openafs.git
8 years agoopr: queue swap
Michael Meffie [Thu, 18 Aug 2011 21:04:14 +0000]
opr: queue swap

Add inline function to pivot two queues.
Implementation by Simon Wilkinson.

Change-Id: I704a1ff3e0d6314e2bfe47c870226cb6ffd44b1b
Reviewed-on: http://gerrit.openafs.org/5742
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

8 years agoexternal: trigger git commit-msg hooks on import
Andrew Deason [Thu, 3 Nov 2011 17:33:55 +0000]
external: trigger git commit-msg hooks on import

When we import commits, the new commit does not have a Change-Id in
the commit message. Usually a git commit-msg hook will be installed to
automatically insert a Change-Id into the commit message if one does
not already exist, so trigger the hook by performing a no-op 'git
commit --amend'.

Change-Id: Ie5e83c529b83c9f3d269df6742a59e965dbc6407
Reviewed-on: http://gerrit.openafs.org/5797
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

8 years agoexternal: commit imports with --no-verify
Andrew Deason [Thu, 3 Nov 2011 17:27:22 +0000]
external: commit imports with --no-verify

Currently, importing external commits with import-external-git.pl may
fail if the current repository checks for whitespace errors, and the
external commits contain whitespace errors. So run the commit with
--no-verify, so we can make the commit before fixing up the whitespace
errors immediately afterwards.

Change-Id: I8c9c39550e118956aa847e0ec5c5b1181f42e6cb
Reviewed-on: http://gerrit.openafs.org/5796
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

8 years agoImport setenv.c and unsetenv.c from roken
Andrew Deason [Thu, 3 Nov 2011 17:17:07 +0000]
Import setenv.c and unsetenv.c from roken

Import the roken implementation of setenv and unsetenv, which are
required on HP-UX.

Change-Id: Ia91f407d696f3add27796bf9e2f5998717f4c825
Reviewed-on: http://gerrit.openafs.org/5795
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

8 years agoWindows: fix deadlock in symlink Attrib retrieval
Peter Scott [Wed, 2 Nov 2011 23:29:00 +0000]
Windows: fix deadlock in symlink Attrib retrieval

Commit df22620f66f5ce92776177d4d800fc7f4ae4ae99 introduced a
deadlock when retrieving file attributes for symlink objects.
To correct the deadlock, do not hold the Fcb locks during the
AFSRetrieveFileAttributes() call.

Change-Id: I8d3d2c5f54d4e5c97f690fd1fd6db62128317cf2
Reviewed-on: http://gerrit.openafs.org/5792
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

8 years agoWindows: always set reparse point attrib for symlink
Jeffrey Altman [Wed, 2 Nov 2011 23:27:42 +0000]
Windows: always set reparse point attrib for symlink

Make sure that we always set the reparse point attribute
for symlinks in directory enumerations and attribute queries.

Change-Id: Ie7520cef64793e3385f229934189fc56a3edb584
Reviewed-on: http://gerrit.openafs.org/5793
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

8 years agoWindows: remember when a reparse point is opened
Peter Scott [Wed, 2 Nov 2011 23:25:36 +0000]
Windows: remember when a reparse point is opened

If the mount point or symlink is opened as a reparse point
we must remember that in the Ccb so that later reparse point
processing can be implemented properly.  For now remember
with the CCB_FLAG_MASK_OPENED_REPARSE_POINT flag.

Change-Id: I32997a85701a95a5be1402cdcd3335a938395760
Reviewed-on: http://gerrit.openafs.org/5791
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

8 years agoWindows: Mount points are always directories
Jeffrey Altman [Wed, 2 Nov 2011 14:35:14 +0000]
Windows: Mount points are always directories

Set the directory file attribute for all mount points
since a mount point target is always represented as
a directory.

Change-Id: I57a8e44d1382093311e8263c18d55f22f2f745ea
Reviewed-on: http://gerrit.openafs.org/5784
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Peter Scott <pscott.kd@gmail.com>
Tested-by: Peter Scott <pscott.kd@gmail.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

8 years agoWindows: Retrieve file attributes on symlinks
Peter Scott [Wed, 2 Nov 2011 14:23:48 +0000]
Windows: Retrieve file attributes on symlinks

When a component is a symlink, correctly retrieve file
attributes during a query file information request.

FIXES 130283

Change-Id: I0e9a576ca57acf9c65a5dc0eac8c9d65ddf6ab6f
Reviewed-on: http://gerrit.openafs.org/5782
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

8 years agoWindows: add AFSRetrieveParentPath()
Peter Scott [Wed, 2 Nov 2011 14:18:09 +0000]
Windows: add AFSRetrieveParentPath()

Change-Id: Ic3da7f09f12f0e089b9561a542b2cf04ff61c98d
Reviewed-on: http://gerrit.openafs.org/5783
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agokdump-build: don't install when it hasn't been built for *bsd*
Jonathan A. Kollasch [Tue, 11 Oct 2011 17:56:08 +0000]
kdump-build: don't install when it hasn't been built for *bsd*

Change-Id: Iac9df9eb199433127941590d24633dfe9bbc6815
Reviewed-on: http://gerrit.openafs.org/5596
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoklog.krb5: enforce DES for rxkad
Jeffrey Altman [Fri, 14 Oct 2011 13:10:19 +0000]
klog.krb5: enforce DES for rxkad

0. Always request a TGT regardless of the state of
   writeTicketFile.

1. request des-cbc-crc when requesting a ticket for an
   rxkad service principal

2. check the returned key length to ensure that it matches
   the permitted length of an rxkad key.  If not, generate
   an error instead of overwriting memory and continuing.

FIXES 130278

Change-Id: Ib7d27acc60c1b38dab4e2a56f83643477afb9721
Reviewed-on: http://gerrit.openafs.org/5619
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoWindows: DE Reference count added during create
Peter Scott [Tue, 1 Nov 2011 14:55:37 +0000]
Windows: DE Reference count added during create

A reference count is added under lock to a newly created DE to
ensure it is not removed through deletion while in flight.

Change-Id: If0955c4a6c4cf197a4e071f9c6022875aabb70bb
Reviewed-on: http://gerrit.openafs.org/5775
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: change AFSProcessUserFsRequest default
Peter Scott [Mon, 31 Oct 2011 22:10:07 +0000]
Windows: change AFSProcessUserFsRequest default

Default the return code for AFSProcessUserFsRequest() to
STATUS_INVALID_DEVICE_REQUEST instead of STATUS_INVALID_PARAMETER.

