R. Lindsay Todd [Wed, 28 May 2003 19:18:08 +0000]
start-splint-support-
20030528
FIXES 1493
only part of this bug
Derrick Brashear [Thu, 22 May 2003 16:51:08 +0000]
linux-system-calls-return-long-
20030522
the kernel says they return long. returning int causes sadness. stop being sad.
Derrick Brashear [Thu, 22 May 2003 16:16:19 +0000]
revert-pioctl-return-negative-errors-on-linux-
20030522
syscall stub handles it. this is actually an overflow on 64 bit system
Derrick Brashear [Thu, 22 May 2003 15:22:18 +0000]
pioctl-return-negative-errors-on-linux-
20030522
otherwise we can end up with errors that make no sense.
Derrick Brashear [Thu, 22 May 2003 05:57:31 +0000]
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.
Derrick Brashear [Wed, 21 May 2003 20:56:28 +0000]
linux-group-syscalls-return-long-
20030521
these return long. also call the underlying handler expecting to get a long
back
Thomas Mueller [Wed, 21 May 2003 15:45:05 +0000]
close-cellaliases-
20030521
FIXES 1481
don't leak a ref to CellAliases
Derrick Brashear [Wed, 21 May 2003 14:42:53 +0000]
amd64-copy-args-for-ia32-table-
20030520
first cut at an opteron port
Derrick Brashear [Tue, 20 May 2003 08:02:53 +0000]
amd64-hook-ia32-table-
20030519
first cut at an opteron port
Derrick Brashear [Tue, 20 May 2003 05:01:15 +0000]
amd64-linux-port-
20030428
first cut at an opteron port
Derrick Brashear [Tue, 20 May 2003 04:57:44 +0000]
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.
Derrick Brashear [Tue, 20 May 2003 04:49:29 +0000]
linux-look-for-syscall-h-in-kernel-
20030519
some new kernels appear not to have linux/syscall.h in the kernel. no clue.
Derrick Brashear [Fri, 16 May 2003 18:42:28 +0000]
linux-try-sys-wait4-
20030516
we may have close/wait4 instead of close/chdir or open/exit
Nathan Neulinger [Thu, 15 May 2003 18:10:39 +0000]
stifle-valgrind-
20030515
FIXES 1379
shut up valgrind on zeroing allocations
Derrick Brashear [Thu, 15 May 2003 17:59:12 +0000]
linux-busy-cache-partition-while-afs-running-and-dont-allow-afsd-shutdown-while-afs-mounted-
20030515
FIXES 1454
otherwise you can get into situations where you get strange oopses
and superblock corruption
Derrick Brashear [Thu, 15 May 2003 17:41:25 +0000]
vol-more-logging-
20030515
FIXES 1453
much more to do here. start with this.
Jeffrey Hutzelman [Thu, 15 May 2003 17:23:37 +0000]
salvage-allow-hardlinked-logs-by-date-
20030515
FIXES 1259
a switch causes salvager to create salvagelog hardlinks by date so
logs are preserved forever
James Peterson [Thu, 15 May 2003 16:41:11 +0000]
winnt-add-osi-event-mc-
20030515
FIXES 1373
add missed file
Jeffrey Hutzelman [Thu, 15 May 2003 16:25:08 +0000]
solaris-deal-with-gatekeeper-stupidity-
20030515
FIXES 1261
make gmake not try to link libafs.o and libafs.nonfs.o into another object
Christopher Allen Wing [Thu, 15 May 2003 16:16:26 +0000]
setgroups-return-errors-negative-
20030515
FIXES 1383
return errors negative as linux kernel needs
Alexei Kosut [Thu, 15 May 2003 16:00:27 +0000]
darwin-mount-and-disk-arbitration-enhancements-
20030515
FIXES 1363
* Allow AFS to be mounted multiple times, each mountpoint with its
own root volume. This allows usage similar to the Windows client,
e.g., mounting a drive that contains only the user's home
directory. Our users find this much easier to work with than
having to navigate the entire AFS tree in the Finder or open/save
dialog.
* Have afsd register /afs with the DiskArbitration system
when mounting it. This means that the root afs volume shows up in
the Finder immediately, and prevents the Finder from getting
confused by additional AFS volumes later on.
