You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
vlmcsd/man/vlmcsd.8.unix.txt

649 lines
31 KiB
Plaintext

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

VLMCSD(8) KMS Activation Manual VLMCSD(8)
NAME
vlmcsd - a fully Microsoft compatible KMS server
SYNOPSIS
vlmcsd [ options ]
DESCRIPTION
vlmcsd is a fully Microsoft compatible KMS server that provides product
activation services to clients. It is meant as a drop-in replacement
for a Microsoft KMS server (Windows computer with KMS key entered). It
currently supports KMS protocol versions 4, 5 and 6.
vlmcsd is designed to run on POSIX compatible operating systens. It
only requires a basic C library with a BSD-style sockets API and either
fork(2) or pthreads(7). That allows it to run on most embedded systems
like routers, NASes, mobile phones, tablets, TVs, settop boxes, etc.
Some efforts have been made that it also runs on Windows.
Although vlmcsd does neither require an activation key nor a payment to
anyone, it is not meant to run illegal copies of Windows. Its purpose
is to ensure that owners of legal copies can use their software without
restrictions, e.g. if you buy a new computer or motherboard and your
key will be refused activation from Microsoft servers due to hardware
changes.
vlmcsd may be started via an internet superserver like inetd(8) or
xinetd(8) as well as an advanced init system like systemd(8) or
launchd(8) using socket based activation. If vlmcsd detects that
stdin(3) is a socket, it assumes that there is already a connected
client on stdin that wants to be activated. All options that control
setting up listening sockets will be ignored when in inetd mode.
OPTIONS
Since vlmcsd can be configured at compile time, some options may not be
available on your system.
All options that do no require an argument may be combined with a sin
gle dash, for instance "vlmcsd -D -e" is identical to "vlmcsd -De". For
all options that require an argument a space between the option and the
option argument is optional. Thus "vlmcsd -r 2" and "vlmcsd -r2" are
identical too.
-h or -?
Displays help.
-V Displays extended version information. This includes the com
piler used to build vlmcsd, the intended platform and flags
(compile time options) to build vlmcsd. If you have the source
code of vlmcsd, you can type make help (or gmake help on systems
that do not use the GNU version of make(1) by default) to see
the meaning of those flags.
-L ipaddress[:port]
Instructs vlmcsd to listen on ipaddress with optional port
(default 1688). You can use this option more than once. If you
do not specify -L at least once, IP addresses 0.0.0.0 (IPv4) and
:: (IPv6) are used. If the IP address contains colons (IPv6) you
must enclose the IP address in brackets if you specify the
optional port, e.g. [2001:db8::dead:beef]:1688.
If no port is specified, vlmcsd uses the default port according
to a preceding -P option. If you specify a port, it can be a
number (1-65535) or a name (usually found in /etc/services if
not provided via LDAP, NIS+ or another name service).
If you specify a link local IPv6 address (fe80::/10, usually
starting with fe80::), it must be followed by a percent sign (%)
and a scope id (=network interface name or number) on most
unixoid OSses including Linux, Android, MacOS X and iOS, e.g.
fe80::1234:56ff:fe78:9abc%eth0 or
[fe80::1234:56ff:fe78:9abc%2]:1688. Windows (including cygwin)
does not require a scope id unless the same link local address
is used on more than one network interface. Windows does not
accept a name and the scope id must be a number.
-o level
Sets the level of protection against activations from public IP
addresses. The default is -o0 for no protection.
-o1 causes vlmcsd not to listen on all IP addresses but on pri
vate IP addresses only. IPv4 addresses in the 100.64.0.0/10
range (see RFC6598) are not treated as private since they can be
reached from other users of your ISP. Private IPv4 addresses are
10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16, 169.254.0.0/16 and
127.0.0.0/8. vlmcsd treats all IPv6 addresses not within
2000::/3 as private addresses.
If -o1 is combined with -L, it will listen on all private IP
addresses plus the ones specified by one or more -L statements.