Change-Id: Ie7fb260a19617635c11c92e2daf5947cef73c9be
Reviewed-on: http://gerrit.openafs.org/5774
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: correct ordering of reparse point tests
Peter Scott [Mon, 31 Oct 2011 22:07:23 +0000]
Windows: correct ordering of reparse point tests

Test for whether an object is a reparse point before testing
whether the provided buffer is large enough to hold the result
if it is.

FSCTL_GET_REPARSE_POINT
FSCTL_SET_REPARSE_POINT
FSCTL_DELETE_REPARSE_POINT

Change-Id: If6c1b9b6e5853d7759f169943310321d408190e0
Reviewed-on: http://gerrit.openafs.org/5748
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: log FSCTL_CSC_INTERNAL requests
Peter Scott [Mon, 31 Oct 2011 22:05:31 +0000]
Windows: log FSCTL_CSC_INTERNAL requests

FSCTL_CSC_INTERNAL requests are not supported and will
return STATUS_INVALID_DEVICE_REQUEST.  However, log them
for easier debugging.

FSCTL_CSC_INTERNAL is defined in ntifs.h only for
Vista and above.  Since afsredir is built for XP
and above its value must be conditionally defined
by us.

Change-Id: I39cf1f9d068d28ab6c9267c302d49feece6af700
Reviewed-on: http://gerrit.openafs.org/5773
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: add braces to case statements
Peter Scott [Mon, 31 Oct 2011 21:59:51 +0000]
Windows: add braces to case statements

A number of case statements did not have braces protecting
the block.  Add them.

Change-Id: I313462ca0c7490931e7c994ec45e39be071e782b
Reviewed-on: http://gerrit.openafs.org/5772
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Directory Entry Processing
Peter Scott [Sat, 29 Oct 2011 17:55:46 +0000]
Windows: Directory Entry Processing

Memory corruption was detected due to improper handling
of directory entries and the short name b-tree during
directory revalidation and tear down.  The problem
resulted in a BSOD most frequently when roaming profiles
were written back to the file server or when multiple
clients were actively modifying the contents of a directory
simultaneously.

This patchset makes a number of changes:

 . rename processing no longer handles collisions locally.
   the file server replaces a target atomicly when there
   is a collision.

 . directory validation is no longer delayed until
   a handle is closed

 . data version numbers are checked for implied
   invalidation

Change-Id: Ib4b281bfb5da4b5a5ce735e85d85c50e15fbcde1
Reviewed-on: http://gerrit.openafs.org/5736
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: additional trace logging for afsredirlib
Peter Scott [Sat, 29 Oct 2011 17:29:55 +0000]
Windows: additional trace logging for afsredirlib

A large number of addition trace log entries were added
to assist in diagnosing problems in the dir entry processing.

Change-Id: I85cab4d66680eb5b9c00843c49c33a81acb18bc2
Reviewed-on: http://gerrit.openafs.org/5739
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Reimplement AFSGenerateCRC()
Peter Scott [Sat, 29 Oct 2011 15:34:07 +0000]
Windows: Reimplement AFSGenerateCRC()

Replace the home grown CRC generation with a call to
RtlHashUnicodeString() which is present on Windows XP
and above.

Change-Id: Id98a1223e0aad54b3c43a16653ab888ac883639f
Reviewed-on: http://gerrit.openafs.org/5738
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoLinux: 3.1: update RCU path walking detection in permission i_op
Marc Dionne [Sat, 29 Oct 2011 23:23:07 +0000]
Linux: 3.1: update RCU path walking detection in permission i_op

The permission() inode operation changed again with kernel 3.1,
back to the form it had before 2.6.38.  This compiles fine,
but is missing the new way of detecting when we get called in
RCU path walking mode, resulting in system hangs.

Change-Id: I9a9a0c13aa3b35d8fe2bd592e2bca8187eacacc5
Reviewed-on: http://gerrit.openafs.org/5740
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agorx: Remove the ADAPT_WINDOW code
Simon Wilkinson [Tue, 21 Jun 2011 17:32:02 +0000]
rx: Remove the ADAPT_WINDOW code

RX still has the remnants of an old mechanism for doing RTT and
congestion window detection. This code is #ifdef'd out using
the ADAPT_WINDOW define, but is pretty much unservicable these days,
as it collides with the TCP style implementation (with ADAPT_WINDOW
enabled, both will attempt to manipulate a connections RTT and
window size)

As the current TCP-style RTT and window calculation seem to work
much better in deployment, and there isn't much hope for us being
able to maintain two different congestion mechanisms, just remove
ADAPT_WINDOW. It is in git, if we ever want it back (not that I
think we ever would).

Change-Id: I4a22354e362fde8021f7d8b19e099a0f7b4fba3b
Reviewed-on: http://gerrit.openafs.org/5737
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: cm_Rename break in 9cf7a628c2ff178b2fa236d2c0a41ca9be315036
Jeffrey Altman [Fri, 28 Oct 2011 21:43:24 +0000]
Windows: cm_Rename break in 9cf7a628c2ff178b2fa236d2c0a41ca9be315036

9cf7a628c2ff178b2fa236d2c0a41ca9be315036 broke renaming of
a file to a name that does not exist.  Put back the code that
makes that possible with a comment.

Change-Id: I6c13c0150d2366c9578502d0c8f6188bdc40a9c8
Reviewed-on: http://gerrit.openafs.org/5735
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: out of date version not in current chunk
Jeffrey Altman [Fri, 28 Oct 2011 15:36:10 +0000]
Windows: out of date version not in current chunk

In buf_GetNewLocked(), the comparision to decide whether a
cm_buf_t is a member of the current chunk must take the data
version into account.  If the data version is out of date, it
is not part of the current chunk and is an object that can be
safely recycled.

Change-Id: Id8ca0a897a0a65efc1cabbe39107ca210b48bd4e
Reviewed-on: http://gerrit.openafs.org/5734
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agolinux: Update Packaging to build OpenAFS services for Fedora's systemd
Edward Z. Yang [Tue, 18 Oct 2011 03:16:15 +0000]
linux: Update Packaging to build OpenAFS services for Fedora's systemd

Fedora 15 now uses systemd (see http://fedoraproject.org/wiki/Systemd)
for the OS init system. While it currently has backwards
compatibility with older SysV-style init scripts, future versions of
Fedora may no longer support it, and OS startup tends to be faster
with the systemd service units. Also, systemd runs all the service's
processes within a linux kernel cgroup.
(see http://www.kernel.org/doc/Documentation/cgroups/cgroups.txt)

This change includes an openafs-client.service and
openafs-server.service unit files for the client and server packages
respectively.

