Avoid asymmetry between rx connections and clients, such that one
points at the other but the other doesn't point at the first one.
This can sometimes happen with hosts that change addresses/ports.
if something else ends up trying to e.g. break delayed callbacks over the same
callback that we're playing with in break later callbacks, the manipulation in
CDelPtr becomes important
FIXES 1295
Make cc only take the first depency, so it wont compile header files that
the target is depends on.
Now that there isn't any VPATH, depend on the file with path, not by name.
add warning and don't clean up if host is locked. shouldn't be able to happen
====================
This delta was composed from multiple commits as part of the CVS->Git migration.
The checkin message with each commit was inconsistent.
The following are the additional commit messages.
====================
FIXES 1294
and make it useful
Stop passing around needless arguments to rxi_SendAck.
Also send the proper reason value in ack packets for
packets that were received out-of-sequence.
in rx_pthread.c the
++rxi_pthread_hinum;
has to be protected by the rx_stats_mutex in all three cases. In the file I
attached in the openafs-devel article the last one was accidentally
unprotected.
the FSYNC_sync thread (fssync.c) needs needs to be handled as well.
Don't lock against myself in reclaim vnodeop.
====================
This delta was composed from multiple commits as part of the CVS->Git migration.
The checkin message with each commit was inconsistent.
The following are the additional commit messages.
====================
Various stability fixes:
Call vgone() instead of afs_FlushVCache() when freeing vcache
Acquire AFS_GLOCK in VOP_INACTIVE
Don't bother calling cache_purge() or uvm_vnp_uncache() in reclaim();
vclean() already does this
Acquire xvcache lock in reclaim()
rx_knet fixes for Darwin and FreeBSD:
netreceive: fix memory leak, check return code from soreceive
netsend: remove unnecessary mbuf alloc, remove misleading comment
all: general cleanup and minor bug fixes
Thanks to emoy@apple.com for reporting this bug and testing the fix
rxi_FindIfnet: return NULL on no match, instead of first interface on list
use ifa_ifwithnet on XBSD (was only used on Darwin)
remove dead code and unused args
Avoid a subtle cycle in the locking order hierarchy involving
the rxi_ServerThreadSelectingCall flag, and instead rely on
call->lock to make sure we don't attach the same call twice.
Also some miscellaneous cleanup and code unification.
