STABLE12-linux-null-superblock-if-mount-failed-20040728

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.

(cherry picked from commit e5e7801db895b1e57ff99fb59371aee47ac33b9c)

STABLE12-ubik-multihome-byte-order-problem-20040818

"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"

(cherry picked from commit 6f1cd5a43e1240ecc7a964b8236e25e1d0e46fca)

STABLE12-pr-initialize-dont-leak-ref-20041009

we were leaking an afsconf_Open ref. let's not.

(cherry picked from commit d114c46e85624c1988a7f61a872251660c7d52c4)

STABLE12-solaris-fix-ulimit-write-size-bug-20041012

FIXES 15609

convert nonsensical comparison to one which makes sense in light of code and
comments below it

(cherry picked from commit 512f3b0555c10d26ea6b05695471481df11fe1c9)

STABLE12-attachserverproc-dont-dequeue-if-not-in-queue-take-two-20040818

nWaiting ended up negative, because the dequeuer (GetCall) decremented.

(cherry picked from commit 399785acf166dac15123889c28974ff06d6d2de0)

STABLE12-attachserverproc-dont-dequeue-if-not-in-queue-20040602

FIXES 4699

can't see what else this can be. pile of details in the ticket.

(cherry picked from commit 5d9fcb30347e97471e5604828ef433683ca7a766)

STABLE12-partly-revert-rx-cleanup-20040804

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

(cherry picked from commit b4566d725e1aa4f57d1e6db5821c590a4b6da7c0)

STABLE12-rx-getputconnection-20041011

FIXES 15584

add Get/PutConnection funcs to diddle conn refCounts.

(cherry picked from commit 3b9cdc23b5122e371139828e557732c5e434230a)

STABLE12-cleanup-rx-peer-leak-20041012

if the peer was the second in the chain we could throw away our reference to
the first, effectively leaking it forever.

(cherry picked from commit e106a39fe62a8d338ac7f8cf4ed857c97a9b3484)

STABLE12-h-tossstuff-r-leak-20041013

don't leak host hash objects in h_TossStuff_r

(cherry picked from commit 8be8a671c5dfa478b571c70d6c6a9757ebec553d)

STABLE12-findclient-print-stolen-clients-20040805

when a client is "stolen" due to a race, print it.

STABLE12-callback-rxcon-protection-20031114

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

(cherry picked from commit 4458dd4410207fcdf8dec6789f0397adc6ce7b01)

STABLE12-rpc-rename-avoid-loop-20031201

enumerated solution (n vnodes cannot yield n+1 loop passes) suggested by
jhutz@cmu.edu. avoid looping forever on a corrupt parent.

(cherry picked from commit 1c5a7b4ea135dcccadea1775922a783c5fc42989)

STABLE12-rx-peer-and-conn-refcounts-now-unsigned-int32-20041015

FIXES 15459

in an extreme case, the peer object refcount could wrap

we should probably also log if wrapping is imminent

(cherry picked from commit 85835f169cf47207795e00cbc2dae6fd77950f78)

STABLE12-undo-relnotes-1-2-11-20040110

remove extra fix

STABLE12-relnotes-1-2-11-20040110

add extra fix

STABLE12-undo-release-zero-length-dcache-entries-20040110

as from kolya@mit.edu's rewrite of this code for 1.3.x

STABLE12-release-zero-length-dcache-entries-20040110

as from kolya@mit.edu's rewrite of this code for 1.3.x

STABLE12-openafs-1-2-11-relnotes-20040110

have some release notes

STABLE12-res-search-test-shouldnt-use-res-close-20030915

FIXES 2099

res_close() not part of generally accepted resolver api

(cherry picked from commit ea4a465aa48fd79c6b04a3f9bb438d9c88daa85a)

STABLE12-hello-openafs-1-2-11-20040110

damn it.