* Add a "-nomount" option to afsd, which prevents afsd from
mounting /afs. We use this by default on our Mac OS X client, so
that AFS doesn't show up in the interface at all unless the user
explicitly asks for an AFS volume to be mounted. This is
especially handy for portable or remote users who don't always
have a network connection, since there isn't an AFS filesystem
around to hang the Finder if the AFS servers can't be contacted
(the cache manager may still hang, but since there are no AFS
volumes mounted, the rest of the OS doesn't notice).
* Make it so that /afs (mounted by afsd) will not unmount unless
MNT_FORCE is given. This prevents the user from accidentally
ejecting /afs in the Finder, and also works around a DiskArb bug in
Mac OS X (possibly fixed in 10.2) where autodiskmount would sometimes
unmount /afs when logging out, even though it's marked as
non-ejectable.
Jim Rees [Thu, 15 May 2003 15:53:28 +0000]
pioctl-buffer-size-checking-
20030515
don't check output buffer size against PIGGY
Jim Rees [Wed, 14 May 2003 23:08:25 +0000]
openbsd-
20030515
OpenBSD 3.3 is out now, and seems to work.
Jim Rees [Wed, 14 May 2003 22:47:32 +0000]
openbsd-
20030514
#include afs/sysincludes.h for OpenBSD.
I suspect other systypes will need this too.
Alexei Kosut [Wed, 14 May 2003 16:47:24 +0000]
osx-fake-mode-bits-
20030513
FIXES 1362
give up on waiting for apple to get things right, which probably means they
never will.
Nathan Neulinger [Wed, 14 May 2003 16:14:54 +0000]
more-prototyping-
20030513
FIXES 1351
some ansification in src/ubik, more prototyping
Nathan Neulinger [Wed, 14 May 2003 16:10:23 +0000]
kdump-deal-with-vnode-to-vc-change-
20030513
FIXES 1350
so kdump can compile again
Alexei Kosut [Wed, 14 May 2003 16:06:41 +0000]
darwin-use-perl-instead-of-grep-
20030513
FIXES 1361
perl is installed by default. grep isn't. use perl.
Nathan Neulinger [Wed, 14 May 2003 15:54:53 +0000]
vol-additional-logging-
20030513
FIXES 1387
additional logging for attach/detach.
Douglas Engert [Wed, 14 May 2003 15:37:06 +0000]
solaris8-dont-read-binding-file-
20030513
FIXES 1434
Sun changed read_binding_file in a patch; Make sure we avoid the problem.
We don't need to call it if afs was already in /etc/system.
Matthew N. Andrews [Wed, 14 May 2003 15:30:51 +0000]
rx-bulk-example-cleanup-
20030513
FIXES 1436
make the bulk example work
Jim Rees [Tue, 13 May 2003 22:22:19 +0000]
rx-mtu-params-fix-again-
20030513
remove redundant decl of rxmtu
Nathan Neulinger [Tue, 13 May 2003 05:27:59 +0000]
rxgen-xdr-call-stifle-warnings-
20030513
FIXES 1349
eliminate some warnings in generated code from rxgen
Kris Van Hees [Tue, 13 May 2003 05:23:56 +0000]
i386-user-mode-linux-support-
20030513
FIXES 1396
support for usermode linux
Nathan Neulinger [Tue, 13 May 2003 05:11:01 +0000]
vol-gi-prototyping-
20030513
FIXES 1432
changed slightly by shadow@dementia.org to not need forward decl.
Nathan Neulinger [Tue, 13 May 2003 05:05:39 +0000]
rx-mtu-params-fix-
20030513
FIXES 1433
define needed variables outside ifdef
David Howells [Tue, 13 May 2003 04:55:13 +0000]
pioctl-buffer-size-checking-
20030512
FIXES 1413
buffers are signed, should be checked for negative in addition to smaller than
PIGGYSIZE; also, check out buffer size before calling pioctl handler.
Derrick Brashear [Tue, 13 May 2003 02:58:32 +0000]
volser-ros-use-parent-id-
20030512
FIXES 1448
everything shoudl be done as the parent volume id so the volume group stuff works right.
otherwise, we create things which get stored under one "tag", try to open the,
under another "tag" and they aren't there, because it doesn't hash the same.
Derrick Brashear [Thu, 8 May 2003 19:53:48 +0000]
linux-gcpags-deal-correctly-with-uid-tokens-
20030508
look at the correct uid, so we don't gc in use tokens
Jeffrey Hutzelman [Mon, 5 May 2003 20:58:36 +0000]
winnt-eventlog-newline-fixes-
20030505
Clean up newline conventions in src/WINNT/eventlog/lang/*/event.mc.
