"2050 LA_DEL_CRL" - this entry appear in logfile when you delete cert from Certificate Revocation List. Thats why need to change it.
"2051 LA_SET_CRL" - this entry must appear in logfile when you edit cert in Certificate Revocation List, but it doesn't happen (perhaps it's a bug)
Since 35200a29ea we build complete installers using CMake, meaning that there's no need for BuildUtil anymore.
MSBuild projects that are not migrated to CMake yet are kept for reference.
This commit also updates BUILD_WINDOWS.md so that it mentions Visual Studio 2019 instead of 2017.
To solve the problem that the escape condition of the loop that tries name resolution in UDP mode was reversed in the keep-alive function of the Internet connection, so the name resolution retry is set to 250 msec interval instead of the normal 60 second interval.
This works for all VPN protocols.
In SessionMain(): for DHCPDISCOVER and DHCPREQUEST frames, write the static IP address (which is retrieved from the user notes) in the SIADDR field of DHCPHEADER.
In VirtualDhcpServer(): for DHCPDISCOVER and DHCPREQUEST frames, read the static IP address from the SIADDR field of DHCPHEADER and assign it to the client.
- When building on Windows XP using Visual Studio 2008, I encountered the following issue.
- I did a fresh install of Windows XP SP3 32-bit, then applied updates including .NET 3.5. Next I installed MS Visual Studio 2008, then updated with sp1. All of this according to the documentation in your readme for building on Windows.
- In file src/BuildUtils/VpnBuilder.cs, there are two "if" statements testing the same thing, which is to determine if it is a 32-bit or 64-bit machine/compiler. But the then and else clauses are reversed, so clearly, one of them is wrong. The result I saw is that the SDK path being used to run RC.exe is left as the NULL string and so it fails to run the RC.exe program.
- This happens early in the build process, building the build utils. The two "if" statements are used to set paths for the Visual Studio VC and SDK directories. Depending on the integer pointer size, it uses different paths in the registry.
- When I looked in the registry on my Windows XP machine, there is no key HKLM\SOFTWARE\Wow6432Node, I have only seen that on 64-bit machines.
- For the fix, I consolidated the two "if" statements into one, the existing statement on line 380 would only set a value for Paths.VisualStudioVCDir (which got set correctly). Now I moved the code for also setting Paths.MicrosoftSDKDir, while reversing the values from the incorrectly coded "if" statement.
- I can understand that under certain circumstances, this issue would not be encountered, but should be easily reproducible when installing a clean system.
error: cannot initialize a variable of type 'wchar_t *' with an lvalue of type 'const wchar_t [4]'
wchar_t *protocol_str = (udp ? L"UDP" : L"TCP");
^ ~~~~~~~~~~~~~~~~~~~~~~~
The "session created" and "session deleted" messages were useful when a single OPENVPN_SERVER object handled multiple UDP sessions.
Now that each session has its own OPENVPN_SERVER object and session creations/deletions are logged by PROTO, the messages are redundant.
In future we will change the OpenVPN implementation so that the multi-session handling code is deleted.
The messages were like this:
OpenVPN Module: The OpenVPN Server Module is starting.
OpenVPN Session 1 (192.168.122.211:47390 -> 0.0.0.0:1194): A new session is created. Protocol: UDP
OpenVPN Session 1 (192.168.122.211:47390 -> 0.0.0.0:1194): Deleting the session.
OpenVPN Module: The OpenVPN Server Module is stopped.
ProtoHandleDatagrams() takes care of deleting a session if marked as halted.
However, the check is performed when a packet for that session is received; that never happens if the remote host doesn't send at least a packet.
This commit fixes the issue by moving the check into the loop that iterates through all sessions.
ProtoOptionsGet command - Lists the options for the specified protocol
Help for command "ProtoOptionsGet"
Purpose:
Lists the options for the specified protocol
Description:
This command can be used to retrieve the options for a specific protocol.
Detailed info (e.g. value type) will be shown.
You can change an option's value with the ProtoOptionsSet command.
Usage:
ProtoOptionsGet [protocol]
Parameters:
protocol - Protocol name.
ProtoOptionsSet command - Sets an option's value for the specified protocol
Help for command "ProtoOptionsSet"
Purpose:
Sets an option's value for the specified protocol
Description:
This command can be used to change an option's value for a specific protocol.