Client
- Loading the openafs module was moved into
/etc/sysconfig/modules/openafs-client.modules. This causes the OS to
load the module on boot. This is the preferred way for modules to be
loaded with Fedora. (See
http://docs.fedoraproject.org/en-US/Fedora/15/html/Deployment_Guide/sec-Persistent_Module_Loading.html
for more details)
- The CellServDB file is generated with sed rather than cat.
This change was made because Systemd doesn't execute as a shell
script, but rather executes processes directly. Rather than invoking
a shell to concatenate the CellServDB.* files, they're written to the
CellServDB file using a sed oneliner.
- Do all of the proper kernel module loading and unloading.

Server
- Since systemd uses cgroups, when the service is shut down, all
processes in the openafs-server.service cgroup will be terminated.

The other changes are standard as per:
    http://fedoraproject.org/wiki/Packaging:ScriptletSnippets#Systemd

Original version by Jonathan Billings <jsbillin@umich.edu>.

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

Change-Id: Ifb41790ffe107b319097b9750273aecfe82c3349
Reviewed-on: http://gerrit.openafs.org/5637
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Alex Chernyakhovsky <achernya@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agoWindows: only flush buffers on shutdown if running
Jeffrey Altman [Thu, 27 Oct 2011 21:57:25 +0000]
Windows: only flush buffers on shutdown if running

If a service shutdown message is received prior to the
service entering the running state, do not attempt to
buf_CleanAndReset() because the required data structures
and locks are not initialized.

Change-Id: I29f20dbe91230f078334ca58e8791d2422cdecfd
Reviewed-on: http://gerrit.openafs.org/5733
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Do not EEXIST exact match during rename
Jeffrey Altman [Tue, 25 Oct 2011 19:32:11 +0000]
Windows: Do not EEXIST exact match during rename

AFS Rename operations on the file server will delete a
target file if it exists.  Do not prevent renames because
an exact match of the target name exists in the target
directory.

Change-Id: I680f8886575a9276a8df05965d14bbe7f24fba5d
Reviewed-on: http://gerrit.openafs.org/5731
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoopr: build on Windows after queue.h move
Jeffrey Altman [Tue, 25 Oct 2011 19:42:44 +0000]
opr: build on Windows after queue.h move

55efff93fe2762b58dc70cdd20aaf97e39825929 broke Windows.
update the NTMakefile in src/util and src/opr to correspond
to the move of queue.h.

Change-Id: I3379c131ee39758a36cf8574aff5766f79497236
Reviewed-on: http://gerrit.openafs.org/5732
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Look for 8.3 name when doing a rename
Rod Widdowson [Sat, 22 Oct 2011 15:46:26 +0000]
Windows: Look for 8.3 name when doing a rename

If we are doing a destructive rename we need to find whether the
target file exists.  This is done in the usual way (case sensitive
case insensitive), but the short name is not looked for.

This means that the rename of a file to a short name will not
supersede correctly; rather the service refuses the rename since
the target existed already.

This patch looks the target name up in the shortname tree if the
target name is short and all else has failed.

Change-Id: Ie054516a378edf52938d3465143a16107a99c0f6
Reviewed-on: http://gerrit.openafs.org/5652
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: New Trace for DirEntry tracking problem
Peter Scott [Thu, 20 Oct 2011 18:39:31 +0000]
Windows: New Trace for DirEntry tracking problem

Incuded in this patch is additional trace for logging how
directory entries are added, modified and deleted throughout
the redirector

Change-Id: I7ea80331ca24d543579603986f4eb6bcdd0bf072
Reviewed-on: http://gerrit.openafs.org/5640
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Free work item memory with tag value
Jeffrey Altman [Mon, 17 Oct 2011 13:29:37 +0000]
Windows: Free work item memory with tag value

For consistency throughout the source module, free all work
item memory allocations with ExFreePoolWithTag.

Change-Id: I4e9b275bb8b92a81884d36ce4c8ed754d281bbf8
Reviewed-on: http://gerrit.openafs.org/5627
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: free pointer after last reference
Jeffrey Altman [Mon, 17 Oct 2011 13:28:11 +0000]
Windows: free pointer after last reference

This is a superficial change but is being done for readability.
If given the choice of freeing memory and then testing the pointer
value or vice-versa, test the pointer value first.

Change-Id: I189851c265a257d433df111622837db1d706c054
Reviewed-on: http://gerrit.openafs.org/5626
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: AFSInitPIOCtlDirectoryCB free in try_exit
Jeffrey Altman [Mon, 17 Oct 2011 13:26:13 +0000]
Windows: AFSInitPIOCtlDirectoryCB free in try_exit

Process all conditional release of allocated memory on
failure within the try_exit on failure.

Change-Id: I4a29f6e9d2506c00713558849c49c377a74d607a
Reviewed-on: http://gerrit.openafs.org/5625
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: close leak in AFSPrimaryVolumeWorkerThread
Jeffrey Altman [Tue, 25 Oct 2011 17:30:19 +0000]
Windows: close leak in AFSPrimaryVolumeWorkerThread

Prevent a leak of the non-paged pool and a Resource by freeing all
allocations within the Specific.Directory.PIOCtlDirectoryCB object.

Use AFSExFreePool() instead of ExFreePool() because the objects
are allocated from the AFSRedir.sys driver allocator.

FIXES 130280

Change-Id: I3b1c8fa838cfdf434bf7c08bb8935999cf328357
Reviewed-on: http://gerrit.openafs.org/5730
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: afsredir only wants shortName if not 8.3
Jeffrey Altman [Tue, 25 Oct 2011 01:14:59 +0000]
Windows: afsredir only wants shortName if not 8.3

When performing a directory enumeration for the afs redirector
only specify a shortname if the name is not valid 8.3.

Change-Id: I650be75b9ea8e41721272cd8bd426f1989ca01fb
Reviewed-on: http://gerrit.openafs.org/5727
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Use SetFlag macro when setting flag
Jeffrey Altman [Thu, 20 Oct 2011 18:06:47 +0000]
Windows: Use SetFlag macro when setting flag

Change-Id: Ibda7ac5e30a623dc26f031831e75380fdf700a88
Reviewed-on: http://gerrit.openafs.org/5642
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: minor type in trace output
Jeffrey Altman [Thu, 20 Oct 2011 18:05:59 +0000]
Windows: minor type in trace output

Change-Id: I72b348dea83547260575c0bd2aca327ea4f72532
Reviewed-on: http://gerrit.openafs.org/5641
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: AFSRemoveCaseInsensitiveDirEntry corrections
Peter Scott [Tue, 18 Oct 2011 14:04:48 +0000]
Windows: AFSRemoveCaseInsensitiveDirEntry corrections

If there is a fLink but no parentLink for the DirEntry being
removed, return the fLink as the new value for *RootNode.

