openafs.git
13 years agohcrypto: Add hcrypto EVP support to the Unix CM
Simon Wilkinson [Sun, 12 Sep 2010 17:40:08 +0000]
hcrypto: Add hcrypto EVP support to the Unix CM

This commit adds the files which are necessary to support hcrypto's
EVP interface to the Unix cache manager build. Only a small number
of EVP ciphers and hashes are currently supported -
    * aes_128_cbc
    * aes_256_cbc
    * sha1

Note that the EVP interface is the only supported mechanism to use
the AES cipher - directly calling the underlying crypto functions
is not recommended and may break at any time.

Change-Id: I662073e578b29db1707c6b6433209e75e4db455d
Reviewed-on: http://gerrit.openafs.org/3945
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoUnix CM: Add osi_crypto crypto interface
Simon Wilkinson [Sat, 11 Sep 2010 12:02:49 +0000]
Unix CM: Add osi_crypto crypto interface

Add a new osi file to the cache manager. This file includes functions
which interface hcrypto with the kernel's own cryptography library.

At the moment, the only such function is osi_readRandom which returns
a block of random data from the kernel's PRNG.

Change-Id: Ib728f5bb638ef634efe4822da24bb7cca723983f
Reviewed-on: http://gerrit.openafs.org/3944
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoMakefile whitespace fixes
Andrew Deason [Wed, 16 Feb 2011 15:49:03 +0000]
Makefile whitespace fixes

Some "make"s get tripped up by recipe lines starting with anything but
a tab character. Most of the time we start ours with tabs, but fix a
few instances where they begin with spaces.

Change-Id: I2c17c68deb1b7641ca73e84ec0043b62ce3c54ba
Reviewed-on: http://gerrit.openafs.org/3961
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: Restore inode OS_READ/WRITE
Andrew Deason [Tue, 15 Feb 2011 23:21:49 +0000]
vol: Restore inode OS_READ/WRITE

Commit 335ccb4082657b7d0e4e9af1076356cf115642d2 removed the OS_READ
and OS_WRITE definitions for non-namei code. We need those definitions
to build the pread/pwrite emulation functions, so put them back in.
This allows us to build the inode fileserver backend when we do not
have native positional I/O.

Change-Id: I3c066dacbee20301b38f074142cb6b6bb6b05de7
Reviewed-on: http://gerrit.openafs.org/3959
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agoAutoconf: Add GSSAPI autoconf rules
Simon Wilkinson [Thu, 2 Sep 2010 07:44:27 +0000]
Autoconf: Add GSSAPI autoconf rules

Add rules to autodetect the location of a system's GSSAPI libraries,
using the gssapi.m4 script from rra-c-util 2.6

Change-Id: Idabb8fdb138a2dac6983aa4d5fd8a2f558ee08be
Reviewed-on: http://gerrit.openafs.org/3943
Tested-by: Christof Hanke <christof.hanke@rzg.mpg.de>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoHPUX: Disable positional I/O
Andrew Deason [Mon, 14 Feb 2011 19:53:11 +0000]
HPUX: Disable positional I/O

Some versions of HP-UX have the pread() and pwrite() functions, but
they behave in odd ways; most notably, ignoring the offset argument
when _FILE_OFFSET_BITS is defined to 64.

This is noted in recent gnulib documentation
<http://git.savannah.gnu.org/cgit/gnulib.git/tree/doc/posix-functions/pwrite.texi>,
and slightly less clear references to pread() being broken can be
found on the development mailing list for git itself.

It is not completely clear what specific HP-UX versions are affected
by this. An autoconf run-time test may also be insufficient, because
the same binaries should be usable on machines with broken and
non-broken pread() implementations. So, to be safe, disable positional
I/O on HP-UX unconditionally.

Change-Id: I09b8d9c441622c961d1df90fe27eeccaa948f5c4
Reviewed-on: http://gerrit.openafs.org/3949
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoUpdate 1.6 NEWS entry based on 1.6.0pre2 release notes
Russ Allbery [Mon, 14 Feb 2011 20:36:18 +0000]
Update 1.6 NEWS entry based on 1.6.0pre2 release notes

Change-Id: Id795a74910bc881f8e93c8e80b9cb55ff48c3b15
Reviewed-on: http://gerrit.openafs.org/3950
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoLinux: Fix RX atomics for kernel mode
Simon Wilkinson [Sun, 30 Jan 2011 19:27:17 +0000]
Linux: Fix RX atomics for kernel mode

Fix a mistake with the definition of rx_atomic_set, and add
the missing declaration of rx_dec_and_read for Linux kernel builds

Change-Id: Ieed7a369670c051abc02d8cc831e8736969fb67a
Reviewed-on: http://gerrit.openafs.org/3942
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years ago1.6.0pre2 release notes
Derrick Brashear [Mon, 14 Feb 2011 16:03:17 +0000]
1.6.0pre2 release notes

add release notes for 1.6.0pre2

Change-Id: I05eb7d02b3daa9935eec6cbc7f6b8f1de78c23f0
Reviewed-on: http://gerrit.openafs.org/3940
Tested-by: Derrick Brashear <shadow@dementia.org>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoRedHat: Change the defined initdir path to /etc/rc.d/init.d
Jonathan Billings [Fri, 7 Jan 2011 15:44:27 +0000]
RedHat: Change the defined initdir path to /etc/rc.d/init.d

On Red Hat systems, /etc/init.d is a symlink to /etc/rc.d/init.d.  We
should use the actual path for packaging the init scripts, to avoid
any issues with package verification.

Change-Id: I476193b3fd534c8aa70c67970bf2765a30f4c4d5
Reviewed-on: http://gerrit.openafs.org/3625
Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agoAdd PIC variant for libuafs
Andrew Deason [Mon, 13 Dec 2010 19:11:02 +0000]
Add PIC variant for libuafs

Build a libuafs that is built with PIC, so it can be used in shared
libraries on all platforms. Also build PIC variants for libafsutil and
libcmd, since we need them for a PIC libuafs anyway.

Change-Id: I017fa22c3b4dd661123ea92da7b3e5b69ee5c170
Reviewed-on: http://gerrit.openafs.org/3896
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoauth: Rewrite KeyFile handling code
Simon Wilkinson [Sat, 1 Jan 2011 23:41:29 +0000]
auth: Rewrite KeyFile handling code

Extend the KeyFile API so that we can support arbitrary numbers of
different key types, each with their own key version numbers and
sub types. Completely rewrite the KeyFile implementation with this
in mind, but implement all of the "old" API in terms of the new one.

Given that the existing KeyFile is modified by third party programs,
we retain that as the storage location for all afsconf_rxkad keys.
Only keys with a type of 1, or above are stored in the new extended
keyfile.

Change-Id: I903a1de938544541a1bfecedb2a039ba24bdfdbc
Reviewed-on: http://gerrit.openafs.org/3772
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: Release Notes updates for 1.6pre2
Jeffrey Altman [Sun, 13 Feb 2011 02:10:44 +0000]
Windows: Release Notes updates for 1.6pre2

A fairly thorough review of all sections to ensure that the information
is up to date.