STABLE12-namei-ihandle-leak-20031114

FIXES 2518

flush all dir buffers we have cached when rewriting a volume from a restore

avoids holding an ihandle which has an open fdhandle for an unlinked file

(cherry picked from commit 79b67c7461097ab5ca1dac9b9c1f069dfb4d290d)

STABLE12-vol-linux-dont-assume-ext2-20031208

to deal simply with /usr/include/linux potentially having 2.6 kernel headers
in newer linuxes. given that we might not have ext2 as root, this check
probably wasn't sufficient anyway.

(cherry picked from commit 36ac69eedca1b2de6119ddd19b99dd302d5a496b)

STABLE12-ubik-election-epoch-fix-20040110

ubik's ability to keep quorum broke when unix time passed 2^30 seconds.
Buy us another 2^30 seconds.

(cherry picked from commit 276df5918b057765e50b4c8a48ec8060f8abe5a7)

STABLE12-macos103-20031024

FIXES 2325

support for macos 10.3 for 1.2.10

(cherry picked from commit e2e93aa8920c0b1bfc672a555a59eb4e15dbeaae)

STABLE12-hpux-build-fixes-20030805

(cherry picked from commit 937a911c674551bf66ac440da43e329ba6b0fda2)

STABLE12-update-news-20030729

mention butc

STABLE12-release-notes-1-2-10-20030729

make new release notes

STABLE12-butm-restore-from-large-files-20030729

make restore also work from large files

(cherry picked from commit f7c4f16110497457263851fed22a0c2ed50521c0)

STABLE12-opteron-linux-sys-call-hack-fix-20030728

smp machines need a larger search space?

(cherry picked from commit 42396526a7f141ed1d03ba1b2451fc7475773258)

STABLE12-package-build-rules-20030715

allow building of macos packages from make

(cherry picked from commit cd84e65f972ef53b45df742dfef67c04e94c95fd)

STABLE12-create-openafs-1_2_10-20030725

Begin 1.2.10 release cycle.

STABLE12-use-strerror-if-available-20030725

discover and use strerror instead of sys_errlist and sys_nerr.

(cherry picked from commit 7356b6696e7b22f0d130fe27ef6233bc251b9a0d)

STABLE12-linux-largefile-defns-20030724

make usd library able to write large files on i386 linux

(cherry picked from commit 26891bbac0e4aea5402bf44651d284194d503270)

STABLE12-butc-notice-eot-20030723

if we're going to check for BUTM_EOT, have a chance in hell of finding it.
and not, say, 0.

(cherry picked from commit 1a2fa8bd5cca6d69f13b5da47cc175d6337dabcd)

STABLE12-hpux-make-packaging-target-20030715

FIXES 1315

make the make rule able to make packages

(cherry picked from commit 40d7a8b4f144d13628ac18b35e8c861af35beab6)

STABLE12-winnt-fileinfo-is-directory-20030710

correctly report directories by not setting field in incorrect position
in reply.

(cherry picked from commit 7eeb271d9f12498b77f90a28c063bdfcb22afa39)

STABLE12-linux-usermode-better-test-20030710

test the source tree and not the running kernel

(cherry picked from commit ddac9ce59abee2f53d6243b05c502966dfc9ad56)

STABLE12-package-build-rules-20030709

allow building of macos packages from make

(cherry picked from commit 378de32cbca1bbe3bf605f72a222ec0a655a8c6b)

STABLE12-macos-pkg-update-csdb-20030709

update the cellservdb while we build packages

(cherry picked from commit c59e18c174569393ada98bb2ff1ab80c51c07b2b)

STABLE12-packagings-20030709

pull up hpux and macos packaging

STABLE12-kerneldir-cleanup-20030707

FIXES 1706

"Now that the make variables KERNELDIR and UKERNELDIR are
consistently used with a trailing slash
([grand.central.org #1655] resp. STABLE12-makefile-consistency-20030701)
they should be defined without trailing slash.

This certainly doesn't change any functionality but avoids
funny looking doble slashes in the expanded make rules."

STABLE12-cleanup-dbg-and-ldflags-20030701

FIXES 1653

switch from -g to ${DBG}; remove -g from LDFLAGS.

STABLE12-darwin-kext-versions-20030702

we can't call it 1.2.10, so it's 1.2.9f1. bleah.