This patch set also includes two changes that improve readability
but do not alter the behavior.

Change-Id: I112265798df9d46c83ffacddc7d97ad0f6fce5bf
Reviewed-on: http://gerrit.openafs.org/5638
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Reviewed-by: Peter Scott <pscott.kd@gmail.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: AFSEvaluateTargetByName free buffer if no return
Jeffrey Altman [Mon, 17 Oct 2011 13:22:53 +0000]
Windows: AFSEvaluateTargetByName free buffer if no return

For consistency with other functions in AFSCommSupport
modify AFSEvaluateTargetByName to free the DirEntry on
completion if the caller has not provided an out parameter
to accept it.

Change-Id: Ic5b6cc655fd47b95493e6aadb7b23481d0187d5c
Reviewed-on: http://gerrit.openafs.org/5624
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: remove unused code block
Jeffrey Altman [Mon, 17 Oct 2011 13:19:33 +0000]
Windows: remove unused code block

A small bit of dead code in the redirector.

Change-Id: Ieeea58e7681e8b0bd043794f1aa51c112c212497
Reviewed-on: http://gerrit.openafs.org/5623
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: re-indent afsredir log message params
Jeffrey Altman [Mon, 17 Oct 2011 13:10:17 +0000]
Windows: re-indent afsredir log message params

convert tabs to spaces for consistent indentation

make use of braces be consistent

Change-Id: I76e2c6427c02cb50f1eb4e6ab57e484a4bfa3c65
Reviewed-on: http://gerrit.openafs.org/5621
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Defer deref of a directoryEntry
Rod Widdowson [Sat, 22 Oct 2011 14:00:03 +0000]
Windows:  Defer deref of a directoryEntry

During the handling of SL_OPEN_TARGET opens (usually associated
with a rename) a directory entry was deferenced prior to its
contents being used (to set up a seconding inforation field).

This change just holds on to the reference until after that processing.

Change-Id: I26dbd4bfb6595863109e549893f2367f71ad6404
Reviewed-on: http://gerrit.openafs.org/5651
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows: Set new file index correctly during rename
Rod Widdowson [Fri, 21 Oct 2011 15:57:02 +0000]
Windows: Set new file index correctly during rename

Directory entries are required to have a file index which is used during
directory enumeration.  When inserting into a new directory we have to
update this from the target directory.

This code fixes a bug whereby it was being set from the source FCB, rather
than the target one.  On failure we now also reset the value to its old value.

Change-Id: I726dd2e7ecc84e4a8b28f1f551c89fc6e815d853
Reviewed-on: http://gerrit.openafs.org/5646
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoallow cloning of any volume to any volume with same parent ID
Jacob Thebault-Spieker [Tue, 19 Jul 2011 22:01:26 +0000]
allow cloning of any volume to any volume with same parent ID

remove checks to disallow cloning of ro volumes to rw volumes,
which allows cloning of any volume within the same parent ID
grouping, including allowing destruction of newer version of the
volumes.

Change-Id: I13f600c33934b9af8237f049225d6bacfbbb5690
Reviewed-on: http://gerrit.openafs.org/5049
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoremove check for disallowing clones of backup or ro volumes
Jacob Thebault-Spieker [Tue, 19 Jul 2011 16:02:49 +0000]
remove check for disallowing clones of backup or ro volumes

removes the if-statement ensuring that the volume being cloned is
not a backup volume, nor a read-only volume. This allows clones
from any type of volume to a given volume. Parent volume meta-data
is maintained, only the cloneId value changes.

Change-Id: If8444fcabc9b9106db7db7df881ae07e8d287915
Reviewed-on: http://gerrit.openafs.org/5046
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agobozo: build notifier sample program
Michael Meffie [Thu, 6 Oct 2011 12:33:37 +0000]
bozo: build notifier sample program

Build the smail-notifier sample program.

Change-Id: Ida2fd16e486934c984b2431c95c3a24966b3b2b7
Reviewed-on: http://gerrit.openafs.org/5555
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agobozo: test program update
Michael Meffie [Thu, 6 Oct 2011 11:55:21 +0000]
bozo: test program update

Add -file option to test program to simulate program crashes.

Change-Id: I1ba14152438e2f857fdf5d181023266b3ab64fa3
Reviewed-on: http://gerrit.openafs.org/5533
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoDAFS: Remove VOL_SALVAGE_INVALIDATE_HEADER
Andrew Deason [Mon, 29 Aug 2011 22:41:31 +0000]
DAFS: Remove VOL_SALVAGE_INVALIDATE_HEADER

Currently VRequestSalvage_r takes a flag,
VOL_SALVAGE_INVALIDATE_HEADER, which causes the header for the
specified volume to be freed (via FreeVolumeHeader). This is almost
never safe to do, since there may be other users of the specified
volume that can be accessing the volume header at the same time.

There is also no reason to invalidate the header at the time of the
VRequestSalvage_r call, since the header must be invalidated when we
detach the volume (other utilities may change header information). So,
if there are any problems in the future because we do not invalidate
the header at the time of VRequestSalvage_r, it is the fault of the
detachment/offlining logic.

So, remove VOL_SALVAGE_INVALIDATE_HEADER and all of its users. Take
this opportunity to correctly document the VRequestSalvage_r headers
in the VRequestSalvage_r comment, as it was previously missing the
VOL_SALVAGE_NO_OFFLINE flag.

Change-Id: I65c4cc0e81171aafa7b78ae9e1f503edb2f3f2c3
Reviewed-on: http://gerrit.openafs.org/5319
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agoDAFS: fssync online requires a partition name argument
Michael Meffie [Thu, 13 Oct 2011 16:23:35 +0000]
DAFS: fssync online requires a partition name argument

fssync-debug online silently fails when run without a partition name.
Check for the required partition name on the server side and the client
side.  Report errors back to the client when the server side fails to
pre-attach the volume.