Existing files were in CVS with CRLF newlines, and so some Windows tools
would check them out with CRCRLF. Unfortunately, the mc-compiler can't
handle CRCRLF newlines. Fixing the files in CVS to have LF newlines
(like all proper text files) makes the problem go away -- they end up
with CRLF if checked out using tools that do newline conversion, and
LF-only if not, and the compiler accepts either.
Derrick Brashear [Mon, 5 May 2003 17:13:25 +0000]
ubik-dont-copy-stack-garbage-when-netinfo-forces-primary-address-change-
20030505
don't mess up count by reusing it
Derrick Brashear [Mon, 5 May 2003 16:40:25 +0000]
mapname-allocate-enough-space-for-pr-nametoid-
20030505
the alternative is to crash when we get a foreign user here
Nickolai Zeldovich [Wed, 30 Apr 2003 02:14:43 +0000]
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.
Jim Rees [Tue, 29 Apr 2003 22:39:36 +0000]
freebsd-50-
20030429
FreeBSD 5.0 kernel module compiles now, but not likely to work yet.
Derrick Brashear [Tue, 29 Apr 2003 00:30:09 +0000]
no-sysincludes-in-rx-
20030428
not the correct fix but without it linux/vnode.h conflicts with osi_vfs.h
when both get included.
Derrick Brashear [Tue, 29 Apr 2003 00:05:57 +0000]
autoconf-deal-with-res-search-as-macro-
20030428
on some platforms, res_search is a macro for __res_search, find it anyway
Derrick Brashear [Mon, 28 Apr 2003 22:38:20 +0000]
amd64-linux-port-
20030428
first cut at an opteron port
Jim Rees [Fri, 25 Apr 2003 16:54:48 +0000]
proto-
20030425
add missing prototype for ClearHostCallbacks_r()
Nickolai Zeldovich [Thu, 24 Apr 2003 20:21:31 +0000]
uae-dont-translate-zero-
20030424
Leave a zero return value as zero, rather than translating to UAEMEDIUMTYPE.
Rainer Toebbicke [Thu, 24 Apr 2003 14:37:15 +0000]
viced-shutdown-avoid-deadlock-
20030424
don't deadlock on H_LOCK if we need to shut down in an emergency
Jim Rees [Thu, 24 Apr 2003 00:03:45 +0000]
freebsd-50-
20030423
more FreeBSD 5.0 kernel module work.
Derrick Brashear [Wed, 23 Apr 2003 05:31:14 +0000]
afs-analyze-avoid-potential-loop-
20030423
avoid potentially looping forever in e.g. afs_FetchStatus because
we exit afs_Analyze too early to increment busyCount, and despite having
areq->busyCount non-zero we have servers still not_busy which are already down
that we'll never talk to again. that combined with no conn means we never
touch most of afs_Analyze, exiting early with shouldRetry set.
incrementing busyCount whenever we exit with shouldRetry set should
ensure we can never end up looping forever.
Rolf Sattler [Tue, 22 Apr 2003 17:45:07 +0000]
winnt-xdr-mem-needs-limits-h-
20030422
as received from scoot williams
Chas Williams [Tue, 22 Apr 2003 16:45:40 +0000]
fs-include-string-function-prototypes-
20030422
so fs won't segfault on ia64 linux
Derrick Brashear [Tue, 22 Apr 2003 06:27:17 +0000]
pass-pointers-for-uuid-printing-
20030422
pass pointers for uuid string building
Derrick Brashear [Tue, 22 Apr 2003 06:25:50 +0000]
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
Chaskiel M Grundman [Mon, 21 Apr 2003 17:37:08 +0000]
freebsd-
20030421
add .PHONY target for libafs
Jim Rees [Thu, 17 Apr 2003 23:14:30 +0000]
openbsd-
20030417
use the correct group list entries (1,2) for pag (duh)
Kevin Coffman [Thu, 17 Apr 2003 16:05:58 +0000]
aix-
20030417
I had to do the following to do a 'make dest' on AIX in a directory
separate from the source:
Jim Rees [Wed, 16 Apr 2003 23:28:52 +0000]
freebsd-50-
20030416
More FreeBSD 5.0. Most of it builds now.