Significant reformatting caused by XML editor.

Change-Id: Iad988b91091a02a7b5de2a1d1fae052fc5ee897f
Reviewed-on: http://gerrit.openafs.org/3936
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoLinux: 2.6.38: dentry->d_count is not an atomic
Marc Dionne [Thu, 3 Feb 2011 02:55:27 +0000]
Linux: 2.6.38: dentry->d_count is not an atomic

d_count is now an int protected by the dentry's d_lock.
Take the lock when we use it, instead of using an atomic_*
function.

Change-Id: Ib70e4a5315cc343518fa983e47bc7ff925acfc7f
Reviewed-on: http://gerrit.openafs.org/3883
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: ChangeLog updates for 1.6.pre1
Jeffrey Altman [Sat, 12 Feb 2011 19:24:02 +0000]
Windows: ChangeLog updates for 1.6.pre1

Change-Id: I17de717ffd597c45720a4fbd39a14c5df67f870c
Reviewed-on: http://gerrit.openafs.org/3926
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: Fix GetIoctlHandle path construction
Jeffrey Altman [Sat, 12 Feb 2011 16:52:12 +0000]
Windows: Fix GetIoctlHandle path construction

GetIoctlHandle() is used to construct the magic pioctl file
path used to initiate pioctl operations with the cache manager.
The first error introduced double directory separators.  The
second error was testing an uninitialized value which could
have resulted in a missing directory separator.

Change-Id: I4e863b854ea12b69858be712455dc6b6a0d36c39
Reviewed-on: http://gerrit.openafs.org/3925
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: Fix symlink and mount point make \\afs\xxx handling
Jeffrey Altman [Sat, 12 Feb 2011 16:45:15 +0000]
Windows: Fix symlink and mount point make \\afs\xxx handling

When processing a request to make a symlink or a mount point
in the afs root volume (\\AFS) the smb redirector will fail
the request because a server name by itself is not a valid path.
Therefore, we insert the "all" share component to refer to the
root volume as a valid path.  \\AFS\foobar becomes \\AFS\all\foobar.

A recent change stripped the trailing slash from the string
returned by Parent().  This broke the test that determines
whether or not the provided path that failed the IsAFS() test
is in fact referring to the \\AFS server and requires the insertion
of the "all" share name.

This patchset permits the test to work with \\AFS or \\AFS\
and removes extraneous directory separators from the generated
path containing the "all" share.

Change-Id: Ia3b28d83302d9737bd149b4a22bc3ee3a6c8da46
Reviewed-on: http://gerrit.openafs.org/3924
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoutil: Actually install thread_pool_types.h
Simon Wilkinson [Wed, 9 Feb 2011 19:37:18 +0000]
util: Actually install thread_pool_types.h

The action portion of the rule for thread_pool_types.h was missing,
so the file was never installed. As that file is a dependency for the
whole of the util directory, it would result in util being completely
rebuild with every make invocation.

Add the missing installation rule, so we don't rebuild quite so often!

Change-Id: Idb201c239b6b37a25567cf231059ef05536fc258
Reviewed-on: http://gerrit.openafs.org/3921
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoscout: restore parallel make
Marc Dionne [Wed, 9 Feb 2011 22:06:00 +0000]
scout: restore parallel make

Restore the + in the scout Makefile section, lost with commit
4a6a00d6.
Lets scout benefit from a parallel make and avoids a warning.

Change-Id: Ie4c35c93afae4bfefaae0a0db9d676200cb3ee7b
Reviewed-on: http://gerrit.openafs.org/3910
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoubik: always prefer a dirty cache page for write transactions
Marc Dionne [Fri, 28 Jan 2011 00:07:32 +0000]
ubik: always prefer a dirty cache page for write transactions

If a write transaction is running concurrently with a read transaction,
a DRead in the write transaction may return a clean cache page brought
in by the read transaction, instead of a previously written dirty page
from the same transaction.  This can result in loss of the written data.

Fix by making sure there is not a dirty version of the requested page
before returning a clean one.

Bug spotted by Jeffrey Hutzelman.

Change-Id: I20543693c98218d8ec1f791508d9404043819376
Reviewed-on: http://gerrit.openafs.org/3764
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoEarly dispose of replies in rx_Multi
Rainer Toebbicke [Thu, 27 Jan 2011 14:03:57 +0000]
Early dispose of replies in rx_Multi

Handles replies as soon as they arrive in rx_Multi, limiting
strain on RX packet queues on massive rx_Multi loops.

Change-Id: I345f7dba1707beeb75bcc65bb519afb9dc741973
Reviewed-on: http://gerrit.openafs.org/3765
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoevalmountdata null pointer before use
Derrick Brashear [Mon, 31 Jan 2011 16:04:08 +0000]
evalmountdata null pointer before use

volnamep can be used uninitialized. fix that.

Change-Id: I4aafb3bc56241b4cfc7ccd0a9cfa9e57bd97157a
Reviewed-on: http://gerrit.openafs.org/3777
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agoLinux: 2.6.38: deal with dcache_lock removal
Marc Dionne [Sat, 29 Jan 2011 01:59:17 +0000]
Linux: 2.6.38: deal with dcache_lock removal

dcache_lock is gone in 2.6.38, and some of the vfs locking rules
have changed.

Of interest for openafs:
- inode->i_lock protects the d_alias list
- dentry->d_lock protects d_unhashed()

Add a new configure test for dcache_lock, and replace its use by
the appropriate new lock(s).

Change-Id: Ic8d860d4f9f6438e966c15f5178c0d82343e3e1c
Reviewed-on: http://gerrit.openafs.org/3771
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoLinux: 2.6.38: Adjust for permission inode operation changes
Marc Dionne [Sat, 29 Jan 2011 00:41:32 +0000]
Linux: 2.6.38: Adjust for permission inode operation changes

The permission i_op has a new signature with a flags argument, and
must now deal with RCU path walking.
- Fix existing configure test for this i_op, it succeeds when it
shouldn't
- Add a new configure test for the new signature
- Make our permission i_op "RCU-walk aware" - return ECHILD if
called in that mode

Change-Id: I42a171694717f4621f29f9f59e4f6049926862a3
Reviewed-on: http://gerrit.openafs.org/3770
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoLinux: allow compile flags to be passed to AC_CHECK_LINUX_BUILD
Marc Dionne [Sat, 29 Jan 2011 00:03:39 +0000]
Linux: allow compile flags to be passed to AC_CHECK_LINUX_BUILD

Some linux autoconf tests require particular compile flags such as
-Werror.  Add a parameter to the AC_CHECK_LINUX_BUILD macro that
lets the caller pass in any needed special flags.

Adapt a few existing tests that were still using AC_TRY_KBUILD
directly.