If -o1 is combined with -P, only the last -P statement will be
used.
Using -o1 does not protect you if you enable NAT port forwarding
on your router to your vlmcsd machine. It is identical to using
multiple -L statements with all of your private IP addresses.
What -o1 does for you, is automatically enumerating your private
IP addresses.
-o2 does not affect the interfaces, vlmcsd is listening on. When
a clients connects, vlmcsd immediately drops the connection if
the client has a public IP address. Unlike -o1 clients will be
able to establish a TCP connection but it will be closed without
a single byte sent over the connection. This protects against
clients with public IP addresses even if NAT port forwarding is
used. While -o2 offers a higher level of protection than -o1,
the client sees that the KMS TCP port (1688 by default) is actu
ally accepting connections.
If vlmcsd is compiled to use MS RPC, -o2 can only offer very
poor protection. Control is passed from MS RPC to vlmcsd after
the KMS protocol has already been negotiated. Thus a client can
always verify that the KMS protocol is available even though it
receives an RPC_S_ACCESS_DENIED error message. vlmcsd will issue
a warning if -o2 is used with MS RPC. For adaequate protection
do not use a MS RPC build of vlmcsd with -o2.
-o3 combines -o1 and -o2. vlmcsd listens on private interfaces
only and if a public client manages to connect anyway due to NAT
port forwarding, it will be immediately dropped.
If you use any form of TCP level port forwarding (e.g. nc(1),
netcat(1), ssh(1) port forwarding or similar) to redirect KMS
requests to vlmcsd, there will be no protection even if you use
-o2 or -o3. This is due to the simple fact that vlmcsd sees the
IP address of the redirector and not the IP address of the
client.
-o1 (and thus -o3) is not (yet) available in some scenarios:
FreeBSD: There is a longtime unfixed bug ⟨https://
bugs.freebsd.org/bugzilla/show_bug.cgi?id=178881⟩ in the
32-bit ABI of the 64-bit kernel. If you have a 64-bit Free
BSD kernel, you must run the 64-bit version of vlmcsd if
you use -o1 or -o3. The 32-bit version causes undefined
behavior up to crashing vlmcsd. Other BSDs (NetBSD, Open
BSD, Dragonfly and Mac OS X) work correctly.
If vlmcsd was started by an internet superserver or was
compiled to use Microsoft RPC (Windows only) or simple
sockets, -o1 and -o3 are not available by design.
-P port
Use TCP port for all subsequent -L statements that do not
include an optional port. If you use -P and -L, -P must be spec
ified before -L.
-F0 and -F1
Allow (-F1) or disallow (-F0) binding to IP addresses that are
currently not configured on your system. The default is -F0. -F1
allows you to bind to an IP address that may be configured after
you started vlmcsd. vlmcsd will listen on that address as soon
as it becomes available. This feature is only available under
Linux (IPv4 and IPv6) and FreeBSD (IPv4 only). FreeBSD allows
this feature only for the root user (more correctly: processes
that have the PRIV_NETINET_BINDANY privilege). Linux does not
require a capability for this.
-t seconds
Timeout the TCP connection with the client after seconds sec
onds. After sending an activation request. RPC keeps the TCP
connection for a while. The default is 30 seconds. You may spec
ify a shorter period to free ressources on your device faster.
This is useful for devices with limited main memory or if you
used -m to limit the concurrent clients that may request activa
tion. Microsoft RPC clients disconnect after 30 seconds by
default. Setting seconds to a greater value does not make much
sense.
-m concurrent-clients
Limit the number of clients that will be handled concurrently.
This is useful for devices with limited ressources or if you are
experiencing DoS attacks that spawn thousands of threads or
forked processes. If additional clients connect to vlmcsd, they
need to wait until another client disconnects. If you set con
current-clients to a small value ( <10 ), you should also select
a reasonable timeout of 2 or 3 seconds with -t. The default is
no limit.
-d Disconnect each client after processing one activation request.
