nfs-utils.git
13 years agoMerge branch 'master' of git://linux-nfs.org/nfs-utils
Greg Banks [Mon, 3 Jul 2006 04:21:48 +0000 (14:21 +1000)]
Merge branch 'master' of git://linux-nfs.org/nfs-utils

13 years agonfs-utils 1.0.9-pre1 nfs-utils-1-0-9-pre1
Neil Brown [Mon, 3 Jul 2006 00:02:03 +0000 (10:02 +1000)]
nfs-utils 1.0.9-pre1

Update change log and change version number

13 years agoAllow rpc.nfsd to suppress tcp or udp, and listen on a specific address.
Steve Dickson [Sun, 2 Jul 2006 23:52:00 +0000 (09:52 +1000)]
Allow rpc.nfsd to suppress tcp or udp, and listen on a specific address.

 -T       - will suppressing listening for TCP connection.
 -U       - will suppress UDP
 -H host  - will only listen on that local address
 -p port  - will listen on that port.

This requires kernel patches which will hopefully be in 2.6.19 and possibly some
earlier test and vendor kernels.

13 years agoComment out unused variable.
Greg Banks [Tue, 27 Jun 2006 11:30:18 +0000 (21:30 +1000)]
Comment out unused variable.

13 years agoComment out the decades-old SCCS id strings from the original Sun
Greg Banks [Tue, 27 Jun 2006 11:22:59 +0000 (21:22 +1000)]
Comment out the decades-old SCCS id strings from the original Sun
distribution.  They cause compile warnings, there is no longer any
reason to try to build them into the binaries, and gcc seems to be
eliding some of them anyway.

13 years agoDetect if glibc provides socklen_t and use that instead
Greg Banks [Tue, 27 Jun 2006 10:55:07 +0000 (20:55 +1000)]
Detect if glibc provides socklen_t and use that instead
of int in those cases which generate compile warnings,
e.g. the last argument of recvfrom().

13 years agoReplace the deprecated sigblock() with more modern
Greg Banks [Tue, 27 Jun 2006 10:28:02 +0000 (20:28 +1000)]
Replace the deprecated sigblock() with more modern
signal functions to avoid compile warnings.

13 years agoAdd support for suppressing different NFS versions.
Steve Dickson [Mon, 26 Jun 2006 05:23:19 +0000 (15:23 +1000)]
Add support for suppressing different NFS versions.

e.g.  -N 2
means that NFSv2 won't be supported, just v3 and v4 (if the kernel
supports them).

13 years agoFurther coverity related cleanups.
Neil Brown [Fri, 23 Jun 2006 07:10:56 +0000 (17:10 +1000)]
Further coverity related cleanups.

Greg Banks suggested some variations, particularly improved
use of xmalloc/xstrdup functions.  Thanks.

13 years agoFix various issues discovered by Coverity
Neil Brown [Fri, 23 Jun 2006 04:38:33 +0000 (14:38 +1000)]
Fix various issues discovered by Coverity

Thanks to Michael Halcrow for finding them.

13 years agoFix comment parsing (again)
Neil Brown [Fri, 23 Jun 2006 04:06:00 +0000 (14:06 +1000)]
Fix comment parsing (again)

Bruce Fields noticed that I broke comment parsing...

as xskip() is always called before xgettok(), that is the
best place to put xskipcomment and still maintain proper
semantics of xskip and xgettok.

13 years agoChange mount configure option to --enable-mount
Amit Gud [Thu, 22 Jun 2006 16:51:04 +0000 (12:51 -0400)]
Change mount configure option to --enable-mount

Change the configure option from --with-mount to --enable-mount.

Signed-off-by: Amit Gud <agud@redhat.com>
Signed-off-by: Steve Dickson <steved@redhat.com>
13 years agoMerge nfsmount.x and mount.x into mount.x
Amit Gud [Thu, 22 Jun 2006 16:49:24 +0000 (12:49 -0400)]
Merge nfsmount.x and mount.x into mount.x

Merge utils/mount/nfsmount.x and support/export/mount.x into support/export/mount.x.