Change-Id: I4a5e3db74f6fb9c35bfa7b2964a294f646e78626
Reviewed-on: http://gerrit.openafs.org/3769
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoConvertROtoRW: Use old copyDate for creationDate
Andrew Deason [Thu, 3 Feb 2011 21:40:48 +0000]
ConvertROtoRW: Use old copyDate for creationDate

When we convert an RO volume to an RW, currently we just copy the
copyDate and creationDate from the RO metadata into the RW. But the
copyDate and creationDate fields have different meanings for RW and RO
volumes: for ROs, the creationDate is merely the last time the data
was updated from the RW during a release operation.

So, if the copyDate is older than the creationDate, use the copyDate
as the new RW creationDate instead. This will probably not match the
creationDate of the original RW, but it will be closer to it, and it
will more accurately represent the conceptual "created time" of the
new RW.

Doing this can avoid forcing an unnecessary full dump on a subsequent
release of the resultant RW volume, since the creationDate is more
accurate.

Change-Id: Ibfffbdc4a79a568261e9e116f6991a3af9db5e0c
Reviewed-on: http://gerrit.openafs.org/3891
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: correct pthread_xxx_init semantics
Jeffrey Altman [Mon, 7 Feb 2011 21:44:09 +0000]
Windows: correct pthread_xxx_init semantics

pthread lock and conditional initialization semantics
do not require that the lock structure be zeroed before
pthread_xxxx_init() functions are called.  Since the Windows
CriticalSection initialization does require that the memory
be zeroed, the pthread_xxxx_init() functions must zero the
memory just in case before performing the CriticalSection
initialization.

Change-Id: I61e78ca7cbc10f7d4144b8efcbb38f67bb8fd695
Reviewed-on: http://gerrit.openafs.org/3904
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoafscp: Fix -s option for writes
Andrew Deason [Thu, 27 Jan 2011 19:13:21 +0000]
afscp: Fix -s option for writes

When writing to AFS with afscp, the -s option was sleeping before any
StoreData RPCs actually got issued to the fileserver. Move the sleep
to after we have done one rx_Read/rx_Write, so we sleep after starting
to contact the fileserver, to make sleeping while writing more
consistent with sleeping while reading.

Change-Id: Ib4e93b65f50cbad30d057ad7905617d00070c1a4
Reviewed-on: http://gerrit.openafs.org/3762
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: remove duplicate advapi32.lib references
Jeffrey Altman [Thu, 3 Feb 2011 00:24:00 +0000]
Windows: remove duplicate advapi32.lib references

advapi32.lib is included in the conlibsdll macro.
No need to include it in the link list twice.

Change-Id: I2f9725f8b48b26afc959477d2fe2f2925f357ee7
Reviewed-on: http://gerrit.openafs.org/3882
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agospelling/grammar fixes for manpages
Ken Dreyer [Thu, 3 Feb 2011 20:22:02 +0000]
spelling/grammar fixes for manpages

Several small fixes for spelling and grammar.

Change-Id: I5c1cbc933cd683f997b8bb649408d57d84a110f2
Reviewed-on: http://gerrit.openafs.org/3890
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agospelling/grammar fixes for vos_clone manpage
Ken Dreyer [Wed, 2 Feb 2011 16:49:31 +0000]
spelling/grammar fixes for vos_clone manpage

Small spelling/grammar fixes for the vos_clone manpage.

Change-Id: If312799b9e95476a1de2d386e6d51cd800c7e069
Reviewed-on: http://gerrit.openafs.org/3879
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoRx: Do not stop keepalives on ACKALL receipt
Andrew Deason [Mon, 31 Jan 2011 21:00:09 +0000]
Rx: Do not stop keepalives on ACKALL receipt

We need to still keep sending periodic keepalives after receiving an
ACKALL, since the call is not done yet. ACKALLs can be received when
the peer has received all data from us, well before the call has
finished. This is particularly noticeable for long-lived calls that
have little data transfer, such as AFSVolForward and
AFSVolForwardMultiple.

This reverts commit b1549f29a9515de09d4a64bfb28bf5afdfee2615, which
was intended to be part of a fix for RT 20727.

FIXES 128848

Change-Id: I319e6e95d699ea87ecc72888452b5e1e6fcd88b0
Reviewed-on: http://gerrit.openafs.org/3876
Tested-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoafsd: Do not check for /afs if -nomount
Andrew Deason [Mon, 31 Jan 2011 17:18:39 +0000]
afsd: Do not check for /afs if -nomount

If -nomount was passed, we don't try to mount /afs. So, there is no
reason to require its existence.

Change-Id: Id426f345b94de8f2d673af62fe14f803d1c77752
Reviewed-on: http://gerrit.openafs.org/3822
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: No NCBRESET when probing Loopback after start
Jeffrey Altman [Mon, 31 Jan 2011 16:43:39 +0000]
Windows: No NCBRESET when probing Loopback after start

The Netbios NCBRESET command resets all of the Netbios state
associated with the LAN adapter including the name bindings.
In response to a detected LAN adapter IP address change, the
smb_LanAdapterChange() function is called to determine if any
Netbios LAN adapter bindings that were in use or should be in
use by afsd_service were altered.  As part of the check,
lana_GetUncServerNameEx() is called which in turn calls
lana_FindLoopback() which in turn issued a lana_IsLoopback()
for each LAN adapter with the 'reset adapter' flag set to TRUE.

Calling lana_IsLoopback() with 'reset' equal TRUE was fine
when lana_GetUncServerNameEx() was only called from smb_Init(),
but it is not fine when called after the service is processing
calls.  By resetting the adapter the binding of the netbios name
"AFS" (or "<MACHINE>-AFS") is removed and all outstanding calls
are canceled.  If the SMB redirector attempts a reconnect during
the window before NCBADDNAM is called to re-bind the name, a
negative cache entry will be placed in the netbios name lookup
table that will prevent the SMB redirector from connecting to
the client for several minutes.

If the environment is one in which frequent IP address change
events are triggered, it is possible that the SMB redirector
will never be able to reconnect to the service.

This patchset adds a flag, LANA_NETBIOS_NO_RESET, to the
lana_GetUncServerEx interface which permits smb_LanAdapterChange()
to avoid the undesirable reset.  This negative flag was selected
in order to avoid changing the current default behavior as the
lanahelper library is used by out of tree installers and it
is preferred that OpenAFS avoid breaking them unnecessarily.

Change-Id: I094af7b342d27c65025235888705af8af1a7d56e
Reviewed-on: http://gerrit.openafs.org/3821
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoevalmountdata: put back colon in .:mount syntax only if we removed it
Derrick Brashear [Mon, 31 Jan 2011 20:07:45 +0000]
evalmountdata: put back colon in .:mount syntax only if we removed it

we looked for a set character (not NULL) at cpos before, instead
of looking for a non-NULL pointer. fix it.

Change-Id: Iaa370980b6482f1c50e11f161e3208c039043a6f
Reviewed-on: http://gerrit.openafs.org/3874
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agotests: Fix auth/superuser-t.c to work on Linux
Simon Wilkinson [Sun, 30 Jan 2011 20:54:30 +0000]
tests: Fix auth/superuser-t.c to work on Linux

Fix the superuser tests so that they can be run on Linux. This
requires explicitly including sys/wait.h so we can use waitpid,
and changing some initialisation ordering so that we initialise
the RX library before we try and fake an rxkad token.