Change-Id: I454b202ac83893b1ee54d586a46d891773090a64
Reviewed-on: http://gerrit.openafs.org/5615
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agovolser: Remove ExtractVolId
Andrew Deason [Tue, 11 Oct 2011 15:51:14 +0000]
volser: Remove ExtractVolId

volser was using its own function to extract a volume ID from a
filename string, and was using atol to do so. The ato* family of
functions can have problems with larger volume IDs, not to mention a
lack of error checking, so don't use it. Since we already have the
function VolumeNumber in the vol package to do the very same thing,
just use that instead.

Change-Id: I40953d3533454503583685eb3adeb0079137c8a1
Reviewed-on: http://gerrit.openafs.org/5594
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agoviced: Check for HOSTDELETED in stillborn check
Andrew Deason [Mon, 3 Oct 2011 18:10:44 +0000]
viced: Check for HOSTDELETED in stillborn check

h_FindClient_r checks the connection rock for a client object twice.
First it sees if we already have a client object, and if we don't, we
effectively create one (or find a suitable one). Then we check again,
to see if someone else set the rock while we were creating a client
structure.

Currently, the first check checks if client->host->hostFlags has
HOSTDELETED set, but the second check does not. So, if the host
associated with the client has been deleted by someone else, currently
we will unnecessarily log a "stillborn client" message, and we will
continue to use the deleted host. If the host continues to be held by
someone, we will run into the same situation repeatedly on future
requests until all of the host references go away.

To fix this, also ignore HOSTDELETED clients when performing the
stillborn race check.

Change-Id: Id72e83cb55b99a686d4179905116592e80eb9ac7
Reviewed-on: http://gerrit.openafs.org/5614
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agovlserver: Log to VLLog by default, not PtLog
Andrew Deason [Thu, 13 Oct 2011 16:45:58 +0000]
vlserver: Log to VLLog by default, not PtLog

Change-Id: I25c13a4461fc5a6ba2809b6937cbc18ed9c6ed53
Reviewed-on: http://gerrit.openafs.org/5616
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agovos offline: Bring volume back online for -busy
Andrew Deason [Fri, 14 Oct 2011 16:32:34 +0000]
vos offline: Bring volume back online for -busy

vos offline is supposed to bring a volume back online from "busy"
status before exiting, as volumes should not be in "busy" status for
extended periods of time. This was being enforced by required that
-sleep be specified; however, -sleep only results in the volume being
brought back online if a non-zero sleep time was specified. So, make
sure the volume is brought back online if -busy was specified.

Change-Id: I6e605771fcfcabbb09f6dcfd0576c062f64ad93f
Reviewed-on: http://gerrit.openafs.org/5620
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agocache bypass: don't define iovecp for UKERNEL
Marc Dionne [Sat, 22 Oct 2011 13:32:33 +0000]
cache bypass: don't define iovecp for UKERNEL

iovecp is defined but not used for UKERNEL.  Define it conditionally
to avoid gcc warnings and --enable-checking failure.

Change-Id: I109ada36d0ab75e8a477498900dfaf85589e20e2
Reviewed-on: http://gerrit.openafs.org/5650
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoWindows: Remove unused cleanup flag
Rod Widdowson [Sat, 22 Oct 2011 13:27:41 +0000]
Windows: Remove unused cleanup flag

In AFSOpenTargetDirectory the flag bRemoveShare was initialized
FALSE and never set TRUE.  In teardown after failure some code
did listen to the flag, but the operation (IoRemoveShareAccess)
was not protected by the FCB mainlock which it should have been.

Rather than get the locking correct, just remove the flag entirely.

Change-Id: I097f15bb0903f3c0889b62427d2169cf84aa04a0
Reviewed-on: http://gerrit.openafs.org/5649
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agobypasscache: allow arbitrary sized iovecs
Derrick Brashear [Sun, 23 Oct 2011 23:15:16 +0000]
bypasscache: allow arbitrary sized iovecs

instead of binding outselves to PAGE_SIZE, just fill any size

Change-Id: I01385df372ac2bda6d8ea24381bf38102c12615f
Reviewed-on: http://gerrit.openafs.org/5658
Tested-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agobypasscache: do errors correctly
Derrick Brashear [Thu, 6 Oct 2011 08:04:36 +0000]
bypasscache: do errors correctly

do set errors when we bomb out early
do not unlock and return early when we happen to do a correct zero
length read
do set errors the kernel can deal with if we're feeding a page routine

Change-Id: I1dca1f9e3b0f3d24da3e4f55b473775a737370b4
Reviewed-on: http://gerrit.openafs.org/5554
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: get an ip address even when dns and hosts suck
Derrick Brashear [Tue, 11 Oct 2011 19:37:57 +0000]
ukernel: get an ip address even when dns and hosts suck

gethostname plus gethostbyname being useless make things fun

make things less fun

Change-Id: I5595698b0b7f2448b8c86448bedd8b16ab088510
Reviewed-on: http://gerrit.openafs.org/5600
Tested-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agodir: add missing return in DRead
Marc Dionne [Mon, 24 Oct 2011 02:45:21 +0000]
dir: add missing return in DRead

A missing return in the kernel version of DRead causes the code to
think that no entry exists for a dir and proceed to allocate a new
one, if the entry is the third one in the hash chain.

If the existing entry is dirty, its contents are never written back,
and the pending changes to the directory are not seen by the client.

Change-Id: I8b0f86b04d6ac8f75720933308a0983810974ff4
Reviewed-on: http://gerrit.openafs.org/5666
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agotests: volser doesn't need rxgk yet
Simon Wilkinson [Sun, 23 Oct 2011 16:41:27 +0000]
tests: volser doesn't need rxgk yet

An rxgk dependency has crept in to the volser tests. Remove it for
now.

Change-Id: I0439ab5cf6deb335ebe76fb5a6736c9905312eec
Reviewed-on: http://gerrit.openafs.org/5665
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agotests: Add opr library for ktime test
Simon Wilkinson [Sun, 23 Oct 2011 16:40:20 +0000]
tests: Add opr library for ktime test

The ktime tests need libopr for the lcstring routines. Add the
dependency.

