Contains DKMS robustness fixes, improvements to the defaults for the
module build, and cleanup of the openafs-client init script. Updates
the build system for the new demand-attach binary naming and for the
changes to supported configure options. Fixes some issues with
afs-newcell. Forces disabling of the Linux syscall probing in kernel
module builds, since no supported Debian kernel allows this and it
causes problems. Update debhelper to V8, which allows simplification
of debian/rules and debian/module/rules.
Simplify the module naming code in openafs-client init
* In the openafs-client init script, stop looking for openafs.o; Debian
hasn't supported 2.4 kernels for quite some time. Remove all the $MP
handling since we force a module name of openafs.ko using
--with-linux-kernel-packaging.
Use /lib/modules/`uname -r`/source as the default kernel source
* In openafs-modules-source, use /lib/modules/`uname -r`/source as the
default path to the kernel headers rather than /usr/src/linux.
Current kernel header packages will create an appropriate symlink
there. This default is only used when neither module-assistant nor
kernel-package are used.
Update bos create man page for new naming of demand-attach binaries
The demand-attach fileserver binaries now have a "da" prefix. Adjust
the documentation in the man page for bos create accordingly, and add
the new binaries to SEE ALSO.
Update bos create man page for new naming of demand-attach binaries
The demand-attach fileserver binaries now have a "da" prefix. Adjust
the documentation in the man page for bos create accordingly, and add
the new binaries to SEE ALSO.
The two commands are documented identically for right now, so just link
the dafssync-debug man page to the fssync-debug man page. Remove the
incorrect statement in the man page that fssync-debug only works with
demand-attach.
Simon Wilkinson [Thu, 23 Sep 2010 07:58:21 +0000 (08:58 +0100)]
libuafs: Don't #define user
libuafs used to #define user to usr_user, so that any references to
'struct user' would become 'struct usr_user'. However, none of the
kernel code uses struct user, and this #define conflicts with the
definitions in sys/user.h on Linux.
So, just remove it.
Thanks to Russ Allbery for the original problem report.
The two commands are documented identically for right now, so just link
the dafssync-debug man page to the fssync-debug man page. Remove the
incorrect statement in the man page that fssync-debug only works with
demand-attach.
Simon Wilkinson [Thu, 23 Sep 2010 07:58:21 +0000 (08:58 +0100)]
libuafs: Don't #define user
libuafs used to #define user to usr_user, so that any references to
'struct user' would become 'struct usr_user'. However, none of the
kernel code uses struct user, and this #define conflicts with the
definitions in sys/user.h on Linux.
So, just remove it.
Thanks to Russ Allbery for the original problem report.
Michael Meffie [Thu, 23 Sep 2010 14:15:57 +0000 (10:15 -0400)]
scout: display fetch and store counts as unsigned
Fetches and stores are already defined as unsigned, so format
them as unsigned values when displaying in scout. This fixes
the bug where scout shows those counts as negative values on
busy servers which have been running for a while.
Simon Wilkinson [Thu, 23 Sep 2010 16:41:47 +0000 (17:41 +0100)]
rx: Big windows make us sad
The commit which took our Window size to 128 caused rxperf to run
40 times slower than before. All of the recent rx improvements have
reduced this to being around 2x slower than before, but we're still
not ready for large window sizes.
As 1.6 is nearing release, reset back to the old, fast, window size
of 32. We can revist this as further performance improvements and
restructuring happen on master.
Matt Smith [Thu, 16 Sep 2010 20:44:04 +0000 (15:44 -0500)]
Updates to the Cache Manager to include NetBSD5 support
LKM currently builds and will mount when forced with the entry point
manually defined. Contents of /afs can be discovered, but when listing
the directory, the system call will not return.