Change-Id: I8439ff6211a50c749ea22819e2d836409a64d2ad
Reviewed-on: http://gerrit.openafs.org/3776
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agobozo: Fix linker problem on Linux
Simon Wilkinson [Sun, 30 Jan 2011 18:35:28 +0000]
bozo: Fix linker problem on Linux

Some unconnected changes break the Linux linker when faced with this
library list. librx is included too early, and liblwp isn't included
after the second inclusion of librx. Reorder the library list to fix
this problem, and so it matches the library order in the rest of our
modules.

Change-Id: Iba19111f613c70c8a6e66539b437a99d58faf7f4
Reviewed-on: http://gerrit.openafs.org/3775
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: Install rx_atomic.h
Simon Wilkinson [Sun, 30 Jan 2011 19:11:57 +0000]
Windows: Install rx_atomic.h

Install the rx_atomic.h header into the build-wide header directory,
so that modules other than RX can make us of it.

Change-Id: Id3c3d0c486412942fb0772d7a8cf5d2fc1bf8876
Reviewed-on: http://gerrit.openafs.org/3774
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoUnix CM: Move kernel crypto include files
Simon Wilkinson [Fri, 28 Jan 2011 15:20:25 +0000]
Unix CM: Move kernel crypto include files

Move the crypto include files which are used for kernel builds into
the hcrypto directory where they logically belong.

Change-Id: Ic904e63e48d09e6a2049f4e1c4433980ecf0fa1e
Reviewed-on: http://gerrit.openafs.org/3767
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: out of order locks cm_CheckCBExpiration
Jeffrey Altman [Sun, 30 Jan 2011 04:24:16 +0000]
Windows: out of order locks cm_CheckCBExpiration

The recent refactoring of cm_CheckCBExpiration introduced
a lock ordering error between the cm_scache_t rw lock and the
cm_scacheLock.  This patchset fixes the error by dropping the
cm_scacheLock as each cm_scache_t is being processed.

Change-Id: Ib9e45abc5a43ca550d4a2a7923e3b30017e9fbf9
Reviewed-on: http://gerrit.openafs.org/3773
Tested-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoaklog: Use correct CFLAGS
Simon Wilkinson [Fri, 28 Jan 2011 15:01:21 +0000]
aklog: Use correct CFLAGS

The aklog Makefile was setting XCFLAGS, but then not actually using them
which meant that we never used the correct include path for the Kerberos
libraries, or defined ALLOW_REGISTER

Change-Id: Iae573bc216b42bdf9aa49e9f5f42a883e08b6265
Reviewed-on: http://gerrit.openafs.org/3766
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoDo not compare an FD_t < 0
Rod Widdowson [Fri, 28 Jan 2011 16:49:03 +0000]
Do not compare an FD_t < 0

FD_t is an abstract type.  Invalid file decriptors are == INVALID_FD.
In most places this module does the right thing.  Fix the last
stragglers where it is testing as less than zero.

Change-Id: I9ae3db8775ce69da070a4ebef5a532e1db035977
Reviewed-on: http://gerrit.openafs.org/3768
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agolinux: 2.6.38: Make d_revalidate RCU-walk aware
Marc Dionne [Tue, 25 Jan 2011 23:18:56 +0000]
linux: 2.6.38: Make d_revalidate RCU-walk aware

Signal to the vfs that we don't support path RCU walk in
d_revalidate.  It will call back in non-RCU mode.

Change-Id: I15d0d30131b205a38577666122b067807282a7cd
Reviewed-on: http://gerrit.openafs.org/3759
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agolinux: 2.6.38: New d_op handling
Marc Dionne [Tue, 25 Jan 2011 22:17:21 +0000]
linux: 2.6.38: New d_op handling

In 2.6.38, the super block structure has a new field to hold the
default dentry ops.  The vfs will automatically set it for new
dentries in most cases.

Set s_d_op to our set of operations, and omit setting the dentry
ops where the vfs will already do it (and where new locking rules
prohibit it).

Change-Id: I9d738bdd528577d0a7ccd307246b5e935379c84d
Reviewed-on: http://gerrit.openafs.org/3758
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoRX: Avoid retrying calls on busy channels
Andrew Deason [Thu, 2 Dec 2010 23:13:17 +0000]
RX: Avoid retrying calls on busy channels

When we receive an RX_PACKET_TYPE_BUSY packet, we currently ignore it.
This is a problem when the server has a long-running call on that same
call channel that we don't know about, since we will then keep
retrying the call on the same channel and keep getting
RX_PACKET_TYPE_BUSY responses.

Try to avoid this by returning the RX_CALL_TIMEOUT error when we get a
BUSY packet and there are other free call channels available on the
conn. When the application gets the error and retries the call, we
avoid using the same call channel again where possible. When all of
the call channels appear busy, we revert to effectively the old
behavior of retrying the call on the same channel until we get an
RX_CALL_DEAD (or similar) error.

FIXES 128671

Change-Id: I29b84aa36dcc856528c30d26811e5d6c4b78a1ca
Reviewed-on: http://gerrit.openafs.org/3431
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoImport of code from heimdal
Heimdal Developers [Thu, 27 Jan 2011 12:18:44 +0000]
Import of code from heimdal

This commit updates the code imported from heimdal to
8ce821c38744b30e226ff024f675668b991b984d (switch-from-svn-to-git-1998-g8ce821c)

Upstream changes are:

Asanka C. Herath (2):
      Revert use of backslash as an escape for double quote in config strings
      Condition roken's ssize_t definition for compatibility

Love Hornquist Astrand (1):
      don't undefined SLIST

Change-Id: I8855e060e88008f95eb2b24b029843081702fc04
Reviewed-on: http://gerrit.openafs.org/3761
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: Correct cm_volume locking
Jeffrey Altman [Thu, 27 Jan 2011 01:10:57 +0000]
Windows: Correct cm_volume locking

cm_volume_t flags was used for two categories of flags.  The first
protected by the cm_volume_t->rw lock.  The second protected by
the global cm_volumeLock.  Separate the flags field into two
afs_uint16 fields and break the flag space into FLAG and QFLAG.

Add assertions to the volume LRU functions to ensure that they
are always called with cm_volumeLock write-locked.

Correct two locations where cm_AdjustVolumeLRU() was called
read-locked instead of write-locked.

Change-Id: I7991b995a3c981cd5d08d9cbba09badc81518a5a
Reviewed-on: http://gerrit.openafs.org/3760
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agovol-salvage: Only delete bad vnodes during !check
Andrew Deason [Tue, 25 Jan 2011 16:46:38 +0000]
vol-salvage: Only delete bad vnodes during !check

In SalvageIndex, if check is true, we're only supposed to check for
consistency and not actually change anything (this is enforced by an
assert later on). The check for vnode magic consistency breaks this
assumption, and just always fixes the vnode.

Change this so we just error out if check is set, and the vnode magic
is wrong.