Jim Rees [Wed, 16 Apr 2003 20:47:07 +0000]
ignore-
20030416
ignore all MakefileProtos in libuafs
Joakim Fallsjö [Mon, 14 Apr 2003 16:45:54 +0000]
solaris-x86-port-already-htons-
20030414
rx_port is already htons() and htons is a NULL-op on sparc but not on x86...
make shutdown actually work on x86 solaris
Garry Zacheiss [Mon, 14 Apr 2003 00:53:07 +0000]
update-news-file-
20030413
Note minimum version requirement for 2.2.x series Linux kernels, and
also note incompatability between 1.2.9 afsd and older kernel modules.
Derrick Brashear [Fri, 11 Apr 2003 18:50:12 +0000]
softsig-dont-block-critical-signals-
20030411
blocking these seems to prevent all threads from dying when the kernel
gives e.g. SIGSEGV to one of the processes. since softsig is to protect us
from what signals we handle might do in the handler, this should be fine
Garry Zacheiss [Fri, 11 Apr 2003 01:58:23 +0000]
update-news-
20030410
Update NEWS file with important changes going into 1.2.9.
Derrick Brashear [Thu, 10 Apr 2003 20:57:22 +0000]
linux-try-more-kernel-paths-
20030410
FIXES 1364
try /usr/src/linux-2.4 in addition to /usr/src/linux
Garry Zacheiss [Thu, 10 Apr 2003 20:23:23 +0000]
darwin-map-more-versions-
20030410
Reocgnize MacOS X 10.2.5 as ppc_darwin_60.
Love Hörnquist-Åstrand [Thu, 10 Apr 2003 19:55:03 +0000]
rxkad-2b-deal-with-expiry-
20030410
FIXES 1369
return the actual error and not RXKADBADTICKET always
Kevin Coffman [Thu, 10 Apr 2003 15:55:32 +0000]
static-return-from-structname-
20030410
return pointer to static data from structname()
needed by aix, probably others too.
This bug has been around forever!
Love Hörnquist-Åstrand [Tue, 8 Apr 2003 22:38:18 +0000]
viced-file-readonly-fs-support-
20030408
add missed curly braces so VannilaUsers don't get EACCES
Jim Rees [Mon, 7 Apr 2003 21:24:39 +0000]
afsutil-protos-
20030407
remove redundant prototypes and move to afsutil_prototypes.h
fix protos for int64_to_flipbase64
thanks to Alexei Kosut <akosut@cs.stanford.edu>
Derrick Brashear [Mon, 7 Apr 2003 21:11:35 +0000]
h-gethost-r-race-
20030401
FIXES 1308
don't deref null identP if code was -1
Michael Niksch [Mon, 7 Apr 2003 18:56:28 +0000]
salvager-aix-update-
20030407
make it work with namei
Derrick Brashear [Mon, 7 Apr 2003 18:38:56 +0000]
flushhostcps-
20030407
make it so flushhostcps can't try to Release when no host was found,
which was apparently the problem dlc@cs really observed.
Jim Rees [Fri, 4 Apr 2003 21:15:54 +0000]
openbsd-33-
20030404
param file for OpenBSD 3.3. untested but "should work."
Jim Rees [Fri, 4 Apr 2003 19:04:22 +0000]
freebsd-5-
20030404
preliminary work on freebsd 5.0 kernel module
Jim Rees [Fri, 4 Apr 2003 17:49:33 +0000]
openbsd-
20030404
changes suggested by mcmer@gmx.net
Jim Rees [Fri, 4 Apr 2003 17:09:21 +0000]
remove-redundant-proto-
20030404
remove redundant rxk_NewSocket proto
Love Hörnquist-Åstrand [Fri, 4 Apr 2003 16:53:00 +0000]
freebsd-5-update-
20030404
re-apply rev 1.6 patch, lost in rev 1.7
Jim Rees [Thu, 3 Apr 2003 22:46:51 +0000]
openbsd-
20030403
these never worked, and it seems unlikely anyone will ever use them
Derrick Brashear [Thu, 3 Apr 2003 22:42:22 +0000]
obsd-create-file-so-configure-shuts-up-
20030403
so configure won't whine
Jim Rees [Thu, 3 Apr 2003 15:48:53 +0000]
fix-window-size-
20030403
print actual limits if window too small
Derrick Brashear [Wed, 2 Apr 2003 01:22:16 +0000]
h-gethost-r-race-
20030401
FIXES 1308
Thanks to Chaskiel Grundman for explaining what was happening:
- the connection is old and pre-existing, but has no host structure.