This is a direct violation of DCE RPC but may help if you
receive malicous fake RPC requests that block your threads or
forked processes. Some other KMS emulators (e.g. py-kms) behave
this way.
-k Do not disconnect clients after processing an activation
request. This selects the default behavior. -k is useful only if
you used an ini file (see vlmcsd.ini(5) and -i). If the ini file
contains the line "DisconnectClientsImmediately = true", you can
use this switch to restore the default behavior.
-N0 and -N1
Disables (-N0) or enables (-N1) the use of the NDR64 transfer
syntax in the RPC protocol. Unlike Microsoft vlmcsd supports
NDR64 on 32-bit operating systems. Microsoft introduced NDR64 in
Windows Vista but their KMS servers started using it with Win
dows 8. Thus if you choose random ePIDs, vlmcsd will select
ePIDs with build numbers 9200 and 9600 if you enable NDR64 and
build numbers 6002 and 7601 if you disable NDR64. The default is
to enable NDR64.
-B0 and -B1
Disables (-B0) or enables (-B1) bind time feature negotiation
(BTFN) in the RPC protocol. All Windows operating systems start
ing with Vista support BTFN and try to negotiate it when initi
ating an RPC connection. Thus consider turning it off as a debug
/ troubleshooting feature only. Some older firewalls that selec
tively block or redirect RPC traffic may get confused when they
detect NDR64 or BTFN.
-l filename
Use filename as a log file. The log file records all activations
with IP address, Windows workstation name (no reverse DNS
lookup), activated product, KMS protocol, time and date. If you
do not specify a log file, no log is created. For a live view of
the log file type tail -f file.
If you use the special filename "syslog", vlmcsd uses syslog(3)
for logging. If your system has no syslog service (/dev/log)
installed, logging output will go to /dev/console. Syslog log
ging is not available in the native Windows version. The Cygwin
version does support syslog logging.
-T0 and -T1
Disable (-T0) or enable (-T1) the inclusion of date and time in
each line of the log. The default is -T1. -T0 is useful if you
log to stdout(3) which is redirected to another logging mecha
nism that already includes date and time in its output, for
instance systemd-journald(8). If you log to syslog(3), -T1 is
ignored and date and time will never be included in the output
sent to syslog(3).
-D Normally vlmcsd daemonizes and runs in background (except the
native Windows version). If -D is specified, vlmcsd does not
daemonize and runs in foreground. This is useful for testing and
allows you to simply press <Ctrl-C> to exit vlmcsd.
The native Windows version never daemonizes and always behaves
as if -D had been specified. You may want to install vlmcsd as a
service instead. See -s.
-e If specified, vlmcsd ignores -l and writes all logging output to
stdout(3). This is mainly useful for testing and debugging and
often combined with -D.
-v Use verbose logging. Logs every parameter of the base request
and the base response. It also logs the HWID of the KMS server
if KMS protocol version 6 is used. This option is mainly for
debugging purposes. It only has an effect if some form of log
ging is used. Thus -v does not make sense if not used with -l,
-e or -f.
-q Do not use verbose logging. This is actually the default behav
ior. It only makes sense if you use vlmcsd with an ini file (see
-i and vlmcsd.ini(5)). If the ini file contains the line
"LogVerbose = true" you can use -q to restore the default behav
ior.
-p filename
Create pid file filename. This has nothing to do with KMS ePIDs.
A pid file is a file where vlmcsd writes its own process id.
This is used by standard init scripts (typically found in
/etc/init.d). The default is not to write a pid file.
-u user and -g group
Causes vlmcsd to run in the specified user and group security
context. The main purpose for this is to drop root privileges
after it has been started from the root account. To use this
feature from cygwin you must run cyglsa-config and the account
from which vlmcsd is started must have the rights "Act as part
of the operating system" and "Replace a process level token".
The native Windows version does not support these options.
The actual security context switch is performed after the TCP
sockets have been created. This allows you to use privileged
ports (< 1024) when you start vlmcsd from the root account.