Change-Id: I9e2d3bd5e9e340a9b2ac66d70d52aa110b512561
Reviewed-on: http://gerrit.openafs.org/3756
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agoMacOS: fix SetFile call in growlagent makefile
Derrick Brashear [Tue, 25 Jan 2011 18:32:38 +0000]
MacOS: fix SetFile call in growlagent makefile

somehow this got corrupted earlier. fix it to use the right arg.

Change-Id: Ic597412819a9186d7b4b0b0c7e93e7c75b9a979a
Reviewed-on: http://gerrit.openafs.org/3755
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: cm_GiveUpAllCallBacksAllServersMulti()
Jeffrey Altman [Thu, 2 Dec 2010 05:36:58 +0000]
Windows: cm_GiveUpAllCallBacksAllServersMulti()

Implement a multi_RXAFS_GiveUpAllCallBacks version of
cm_GiveUpAllCallBacksAllServers() and use it in place of
the non-multi version.

Change-Id: I6d630ab7a5cf3a3434d088a9e8cde046b541cd3f
Reviewed-on: http://gerrit.openafs.org/3648
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: fix parameters and return value from nt_seek
Rod Widdowson [Sun, 23 Jan 2011 14:29:51 +0000]
Windows: fix parameters and return value from nt_seek

SetFilePointerEx takes specific values
(FILE_BEGIN/FILE_CURRENT/FILE_END) whilse fseek requires SEEK_SET,
SEK_END, SEEK_CUR.  It turns out that these overlap, but we should
not let that pass unchallenged.

SetFilePointerEx returns nonzero for success zero for failure. fseek
returns the other way around.

Neither of these changes currently matter, but we should fix them.

Change-Id: Ib31cf6265fa1e714232ff0d2e099c657e41f17e9
Reviewed-on: http://gerrit.openafs.org/3746
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: read and write take void* buffers, open takes a const char*
Rod Widdowson [Sun, 23 Jan 2011 10:59:28 +0000]
Windows: read and write take void* buffers, open takes a const char*

nt_read and nt_write were defined to take a char* buffer which was
then cast to a void *.  Meantime every call of OS_READ and
OS_WRITE were casting to a char*.

Equally every call of OS_OPEN was passing down a const char*,
causing warnings.

This checkin fixes this:
 nt_read : char* to void*
 nt_write: char* to const void*
 nt_open char*  to const char*

Change-Id: Id4e138b9d347e1a9f35241e162a105d5f462b168
Reviewed-on: http://gerrit.openafs.org/3744
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoOpenBSD: Eliminate complaint about built-in malloc.
Antoine Verheijen [Mon, 24 Jan 2011 07:38:03 +0000]
OpenBSD: Eliminate complaint about built-in malloc.

With OpenBSD 4.8, OpenBSD now uses gcc 4. With its new
defaults, the OpenAFS compile of the kernel module now
complains incessantly about the conflict between the
built-in malloc versus the kernel version (which has
extra parameters). Therefore, set -fno-builtin-malloc
when compiling the kernel module to remove the noise
since the differences can't be reconciled otherwise.

Change-Id: I7e28e6275ca90f73a8d1f65d82f1ccf2cc3a0ead
Reviewed-on: http://gerrit.openafs.org/3751
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Remove user.h from dir.c for OpenBSD 4.8
Antoine Verheijen [Mon, 24 Jan 2011 07:34:56 +0000]
OpenBSD: Remove user.h from dir.c for OpenBSD 4.8

OpenBSD 4.8 no longer needs to include h/user.h in dir.c
and it's presence now gets in the way of the compile so
remove it.

Change-Id: Ie270f94db69f344c5c1a3942db395dbe3a47d5e9
Reviewed-on: http://gerrit.openafs.org/3750
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: curproc has moved in OpenBSD 4.8
Antoine Verheijen [Mon, 24 Jan 2011 07:17:46 +0000]
OpenBSD: curproc has moved in OpenBSD 4.8

With OpenBSD 4.8, curproc has moved from h/proc.h to
h/systm.h. Fix rx_event.c to reflect this change.

Change-Id: I38d4676d445bfafa47f2ae973d789d0b9f6a687a
Reviewed-on: http://gerrit.openafs.org/3749
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Add support for OpenBSD 4.8
Antoine Verheijen [Mon, 24 Jan 2011 07:16:02 +0000]
OpenBSD: Add support for OpenBSD 4.8

Add param headers and other config info for OpenBSD 4.8
support.

Change-Id: I3af37dd79260d29bb7f3793c9a965bb84faab31f
Reviewed-on: http://gerrit.openafs.org/3748
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: remove unused label in ntops.c
Rod Widdowson [Sun, 23 Jan 2011 14:32:04 +0000]
Windows: remove unused label in ntops.c

The label 'retry' in nt_open is not used and was causing a warning.  Remove it.

Change-Id: I964d954e4f9cdb672c689a6e757a5f62bea8f65c
Reviewed-on: http://gerrit.openafs.org/3747
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agoWindows: remove faulty assumptions about device names in vol-salvage
Rod Widdowson [Sun, 23 Jan 2011 12:04:59 +0000]
Windows: remove faulty assumptions about device names in vol-salvage

The implementation has an assumption that all disk volumes have an
object name of \Device\HarddiskXXX (where XXX is a number).  This is
wrong since the name is purely a convention and since about WXP they
have been called \Device\HarddiskVolumeXXX.

Either way it is spurious to assume the format and then try to compare
the XXX.  This change just compares the strings.  This is done in a
case insenstive manner which is the safer option.  It is quite
feasible, but very unlikely that someone will uses 'case sensitively
different' object names.

Change-Id: Ifa91c88f2b17f747f30541b8833b722cf5993e48
Reviewed-on: http://gerrit.openafs.org/3745
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: more exports afsauthent.dll
Jeffrey Altman [Sat, 22 Jan 2011 01:13:45 +0000]
Windows: more exports afsauthent.dll

pr_SIdToName
afsconf_GetExtendedCellInfo
afsconf_UpToDate

required for new Explorer Shell functionality

Change-Id: I085d32b9786d9bf26d527cc3dd49c2d51ab1cb56
Reviewed-on: http://gerrit.openafs.org/3739
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Stefan Kueng <tortoisesvn@gmail.com>
Tested-by: Stefan Kueng <tortoisesvn@gmail.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: log and invalidate invalid dir pages
Jeffrey Altman [Sat, 22 Jan 2011 17:15:03 +0000]
Windows: log and invalidate invalid dir pages

If when parsing a directory page in cm_ApplyDir the consistency
checks fail, log a message and invalidate the page in case
it was a transient failure.

Change-Id: Iab690cd876a717fbdef30fb78e4ed559fa7e0a5a
Reviewed-on: http://gerrit.openafs.org/3742
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoOpenBSD: Make OpenBSD 4.7 param headers consistent
Antoine Verheijen [Fri, 21 Jan 2011 16:25:22 +0000]
OpenBSD: Make OpenBSD 4.7 param headers consistent

OpenBSD follows the policy of using a separate common
param file (post 1.4.x). However, when introducing support
for i386 OpenBSD 4.7, this was not done and a single param
config header file was created instead. This patch changes
the param files for OpenBSD 4.7 so that they are
consistent with the other OpenBSD versions by using a
separate common param header file.