- 2 calls come in
- the first one enters h_GetHost_r, and h_Lookup_r returns null (but identP is
non-null, since rx keeps it around until it gc's the connection)
The first thread calls WhoAreYou, which succeeds, it then calls
InitCallBackState3 (after H_UNLOCK) note that the host has been inserted into
the hashtable
- the second thread enters h_GetHost_r, and calls rx_GetSpecific. it then
calls h_Lookup_r. h_Lookup_r will block (new host is locked), but eventually
returns the new host
- InitCallBackState3 returns, and the frees the old identP, replaces it, and
unlocks the host.
- the first thread returns from h_Lookup_r. boom.
the changes:
-call rx_GetSpecific after h_Lookup_r returns (and potentially slept)
-removes an if wrapping which always is true (since !interfValid is always
true)
-don't realloc identP if it exists
-don't free an old one by calling rx_SetSpecific either
Garry Zacheiss [Mon, 31 Mar 2003 19:22:04 +0000]
darwin-map-more-version-
20030331
Recognize MacOS X 10.2.3 and 10.2.4.
Derrick Brashear [Fri, 28 Mar 2003 21:26:20 +0000]
solaris-fs-rolled-
20030328
for solaris 9 u2
Onime Ehimika Ohireime [Fri, 28 Mar 2003 17:41:04 +0000]
linux-execsorwrites-at-close-
20030328
FIXES 1215
fixes bad interaction with openmosix
comment from derrick: i'm still not sure what's going on here.
====================
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 1215
fixes bad interaction with openmosix
somehow diff screwed up the first time and applied one hunk wrong.
comment from derrick: i'm still not sure what's going on here.
Nathan Neulinger [Fri, 28 Mar 2003 09:35:53 +0000]
revert-large-file-support-
20030328
FIXES 1341
sadly there are problems. how long before this comes to the fore again
i don't know.
Nathan Neulinger [Fri, 28 Mar 2003 09:30:53 +0000]
rxstat-fix-objdir-builds-
20030328
FIXES 1340
change where we include headers from
Nathan Neulinger [Fri, 28 Mar 2003 09:27:32 +0000]
volser-cleanup-clone-trans-
20030328
FIXES 1339
so we don't leave an idle transaction around
Daryl Clevenger [Fri, 28 Mar 2003 09:21:13 +0000]
flushhostcps-init-held-
20030328
init held so flushhostcps is more useful.
Jim Rees [Thu, 27 Mar 2003 20:29:51 +0000]
fix-window-size-
20030327
Fix off-by-one error in window size calculation
Derrick Brashear [Wed, 26 Mar 2003 17:23:16 +0000]
shlibafsrpc-export-missing-des-symbols-
20030326
symbols libafsauthent wants but we don't have
Derrick Brashear [Wed, 26 Mar 2003 09:10:59 +0000]
linux-rxk-listener-shutdown-cleanup-
20030323
get rid of need for sys_kill
Nathan Neulinger [Tue, 25 Mar 2003 21:10:40 +0000]
rxkad2b-allow-no-kvno-
20030325
allow a missing kvno field to work usefully
Derrick Brashear [Sun, 23 Mar 2003 19:43:47 +0000]
move-up-cell-initialization-in-cachemgr-
20030323
may be related to an oops at startup
Chaskiel M Grundman [Sun, 23 Mar 2003 06:44:26 +0000]
linux-kthreads-reparent-
20030323
so we don't get orphans on afs shutdown
Jim Rees [Fri, 21 Mar 2003 20:06:16 +0000]
openbsd-
20030321
Restore lost -O2 CFLAG
Dan Hyde [Fri, 21 Mar 2003 19:26:36 +0000]
xdr-mem-take-II-
20030321
Fix pasto in xdr_mem security patch.
Chaskiel M Grundman [Fri, 21 Mar 2003 18:28:31 +0000]
linux-hold-tasklist-lock-in-get-page-offset-
20030321
tasklist_lock must be held to traverse list
Jim Rees [Fri, 21 Mar 2003 17:59:28 +0000]
xdrmem-getbytes-
20030321
See, for example:
http://web.mit.edu/kerberos/www/advisories/MITKRB5-SA-2003-003-xdr.txt
This source is unused but it seems like a good idea to patch it anyway.