Simon Wilkinson [Wed, 22 Sep 2010 08:03:27 +0000 (09:03 +0100)]
Add an LWP version of the hcrypto library
hcrypto uses a single pthread mutex, which is only required when
we're running in a pthreaded world (in a cooperative threading world
such as LWP, there's no way that two processes can both access the
Fortuna PRNG at the same time)
So, build an LWP version of hcrypto which just disables the mutexes.
Simon Wilkinson [Thu, 15 Apr 2010 16:35:37 +0000 (17:35 +0100)]
Move des/stats.h to rxkad directory
The structures defined in des/stats.h are rxkad specific, and so
should be handled by an rxkad header file, rather than by something
in the eventually-to-be-removed DES directory.
The structure shouldn't be initialised in libutil. Move initialisation
to rxkad_common, where it is already initialised in the pthread case.
The two commands are documented identically for right now, so just link
the dafssync-debug man page to the fssync-debug man page. Remove the
incorrect statement in the man page that fssync-debug only works with
demand-attach.
Update bos create man page for new naming of demand-attach binaries
The demand-attach fileserver binaries now have a "da" prefix. Adjust
the documentation in the man page for bos create accordingly, and add
the new binaries to SEE ALSO.
Simon Wilkinson [Thu, 23 Sep 2010 07:58:21 +0000 (08:58 +0100)]
libuafs: Don't #define user
libuafs used to #define user to usr_user, so that any references to
'struct user' would become 'struct usr_user'. However, none of the
kernel code uses struct user, and this #define conflicts with the
definitions in sys/user.h on Linux.
So, just remove it.
Thanks to Russ Allbery for the original problem report.
Ensure we set up debian/control before doing module builds
In the openafs-modules-source package, we have to be sure to generate
a debian/control file listing our module package before we run any
dh commands, since otherwise dh gives up and quits.
Remove bison and flex dependencies for module builds
* Remove bison and flex dependencies for the module build packages. We
now ship the results of rxgen, so don't need all the tools required to
build the full source tree.
* Add ${perl:Depends} to openafs-modules-dkms and perl to the
dependencies of openafs-modules-source, since Perl is used as part of
the module build system now.
Simon Wilkinson [Wed, 1 Sep 2010 13:38:58 +0000 (14:38 +0100)]
rxgen: Handle complex structures
Servers built using rxgen will break if they take complex
structures as RPC arguments. A complex structure, in this case, is
one which contains an array.
For example an RPC which takes as an argument:
struct MyData {
opaque somebytes<>;
}
... will cause memory corruption on the server whenever it is called.
This is becase the server stubs emitted by rxgen do not zero out the
contents of the MyData structure, leaving it with whatever garbage may
be on the stack. When XDR comes to populate the somebytes opaque
array, it sees that MyData.somebytes.somebytes_val is non-zero, and
assumes that this is a pre-allocated block into which it can record
the data from the wire. However, it's really just stack garbage, and
so we overwrite memory.
As a fix, this patch creates a new list of 'complex' structures, which
are identified as structures which contain arrays. When a server stub
is created for a function that takes a complex structure, the structure
is set to zero before use, and marked to be freed afterwards.
I suspect that there may be a wider class of complex structures than are
caught by this routine, but this is a start...
The previous value, 350ms, is historical. Now that networks are
so much faster, an artificially high timeout value when backed off
results in an extremely long delay before communication can resume.
Rx: Do not hold call lock across memcpy in rx_ReadProc/rx_WriteProc
1.4.x does not hold the call lock across memcpy operations in
rx_ReadProc, rx_ReadProc32, rx_WriteProc, rx_WriteProc32. The
claim is that the call curpos, curlen, and nLeft fields which
refer to the current packet being processed will not be touched
by any other thread. Therefore it is safe to drop the call lock
to permit another thread to add packets to the call while the memcpy
is performed in parallel.
This patchset continues to hold the call lock longer than the
original implementation but does drop it for the length of time
it takes to copy data from the packet buffer to the application
buffer.