However if you use an ini, pid or log file, you must ensure that
the unprivileged user has access to these files. You can always
log to syslog(3) from an unprivileged account on most platforms
(see -l).
-w ePID
Use ePID as Windows ePID. If specified, -r is disregarded for
Windows.
-0 ePID
Use ePID as Office 2010 ePID (including Project and Visio). If
specified, -r is disregarded for Office 2010.
-3 ePID
Use ePID as Office 2013 ePID (including Project and Visio). If
specified, -r is disregarded for Office 2013.
-6 ePID
Use ePID as Office 2016 ePID (including Project and Visio). If
specified, -r is disregarded for Office 2016.
-H HwId
Use HwId for all products. All HWIDs in the ini file (see -i)
will not be used. In an ini file you can specify a seperate HWID
for each application-guid. This is not possible when entering a
HWID from the command line.
HwId must be specified as 16 hex digits that are interpreted as
a series of 8 bytes (big endian). Any character that is not a
hex digit will be ignored. This is for better readability. The
following commands are identical:
vlmcsd -H 0123456789ABCDEF
vlmcsd -H 01:23:45:67:89:ab:cd:ef
vlmcsd -H "01 23 45 67 89 AB CD EF"
-i filename
Use configuration file (aka ini file) filename. Most configura
tion parameters can be set either via the command line or an ini
file. The command line always has precedence over configuration
items in the ini file. See vlmcsd.ini(5) for the format of the
configuration file.
If vlmcsd has been compiled to use a default configuration file
(often /etc/vlmcsd.ini), you may use -i- to ignore the default
configuration file.
-r0, -r1 (default) and -r2
These options determine how ePIDs are generated if
- you did not sprecify an ePID in the command line and
- you haven't used -i or
- the file specified by -i cannot be opened or
- the file specified by -i does not contain an ePID for the KMS
request
-r0 means there are no random ePIDs. vlmcsd simply issues
default ePIDs that are built into the binary at compile time.
Pro: behaves like real KMS server that also always issues the
same ePID. Con: Microsoft may start blacklisting again and the
default ePID may not work any longer.
-r1 instructs vlmcsd to generate random ePIDs when the program
starts or receives a SIGHUP signal and uses these ePIDs until it
is stopped or receives another SIGHUP. Most other KMS emulators
generate a new ePID on every KMS request. This is easily
detectable. Microsoft could just modify sppsvc.exe in a way that
it always sends two identical KMS requests in two RPC requests
but over the same TCP connection. If both KMS responses contain
the different ePIDs, the KMS server is not genuine. -r1 is the
default mode. -r1 also ensures that all three ePIDs (Windows,
Office 2010 and Office 2013) use the same OS build number and
LCID (language id).
If vlmcsd has been started by an internet superserver, -r1 works
almost identically to -r2. The only exception occurs if you send
more than one activation request over the same TCP connection.
This is simply due to the fact that vlmcsd is started upon a
connection request and does not stay in memory after servicing a
KMS request. Consider using -r0 or -w, -0, -3 and -6 when start
ing vlmcsd by an internet superserver.
-r2 behaves like most other KMS server emulators with random
support and generates a new random ePID on every request. -r2
should be treated as debugging option only because it allows
very easy emulator detection.
-C LCID
Do not randomize the locale id part of the ePID and use LCID
instead. The LCID must be specified as a decimal number, e.g.
1049 for "Russian - Russia". This option has no effect if the
ePID is not randomized at all, e.g. if it is selected from the
command line or an ini file.
By default vlmcsd generates a valid locale id that is recognized
by .NET Framework 4.0. This may lead to a locale id which is
unlikely to occur in your country, for instance 2155 for "Quecha
- Ecuador". You may want to select the locale id of your country
instead. See MSDN ⟨http://msdn.microsoft.com/en-us/goglobal/
bb964664.aspx⟩ for a list of valid LCIDs. Please note that some
of them are not recognized by .NET Framework 4.0.
Most other KMS emulators use a fixed LCID of 1033 (English -
US). To achive the same behavior in vlmcsd use -C 1033.