Signed-off-by: Amit Gud <agud@redhat.com>
Signed-off-by: Steve Dickson <steved@redhat.com>
13 years agoTry to make sure that clientid used for NFSv4 is reliable.
Neil Brown [Fri, 23 Jun 2006 03:37:08 +0000 (13:37 +1000)]
Try to make sure that clientid used for NFSv4 is reliable.

We need to give an IP address to identify this client to the
server.
The current code does a gethostbyname of the hostname. One
some systems this returns 127.0.0.1 or similar, which is not useful.

Instead, use getsockname of the sock used to connect to the server
to confirm that the server is working.  This gives the address on the
interface that was chosen to talk to that server, which is the
best address we can find (if there is a NAT in the way, it might
still not work, but in that case there is nothing we can do).

13 years agoFix a number of the easier compile warnings: unused variables,
Greg Banks [Thu, 22 Jun 2006 08:01:10 +0000 (18:01 +1000)]
Fix a number of the easier compile warnings: unused variables,
unused labels, constness, signedness.

13 years agoDetect presence of nfs4_set_debug() in libnfsidmap and
Greg Banks [Thu, 22 Jun 2006 07:31:24 +0000 (17:31 +1000)]
Detect presence of nfs4_set_debug() in libnfsidmap and
don't bother calling it if it's missing.

13 years agoRemove useless m4 quoting around args to PKG_CHECK_MODULES().
Greg Banks [Thu, 22 Jun 2006 07:19:17 +0000 (17:19 +1000)]
Remove useless m4 quoting around args to PKG_CHECK_MODULES().

13 years agoDo m4 quoting of AC_MSG_ERROR() and AC_MSG_WARN() as
Greg Banks [Thu, 22 Jun 2006 07:16:19 +0000 (17:16 +1000)]
Do m4 quoting of AC_MSG_ERROR() and AC_MSG_WARN() as
the autoconf people intended.  This avoids errors in
autoconf on SLES10.

13 years agoAdd a .gitignore file to suppress the files generated
Greg Banks [Thu, 22 Jun 2006 06:51:44 +0000 (16:51 +1000)]
Add a .gitignore file to suppress the files generated
during autogen, configure and build.

13 years agomultiple threads for mountd
Greg Banks [Wed, 14 Jun 2006 12:48:10 +0000 (22:48 +1000)]
multiple threads for mountd

How about the attached patch against nfs-utils tot?  It
adds a -t option to set the number of forked workers.
Default is 1 thread, i.e. the old behaviour.

I've verified that showmount -e, the Ogata mount client,
and a real mount from Linux and IRIX boxes work with and
without the new option.

I've verified that you can manually kill any of the workers
without the portmap registration going away, that killing
all the workers causes the manager process to wake up and
unregister, and killing the manager process causes the
workers to be killed and portmap unregistered.

I've verified that all the workers have file descriptors
for the udp socket and the tcp rendezvous socket, that
connections are balanced across all the workers if service
times are sufficiently long, and that performance is
improved by that parallelism, at least for small numbers
of threads.  For example, with 60 parallel MOUNT calls
and a testing patch to make DNS lookups take 100 milliseconds
time to perform all mounts (averaged over 5 runs) is:

num elapsed
threads time (sec)
 ------ ----------
1 13.125
2   6.859
3  4.836
4    3.841
5  3.303
6  3.100
7  3.078
8  3.018

Greg.
--
Greg Banks, R&D Software Engineer, SGI Australian Software Group.
I don't speak for SGI.

13 years agoRemove some temporary files that shouldn't be in 'git'.
Neil Brown [Fri, 16 Jun 2006 03:16:09 +0000 (13:16 +1000)]
Remove some temporary files that shouldn't be in 'git'.

deleted:  compile
deleted:  config.guess
deleted:  config.sub
deleted:  depcomp
deleted:  install-sh
deleted:  ltmain.sh
deleted:  missing

13 years agoAdd support to auto-generate nfsmount* files for new nfs.mount program
Neil Brown [Fri, 16 Jun 2006 03:09:26 +0000 (13:09 +1000)]
Add support to auto-generate nfsmount* files for new nfs.mount program

13 years agoMove NFS mount code from util-linux to nfs-utils - part 2
Amit Gud [Mon, 12 Jun 2006 23:08:27 +0000 (19:08 -0400)]
Move NFS mount code from util-linux to nfs-utils - part 2

Adds the support functions needed for mount and umount. This
functionality will someday be available in the form of shared mount
library.