If a volume lookup returns VL_NOENT or VL_BADNAME, cache the negative
response for five minutes. This prevents volume lookup storms caused
by the same volume lookup being performed repeated during a short
time period. This can happen if mount points to volumes that do not
exist are present in a directory that is being evaluated by Windows
Explorer or Common Control File Dialogs.
This functionality is implemented by storing the most recent update
time for the volume group as part of the cm_volume_t. A non-existing
volume group is identified with a new CM_VOLUMEFLAG_NOEXIST flag.
The presence of the lastUpdateTime value also permits volume location
information to expire at lastUpdateTime + lifetime instead of expiring
all volume information simultaneously each lifetime period.
AFS users not otherwise familiar with Kerberos may not realize that
one sets KRB5CCNAME to use an alternative ticket cache. Mention the
variable in the aklog man page, although defer to the Kerberos
documentation for most details.
Change-Id: I13133285f153f86e0aa1e95b24715e01d576075d
Reviewed-on: http://gerrit.openafs.org/2761 Reviewed-by: Andrew Deason <adeason@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com> Reviewed-by: Russ Allbery <rra@stanford.edu>
Windows: Export additional RX debugging variables from afsrpc.dll
Export
rxi_nRecvFrags @2008 DATA
rxi_nSendFrags @2009 DATA
rx_initReceiveWindow @2010 DATA
rx_initSendWindow @2011 DATA
rx_intentionallyDroppedPacketsPer100 @2012 DATA
rx_intentionallyDroppedOnReadPer100 @2013 DATA
so they can be referenced from pthreaded builds of src/rx/test tools.
Exported variables must be present in both FREE and CHECKED builds.
Rx: PrintTheseStats should not be dependent on RXDEBUG
When RXDEBUG is not defined, PrintTheseStats generates an error
even though the statistics are in fact available. The global
variable rx_packetTypes was not being defined without RXDEBUG.
Make rx_packetTypes defined always and permit statistics to
always be printed.
The global dataPacketsReSent statistic should be the sum of all
peer->reSends and dataPacketsSent should not include the count of
resent packets. Prior to this patchset, dataPacketsSent included
the resent packets and dataPacketsReSent was computed as the number
of requests for Ack instead of the number of packets resent.
If a packet is missing, the peer timeout is backed off to provide
a new starting point for timeout computation. The backoff state
must be stored in the peer object to ensure that multiple failures
do not result in more than one backoff before a successfully received
packet is available for recomputation.
Rx: only compute peer bytes sent and received if rx_stats_active
Computing the bytes sent and received is an expensive operation.
If rx statistics collection has been disabled we should not collect
the peer data. The most expensive operation is the rx_FindPeer()
call that is performed during rxi_ReadPacket(). rxi_ReadPacket()
is processed by the rx listener thread which must be as fast as
possible.
rxi_ReceiveAckPacket can acquire and drop the conn_data_lock several
times and acquires and drops the peer_lock unnecessarily. This patchset
adds a variable to track whether the conn_data_lock is held in order
to avoid the need to drop it and reacquire it based upon conditional
operations. It also relocates the peer->maxPacketSize computations
in order to consolidate the work performed under the peer_lock.
rxperf made assumptions that it was built against LWP, used buffer
sizes for read/write that were too small, made use of non-portable
types, and set signal handlers that are unsupported.
Simon Wilkinson [Sat, 11 Sep 2010 22:17:24 +0000 (23:17 +0100)]
viced: Don't fall back to tokens
I can't think of any circumstances in which it is appropriate for
the fileserver to fallback to using tokens from the kernel, rather than
those generated from keys on disk. So, remove that fallback.
Simon Wilkinson [Sat, 11 Sep 2010 07:21:10 +0000 (08:21 +0100)]
ubik: Remove unused error codes
The error codes in ubik.p.h have been replaced by ones defined by
com_err. Remove the redundant, #if 0, definitions of these, as they're
just confusing when grepping the code.