Change-Id: I6902537236fbf35a654a3a4d70fa8f2df89e2a92
Reviewed-on: http://gerrit.openafs.org/5664
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agorpm: Add new libraries
Simon Wilkinson [Sun, 23 Oct 2011 23:09:32 +0000]
rpm: Add new libraries

Add the new libopr.a and libubik_pthread.a libraries

Change-Id: Ia91c67d72f1ac98060af60b6535db008911d3b1a
Reviewed-on: http://gerrit.openafs.org/5663
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agorpm: Turn on debugging
Simon Wilkinson [Sun, 23 Oct 2011 23:07:33 +0000]
rpm: Turn on debugging

Now that we build with a blank CFLAGS line, we need to make sure and
actually turn on debugging in the build system, so that our debuginfo
files are vaguely useful

Change-Id: Ic2ebd0ebe866d511c4b0aef15845a85d95bacb5b
Reviewed-on: http://gerrit.openafs.org/5662
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoroken: Add *progname() to shared library exports
Simon Wilkinson [Sun, 23 Oct 2011 21:49:08 +0000]
roken: Add *progname() to shared library exports

Add rk_getprogname and rk_setprogname to the set of shared
library exports in libroken

Change-Id: I1fd26c59b26a256b2070bfdc14dfaa5f82fb8002
Reviewed-on: http://gerrit.openafs.org/5661
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agorx: Define afs_kmutex_t for LWP too
Simon Wilkinson [Sun, 23 Oct 2011 20:23:34 +0000]
rx: Define afs_kmutex_t for LWP too

afs_kmutex_t is used for lock definitions in the kernel, and in
pthreaded builds. LWP doesn't have any equivalent, and all structure
members using this type have to be protected with RX_ENABLE_LOCKS, which
starts to become untidy.

Just make afs_kmutex_t an int for LWP, so that we can simplify our
headers, at the expense of some additional storage on LWP builds (which
are going away at some point, anyway)

Change-Id: Iaa5683bdfd932c78a9a5d35b8a4530cecf5bdb37
Reviewed-on: http://gerrit.openafs.org/5660
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agodir: Don't leak a buffer on a failed Enumerate
Simon Wilkinson [Sun, 23 Oct 2011 15:38:13 +0000]
dir: Don't leak a buffer on a failed Enumerate

If, for some reasons, Enumerate encounters a hash object with a NULL
buffer pointer, that's no reason to leak the hash object. Make sure
that we DRelease it before failing

Change-Id: I9ba6c69ca524937a6068d33a0ec0a90eef5b88ea
Reviewed-on: http://gerrit.openafs.org/5659
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agorx: ackall handling
Simon Wilkinson [Wed, 12 Oct 2011 13:50:18 +0000]
rx: ackall handling

If we ACKALL a stream, then we're sending a hard ACK for all of the
packets in the stream. We shouldn't send that hard ACK, and then a
load of soft ACKs for packets that don't actually exist.

Change-Id: Ibe0c76b4e92c5baa8801cfd12f961fabce223039
Reviewed-on: http://gerrit.openafs.org/5604
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: BuildBot <buildbot@rampaginggeek.com>

9 years agoLINUX: Fix afs_linux_pag_to_groups signature
Andrew Deason [Fri, 21 Oct 2011 20:05:06 +0000]
LINUX: Fix afs_linux_pag_to_groups signature

"new" is a struct group_info**, not a struct group_info*

Change-Id: I606afce74be6820b99b151b6077cf23d06e8e2d6
Reviewed-on: http://gerrit.openafs.org/5648
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoLINUX: Revert group changes on keyring failure
Andrew Deason [Fri, 12 Aug 2011 19:50:26 +0000]
LINUX: Revert group changes on keyring failure

On Linux kernels that support keyrings, when we setpag we try to add
the PAG to the session keyring and to the supplemental group list.
Currently, if we fail to add the PAG to the keyring (which may happen
due to key quotas, or possibly other reasons), we return failure but
the group list is still modified with the new PAG in it.

Therefore, if the keyring-based approach fails, the new PAG may still
be in use, but there are no keyring keys associated with that PAG, so
the PAG may never get destroyed. This can cause a large number of PAGs
to accumulate over time, causing performance problems.

So, change this so that, in the event that keyring installation fails,
we revert the group list back to what it was before we touched it.
Also mark all unixusers with the new PAG as expired, in case one got
created during processing. Thus, the new PAG never gets used.

Change-Id: I61993edf705209e1a5d8dee638a9003690017c35
Reviewed-on: http://gerrit.openafs.org/5238
Tested-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoviced: Do not swallow errors on StoreData recovery
Andrew Deason [Thu, 20 Oct 2011 21:57:14 +0000]
viced: Do not swallow errors on StoreData recovery

When we encounter any error in the StoreData fetch/store loop, we
reset the disk usage to ensure it remains correct, even in the face of
unexpected errors. However, when we do so, we use the errorCode from
VAdjustDiskUsage as our return value; if it is 0, we return success,
ignoring the error that got us in this code path in the first place.

Instead, keep track of a temporary errorCode for the disk usage
adjustment, and do not override our return value if there was no error
in the disk usage numbers.

Change-Id: Ifbe2377f2750b508d48e535215be3583d1d78e65
Reviewed-on: http://gerrit.openafs.org/5645
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoopr: Move queue header out of util
Simon Wilkinson [Sat, 22 Oct 2011 08:43:41 +0000]
opr: Move queue header out of util

Move the header which is installed as opr/queues.h out of util/ and
into the new, top level, opr/ directory. Similarly move the tests out
of the util/ test suite, and into the opr/ tests

Change-Id: I81af487f09e1f0f4b25654a1f64c5ac75fd5a95b
Reviewed-on: http://gerrit.openafs.org/5656
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Simon Wilkinson <sxw@inf.ed.ac.uk>

9 years agodir: Remove double release in FindBlobs
Simon Wilkinson [Tue, 11 Oct 2011 00:01:26 +0000]
dir: Remove double release in FindBlobs

When DRead() fails, we DRelease the entrybuf, then break. However,
this break takes us to the end of the function, where we promptly
DRelease again, causing a double free

Change-Id: I5a0293990e01376525d8adae25477c64e2cc93ad
Reviewed-on: http://gerrit.openafs.org/5655
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agolinux: makesrpm should cope with releases with letters
Simon Wilkinson [Wed, 12 Oct 2011 13:44:37 +0000]
linux: makesrpm should cope with releases with letters

don't just deal with numbered releases.