-K0, -K1, -K2 and -K3
Sets the whitelisting level to determine which products vlmcsd
activates or refuses. The default is -K0.
-K0: activate all products with an unknown, retail or
beta/preview KMS ID.
-K1: activate products with a retail or beta/preview KMS ID
but refuse to activate products with an unknown KMS ID.
-K2: activate products with an unknown KMS ID but refuse
products with a retail or beta/preview KMS ID.
-K3: activate only products with a known volume license RTM
KMS ID and refuse all others.
The SKU ID is not checked. Like a genuine KMS server vlmcsd
activates a product that has a random or unknown SKU ID. If you
select -K1 or -K3, vlmcsd also checks the Application ID for
correctness. If Microsoft introduces a new KMS ID for a new
product, you cannot activate it if you used -K1 or -K3 until a
new version of vlmcsd is available.
-c0 and -c1
-c1 causes vlmcsd to check if the client time differs no more
than four hours from the system time. -c0 (the default) disables
this check. -c1 is useful to prevent emulator detection. A
client that tries to detect an emulator could simply send two
subsequent request with two time stamps that differ more than
four hours from each other. If both requests succeed, the server
is an emulator. If you specify -c1 on a system with no reliable
time source, activations will fail. It is ok to set the correct
system time after you started vlmcsd.
-R renewal-interval
Instructs clients to renew activation every renewal-interval.
The renewal-interval is a number optionally immediately followed
by a letter indicating the unit. Valid unit letters are s (sec
onds), m (minutes), h (hours), d (days) and w (weeks). If you do
not specify a letter, minutes is assumed.
-R3d for instance instructs clients to renew activation every 3
days. The default renewal-interval is 10080 (identical to 7d and
1w).
Due to poor implementation of Microsofts KMS Client it cannot be
guaranteed that activation is renewed on time as specfied by the
-R option. Don't care about that. Renewal will happen well
before your activation expires (usually 180 days).
Even though you can specify seconds, the granularity of this
option is 1 minute. Seconds are rounded down to the next multi
ple of 60.
-A activation-interval
Instructs clients to retry activation every activation-interval
if it was unsuccessful, e.g. because it could not reach the
server. The default is 120 (identical to 2h). activation-inter
val follows the same syntax as renewal-interval in the -R
option.
-s Installs vlmcsd as a Windows service. This option only works
with the native Windows version and Cygwin. Combine -s with
other command line options. These will be in effect when you
start the service. The service automatically starts when you
reboot your machine. To start it manually, type "net start vlm
csd".
If you use Cygwin, you must include your Cygwin system DLL
directory (usually C:\Cygwin\bin or C:\Cygwin64\bin) into the
PATH environment variable or the service will not start.
You can reinstall the service anytime using vlmcsd -s again,
e.g. with a different command line. If the service is running,
it will be restarted with the new command line.
When using -s the command line is checked for basic syntax
errors only. For example "vlmcsd -s -L 1.2.3.4" reports no error
but the service will not start if 1.2.3.4 is not an IP address
on your system.
-S Uninstalls the vlmcsd service. Works only with the native Win
dows version and Cygwin. All other options will be ignored if
you include -S in the command line.
-U [domain\]username
Can only be used together with -s. Starts the service as a dif
ferent user than the local SYSTEM account. This is used to run
the service under an account with low privileges. If you omit
the domain, an account from the local computer will be used.
You may use "NT AUTHORITY\NetworkService". This is a pseudo user
with low privileges. You may also use "NT AUTHORITY\LocalSer
vice" which has more privileges but these are of no use for run
ning vlmcsd.
Make sure that the user you specify has at least execute permis
sion for your executable. "NT AUTHORITY\NetworkService" normally
has no permission to run binaries from your home directory.
For your convenience you can use the special username "/l" as a
shortcut for "NT AUTHORITY\LocalService" and "/n" for "NT
AUTHORITY\NetworkService". "vlmcsd -s -U /n" installs the ser
vice to run as "NT AUTHORITY\NetworkService".