STABLE12-gcc-warnings-cleanup-20030701

FIXES 1650

gcc warnings cleanup; add headers, fix castings

(cherry picked from commit 88e2d8e292e687d3193dcaff62ffd5f2ab4f72ad)

STABLE12-windows-file-versioning-20030619

FIXES 1488

pull up versioning for windows files

(cherry picked from commit bc9ff0f64010b5ed886389ed584fb1c7c8169dc5)

STABLE12-translate-more-vol-errors-20030701

based on complaint from Ryan Underwood on openafs-info.
translate to EDQUOT and ENOSPC actually instead of just returning random
numbers

(cherry picked from commit a633a9920803860e8b4167501f0ffa171fcf40c2)

STABLE12-linux-i-alloc-sem-usage-20030701

modern kernels appear to use i_alloc_sem around truncation

(cherry picked from commit 239067b1476801254c502c348893bb49553b5e74)

STABLE12-winnt-avoid-nosuchvolume-for-down-servers-20030619

FIXES 1568

a volume could be marked "no such volume" and hence disappear if a server
went down until fs checkv.

(cherry picked from commit 43011a3a0013aa17a941f8ac0abf87f3efe7df7b)

STABLE12-winnt-reset-smb-listener-flags-20030701

FIXES 1641

reset flags on next pass through loop

STABLE12-winnt-enable-global-drive-mappings-20030701

FIXES 1614

This will allow access to AFS from system processes and scheduled tasks.
It also means that each user does not have to map AFS to his own
drive.

(cherry picked from commit 4bda27a286acb45ce87131eb43d80dbcbe5e5b39)

STABLE12-ptserver-make-main-int-20030701

FIXES 1645

don't declare main void.

(cherry picked from commit f661ad4e58f6ad7fd9aea9755c141e0a2d742b66)

STABLE12-fix-static-decl-conflict-for-localcell-20030701

FIXES 1646

declared static, make header match

STABLE12-libadmin-pts-pointer-fixes-20030701

FIXES 1647

fix pointers, kill warnings

(cherry picked from commit 9d1570615849e9f83e78d365fb98df743f512550)

STABLE12-use-lncurses-instead-of-path-20030701

FIXES 1648

use -lncurses instead of explicit path

(cherry picked from commit 4fc1bbdf9406bf28e9027d68d3b861ded9b65fd5)

STABLE12-ignore-pam-module-if-no-headers-20030701

FIXES 1649

we check for pam headers; use the result

(cherry picked from commit 96c7261a229207a44e483cce52478fbb1b35fe96)

STABLE12-eliminate-bogus-rebuilding-20030701

FIXES 1651

so makefiles have proper dependancies

(cherry picked from commit 010af620459e2af0841810f1b232eccd7bc4ccb1)

STABLE12-kill-usr-include-20030701

FIXES 1652

kill explicit makefile refs to -I/usr/include.

STABLE12-cleanup-dbg-and-ldflags-20030701

FIXES 1653

switch from -g to ${DBG}; remove -g from LDFLAGS.

STABLE12-kdump-build-cleanup-20030701

FIXES 1654

don't rebuild kdump on each build

(cherry picked from commit a75f0516f52d967a1ffa0088fddaff88fe3f40bb)

STABLE12-makefile-consistency-20030701

FIXES 1655

apply leading / ourselves for consistency

STABLE12-bozo-ubik-needs-rx-20030627

make ubik depend on rx, not vice-versa

(cherry picked from commit 819df73c887c7999cf8163e0e95477a88a67490a)

STABLE12-rx-xdr-rename-userland-symbols-also-on-amd64-linux-20030627

pull up fix for amd64 xdr problem

(cherry picked from commit def59db27bb8a6b2b123547b30054d8ff380366e)

STABLE12-use-mkstemp-correctly-on-linux-20030619

FIXES 1603

might as well fix it all.