You can retrieve the options using the ProtoOptionsGet command.
To execute this command, you must have VPN Server administrator privileges.
Usage:
ProtoOptionsSet [protocol] [/NAME:option_name] [/VALUE:string/true/false]
Parameters:
protocol - Protocol name.
/NAME - Option name.
/VALUE - Option value. Make sure to write a value that is accepted by the specified protocol!
PROTO_OPTION is a structure that describes an option (who would've guessed?).
It's designed in a way that allows it to occupy as low memory as possible, while providing great flexibility.
The idea is similar to the one implemented in LIST for trivial types, with the difference that PROTO_OPTION doesn't require casting due to the use of union.
The reason why we don't build these two targets is that they're not used.
More specifically: they require proper configuration to work correctly, which is currently missing.
While vpninstall may be worth salvaging, vpnweb is definitely a relict of the past because it relies on ActiveX.
vpndrvinst is the name of the target and thus the default output name, let's use it.
`vpndrvinst.exe" also sounds less fishy than "driver_installer.exe"...
BuildUtil compiles the project as 32 bit and 64 bit, before building the installer package.
64 bit binaries have the "_x64" suffix and are added to the package alongside the 32 bit ones (that have no suffix).
The CMake project compiles the binaries for a single architecture and they have no suffix.
We decided that providing two separate installers is the best solution.
As for the binaries with the "_ia64" suffix: they never existed during the this repository's lifespan.
The MSBuild project built the binary into "src/bin/hamcore", causing it to be added to "hamcore.se2".
As hinted by the name of the file ("vpnsetup_nosign.exe"), it is not signed by BuildUtil, possibly to save time (the setup package is signed).
The CMake project builds the binary in the same directory as the other ones, allowing the setup to install them without the need to build a package.
Previously, the file needed to be present in order for the setup to work.
This commit removes the requirement so that the setup can be ran from the build directory without the need to copy the file (which is now removed from the repository).
The inclusion of the headers is probably a very old leftover, from when OpenSSL was not encapsulated into Mayaqua yet.
In fact, there was a "HAM_C" (defined in vpndrvinst.c) definition check in Mayaqua/Encrypt.h preventing the redefinition of OpenSSL types.
When "VPN_EXE" is defined, Mayaqua.h defines WinMain(), which handles arguments in a special way.
This commit passes "WIN32" to add_executable(), so that WinMain() is used as entry point instead of main().
The use of main() instead of WinMain() was causing service mode not to work due to the "/service" argument being discarded.
Our CMake project used to forcefully create and use two different build directories: "build" and "tmp".
This commit changes the behavior so that only the build directory CMake is ran in is used.
The "configure" script now runs CMake in "build" by default, instead of "tmp".
strtok() and wcstok() are considered unsafe functions.
A segmentation fault caused by the use of strtok() was recently reported.
Co-authored-by: Takuho NAKANO <takotakot@users.noreply.github.com>
The SSTP implementation doesn't provide packet identification, because it's not required: the protocol is identified by the HTTP header it sends to the server.
"ClientOption", as the name implies, is only used in a client context.
The issue was introduced in 235bd07e67. Before that, an unrelated check prevented UnixVLanSetState() from being called in a server context.
SoftEther VPN originally created the NIC in the UP state and never changed it, even when the the client was not connected.
The behavior was changed in 59e1483dbf, which also added the NicDownOnDisconnect option
The option was disabled by default for backwards compatibility with scripts that don't check whether the NIC is down, but it's not ideal.
This commit forces the correct behavior and removes the commands "TUNDownOnDisconnectEnable", "TUNDownOnDisconnectDisable" and "TUNDownOnDisconnectGet".
PortsUDPSet: This command can be used to specify a single or multiple UDP ports the server should listen on. "0" can be specified to disable the UDP listener.
Administrator privileges are required to execute the command.
PortsUDPGet: This command can be used to retrieve the UDP ports the server is listening on.
The two commands replace the functionality that was previously provided by OpenVpnEnable and OpenVpnGet, respectively.
Originally, StrToPortList() returned NULL when it encountered a number equal to 0 or higher than 65535.