Signed-off-by: Amit Gud <agud@redhat.com>
Signed-off-by: Steve Dickson <steved@redhat.com>
13 years agoMove NFS mount code from util-linux to nfs-utils - part 1
Amit Gud [Mon, 12 Jun 2006 23:06:36 +0000 (19:06 -0400)]
Move NFS mount code from util-linux to nfs-utils - part 1

Adds the mount directory and the code to mount and umount the NFS file system.

Signed-off-by: Amit Gud <agud@redhat.com>
Signed-off-by: Steve Dickson <steved@redhat.com>
13 years agoMinor man page tidyup
Neil Brown [Mon, 5 Jun 2006 02:54:42 +0000 (12:54 +1000)]
Minor man page tidyup

13 years agofix nhfsrun signal
Neil Brown [Mon, 5 Jun 2006 02:47:00 +0000 (12:47 +1000)]
fix nhfsrun signal

  nhfsrun is supposed to be able to be signalled with SIGUSR1, but
  the signal trapped is number 30, which is something else
  entirely (SIGPWR). This patch simply changes it to say "USR1",
  which gets it right no matter what the value is.
         "Steinar H. Gunderson" <sesse@debian.org>

13 years agomountd state directory
Neil Brown [Mon, 5 Jun 2006 02:45:01 +0000 (12:45 +1000)]
mountd state directory

  Let the user select (via a new parameter) the path to the NFS
  state directory for mountd, to match the statd functionality.
         "Steinar H. Gunderson" <sesse@debian.org>

13 years agodocument sync option:
Neil Brown [Mon, 5 Jun 2006 02:39:07 +0000 (12:39 +1000)]
document sync option:
  Document the 'sync' option in the exports(5) man page -- ATM
  only the 'async' option is documented, which is not very
  symmetric. :-)  "Steinar H. Gunderson" <sesse@debian.org>

13 years agoOnly treat '#' as starting a comment when at the start of a token
Neil Brown [Mon, 5 Jun 2006 02:36:52 +0000 (12:36 +1000)]
Only treat '#' as starting a comment when at the start of a token

otherwise '#' in filenames cannot be read.

13 years agoescape hashes in exports
Neil Brown [Mon, 5 Jun 2006 02:27:35 +0000 (12:27 +1000)]
escape hashes in exports
  Makes sure any # signs in the printed-out exports file are
  escaped (as with quotes, spaces, etc.), so they won't be treated
  as a comment when they're read back in again.
        "Steinar H. Gunderson" <sesse@debian.org>

13 years agofix exportfs with multiple matches
Neil Brown [Mon, 5 Jun 2006 02:16:20 +0000 (12:16 +1000)]
fix exportfs with multiple matches

  Fixes a problem with exportfs -o
  and multiple entries of the same type for the same patch that matches
  a given client. The entire rationale and problem description can be found
  at http://bugs.debian.org/245449 (fumihiko kakuma <kakmy@mvh.biglobe.ne.jp>)

13 years agoRemove debian/
Neil Brown [Mon, 5 Jun 2006 02:13:31 +0000 (12:13 +1000)]
Remove debian/

.. at request of Debian maintainer "Steinar H. Gunderson" <sesse@debian.org>

13 years ago Remove **/Makefile.in, aclocal.m4, configure, and
Neil Brown [Mon, 17 Apr 2006 07:07:39 +0000 (17:07 +1000)]
Remove **/Makefile.in, aclocal.m4, configure, and
 support/include/config.h.in from source control
These are auto autogenerated by
  aclocal -I aclocal ; autoheader ; automake ; autoconf

13 years agoDefine and use HIAVE_IFADDRS_H
Neil Brown [Mon, 17 Apr 2006 06:54:54 +0000 (16:54 +1000)]
Define and use HIAVE_IFADDRS_H

13 years ago1.0.8 nfs-utils-1-0-8
neilbrown [Wed, 12 Apr 2006 02:55:50 +0000 (02:55 +0000)]
1.0.8

13 years ago2006-04-10 NeilBrown <neilb@suse.de>
neilbrown [Mon, 10 Apr 2006 09:57:17 +0000 (09:57 +0000)]
2006-04-10 NeilBrown <neilb@suse.de>
Various paranoia checks:
gssd_proc.c: pass max_field sizes to sscanf to avoid buffer
  overflow
svcgssd_proc.c: range_check name.length, to ensure name.length+1
  doesn't wrap
idmapd.c(nfsdcb): make sure at least one byte is read before
 zeroing the last byte that was read, otherwise memory corruption
 is possible.