(cherry picked from commit 7f16752202c603eadb6d88ae4cee8b55e8c66436)

STABLE12-check-mkstemp-in-config-20030617

check for mkstemp in configure
use HAVE_MKSTEMP in src/butc/tcudbprocs.c

(cherry picked from commit a8dd10c862abeba2c05809b9fd921293b7e5e5e6)

STABLE12-use-mkstemp-20030617

have the BSDs use mkstemp(3)

(cherry picked from commit ae5e2f7865ce1dee9649aa6992f49fc89eb52669)

STABLE12-rxkad-check-tkt-length-20030610

tkt_DecodeTicket didn't seem to check that ticketLen < MAXKTCTICKETLEN

didn't matter since no callers failed to check, but we should enforce this where the limit is

(cherry picked from commit 8a33c2d8a590d92df27d7febf708449feae17254)

STABLE12-opteron-no-modversions-ia32-sys-call-table-fun-20030610

attempt to deal if modversions is disabled

(cherry picked from commit 10a00664e344fd8d8bb2089c5e8a43a62050a5ed)

STABLE12-winnt-give-better-debugging-20030606

offer better messages in certain cases

(cherry picked from commit b609b807d24d336dfe8a0939249001dc8874c69a)

STABLE12-linux-rx-listener-flush-signals-20030605

FIXES 1409

make sure we flush signals before calling force_sig. i don't know why this
matters, but it does.

(cherry picked from commit ad61a30d57eaf91d6c4edde632459b7e3d9f8700)

STABLE12-windows-version-20030605

update version now

STABLE12-news-20030605

update news to be current

STABLE12-rx-avoid-leaking-nwaiting-20030605

thanks to kolya@mit.edu for suggesting moving rather than adding the
decrement and confirming sanity.

basically if a call ended up call->error it was possible to queue_Remove()
it, drop the call, and not rx_nWaiting--; result should be observed as
rxdebug reporting waitprocs when threads were idle.

(cherry picked from commit 3fd6efb40fb4b81874cd1d48a1b529a175383e55)

STABLE12-amd64-hook-ia32-table-20030519

if sys_call_table is exported and ia32_sys_call_table isn't, we need it this way

(cherry picked from commit 718f25e62e7249682f8e9deddb1eb72408f4df9e)

STABLE12-amd64-hook-ia32-table-20030521

keep track of the ia32 table separately. annoying but needed.

====================
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.
====================

provide the variables we need for ia32 sys call table if
sys_call_table is exported

(cherry picked from commit 2acd0d9937f6040ab8b9dc0ce3304706136fac87)

STABLE12-softsig-update-20030602

FIXES 1435

changes to help us die and to unblock all signals to wait on

(cherry picked from commit ee00daa36432ba6eac6004c6c691aaab98c0e62f)

STABLE12-i386-user-mode-linux-support-20030513

FIXES 1396

usermode linux support for 1.2.x

(cherry picked from commit fcdc120ce70a189ca7ed908688721b72ab3ffe77)

STABLE12-closer-to-vc7-support-20030528

FIXES 1334

VC7 dropped support for the LARGE_INTEGER functions (linking against
largeint.lib doesn't seem to have any effect)

(cherry picked from commit 5ecdab3b75c02784ae6d87e93a92d1a3f705c83d)

STABLE12-hpux11i-ia64-hpux-1122-20030523

FIXES 1490

so 1.2.10 will support hpux 11i and 11.22

STABLE12-linux-system-calls-return-long-20030522

the kernel says they return long. returning int causes sadness. stop being sad.

(cherry picked from commit ce4f08f0a9ab6fc7a2c40267aebc48934ec6dfb9)

STABLE12-revert-pioctl-return-negative-errors-on-linux-20030522

syscall stub handles it. this is actually an overflow on 64 bit system

(cherry picked from commit 20656c4a44c4ae294e99eaa60cf8137a2fda51d3)

STABLE12-pioctl-return-negative-errors-on-linux-20030522

otherwise we can end up with errors that make no sense.

(cherry picked from commit 7d710fd45f7f6fa06de22d4f524342d02ae4afcd)

STABLE12-switch-to-strerror-20030522

FIXES 1458

your brights brighter, your whites whiter. well, at least compilers won't bitch
that sys_errlist[] isn't supported anymore.