This commit adds a new parameter to the function called "limit_range":
- When its value is true, the function retains the original behavior.
- When its value is false, the function doesn't check whether the number is in the network port number range (1-65535).
The change is required because the command to set the UDP ports will allow to remove all ports by specifying "0" as the port number.
Now that Proto supports UDP, the server can handle multiple protocols on each UDP port.
The UDP ports are specified by the "OpenVPN_UdpPortList" configuration setting, because:
- OpenVPN is currently the only UDP protocol supported by SoftEther VPN to allow a custom port number.
- Before Proto was introduced, a unified interface for the protocols didn't exist; each protocol implementation had to create its own listener.
In preparation for the upcoming WireGuard implementation, this commit renames "OpenVPN_UdpPortList" to "PortsUDP", which should clarify that the setting is global.
The change is reflected in the code. Also, the ports are now stored in a LIST rather than a string. The conversion between string and LIST only happens when loading/saving the configuration.
The default UDP ports are now the same as the TCP ones (443, 992, 1194, 5555).
*** CID 358434: Null pointer dereferences (REVERSE_INULL)
/src/Cedar/Proto.c: 451 in ProtoHandleDatagrams()
445 void ProtoHandleDatagrams(UDPLISTENER *listener, LIST *datagrams)
446 {
447 UINT i;
448 HASH_LIST *sessions;
449 PROTO *proto = listener->Param;
450
>>> CID 358434: Null pointer dereferences (REVERSE_INULL)
>>> Null-checking "listener" suggests that it may be null, but it has already been dereferenced on all paths leading to the check.
451 if (proto == NULL || listener == NULL || datagrams == NULL)
452 {
453 return;
454 }
455
456 sessions = proto->Sessions;
As a side effect, the DH parameter is now applied to the TCP server as well.
Previously, the default value was always used, ignoring the one from the configuration.
When a datagram is received, the matching session is looked up in a hash list; if it's not found, a new session is created.
This method allows to use a single UDP port for multiple protocols, as we do with TCP.
Also, each session has its own dedicated thread, used to process the received datagrams and generate the ones that are then sent through the UDP listener.
In addition to guaranteeing constant performance, separate threads also prevent a single one from blocking all sessions.
This allows to stop a UDP listener without deleting it.
It's especially useful when no datagrams should be received anymore, but there are other threads accessing the listener.
- An additional parameter is added to IsPacketForMe(), used to specify the protocol type (currently either TCP or UDP).
- SupportedModes() is dropped because it's now redundant.
- IsOk() and EstablishedSessions() are dropped because error checking should be handled by the implementation.
- ProtoImplDetect() now takes a buffer and its size rather than a SOCK, so that it can be used to detect UDP protocols.
- The OpenVPN toggle check is moved to ProtoImplDetect(), so that we don't have to duplicate it once UDP support is implemented.
The PROTO structure is now used to identify the system as a whole, rather than a single protocol. It's stored and initialized in Server.
ProtoCompare(), ProtoAdd() and ProtoDetected() are renamed to make the difference between PROTO and PROTO_IMPL more clear.
ProtoGet() and ProtoNum() are removed because the related list can now be accessed directly by Server.
From https://community.openvpn.net/openvpn/wiki/Openvpn23ManPage:
--block-outside-dns
Block DNS servers on other network adapters to prevent DNS leaks.
This option prevents any application from accessing TCP or UDP port 53 except one inside the tunnel.
It uses Windows Filtering Platform (WFP) and works on Windows Vista or later.
This option is considered unknown on non-Windows platforms and unsupported on Windows XP, resulting in fatal error.
You may want to use --setenv opt or --ignore-unknown-option (not suitable for Windows XP) to ignore said error.
Note that pushing unknown options from server does not trigger fatal errors.
On uClibc, the ifaddrs.h support is optional. While the default
Buildroot uClibc configuration has it enabled, some external
toolchains may not. Therefore this patch detects that and adjusts
softether usage of ifaddrs accordingly.
Based on an initial patch from Bernd Kuhls.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[Retrieved from:
https://git.buildroot.net/buildroot/tree/package/softether/0009-uclibc-ai-addrconfig.patch]
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Formerly, SKIP_CPU_FEATURES is automatically detected by system
processor. However, "^(armv7l|aarch64|s390x)$" does not cover all
processors that cpu_features should be skipped.