Found by SuSE security audit.

13 years ago2006-04-10 "Kevin Coffman" <kwc@citi.umich.edu>
neilbrown [Mon, 10 Apr 2006 04:00:31 +0000 (04:00 +0000)]
2006-04-10 "Kevin Coffman" <kwc@citi.umich.edu>
Check for sufficient version of librpcsecgss and libgssapi
in configure.in

13 years ago2006-04-10 "Kevin Coffman" <kwc@citi.umich.edu>
neilbrown [Mon, 10 Apr 2006 03:53:14 +0000 (03:53 +0000)]
2006-04-10 "Kevin Coffman" <kwc@citi.umich.edu>
Update aclocal/tcp-wrappers.m4 to define HAVE_LIBWRAP and
HAVE_TCP_WRAPPERS as appropriate.

13 years ago Add checking for innetgr back to configure.in
neilbrown [Mon, 10 Apr 2006 01:39:14 +0000 (01:39 +0000)]
Add checking for innetgr back to configure.in

13 years ago2006-04-10 kwc@citi.umich.edu
neilbrown [Sun, 9 Apr 2006 23:51:59 +0000 (23:51 +0000)]
2006-04-10 kwc@citi.umich.edu
Update calls to gss_export_lucid_sec_context()

Change the calls to gss_export_lucid_sec_context() to match the corrected
interface definition in libgssapi-0.9.

13 years ago2006-04-10 kwc@citi.umich.edu
neilbrown [Sun, 9 Apr 2006 23:50:42 +0000 (23:50 +0000)]
2006-04-10 kwc@citi.umich.edu
Plug memory leaks in svcgssd

Various memory leaks in the svcgssd context processing are eliminated.

13 years ago2006-04-10 kwc@citi.umich.edu
neilbrown [Sun, 9 Apr 2006 23:49:36 +0000 (23:49 +0000)]
2006-04-10 kwc@citi.umich.edu
Fix memory leak of the AUTH structure on context negotiations

Free AUTH structure after completing context negotiation and sending
context information to the kernel.

13 years ago2006-04-10 kwc@citi.umich.edu
neilbrown [Sun, 9 Apr 2006 23:47:08 +0000 (23:47 +0000)]
2006-04-10 kwc@citi.umich.edu
Fix support/include/config.h.in such as would be done be running autoheader.

13 years agoaclocal/autoconf/automake, properly this time.
neilbrown [Tue, 28 Mar 2006 04:22:44 +0000 (04:22 +0000)]
aclocal/autoconf/automake, properly this time.

13 years agoUpdate version, autoconf, automake
neilbrown [Tue, 28 Mar 2006 01:28:14 +0000 (01:28 +0000)]
Update version, autoconf, automake

13 years ago Use PKGCONFIG to locate gssapi and rpcsecgss header files
neilbrown [Tue, 28 Mar 2006 00:53:30 +0000 (00:53 +0000)]
Use PKGCONFIG to locate gssapi and rpcsecgss header files

Instead of having separate copies of the gssapi and rpcsecgss
header files, or depending on the Kerberos gssapi header,
locate the headers now installed with the libgssapi and librpcsecgss
libraries.

Remove local copies of the gssapi and rpcsecgss header files.

This depends on the configure_use_autotools patch.