Change-Id: I9d828d2447a2d018eeea0fbe9f4c047ac819d209
Reviewed-on: http://gerrit.openafs.org/3725
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: fix CreateFile params nt_unlink and nt_open
Jeffrey Altman [Thu, 20 Jan 2011 07:15:18 +0000]
vol: fix CreateFile params nt_unlink and nt_open

nt_unlink: when opening a file handle to assign delete on close
status the caller must request DELETE permission.

nt_open: make sure that DELETE permission along with FILE_SHARE_DELETE
is requested if we wish to permit another CreateFile call in the future
to assign delete on close.

Change-Id: Ib971ecc098dbc5eb64dcbfed93c334397c9d8ac9
Reviewed-on: http://gerrit.openafs.org/3711
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: namei_ops improve readability; fix namei_create on Windows
Jeffrey Altman [Thu, 20 Jan 2011 07:19:03 +0000]
vol: namei_ops improve readability; fix namei_create on Windows

use NAMEI_SPECDIRC, NAMEI_INODESPECIAL, INVALID_FD and other
preprocessor symbols in place of numeric constants.

namei_create() on Windows must call namei_CreateDataDirectories()
and retry if the afs_open() call fails with either ENOTDIR or
ENOENT.

modify namei_HandleToName() on Windows to make its decision
process clearer.

Change-Id: Ib1dfbaa3afe81f57685c95cdcae749c67cc95598
Reviewed-on: http://gerrit.openafs.org/3712
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: add comment nt_unlink cannot with fopen handles
Jeffrey Altman [Thu, 20 Jan 2011 07:06:12 +0000]
vol: add comment nt_unlink cannot with fopen handles

On Windows, files opened via the CRT fopen or open
POSIX compatibility functions cannot be marked for delete
on close.  Add a comment until such time as the use of
fopen and open on Windows can be replaced.

Change-Id: I9cec46bef332c981b2f64ba85e310783e3c71bc9
Reviewed-on: http://gerrit.openafs.org/3707
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>

13 years agovol: remove potential data loss warnings in vol-salvage.c
Jeffrey Altman [Thu, 20 Jan 2011 07:03:56 +0000]
vol: remove potential data loss warnings in vol-salvage.c

afs_sfsize_t is not always the same size as 'int'.
Avoid warnings by using afs_sfsize_t in the appropriate locations.

Change-Id: I88f7458421c3a9c2081597e5481d352bd5a2927a
Reviewed-on: http://gerrit.openafs.org/3706
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>

13 years agovol: use correct file name base for temporary file
Jeffrey Altman [Thu, 20 Jan 2011 07:01:38 +0000]
vol: use correct file name base for temporary file

Inconsistent use of temporary file name bases resulted in
the temporary files never being deleted.

Change-Id: I0be59282488bb2db0f1b7b5200df5bd6ac285711
Reviewed-on: http://gerrit.openafs.org/3705
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: use OS_UNLINK instead of unlink
Jeffrey Altman [Thu, 20 Jan 2011 06:59:46 +0000]
vol: use OS_UNLINK instead of unlink

Use OS_UNLINK in order to ensure that platform specific
unlink functions are called.  OS_UNLINK is unlink on Unix.

Change-Id: Idcfc18345abe5532e9ebe97fc757d6f435c2af52
Reviewed-on: http://gerrit.openafs.org/3704
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: fix namei_ListAFSSubDirs on Windows
Jeffrey Altman [Thu, 20 Jan 2011 06:56:33 +0000]
vol: fix namei_ListAFSSubDirs on Windows

The directory level represented by 'dirp2' does not exist on
Windows.  Do not attempt to close the DIR object that was
never opened.

Remove unnecessary function prototypes.

Change-Id: Ib6b27e52f0120115173cb04893092ba6f2da32fe
Reviewed-on: http://gerrit.openafs.org/3703
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: use OS_DIRSEP in many more places
Jeffrey Altman [Thu, 20 Jan 2011 06:51:26 +0000]
vol: use OS_DIRSEP in many more places

For consistency use OS_DIRSEP and OS_DIRSEPC throughout the
package.

Change-Id: I7bb93b4fe7324492e71a6d303e2b74723aaefc5a
Reviewed-on: http://gerrit.openafs.org/3702
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: fix _namei_examine_reg DELETE_ZLC usage
Jeffrey Altman [Thu, 20 Jan 2011 06:45:29 +0000]
vol: fix _namei_examine_reg DELETE_ZLC usage

replace 'i' with 'dirl' and assign the new variable a value
before using it.

Change-Id: Ia93b6c75154b6f2a9238f0e9c2881d6c79c36540
Reviewed-on: http://gerrit.openafs.org/3701
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: make it clearer that SetOGM is not impl on Windows
Jeffrey Altman [Thu, 20 Jan 2011 06:39:11 +0000]
vol: make it clearer that SetOGM is not impl on Windows

Separate the SetOGM function into two implementations to
match the other OGM functions.  Implement the Windows version
as returning -1 always.

In the future, the OGM functions could be renamed to express
their purpose: saving and restoring the volume id and tag
associated with a file.   Windows currently encodes this
information using the CreationDate and the file name but
in the future this information might be stored in an alternate
data stream.

Change-Id: I7e60cceb4548b7ad3a1e045f390e7052fece4926
Reviewed-on: http://gerrit.openafs.org/3700
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: avoid double dir separators from addtoname
Jeffrey Altman [Thu, 20 Jan 2011 06:35:08 +0000]
vol: avoid double dir separators from addtoname

The manner in which addtoname() is used can often result
in double dir separators in path names.  Check for it and
avoid doubling the separator in addtoname()

Change-Id: Ie33459bec3bcc93f81f13020cd35ed8c473b6ca5
Reviewed-on: http://gerrit.openafs.org/3699
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: clear ih_synced before dropping lock
Jeffrey Altman [Thu, 20 Jan 2011 06:23:37 +0000]
vol: clear ih_synced before dropping lock

in ih_reallyclose() clear ihP->ih_synced before dropped
the IH_LOCK.  ih_reallyclose() is already committed to
performing the OS_SYNC() call.  Failing to clear the ih_synced
flag before dropping the lock permits another thread, the ih_sync_all()
thread for example, to also conclude that it must sync the
file.

Change-Id: Icb928f7d2ec2342a57ed21b440ae0580c980e9d2
Reviewed-on: http://gerrit.openafs.org/3696
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoMacOS: don't allow krb5 at login when AD plugin authenticates
Derrick Brashear [Thu, 20 Jan 2011 03:56:12 +0000]
MacOS: don't allow krb5 at login when AD plugin authenticates

if AD is being used to verify authentication (e.g. via builtin)
don't allow get krb5 at login to succeed. since a helper does this we can't
grey the option, but we can decline to act on it.