(cherry picked from commit 580f1ffc506bb925c28af824e08f84318f1dca64)

STABLE12-linux-group-syscalls-return-long-20030521

these return long. also call the underlying handler expecting to get a long
back

(cherry picked from commit 4b88ff73550199014d552f9fddc74785fa5eccdb)

STABLE12-close-cellaliases-20030521

FIXES 1481

don't leak a ref to CellAliases

(cherry picked from commit ad98c8d9562f8d56ba8497467480a2de49e1527a)

STABLE12-amd64-copy-args-for-ia32-table-20030520

so the ia32 emulation might work

====================
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.
====================

first cut at an opteron port

(cherry picked from commit 055e0b99f911a1c821cf6264f72bad61d8e63d11)

STABLE12-amd64-hook-ia32-table-20030519

first cut at an opteron port

(cherry picked from commit 831e17246314f6204b1fe05c8f8a23a3be6b728a)

STABLE12-amd64-linux-port-20030428

first cut at an opteron port

(cherry picked from commit 9dfb5d60479847c630d46bf67a6cd9570a5b1a21)

STABLE12-linux-use-kernel-net-include-dir-for-net-20030519

since we can end up trying to include things which exist in include/net
but not include/linux, try to rationalize.

(cherry picked from commit a9b72c443e4bd074ecabe7ed005354e55c69b53d)

STABLE12-linux-look-for-syscall-h-in-kernel-20030519

some new kernels appear not to have linux/syscall.h in the kernel. no clue.

(cherry picked from commit 7fa099db6f796eb1eb0528cf8b7b68d27f510dd7)

STABLE12-uss-fix-bogus-grammar-20021026

FIXES 1451

newer bison actually wants you to write correct grammars apparently.

(cherry picked from commit c7f84a5867a5c1e6611dc8ea6e929031553a7818)

STABLE12-linux-try-sys-wait4-20030516

we may have close/wait4 instead of close/chdir or open/exit

(cherry picked from commit a8b536e7d3f45abd20fa0ea293f5f313189114d4)

STABLE12-linux-busy-cache-partition-while-afs-running-and-dont-allow-afsd-shutdown-while-afs-mounted-20030515

FIXES 1454

to preclude superblock corruption and stramge oopses

(cherry picked from commit 6c4bb9b335a3e5464007c61b6c4e4f286281ef0b)

STABLE12-salvage-allow-hardlinked-logs-by-date-20030515

FIXES 1259

so salvagelogs can be made persistent (by switch)

(cherry picked from commit 5012151295bef1f99944041fd19083ffc7c3f912)

STABLE12-amd64-linux-port-20030428

first cut at an opteron port

(cherry picked from commit 509c965d0aef031293edc0fcf3e97587d1deac73)

STABLE12-pass-pointers-for-uuid-printing-20030422

pass pointers for uuid string building

(cherry picked from commit ddd9554560a44314a029d7d2d59af8b7aaaec8ca)

STABLE12-cellconfig-unlock-mutex-on-error-20030422

clean up my own mess (i bet).
certain error cases weren't being cleaned up properly before we returned

(cherry picked from commit 5cebfdc757655ba0762b3edd5d9dbdd3dc5f25dd)

STABLE12-viced-shutdown-avoid-deadlock-20030424

don't deadlock on H_LOCK if we need to shut down in an emergency

(cherry picked from commit 05be212fea360747dbe5a9681ca81459b3d6ac62)

STABLE12-autoconf-deal-with-res-search-as-macro-20030428

on some platforms, res_search is a macro for __res_search, find it anyway

(cherry picked from commit f5df7b1d97572b6a0cd5f8a3a7b0f752ae111905)

STABLE12-add-afsdb-aliases-for-cellservdb-cells-20030429

Add cell aliases we find using AFSDB for cells that have been
statically configured via CellServDB.  This preserves the AFSDB
implementation behavior from before the cell code rewrite.

(cherry picked from commit e9993b1fce5a9d0260d94e95811a5b5dd79ef4ad)