13 years ago Add debugging to better detect negotiation of enctype not supported by kernel
neilbrown [Tue, 28 Mar 2006 00:51:14 +0000 (00:51 +0000)]
Add debugging to better detect negotiation of enctype not supported by kernel

Print debugging message indicating the type of encryption keys being sent
down to the kernel.  This should make it easier to detect cases where
unsupported encryption types are being negotiated.
(really this time)

13 years ago Don't close and reopen all pipes on every DNOTIFY signal.
neilbrown [Tue, 28 Mar 2006 00:50:44 +0000 (00:50 +0000)]
Don't close and reopen all pipes on every DNOTIFY signal.

From: Vince Busam <vbusam@google.com>
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Don't unnecessarily close and re-open all pipes after every DNOTIFY
signal.  These unnecessary closes were triggering a kernel Oops.
Original patch modified to correct segfault when unmounting last
NFSv4 mount.

13 years ago Add option to specify directory to search for credentials cache files
neilbrown [Tue, 28 Mar 2006 00:50:03 +0000 (00:50 +0000)]
Add option to specify directory to search for credentials cache files

From: Vince Busam <vbusam@google.com>
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Add command line option to specify which directory should be searched
to find credentials caches.
(really this time)

13 years ago Must still use knowledge of the glue context for pre-1.4 versions of MIT krb5
neilbrown [Tue, 28 Mar 2006 00:49:29 +0000 (00:49 +0000)]
Must still use knowledge of the glue context for pre-1.4 versions of MIT krb5

We need to get access to the internal krb5 context pointer for
older (pre-1.4) versions of MIT Kerberos.  We get a pointer to
the gss glue's context.  Get the right pointer before accessing
the context information.
(really this time)

13 years ago Remove unused groups variable from get_ids() which was causing a compiler warning.
neilbrown [Tue, 28 Mar 2006 00:48:56 +0000 (00:48 +0000)]
Remove unused groups variable from get_ids() which was causing a compiler warning.
(really this time)

13 years ago Update krb5 code to use glue routine lucid context functions
neilbrown [Tue, 28 Mar 2006 00:48:27 +0000 (00:48 +0000)]
Update krb5 code to use glue routine lucid context functions

The gssd code should not know about the glue layer's context structure.
A previous patch added gss_export_lucid_sec_context() and
gss_free_lucid_sec_context() functions to the gssapi glue layer.
Use these functions rather than calling directly to the Kerberos
gssapi code (which requires the Kerberos context handle rather
than the glue's context handle).

(really this time)

13 years agoadd missing files from previous update
neilbrown [Tue, 28 Mar 2006 00:47:06 +0000 (00:47 +0000)]
add missing files from previous update

13 years ago Separate out context handling code for MIT Kerberos and SPKM3
neilbrown [Tue, 28 Mar 2006 00:45:33 +0000 (00:45 +0000)]
Separate out context handling code for MIT Kerberos and SPKM3
into their own file.
(Really this time)

13 years ago User-selectable idmapping cache lifetime
neilbrown [Tue, 28 Mar 2006 00:33:55 +0000 (00:33 +0000)]
User-selectable idmapping cache lifetime

Read and process new configuration option, Cache-Expiration, and use
the value to determine how long idmapping entries are cached.
(Really this time)

13 years agoautoconf/automake
neilbrown [Mon, 27 Mar 2006 00:41:10 +0000 (00:41 +0000)]
autoconf/automake

13 years agoupdate version to -rc3
neilbrown [Mon, 27 Mar 2006 00:34:34 +0000 (00:34 +0000)]
update version to -rc3

13 years agoFix up the svcgss mess I made, and run autoconf/automake
neilbrown [Mon, 27 Mar 2006 00:29:56 +0000 (00:29 +0000)]
Fix up the svcgss mess I made, and run autoconf/automake

13 years agorun autoconf
neilbrown [Mon, 27 Mar 2006 00:04:57 +0000 (00:04 +0000)]
run autoconf

13 years ago Add debugging to better detect negotiation of enctype not supported by kernel
neilbrown [Sun, 26 Mar 2006 23:58:51 +0000 (23:58 +0000)]
Add debugging to better detect negotiation of enctype not supported by kernel

Print debugging message indicating the type of encryption keys being sent
down to the kernel.  This should make it easier to detect cases where
unsupported encryption types are being negotiated.

13 years ago Don't close and reopen all pipes on every DNOTIFY signal.
neilbrown [Sun, 26 Mar 2006 23:58:31 +0000 (23:58 +0000)]
Don't close and reopen all pipes on every DNOTIFY signal.