-W password
Can only be used together with -s. Specifies a password for the
corresponding username you use with -U. SYSTEM, "NT AUTHOR
ITY\NetworkService", "NT AUTHORITY\LocalService" do not require
a password.
If you specify a user with even lower privileges than "NT
AUTHORITY\NetworkService", you must specify its password. You
also have to grant the "Log on as a service" right to that user.
SIGNALS
The following signals differ from the default behavior:
SIGTERM, SIGINT
These signals cause vlmcsd to exit gracefully. All global sema
phores and shared memory pages will be released, the pid file
will be unlinked (deleted) and a shutdown message will be
logged.
SIGHUP Causes vlmcsd to be restarted completely. This is useful if you
started vlmcsd with an ini file. You can modify the ini file
while vlmcsd is running and then sending SIGHUP, e.g. by typing
"killall -SIGHUP vlmcsd" or "kill -SIGHUP `cat /var/run/vlm
csd.pid`".
The SIGHUP handler has been implemented relatively simple. It is
virtually the same as stopping vlmcsd and starting it again
immediately with the following exceptions:
— The new process does not get a new process id.
— If you used a pid file, it is not deleted and recreated
because the process id stays the same.
— If you used the 'user' and/or 'group' directive in an ini
file these are ignored. This is because once you switched to
lower privileged users and groups, there is no way back. Any
thing else would be a severe security flaw in the OS.
Signaling is not available in the native Windows version and in the
Cygwin version when it runs as Windows service.
SUPPORTED OPERATING SYSTEMS
vlmcsd compiles and runs on Linux, Windows (no Cygwin required but
explicitly supported), Mac OS X, FreeBSD, NetBSD, OpenBSD, Dragonfly
BSD, Minix, Solaris, OpenIndiana, Android and iOS. Other POSIX or
unixoid OSses may work with unmodified sources or may require minor
porting efforts.
SUPPORTED PRODUCTS
vlmcsd can answer activation requests for the following products: Win
dows Vista, Windows 7, Windows 8, Windows 8.1, Windows 10 (up to 1607),
Windows Server 2008, Windows Server 2008 R2, Windows Server 2012, Win
dows Server 2012 R2, Windows Server 2016, Office 2010, Project 2010,
Visio 2010, Office 2013, Project 2013, Visio 2013, Office 2016, Project
2016, Visio 2016. Newer version may work as long as the KMS protocol
does not change. A complete list of fully supported products can be
obtained using the -x option of vlmcs(1).
Office, Project and Visio must be volume license versions.
FILES
vlmcsd.ini(5)
EXAMPLES
vlmcsd -De
Starts vlmcsd in foreground. Useful if you use it for the first
time and want to see what's happening when a client requests
activation.
vlmcsd -l /var/log/vlmcsd.log
Starts vlmcsd as a daemon and logs everything to /var/log/vlm
csd.log.
vlmcsd -L 192.168.1.17
Starts vlmcsd as a daemon and listens on IP address 192.168.1.17
only. This is useful for routers that have a public and a pri
vate IP address to prevent your KMS server from becoming public.
vlmcsd -s -U /n -l C:\logs\vlmcsd.log
Installs vlmcsd as a Windows service with low privileges and
logs everything to C:\logs\vlmcsd.log when the service is
started with "net start vlmcsd".
BUGS
An ePID specified in an ini file must not contain spaces.
INTENTIONAL BUGS
vlmcsd always reports enough active clients to satisfy the N count pol
icy of the request.
AUTHOR
Written by crony12, Hotbird64 and vityan666. With contributions from
DougQaid.
CREDITS
Thanks to CODYQX4, deagles, eIcn, mikmik38, nosferati87, qad, Rati
borus, ...
SEE ALSO
vlmcsd.ini(5), vlmcsd(7), vlmcs(1), vlmcsdmulti(1)
Hotbird64 October 2016 VLMCSD(8)