libsgx-dcap-quote-verify-dev
, libsgx-dcap-ql-dev
, libsgx-uae-service
Please follow the command to build RATS TLS from the latested source code on your system.
mkdir -p "$WORKSPACE"
cd "$WORKSPACE"
git clone https://github.com/inclavare-containers/rats-tls
cd rats-tls
cmake -DBUILD_SAMPLES=on -H. -Bbuild
make -C build install
{rats-tls-server,rats-tls-client}
will be installed to /usr/share/rats-tls/samples/{rats-tls-server,rats-tls-client}
on your system. All instances are placed in /usr/local/lib/rats-tls
.
If you want to build instances related to sgx(sgx_ecdsa, sgx_ecdsa_qve, sgx_la), please type the following command.
cmake -DRATS_TLS_BUILD_MODE="sgx" -DBUILD_SAMPLES=on -H. -Bbuild
make -C build install
If you want to run instances on libos occlum, please type the following command.
cmake -DRATS_TLS_BUILD_MODE="occlum" -DBUILD_SAMPLES=on -H. -Bbuild
make -C build install
If you want to run TDX instances, please type the following command.
cmake -DRATS_TLS_BUILD_MODE="tdx" -DBUILD_SAMPLES=on -H. -Bbuild
make -C build install
Note that SGX LVI mitigation is enabled by default. You can set macro SGX_LVI_MITIGATION
to 0
to disable SGX LVI mitigation.
Right now, RATS TLS supports the following instance types:
Priority | Tls Wrapper instances | Attester instances | Verifier instances | Crypto Wrapper Instance |
---|---|---|---|---|
0 | nulltls | nullattester | nullverifier | nullcrypto |
15 | openssl | sgx_la | sgx_la | openssl |
20 | openssl | csv | csv | openssl |
35 | openssl | sev | sev | openssl |
42 | openssl | sev_snp | sev_snp | openssl |
42 | openssl | tdx_ecdsa | tdx_ecdsa | openssl |
52 | openssl | sgx_ecdsa | sgx_ecdsa | openssl |
53 | openssl | sgx_ecdsa | sgx_ecdsa_qve | openssl |
For instance priority, the higher, the stronger. By default, RATS TLS will select the highest priority instance to use.
cd /usr/share/rats-tls/samples
./rats-tls-server
By default, the listening port is 1234. -p option can be used to specify the listening port.
cd /usr/share/rats-tls/samples
./rats-tls-client
Notice: special prerequisites for TDX remote attestation in bios configuration and hardware capability.
Check msr 0x503, return value must be 0:
sudo rdmsr 0x503s
Note that if you want to run SEV-SNP remote attestation, please refer to link to set up the host and guest Linux kernel, qemu and ovmf bios used for launching SEV-SNP guest.
Notice: special prerequisites for SEV(-ES) remote attestation in software capability.
Notice: special prerequisites for CSV(2) remote attestation in software capability.
The options of rats-tls-server are as followed:
OPTIONS:
--attester/-a value set the type of quote attester
--verifier/-v value set the type of quote verifier
--tls/-t value set the type of tls wrapper
--crypto/-c value set the type of crypto wrapper
--mutual/-m set to enable mutual attestation
--log-level/-l set the log level
--ip/-i set the listening ip address
--port/-p set the listening tcp port
--product-enclave/-P set to enable product enclave
--verdictd/-E set to connect verdictd based on EAA protocol
You can set command line parameters to specify different configurations.
For example:
./rats-tls-server --tls openssl
./rats-tls-server --attester sgx_ecdsa
./rats-tls-server --attester sgx_ecdsa_qve
./rats-tls-server --attester sgx_la
./rats-tls-server --attester tdx_ecdsa
./rats-tls-server --crypto openssl
RATS TLS's log level can be set through -l
option with 6 levels: off
, fatal
, error
, warn
, info
, and debug
. The default level is error
. The most verbose level is debug
.
For example:
./rats-tls-server -l debug
RATS TLS server binds 127.0.0.1:1234
by default. You can use -i
and -p
options to set custom configuration.
./rats-tls-server -i [ip_addr] -p [port]
You can use -m
option to enable mutual attestation.
./rats-tls-server -m
./rats-tls-client -m
In the early bootstrap of rats-tls, the debug message is mute by default. In order to enable it, please explicitly set the environment variable RATS_TLS_GLOBAL_LOG_LEVEL=<log_level>
, where \<log_level> is same as the values of the option -l
.
Please refer to this guide to run Rats Tls with Occlum and rune.
In non-sgx enviroment, it's possible to show the error messages as below when running the command ./rats-tls-client --attester sgx_ecdsa
. According to Intel DCAP's implementation, when calling to sgx_qv_get_quote_supplemental_data_size(),
if the libsgx_urts library is present, it will try to load QvE firstly. If failed, the verification will be launched by QVL. So the error info can be ignored and have no impact on the final attestation result.
[load_qve ../sgx_dcap_quoteverify.cpp:209] Error, call sgx_create_enclave for QvE fail [load_qve], SGXError:2006.
[sgx_qv_get_quote_supplemental_data_size ../sgx_dcap_quoteverify.cpp:527] Error, failed to load QvE.
Direct Dependencies
Name | Repo URL | Licenses |
---|---|---|
openssl | https://github.com/openssl/openssl | Apache |
linux-sgx | https://github.com/intel/linux-sgx | BSD-3-clause |
SGXDataCenterAttestationPrimitives | https://github.com/intel/SGXDataCenterAttestationPrimitives | BSD-3-clause |
GNU C library | C library | GNU General Public License version 3 |