Change-Id: I18b827a6e0532a602f904bb6c1e34c3f63c2e08d
Reviewed-on: http://gerrit.openafs.org/2757 Reviewed-by: Andrew Deason <adeason@sinenomine.net> Reviewed-by: Rod Widdowson <rdw@steadingsoftware.com> Reviewed-by: Derrick Brashear <shadow@dementia.org> Tested-by: Derrick Brashear <shadow@dementia.org>
Simon Wilkinson [Sat, 11 Sep 2010 19:49:55 +0000 (20:49 +0100)]
ptserver: Merge WhoIsThis and WhoIsThisWithName
The functions WhoIsThis and WhoIsThisWithName are pretty much identical,
with WhoIsThisWithName just having additional logic for foreign user
registration. Merge them together, so that there's only one copy of
the code for determining who a user is.
Simon Wilkinson [Mon, 13 Sep 2010 07:54:55 +0000 (08:54 +0100)]
userok: Don't double check for expiry
rxkad_GetServerInfo returns an error if you attempt to get server
information using a connection whose tokens have already expired.
The additional check for expiry in the userOK routines is therefore
not required - they'll never be reached, because GetServerInfo will
already have errored out.
Simon Wilkinson [Sat, 11 Sep 2010 15:39:23 +0000 (16:39 +0100)]
auth: Restructure userok
Restructure the userok logic in auth to split it into smaller functions,
and abstract out common code. This will make it easier to add support
for other security layers in future commits.
Simon Wilkinson [Sat, 11 Sep 2010 11:51:52 +0000 (12:51 +0100)]
rx: Call rxgen_consts.h by its proper name
rxgen_consts.h is actually in the afs/ directory. Include it from
their, rather than requiring that every user have $(TOP_INCDIR)/afs
in their search path.
Simon Wilkinson [Sat, 11 Sep 2010 09:11:57 +0000 (10:11 +0100)]
auth: Add the ktc_ListTokensEx function
Add a ktc_ListTokensEx function which uses the new GetToken pioctl
to implement the same functionality as the old ktc_ListTokens call.
As with ktc_ListTokens this is hugely inefficient, as it gets a
compelete token structure from the kernel, then throws it away to
return just the cell which the token is for.
Windows: Add new token interface with stub for ktc_GetTokenEx
The Windows build was broken by the addition of dependencies on
token_FreeSet, ktc_GetTokenEx, etc. Permit the build to continue.
The ktc_GetTokenEx implementation will depend on the creation of
a new Windows pioctl for the call.
Simon Wilkinson [Thu, 16 Sep 2010 06:09:20 +0000 (07:09 +0100)]
Rename kauth/token.c as kauth/katoken.c
The kauth 'token.c' collides with the same file in auth when doing
Windows builds. As kauth is the legacy package, rename its source
file, and update the build system to take account of this.
Simon Wilkinson [Sat, 4 Sep 2010 14:29:03 +0000 (15:29 +0100)]
RX: Make rxi_Alloc return (void *)
rxi_Alloc returns a pointer to an anonymous data block. Make its
return value (void *) rather than (char *), so that it can be
called in the same way as malloc(), and not require casting.
Andrew Deason [Tue, 14 Sep 2010 16:15:22 +0000 (12:15 -0400)]
volser: Delete timed-out temporary volumes
When a transaction times out on a volume, delete the volume if it is a
temporary volume (destroyMe is set). This prevents half-created
volumes from accumulating, which can take up space and screw up
certain vol ops in some versions.
Andrew Deason [Wed, 15 Sep 2010 16:19:33 +0000 (12:19 -0400)]
libafs: Fix pioctl get/putInt alignment issues
We don't know if the buffer for pioctl data is aligned to anything, so
we can't just dereference the given pointer as an int or anything
else. So, just memcpy the data in for ints and such; conveniently,
afs_pd_getBytes and afs_pd_putBytes can do this for us, so just use
that.
projects / packages / o / openafs.git / log