dementia.org 2004-10-18 03:31 shadow * src/rx/rx.c: DELTA STABLE12-attachserverproc-dont-dequeue-if-not-in-queue-20040602 AUTHOR shadow@dementia.org FIXES 4699 can't see what else this can be. pile of details in the ticket. 2004-10-18 03:29 shadow * src/rx/rx.c: DELTA STABLE12-partly-revert-rx-cleanup-20040804 AUTHOR shadow@dementia.org if there's a callback connection to a multihomed client, you need this or you end up with multiple connections, one per IP, being made from the single connection 2004-10-18 03:27 shadow * src/rx/: rx.c, rx.h: DELTA STABLE12-rx-getputconnection-20041011 AUTHOR jhutz@cmu.edu FIXES 15584 add Get/PutConnection funcs to diddle conn refCounts. 2004-10-18 03:24 shadow * src/rx/rx.c: DELTA STABLE12-cleanup-rx-peer-leak-20041012 AUTHOR jhutz@cmu.edu FIXES if the peer was the second in the chain we could throw away our reference to the first, effectively leaking it forever. 2004-10-18 03:22 shadow * src/viced/host.c: DELTA STABLE12-h-tossstuff-r-leak-20041013 AUTHOR cg2v@andrew.cmu.edu FIXES don't leak host hash objects in h_TossStuff_r 2004-10-18 03:21 shadow * src/viced/host.c: DELTA STABLE12-findclient-print-stolen-clients-20040805 AUTHOR shadow@dementia.org FIXES when a client is "stolen" due to a race, print it. 2004-10-18 03:19 shadow * src/viced/callback.c: DELTA STABLE12-callback-rxcon-protection-20031114 AUTHOR shadow@dementia.org FIXES 2498 CheckHost() sets the HOSTDELETED flag on a host, does its touch, calls H_UNLOCK at the end of its function; AddCallBCal1 in a different thread grabs the lock, does its thing. Gets to GetSomeSpace_r(), which calls h_Enumerate_r; the host in question isn't held, so h_Enumerate_r calls h_Hold_r and h_Release_r on the host that CheckHost() set the HOSTDELETED flag on; h_Release_r sees the HOSTDELETED flag and calls h_TossStuff_r, poof, we have our broken host entry for ClearHostCallbacks_r to trip 2004-10-18 03:15 shadow * src/viced/afsfileprocs.c: DELTA STABLE12-rpc-rename-avoid-loop-20031201 AUTHOR shadow@dementia.org FIXES enumerated solution (n vnodes cannot yield n+1 loop passes) suggested by jhutz@cmu.edu. avoid looping forever on a corrupt parent. 2004-10-18 03:09 shadow * src/rx/: rx.c, rx.h: DELTA STABLE12-rx-peer-and-conn-refcounts-now-unsigned-int32-20041015 AUTHOR shadow@dementia.org FIXES 15459 in an extreme case, the peer object refcount could wrap we should probably also log if wrapping is imminent