Don't unnecessarily close and re-open all pipes after every DNOTIFY
signal.  These unnecessary closes were triggering a kernel Oops.
Original patch modified to correct segfault when unmounting last
NFSv4 mount.

13 years ago Add option to specify directory to search for credentials cache files
neilbrown [Sun, 26 Mar 2006 23:58:10 +0000 (23:58 +0000)]
Add option to specify directory to search for credentials cache files

Add command line option to specify which directory should be searched
to find credentials caches.

13 years ago Must still use knowledge of the glue context for pre-1.4 versions of MIT krb5
neilbrown [Sun, 26 Mar 2006 23:57:47 +0000 (23:57 +0000)]
Must still use knowledge of the glue context for pre-1.4 versions of MIT krb5

We need to get access to the internal krb5 context pointer for
older (pre-1.4) versions of MIT Kerberos.  We get a pointer to
the gss glue's context.  Get the right pointer before accessing
the context information.

13 years ago Remove unused variable causing compile warning
neilbrown [Sun, 26 Mar 2006 23:57:23 +0000 (23:57 +0000)]
Remove unused variable causing compile warning

Remove unused groups variable from get_ids() which was causing a compiler warning.

13 years ago Update krb5 code to use glue routine lucid context functions
neilbrown [Sun, 26 Mar 2006 23:57:00 +0000 (23:57 +0000)]
Update krb5 code to use glue routine lucid context functions

The gssd code should not know about the glue layer's context structure.
A previous patch added gss_export_lucid_sec_context() and
gss_free_lucid_sec_context() functions to the gssapi glue layer.
Use these functions rather than calling directly to the Kerberos
gssapi code (which requires the Kerberos context handle rather
than the glue's context handle).

13 years ago Separate out context handling code for MIT Kerberos and SPKM3
neilbrown [Sun, 26 Mar 2006 23:56:30 +0000 (23:56 +0000)]
Separate out context handling code for MIT Kerberos and SPKM3
into their own file.

13 years ago Consolidate gssd and svcgssd since they share much code
neilbrown [Sun, 26 Mar 2006 23:56:03 +0000 (23:56 +0000)]
Consolidate gssd and svcgssd since they share much code

Remove directory svcgssd which was only created because the old
build system could not handle building two daemons in the same
directory.  This eliminates build complications since gssd and
svcgssd also share many source files.

This patch effectively removes the utils/svcgssd directory, moving
all its files to the utils/gssd directory.  File utils/gssd/Makefile.am
is modified with directions to build both gssd and svcgssd.

13 years ago Use PKGCONFIG to locate gssapi and rpcsecgss header files
neilbrown [Sun, 26 Mar 2006 23:55:32 +0000 (23:55 +0000)]
Use PKGCONFIG to locate gssapi and rpcsecgss header files

Instead of having separate copies of the gssapi and rpcsecgss
header files, or depending on the Kerberos gssapi header,
locate the headers now installed with the libgssapi and librpcsecgss
libraries.

Remove local copies of the gssapi and rpcsecgss header files.

This depends on the configure_use_autotools patch.

13 years ago User-selectable idmapping cache lifetime
neilbrown [Sun, 26 Mar 2006 23:54:56 +0000 (23:54 +0000)]
User-selectable idmapping cache lifetime

Read and process new configuration option, Cache-Expiration, and use
the value to determine how long idmapping entries are cached.

13 years ago Set libnfsidmap library debugging level and logging function.
neilbrown [Sun, 26 Mar 2006 23:54:16 +0000 (23:54 +0000)]
Set libnfsidmap library debugging level and logging function.

This patch adds a call to the new libnfsidmap library function
nfs4_set_debug(), which defines the verbosity level libnfsidmap
should use as well as the logging function.

13 years ago Don't close file descriptor until after calling event_del().
neilbrown [Sun, 26 Mar 2006 23:51:21 +0000 (23:51 +0000)]
Don't close file descriptor until after calling event_del().

Delete event processing for a file descriptor before closing it.
This was causing hangs when used in combination with libevent-1.0b.