"armv6", "armv7", "mips", "mips64" on FreeBSD are examples [1]
that cpu_features is not correctly skipped.
This change intends to build SoftEther without any modifications on
CMakeLists.txt on such processors.
cmake . -DSKIP_CPU_FEATURES=1
[1] https://www.freebsd.org/platforms/
DbDir : directory to store files such as vpn_server.config and backups etc
LogDir : directory to write logs (sub directories is created in this dir)
PidDir : directory to put PID files such as .ctl-* .pid-* .VPN-*
Currently the systemd service unit files are installed
into /lib/systemd/system if that directory exists. This
might not be optimal for every user, e.g. when the build
system is not the target system or when building as an
unprivileged user using CMAKE_INSTALL_PREFIX.
Make this configurable by adding a cached cmake variable
CMAKE_INSTALL_SYSTEMD_UNITDIR. Usage:
- install unit files into /lib/systemd/system if it exists (old
behavior)
cmake
- don't install unit files
cmake -D CMAKE_INSTALL_SYSTEMD_UNITDIR=
- install into absolute path
cmake -D CMAKE_INSTALL_SYSTEMD_UNITDIR=/path
- install into path relative to ${CMAKE_INSTALL_PREFIX}
cmake -D CMAKE_INSTALL_SYSTEMD_UNITDIR=path
The function has been greatly improved, here are some of the changes:
- The required SESSION (c->Session) parameter is checked correctly: the function returns immediately in case it's NULL. Previously, the function didn't return in case the parameter was NULL; multiple checks were in place, but not in all instances where the parameter was dereferenced.
- The resolved IP address is cached with all proxy types.
- The "RestoreServerNameAndPort" variable is documented.
- The Debug() messages have been improved.
This commit moves the generic (not related to our protocol) proxy stuff from Cedar to Mayaqua, in dedicated files.
The functions are refactored so that they all have the same arguments and follow the same logic.
Dedicated error codes are added, in order to indicate clearly why the function(s) failed.
Coverity Scan detected an out-of-bounds access issue: OvsProcessData() checked whether the payload size was bigger than the size of the buffer, instead of checking whether the entire packet size (payload size + 2 bytes) was, resulting in an out-of-bounds access in case the payload size is bigger than 1998.
This commit also improves the variable names, the comments and adds two Debug() lines.
OvsDecrypt() returns 0 when it fails, resulting in "size" rolling over with an end result of 4294967292.
This commit fixes the issue by checking whether "size" is greater than sizeof(UINT) before performing the subtraction.
The bug was caused by a typo in the StrCpy() call: the source buffer was the same as the destination one, meaning that the function didn't do anything.
- Fixed the RADIUS PEAP client to use the standard TLS versioning.
- Implementation of a function to fix the MAC address of L3 VPN protocol by entering e.g. "MAC: 112233445566" in the "Notes" field of the user information.
- Implementation of a function to fix the virtual MAC address to be assigned to the L3 VPN client as a string attribute from RADIUS server when authentication.
Hardcoded paths are used in log file enumeration such as LogFileList
command or GenerateEraseFileList function to delete old log files when
disk free space is lacking.
Fixes: SoftEtherVPN/SoftEtherVPN#972
If SecureNAT is enabled and the hostname of the server
is longer than 16characters, every NETBIOS name resolution
query triggers the buffer overflow. If the server was built
with stack protection, the process will be killed.
This commit adds a protocol interface to the server, its purpose is to manage TCP connections and the various third-party protocols.
More specifically, ProtoHandleConnection() takes care of exchanging the packets between the local and remote endpoint; the protocol implementation only has to parse them and act accordingly.
The interface knows which protocol is the connection for by calling IsPacketForMe(), a function implemented for each protocol.
My previous patch used a wrong if directive, which disabled removed
(de)initialization and threading for LibreSSL. This most likely causes
issues at runtime.
A race condition in the DHCP server caused it to offer the same IP address to multiple clients when they connected at the same time, because an offered IP address was considered free until the final step (DHCP_ACK).
This commit introduces a list to keep track of the pending leases created during DHCP_OFFER, so that an IP address is guaranteed to be offered to a single client.