Change-Id: Iab97e4a7bebebf97d367243501aaf2a0b4b0476c
Reviewed-on: http://gerrit.openafs.org/5602
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoafs: don't try GetDownD if nothing to get
Derrick Brashear [Wed, 12 Oct 2011 18:34:55 +0000]
afs: don't try GetDownD if nothing to get

if we "need" negative slots or caches, don't run. also,
don't stay "too full"

Change-Id: I29ef8d5fd9844a878d0c804d06aed5862a2d0df4
Reviewed-on: http://gerrit.openafs.org/5612
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: don't define rdwr env on linux
Simon Wilkinson [Wed, 12 Oct 2011 19:26:23 +0000]
ukernel: don't define rdwr env on linux

We're not vm rdwr env in ukernel. Don't claim we are, otherwise we
will never push chunks out to the fileserver when our cache is full

Change-Id: I1c52ef502d42a4a2b63e786471fa487b88712e9d
Reviewed-on: http://gerrit.openafs.org/5613
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: set pthread stacksize to an integer multiple of 8k
Derrick Brashear [Wed, 12 Oct 2011 17:09:42 +0000]
ukernel: set pthread stacksize to an integer multiple of 8k

because some pthread stacks require this to be in units of page size,
guess that page sizes will be 4k or 8k, and use the larger as a divisor

Change-Id: I9f89f7e32e2a70e12f1ed7a7fb86cab171b61e07
Reviewed-on: http://gerrit.openafs.org/5606
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: add morepackets check in listener
Simon Wilkinson [Wed, 12 Oct 2011 17:04:28 +0000]
ukernel: add morepackets check in listener

Make the listener loop actually check for more packets needed,
like kernel, pthreads and lwp. Only checking for new packets every
20 seconds isn't sufficient on today's networks!

Change-Id: I80319cfd9d540c9a56b0dd8d948a380665426409
Reviewed-on: http://gerrit.openafs.org/5605
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoafs: on uuid init fail, don't leave garbage behind
Derrick Brashear [Tue, 11 Oct 2011 19:21:12 +0000]
afs: on uuid init fail, don't leave garbage behind

make sure we zero the uuid if we failed

Change-Id: I9d21e003b68eff0ce9bea4634c601c4663a0c078
Reviewed-on: http://gerrit.openafs.org/5599
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: set close-on-exec on our socket
Derrick Brashear [Tue, 11 Oct 2011 18:32:04 +0000]
ukernel: set close-on-exec on our socket

if the process we are in execs a child, it doesn't get our socket

Change-Id: Iaf1831866e0dcda0449e75d89c576557c61090dd
Reviewed-on: http://gerrit.openafs.org/5598
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agorx: Don't clear the receive queue when out of packets
Simon Wilkinson [Wed, 12 Oct 2011 13:47:14 +0000]
rx: Don't clear the receive queue when out of packets

We can end up discarding a receive queue that's been soft acked,
effectively taking back soft acks we sent. Whilst the RX
documentation says that a client can drop soft acked packets at
will, our RX implementation assumes that if the final packet in
a call has been soft acked, we won't clear the queue. If a client
clears the queue in this situation, the call will hang.

What *should* happen is that we should take necessary locks,
confirm that we have not soft-acked all of the packets in a flow,
and then discard, or, if we're just going to discard, error the
call.

Change-Id: Ic8e358b8648c1a6f0154009093468531a9e3cf74
Reviewed-on: http://gerrit.openafs.org/5603
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoauth: Get correct viceid in legacy GetToken
Andrew Deason [Thu, 14 Apr 2011 20:36:50 +0000]
auth: Get correct viceid in legacy GetToken

When ktc_GetTokenEx needs to get tokens via the legacy ktc_GetToken
interface, it was not extracting the viceid. Make it set the viceid so
the caller gets the correct id.

Normally this would require parsing the given client name. To reduce
the amount of times we store and extract the viced from the "AFS ID
%d" string, create a helper GetToken function that can store the
viceid directly, without storing it in a string.

Change-Id: Ib03a419aee6eaed3b253c4d5d575d4dd8d3b8ddc
Reviewed-on: http://gerrit.openafs.org/4482
Tested-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoauth: Force correct evenness on rxkad tokens
Andrew Deason [Thu, 14 Apr 2011 20:05:37 +0000]
auth: Force correct evenness on rxkad tokens

Rxkad tokens historically have forced odd lifetimes when the given
viceid is actually an AFS ID, and even lifetimes when it is not. Force
this when the new token-handling functions are used (so the viceid is
correctly interpreted by users of the old token format), by creating
rxkad tokens with token_importRxkadViceId.

Slightly reworked by Simon Wilkinson to provide a generic token
destructor function.

Change-Id: I9f6aa518b8ae51a3772b69a0722a28bff6b47128
Reviewed-on: http://gerrit.openafs.org/4481
Tested-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agodocs: Refer to dafs binaries by their real names
Simon Wilkinson [Mon, 10 Oct 2011 22:19:13 +0000]
docs: Refer to dafs binaries by their real names

(Most of) the dafs binaries are called da(something). Update the
example in the dafileserver documentation so that we call the binaries
by the names that they are actually installed with on the system.

Change-Id: I05906758aa2b00ea7bc7f9be10414c8ad1a5a11a
Reviewed-on: http://gerrit.openafs.org/5654
Tested-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoptserver: Don't check for noauth before rebuilding
Simon Wilkinson [Mon, 10 Oct 2011 21:09:40 +0000]
ptserver: Don't check for noauth before rebuilding

The ptserver database building scripts would check to see if the server
was running from a bosserver with the noauth flag set before performing
a database rebuild.

This means that you can't start ptserver normally, and then configure
the database using pts -localauth, which is the preferred method for
configuring new cells.

Remove the check for noauth. This is slightly risky, as it means that a
corrupt database could be completely erased upon restart. However, we
already check that the dbheader (65k) is entirely blank - which will
protect us against any single page corruption errors.

Change-Id: I45dbb53d7f80b0b307f88941cb1e061967fcac81
Reviewed-on: http://gerrit.openafs.org/5653
Tested-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoFBSD: typo fix
Ben Kaduk [Sun, 23 Oct 2011 15:22:07 +0000]
FBSD: typo fix

Gerrit/5572 added conditionals on __FreeBSD_version >= 900044, which
is (approximately) when a bunch of kernel API renames happened.
(There has since been a dedicated version bump to 900045 a month
or two post-facto, but 900044 should be fine for now.)
However, 900044 is not 90004.

Change-Id: I566186a7b53fd3b605a0f531e241eaf54a96bba3
Reviewed-on: http://gerrit.openafs.org/5657
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>

9 years agoWindows AFSRDR: Log before decrementing refcount
Rod Widdowson [Wed, 12 Oct 2011 10:04:33 +0000]
Windows AFSRDR: Log before decrementing refcount

The library support package keeps count of the number of times
the library code is active.  When this goes to zero this means
that unload of the library can continue.

Although I cannot see it in the code it seems reasonable to assume
that at that stage the device object might go away so (and if it
doesn't do now it may in the future).  This potentially renders it
unsafe to do anything after InflightLibraryEvent has been signalled.