Change-Id: I1f2bb891377c3ed9765f9e58141c77ec54d3ae22
Reviewed-on: http://gerrit.openafs.org/3690
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: nt_DriveToDev must return a value
Jeffrey Altman [Thu, 20 Jan 2011 07:14:14 +0000]
vol: nt_DriveToDev must return a value

Change-Id: I27cf468e9d61542f9893367a18c2f7dd6854cdf6
Reviewed-on: http://gerrit.openafs.org/3710
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: nt_open should not create missing directories
Jeffrey Altman [Thu, 20 Jan 2011 07:11:57 +0000]
vol: nt_open should not create missing directories

Revert the prior change that added directory creation
to nt_open().  nt_open() should fail and the caller should
create the directories.  Failure to do so results in an
incomplete tree which confuses the salvager.

Change-Id: Ieb2a05baada2531dad1869e95823fe2006bdc2c6
Reviewed-on: http://gerrit.openafs.org/3709
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: Make ntops functions 64-bit capable
Jeffrey Altman [Thu, 20 Jan 2011 07:09:44 +0000]
vol: Make ntops functions 64-bit capable

Add 64-bit offset and length support to ntops functions.

Change-Id: I1804d49df344839598c39084b7763cec4100c8c7
Reviewed-on: http://gerrit.openafs.org/3708
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: avoid defining unused struct on windows
Jeffrey Altman [Thu, 20 Jan 2011 06:33:27 +0000]
vol: avoid defining unused struct on windows

namei_ogm_t is not used on Windows.  There is no equivalent
to owner/group mode.

Change-Id: Ib0675f106bf5e77b006ebe5eae5d67ee232313e0
Reviewed-on: http://gerrit.openafs.org/3698
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: indent cpp definitions; add NAMEI_SPECDIRC
Jeffrey Altman [Thu, 20 Jan 2011 06:31:41 +0000]
vol: indent cpp definitions; add NAMEI_SPECDIRC

Change-Id: Id6d962a7c9e78f6a4b4bb7a6250e77efd40cc3c4
Reviewed-on: http://gerrit.openafs.org/3697
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: remove [UN]LOCKFILE data loss warnings on Windows
Jeffrey Altman [Thu, 20 Jan 2011 06:20:05 +0000]
vol: remove [UN]LOCKFILE data loss warnings on Windows

Cast to DWORD since the results of the mask and shift
are guaranteed to be 32-bit.  This avoids sixteen warnings.

Change-Id: Ifdb9356e207b4b8791b5923985a70efeb457edb1
Reviewed-on: http://gerrit.openafs.org/3695
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoWindows: build mtafsdir.lib and use it
Jeffrey Altman [Thu, 20 Jan 2011 05:57:03 +0000]
Windows: build mtafsdir.lib and use it

The executables built in src/vol are pthreaded.  Therefore, they
require a pthread safe version of the src/dir objects.  Instead
of building the dir object files in yet another directory, create
a mtafsdir.lib in src/dir and use it in src/vol, src/tviced and
src/tsalvaged.

Change-Id: I50017bd56ec741af1ff3fc1dd39208e610369088
Reviewed-on: http://gerrit.openafs.org/3693
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoFreeBSD: properly identify the rxk_Listener so that msleep() returns
Toby Burress [Thu, 20 Jan 2011 04:32:01 +0000]
FreeBSD: properly identify the rxk_Listener so that msleep() returns

afs_osi_Wakeup() was not being called with the proper argument,
causing afs_osi_Sleep() to fail forever while waiting for rxk_Listener
to wake up.

Change-Id: I1b650c57ae507987359d0cb230400d3d23984cfa
Reviewed-on: http://gerrit.openafs.org/3692
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Change code optimization setting
Antoine Verheijen [Wed, 19 Jan 2011 23:06:03 +0000]
OpenBSD: Change code optimization setting

Change the optimization level set when optimization is
enabled during build from the sytem default of '-O' to
'-O2' for OpenBSD systems.

Change-Id: I20962c124da5426409638453e1efdd37b4e5f393
Reviewed-on: http://gerrit.openafs.org/3689
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoMove check for unspecified CFLAGS in configure.ac
Antoine Verheijen [Wed, 19 Jan 2011 22:57:25 +0000]
Move check for unspecified CFLAGS in configure.ac

configure.ac provides a check to see if the user has
specified CFLAGS and if not, it sets CFLAGS to a blank
(not NULL) string so that the resultant configure script
does not set '-g' and/or '-O2' by default. This check
occurs after AC_USE_SYSTEM_EXTENSIONS in the configure.ac
file. However, on at least some systems, such as OpenBSD,
AC_USE_SYSTEM_EXTENSIONS expands to include the code that
configure uses to set '-g -O2' so the check has no effect
and '-g' or '-O2' can not be turned off. This patch moves
the "CFLAGS specified" check so that it precedes the
AC_USE_SYSTEM_EXTENSIONS directive, in which case
everything works.

Change-Id: I68ed5a3c5129a55f7dcd7413c93f35cc7cb14b9c
Reviewed-on: http://gerrit.openafs.org/3688
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: No ruid/rgid in cred structure.
Antoine Verheijen [Wed, 19 Jan 2011 21:14:47 +0000]
OpenBSD: No ruid/rgid in cred structure.

OpenBSD does not retain ruid or rgid values in the
afs_creds_t structure (ucred). This patch, therefore,
removes any references to them in the afs.h file. Since
no attempt is made to actually use the removed code (or
macros) on OpenBSD, at least at this time, this does
not result in any real functional change.

Change-Id: Iaceb792f0c39ba6f7caa205bc66af16a23193549
Reviewed-on: http://gerrit.openafs.org/3687
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Don't call non-existent routines in osi_vfsops.c
Antoine Verheijen [Wed, 19 Jan 2011 20:43:21 +0000]
OpenBSD: Don't call non-existent routines in osi_vfsops.c

The OpenBSD version of osi_vfsops.c makes calls to routines
that do not exist in the system and aren't needed. In the
past, these calls were wrappered by an #ifdef for
AFS_DISCON_ENV which was never defined. That #ifdef is now
gone so this patch removes the calls to these routines,
which are remnants of the NBSD origins for this particular
code.

Change-Id: I8cc97fde8cd5cf6c241dd3074c0f97b5d3e2ccfd
Reviewed-on: http://gerrit.openafs.org/3686
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agoOpenBSD: Fix parameters in call to afs_close()
Antoine Verheijen [Wed, 19 Jan 2011 20:30:24 +0000]
OpenBSD: Fix parameters in call to afs_close()

OpenBSD passes an extra parameter in a call to afs_close()
in its OS-specific code. It went undetected in the past
but now causes compile headaches because of more accurate
prototypes in the header file. This patch removes the
offending extra parameter (which never actually caused any
problems because it was the last one in the parameter list).

Change-Id: I6bde0c6c98867c47a574fa5e3e01ec560deb17b0
Reviewed-on: http://gerrit.openafs.org/3685
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Install no-NFS version of libafs
Antoine Verheijen [Wed, 19 Jan 2011 20:13:09 +0000]
OpenBSD: Install no-NFS version of libafs

The build process for OpenBSD builds the no-NFS version of
the kernel module but, at install time, the NFS version is
subseuquently built and installed while the no-NFS version
is ignored. The NFS version does not load and is not needed
in OpenBSD so this patch makes sure that the no-NFS version
is installed and used.