13 years ago Find krb5-config on SuSE 10
neilbrown [Sun, 26 Mar 2006 23:50:44 +0000 (23:50 +0000)]
Find krb5-config on SuSE 10

SuSE 10.0 puts krb5-config in yet another obscure location.
Look for it there and use it if found.

13 years ago Update debian package information.
neilbrown [Sun, 26 Mar 2006 23:50:03 +0000 (23:50 +0000)]
Update debian package information.

13 years ago Install /var/lib/nfs files using DESTDIR and add rpcsec headers to distribution
neilbrown [Sun, 26 Mar 2006 23:49:00 +0000 (23:49 +0000)]
Install /var/lib/nfs files using DESTDIR and add rpcsec headers to distribution

Add "$(DESTDIR)" to the paths for the "$(statedir)" files so they are
put in the right place when DESTDIR is defined.

Add the rpcsec header files to EXTRA_DIST list.

13 years agohead/tail fixes
neilbrown [Wed, 21 Dec 2005 04:10:33 +0000 (04:10 +0000)]
head/tail fixes

13 years agoHandle new-style quotactl.
neilbrown [Wed, 21 Dec 2005 04:01:08 +0000 (04:01 +0000)]
Handle new-style quotactl.

13 years agorelease 1.0.8-rc2 nfs-utils-1-0-8-rc2
neilbrown [Tue, 20 Dec 2005 06:28:36 +0000 (06:28 +0000)]
release 1.0.8-rc2

13 years agorelease 1.0.8-rc2
neilbrown [Tue, 20 Dec 2005 06:06:32 +0000 (06:06 +0000)]
release 1.0.8-rc2

13 years agoComplete makefile/configure update
neilbrown [Tue, 20 Dec 2005 06:04:18 +0000 (06:04 +0000)]
Complete makefile/configure update

13 years agoAdd some missing files needed for ./configure;make
neilbrown [Tue, 20 Dec 2005 05:26:17 +0000 (05:26 +0000)]
Add some missing files needed for ./configure;make

13 years agoRemove all the Makefiles
neilbrown [Tue, 20 Dec 2005 05:19:23 +0000 (05:19 +0000)]
Remove all the Makefiles

13 years agoMore automake stuff
neilbrown [Tue, 20 Dec 2005 05:17:05 +0000 (05:17 +0000)]
More automake stuff

13 years agoAutogen update
neilbrown [Tue, 20 Dec 2005 04:12:42 +0000 (04:12 +0000)]
Autogen update

13 years agoCorrect documentation of defautl export options nfs-utils-1-0-8-rc1a
neilbrown [Tue, 20 Dec 2005 01:07:15 +0000 (01:07 +0000)]
Correct documentation of defautl export options

13 years agoSet default hostname to '*' rather than leaving it empty
neilbrown [Tue, 20 Dec 2005 00:55:47 +0000 (00:55 +0000)]
Set default hostname to '*' rather than leaving it empty

13 years ago utils/gssd/gssd_proc.c(create_auth_rpc_client): Use service
neilbrown [Mon, 19 Dec 2005 23:49:30 +0000 (23:49 +0000)]
utils/gssd/gssd_proc.c(create_auth_rpc_client): Use service
  portion of clp->servicename rather than hard-coding "nfs".

13 years agoChange version to 1.0.8-rc1 and run autoconf nfs-utils-1-0-8-rc1
neilbrown [Fri, 16 Dec 2005 05:29:14 +0000 (05:29 +0000)]
Change version to 1.0.8-rc1 and run autoconf

13 years agoUpdates from Kevin Coffman at UMich
neilbrown [Fri, 16 Dec 2005 05:24:38 +0000 (05:24 +0000)]
Updates from Kevin Coffman at UMich

13 years agoFix rpc_init so it isn't confused by ssh
neilbrown [Mon, 12 Dec 2005 03:55:53 +0000 (03:55 +0000)]
Fix rpc_init so it isn't confused by ssh

13 years agoidmapd update from Steve Dickson
neilbrown [Thu, 3 Nov 2005 05:33:07 +0000 (05:33 +0000)]
idmapd update from Steve Dickson

13 years agoUnderstand type 2 and type 3 filesystem identifiers.
neilbrown [Fri, 14 Oct 2005 07:38:59 +0000 (07:38 +0000)]
Understand type 2 and type 3 filesystem identifiers.