This patch moves the logging up to above the decrement of the refcount.

Change-Id: I9493d970cec4a025ca9e4c5542e2d838c87f7817
Reviewed-on: http://gerrit.openafs.org/5601
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

9 years agoukernel: don't enforce thread lockers in remove vop
Derrick Brashear [Tue, 11 Oct 2011 18:03:29 +0000]
ukernel: don't enforce thread lockers in remove vop

we enforce locking by pid. when we are ukernel, MyPidXX is stupid.

don't bother.

Change-Id: Iddefd5d63edd6cc35a3c8112cb2cfbf3d47031aa
Reviewed-on: http://gerrit.openafs.org/5597
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: install our sysincludes to root.perf
Derrick Brashear [Mon, 10 Oct 2011 21:56:20 +0000]
ukernel: install our sysincludes to root.perf

otherwise, we can't actually include sysincludes from something
linking libuafs

Change-Id: Ia6565dda8390ef6510aa2170e36ff12ddb19fc9b
Reviewed-on: http://gerrit.openafs.org/5593
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoukernel: add uafs_access
Derrick Brashear [Mon, 10 Oct 2011 20:35:45 +0000]
ukernel: add uafs_access

we don't have an access vop. add one.

Change-Id: I813fb941608b5e49a0f23730414854602630fdee
Reviewed-on: http://gerrit.openafs.org/5589
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agovolinfo: fix bad format string
Derrick Brashear [Mon, 10 Oct 2011 20:57:35 +0000]
volinfo: fix bad format string

we need a format string; currently here we have none, which is no good

Change-Id: Ib097100383dba9d56056b96c5322734abb765f9d
Reviewed-on: http://gerrit.openafs.org/5591
Reviewed-by: Simon Wilkinson <sxw@inf.ed.ac.uk>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agovol_split: avoid using stale open directory vnodes
Hartmut Reuter [Wed, 5 Oct 2011 14:06:05 +0000]
vol_split: avoid using stale open directory vnodes

we could in case of multiple splits end up using a stale open
vnode for a directory; attempt to close and thus force-reopen
any fdhandles backing ihandles.

Change-Id: I50ff39873b523b05287c7a183d84e6f16eb7cd90
Reviewed-on: http://gerrit.openafs.org/5553
Reviewed-by: Hartmut Reuter <reuter@rzg.mpg.de>
Tested-by: Derrick Brashear <shadow@dementix.org>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoFBSD: deal with kernel API rename
Ben Kaduk [Sat, 8 Oct 2011 21:16:26 +0000]
FBSD: deal with kernel API rename

Upstream decided to rename the kernel functions that implement
syscalls to have a sys_prefix (including afs3_syscall!).
We use a couple of them, so we need to conditionalize accordingly.
Unfortunately, __FreeBSD_version was not bumped with the change,
so we use something close to it and hope it's close enough.

Change-Id: I228805a813af70b30f94f659e4da602210a6d6df
Reviewed-on: http://gerrit.openafs.org/5572
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agobozo: bosserver man page updates
Michael Meffie [Thu, 6 Oct 2011 10:03:07 +0000]
bozo: bosserver man page updates

Add the new -pidfiles option to the man page. Add a few other missing
options as well.

Change-Id: I2d72ec9948edb4dce4c6b6d974915d4ac7560b0b
Reviewed-on: http://gerrit.openafs.org/5587
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoWindows: correct use of krb5_init_context in aklog
Jeffrey Altman [Fri, 7 Oct 2011 15:23:57 +0000]
Windows: correct use of krb5_init_context in aklog

A zero return code is success.  Apply ! to obtain proper
conditional test.

Change-Id: Idd60fd2476a36ec36f7b5e250e9f8b8531c80611
Reviewed-on: http://gerrit.openafs.org/5570
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

9 years agoCorrect Heimdal conversion of libadmin/adminutil
Jeffrey Altman [Sat, 8 Oct 2011 08:01:07 +0000]
Correct Heimdal conversion of libadmin/adminutil

Patchset 4251e386aa25bb3fc02fa255e92327fffc8b954d converts to
using Heimdal.  The conversion undid the introduction of the
abstraction function fetch_krb5_error_message() which is
implemented in src/util.  Restore the use of fetch_krb5_error_message()
and modify src/util/krb5_nt.c to use the Kerberos Compat SDK
interface.

Change-Id: I67fe7a309727f67a1da3705e1e485e64747c325c
Reviewed-on: http://gerrit.openafs.org/5571
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>

9 years agobypasscache: free in order
Derrick Brashear [Fri, 7 Oct 2011 12:11:34 +0000]
bypasscache: free in order

we need to free this, but while the pointer is still good

Change-Id: I5afa3505b764c5bf72811f60f63e29c80a29c5af
Reviewed-on: http://gerrit.openafs.org/5562
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agodoc: fileserver synopsis typo
Michael Meffie [Fri, 7 Oct 2011 09:50:37 +0000]
doc: fileserver synopsis typo

Fix a fileserver synopsis typo noticed during Derrick's EACK2011
debugging session.

Change-Id: I73ed6bd93f644689acbdeeb4293a22b3b6752ba6
Reviewed-on: http://gerrit.openafs.org/5558
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agoRedHat: Add xstat_*_test commands to RPMs
Andrew Deason [Fri, 7 Oct 2011 08:04:03 +0000]
RedHat: Add xstat_*_test commands to RPMs

Install the xstat_*_test commands in the 'openafs' RPM.

Change-Id: Idec42e133dd3f27c72f06c5411dd12acc163ef93
Reviewed-on: http://gerrit.openafs.org/5557
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>

9 years agobypasscache for ukernel
Derrick Brashear [Wed, 21 Sep 2011 19:02:40 +0000]
bypasscache for ukernel

simple (read) bypasscache for ukernel.
does not bother trying any buffercache stuff.

Change-Id: Ie75572da7efdb871a4ce807f02fbbb5bd7744c66
Reviewed-on: http://gerrit.openafs.org/5484
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

9 years agoWindows: Explorer shell GetUnixModeBits()
Jeffrey Altman [Tue, 4 Oct 2011 23:01:50 +0000]
Windows: Explorer shell GetUnixModeBits()

actually append the unix mode string elements to the
respective strings.

Change-Id: I32d69f824f01e8cfacbf811a4bb14e01e3b6b9a3
Reviewed-on: http://gerrit.openafs.org/5552
Reviewed-by: Jeffrey Altman <jaltman@openafs.org>
Tested-by: Jeffrey Altman <jaltman@openafs.org>