Change-Id: I1d07334121911c22aea1f6d157f34e1b40e5b780
Reviewed-on: http://gerrit.openafs.org/3684
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoMacOS: panic decoder should check for unloaded kexts
Derrick Brashear [Wed, 19 Jan 2011 19:58:59 +0000]
MacOS: panic decoder should check for unloaded kexts

if the openafs kext was unloaded from under us, try to handle that.

Change-Id: I6a5e56da3c4b6290b71dd10e919a0f374998338e
Reviewed-on: http://gerrit.openafs.org/3683
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Fix use of macros for AFS_KALLOC/AFS_KFREE
Antoine Verheijen [Wed, 19 Jan 2011 18:52:10 +0000]
OpenBSD: Fix use of macros for AFS_KALLOC/AFS_KFREE

The macro definitions for AFS_KALLOC and AFS_KFREE used by
afs_osi_alloc.c to perform OS-specific memory allocation
are only defined by the OpenBSD osi_machdep.h header file
if they already exist, which means: never. This patch fixes
this situtation and makes sure that afs_osi_alloc.c
actually uses them, eliminating a direct OS-specific call
in afs_osi_alloc.c.

Change-Id: Ida1608774de3bc5861b223574f3cbf0d9647266c
Reviewed-on: http://gerrit.openafs.org/3682
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agoOpenBSD: Remove macros definitions for afs_osi_Alloc et al.
Antoine Verheijen [Wed, 19 Jan 2011 18:20:21 +0000]
OpenBSD: Remove macros definitions for afs_osi_Alloc et al.

The afs_osi_Alloc, afs_osi_Free, etc. routines are now
defined in afs_osi_alloc.c so the macro definitions defined
by osi_machdep.h for OpenBSD are no longer needed and, in
fact, interfere with the compile. Remove them.

Change-Id: I98a261b5cd5e44136e4dde677f86d90212b0cfba
Reviewed-on: http://gerrit.openafs.org/3681
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovolser: select() cannot be used to sleep on windows
Jeffrey Altman [Sun, 16 Jan 2011 21:42:15 +0000]
volser: select() cannot be used to sleep on windows

On Windows must use Sleep() instead of select() with no
fd_set parameters.

Change-Id: I693718c74dd9be0b4eb58ec26afac0882845e71c
Reviewed-on: http://gerrit.openafs.org/3675
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>

13 years agoWindows: refactor cm_CheckCBExpiration multihomed
Jeffrey Altman [Sun, 16 Jan 2011 20:49:02 +0000]
Windows: refactor cm_CheckCBExpiration multihomed

cm_CheckCBExpiration() is refactored to make it easier
to read the decision process.  cm_CheckCBExpiration()
determines when a callback is no longer usable and as a
result the object status info should be discarded.

The windows cache manager preserves status info past
callback expiration if all of the sources of a volume
became inaccessible prior to the callback expiration
time.  The cache manager was improperly preserving the
status info for objects when the callback was issued by
a multi-homed file server when only the interface that
issued the callback is down.

A separate cm_server_t object is used to represent
each file server interface.  When one interface goes
down and others are left up, the cache manager will
now replace the down cm_server_t reference for one that
is up.  This substitution is performed as a side effect
of computing the effective downTime in cm_CBServersDownTime().

Change-Id: Ia6336a3bdd0219497fd47460accddd0cd2629f00
Reviewed-on: http://gerrit.openafs.org/3674
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoWindows: use cm_ServerEqual() in cm_Analyze()
Jeffrey Altman [Fri, 14 Jan 2011 05:22:39 +0000]
Windows: use cm_ServerEqual() in cm_Analyze()

In cm_Analyze() when processing VNOVOL, VMOVED or VOFFLINE errors
do not use pointer equivalence as the test to determine if a
server is affected by the error.  Use cm_ServerEqual() instead.

Fix cm_ServerEqual() to work even when both cm_server_t objects
are not UUID aware.

Change-Id: Icb83ba77984075216a44cdeba689bc05e98ed92a
Reviewed-on: http://gerrit.openafs.org/3659
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agoRe-enable rx connection hard timeout
Rainer Toebbicke [Tue, 18 Jan 2011 13:56:17 +0000]
Re-enable rx connection hard timeout

A missing "conn->" in rxi_CheckCall prevents checking hard connection timeouts.

Change-Id: I893a1ba71957e2a830029dfc4e3ac51fa461d6ca
Reviewed-on: http://gerrit.openafs.org/3677
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agovol: Windows requires binary fmode for salvaged
Andrew Deason [Mon, 17 Jan 2011 18:35:46 +0000]
vol: Windows requires binary fmode for salvaged

Same as 98f633c60d20834cee0bbc0770a3146b06642a59, but for the
salvageserver.

Change-Id: I961e31fd6a2a6e08e3bd4381793195bb1e388e7c
Reviewed-on: http://gerrit.openafs.org/3676
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

13 years agovol: use OS_UNLINK() instead of unlink()
Jeffrey Altman [Sat, 15 Jan 2011 16:49:54 +0000]
vol: use OS_UNLINK() instead of unlink()

OS_UNLINK provides the platform specific mechanism
for unlinking objects.

Change-Id: Id0a1f50e57fe3c21c46e137e86507bf28bafa046
Reviewed-on: http://gerrit.openafs.org/3670
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementia.org>

13 years agovol: construct proper VolDir path on Windows
Jeffrey Altman [Sat, 15 Jan 2011 16:54:40 +0000]
vol: construct proper VolDir path on Windows

namei_HandleToInodeDir initializes the path name with
the nt_drive value.  Therefore calling addtoname(name, name->n_drive)
produces the invalid path "D:\D:\\Vol_.....".  Remove the unnecessary
addtoname() call.

Change-Id: I4933d3f4fda76050eecde3caeaa6bc30011cd974
Reviewed-on: http://gerrit.openafs.org/3672
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: fdHandleAllocateChunk should init all fields
Jeffrey Altman [Sat, 15 Jan 2011 16:52:45 +0000]
vol: fdHandleAllocateChunk should init all fields

fdHandleAllocateChunk() failed to initialize the fd_ihnext
and fd_ihprev fields.

Change-Id: I605b6fc3c4cefe6e5c4169762fbd6047fb895b70
Reviewed-on: http://gerrit.openafs.org/3671
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

13 years agovol: use OS_DIRSEP when constructing paths
Jeffrey Altman [Sat, 15 Jan 2011 16:47:47 +0000]
vol: use OS_DIRSEP when constructing paths

Do not assume that / is the directory separator when constructing
paths for internal consumption or log messages.  Use OS_DIRSEP
or OS_DIRSEPC as appropriate.

Change-Id: Ifed618dfa30f2aee27de1ef57de5e504d6701d4d
Reviewed-on: http://gerrit.openafs.org/3669
Reviewed-by: Derrick Brashear <shadow@dementia.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>