13 years agoSmall nfs-utils patch from Olaf Kirch
neilbrown [Fri, 7 Oct 2005 04:45:54 +0000 (04:45 +0000)]
Small nfs-utils patch from Olaf Kirch

13 years agoAssorted changes from Steve Dickson nfs-utils-1-0-7-post6
neilbrown [Thu, 6 Oct 2005 05:20:19 +0000 (05:20 +0000)]
Assorted changes from Steve Dickson

13 years agorquota fix nfs-utils-1-0-7-post5
neilbrown [Thu, 6 Oct 2005 04:24:06 +0000 (04:24 +0000)]
rquota fix

14 years agoCopy dq_dqb into rquota structure more carefully
neilbrown [Fri, 2 Sep 2005 06:29:07 +0000 (06:29 +0000)]
Copy dq_dqb into rquota structure more carefully

14 years agorpc and gssapi not longer exist in support, so remove them from Makefile nfs-utils-1-0-7-post4
neilbrown [Fri, 26 Aug 2005 02:10:18 +0000 (02:10 +0000)]
rpc and gssapi not longer exist in support, so remove them from Makefile

14 years ago Add option to set rpcsec_gss debugging level (if available) nfs-utils-1-0-7-post3
neilbrown [Fri, 26 Aug 2005 02:04:40 +0000 (02:04 +0000)]
Add option to set rpcsec_gss debugging level (if available)

Changes to allow gssd/svcgssd to build when using Hiemdal Kerberos
libraries.  Note that there are still run-time issues preventing
this from working when shared libraries for libgssapi and librpcsecgss
are used.

14 years agoRemove the rpcsec_gss code and rely on an external library instead. nfs-utils-1-0-7-post2
neilbrown [Fri, 26 Aug 2005 02:02:15 +0000 (02:02 +0000)]
Remove the rpcsec_gss code and rely on an external library instead.

14 years agoremove some files.
neilbrown [Fri, 26 Aug 2005 01:51:44 +0000 (01:51 +0000)]
remove some files.

14 years ago2005-08-26 Kevin Coffman <kwc@citi.umich.edu>
neilbrown [Fri, 26 Aug 2005 01:36:14 +0000 (01:36 +0000)]
2005-08-26 Kevin Coffman <kwc@citi.umich.edu>
*utils/mountd/mountd.c:
mountd currently always returns AUTH_NULL and AUTH_SYS as the
allowable flavors in mount replies.  We want it to also return gss
flavors when appropriate. For now as a hack we just have it always
return the KRB5 flavors as well.

*utils/mountd/cache.c:

When attempting to mount an NFSv4 pseudofilesystem (fsid=0) and the
actual exported directory does not exist on the server, rpc.mountd
doesn't check the directory exists (when fsidtype=1, i.e. using fsid,
but does check for fsidtype=0, i.e. using dev/ino).  The non-existent
exported directory path with fsid=0 is written to the kernel via
/proc/net/rpc/nfsd.export/channel, which leads to path_lookup() to
return ENOENT (seems appropriate).  Unfortunately, the new_cache
approach ignores errors returned when writing via the channel file so
that particular error is lost and the mount request is silently ignored.

Assuming it doesn't make sense to revamp the new_cache/up-call method to
not ignore returned errors, it seems appropriate to fix the case where
rpc.mountd doesn't check for the existence of an exported directory with
fsid= semantics.  The following patch does this by moving the stat() up
so it is done for both fsidtype's.  I'm not certain whether the other
tests need to be executed for fsidtype=1, but it doesn't appear to hurt
[Not exactly true: the comparison of inode numbers caused problems so
now it's kept for fsidtype=0 only].

Would it be also desirable to log a warning for every error, if any,
returned by a write to any of the /proc/net/rpc/*/channel files which
would otherwise be ignored (maybe under a debug flag)?

* gssd/mountd/svcgssd: Changes gssd, svcgssd, and mountd to ignore a
SIGHUP rather than dying.

* many: Remove the gssapi code and rely on an external library instead.