Mika Westerberg | 163f151 | 2017-06-06 15:25:18 +0300 | [diff] [blame] | 1 | ============= |
| 2 | Thunderbolt |
| 3 | ============= |
| 4 | The interface presented here is not meant for end users. Instead there |
| 5 | should be a userspace tool that handles all the low-level details, keeps |
| 6 | database of the authorized devices and prompts user for new connections. |
| 7 | |
| 8 | More details about the sysfs interface for Thunderbolt devices can be |
| 9 | found in ``Documentation/ABI/testing/sysfs-bus-thunderbolt``. |
| 10 | |
| 11 | Those users who just want to connect any device without any sort of |
| 12 | manual work, can add following line to |
| 13 | ``/etc/udev/rules.d/99-local.rules``:: |
| 14 | |
| 15 | ACTION=="add", SUBSYSTEM=="thunderbolt", ATTR{authorized}=="0", ATTR{authorized}="1" |
| 16 | |
| 17 | This will authorize all devices automatically when they appear. However, |
| 18 | keep in mind that this bypasses the security levels and makes the system |
| 19 | vulnerable to DMA attacks. |
| 20 | |
| 21 | Security levels and how to use them |
| 22 | ----------------------------------- |
| 23 | Starting from Intel Falcon Ridge Thunderbolt controller there are 4 |
| 24 | security levels available. The reason for these is the fact that the |
| 25 | connected devices can be DMA masters and thus read contents of the host |
| 26 | memory without CPU and OS knowing about it. There are ways to prevent |
| 27 | this by setting up an IOMMU but it is not always available for various |
| 28 | reasons. |
| 29 | |
| 30 | The security levels are as follows: |
| 31 | |
| 32 | none |
| 33 | All devices are automatically connected by the firmware. No user |
| 34 | approval is needed. In BIOS settings this is typically called |
| 35 | *Legacy mode*. |
| 36 | |
| 37 | user |
| 38 | User is asked whether the device is allowed to be connected. |
| 39 | Based on the device identification information available through |
| 40 | ``/sys/bus/thunderbolt/devices``. user then can do the decision. |
| 41 | In BIOS settings this is typically called *Unique ID*. |
| 42 | |
| 43 | secure |
| 44 | User is asked whether the device is allowed to be connected. In |
| 45 | addition to UUID the device (if it supports secure connect) is sent |
| 46 | a challenge that should match the expected one based on a random key |
| 47 | written to ``key`` sysfs attribute. In BIOS settings this is |
| 48 | typically called *One time saved key*. |
| 49 | |
| 50 | dponly |
| 51 | The firmware automatically creates tunnels for Display Port and |
| 52 | USB. No PCIe tunneling is done. In BIOS settings this is |
| 53 | typically called *Display Port Only*. |
| 54 | |
| 55 | The current security level can be read from |
| 56 | ``/sys/bus/thunderbolt/devices/domainX/security`` where ``domainX`` is |
| 57 | the Thunderbolt domain the host controller manages. There is typically |
| 58 | one domain per Thunderbolt host controller. |
| 59 | |
| 60 | If the security level reads as ``user`` or ``secure`` the connected |
| 61 | device must be authorized by the user before PCIe tunnels are created |
| 62 | (e.g the PCIe device appears). |
| 63 | |
| 64 | Each Thunderbolt device plugged in will appear in sysfs under |
| 65 | ``/sys/bus/thunderbolt/devices``. The device directory carries |
| 66 | information that can be used to identify the particular device, |
| 67 | including its name and UUID. |
| 68 | |
| 69 | Authorizing devices when security level is ``user`` or ``secure`` |
| 70 | ----------------------------------------------------------------- |
| 71 | When a device is plugged in it will appear in sysfs as follows:: |
| 72 | |
| 73 | /sys/bus/thunderbolt/devices/0-1/authorized - 0 |
| 74 | /sys/bus/thunderbolt/devices/0-1/device - 0x8004 |
| 75 | /sys/bus/thunderbolt/devices/0-1/device_name - Thunderbolt to FireWire Adapter |
| 76 | /sys/bus/thunderbolt/devices/0-1/vendor - 0x1 |
| 77 | /sys/bus/thunderbolt/devices/0-1/vendor_name - Apple, Inc. |
| 78 | /sys/bus/thunderbolt/devices/0-1/unique_id - e0376f00-0300-0100-ffff-ffffffffffff |
| 79 | |
| 80 | The ``authorized`` attribute reads 0 which means no PCIe tunnels are |
| 81 | created yet. The user can authorize the device by simply:: |
| 82 | |
| 83 | # echo 1 > /sys/bus/thunderbolt/devices/0-1/authorized |
| 84 | |
| 85 | This will create the PCIe tunnels and the device is now connected. |
| 86 | |
| 87 | If the device supports secure connect, and the domain security level is |
| 88 | set to ``secure``, it has an additional attribute ``key`` which can hold |
| 89 | a random 32 byte value used for authorization and challenging the device in |
| 90 | future connects:: |
| 91 | |
| 92 | /sys/bus/thunderbolt/devices/0-3/authorized - 0 |
| 93 | /sys/bus/thunderbolt/devices/0-3/device - 0x305 |
| 94 | /sys/bus/thunderbolt/devices/0-3/device_name - AKiTiO Thunder3 PCIe Box |
| 95 | /sys/bus/thunderbolt/devices/0-3/key - |
| 96 | /sys/bus/thunderbolt/devices/0-3/vendor - 0x41 |
| 97 | /sys/bus/thunderbolt/devices/0-3/vendor_name - inXtron |
| 98 | /sys/bus/thunderbolt/devices/0-3/unique_id - dc010000-0000-8508-a22d-32ca6421cb16 |
| 99 | |
| 100 | Notice the key is empty by default. |
| 101 | |
| 102 | If the user does not want to use secure connect it can just ``echo 1`` |
| 103 | to the ``authorized`` attribute and the PCIe tunnels will be created in |
| 104 | the same way than in ``user`` security level. |
| 105 | |
| 106 | If the user wants to use secure connect, the first time the device is |
| 107 | plugged a key needs to be created and send to the device:: |
| 108 | |
| 109 | # key=$(openssl rand -hex 32) |
| 110 | # echo $key > /sys/bus/thunderbolt/devices/0-3/key |
| 111 | # echo 1 > /sys/bus/thunderbolt/devices/0-3/authorized |
| 112 | |
| 113 | Now the device is connected (PCIe tunnels are created) and in addition |
| 114 | the key is stored on the device NVM. |
| 115 | |
| 116 | Next time the device is plugged in the user can verify (challenge) the |
| 117 | device using the same key:: |
| 118 | |
| 119 | # echo $key > /sys/bus/thunderbolt/devices/0-3/key |
| 120 | # echo 2 > /sys/bus/thunderbolt/devices/0-3/authorized |
| 121 | |
| 122 | If the challenge the device returns back matches the one we expect based |
| 123 | on the key, the device is connected and the PCIe tunnels are created. |
| 124 | However, if the challenge failed no tunnels are created and error is |
| 125 | returned to the user. |
| 126 | |
| 127 | If the user still wants to connect the device it can either approve |
| 128 | the device without a key or write new key and write 1 to the |
| 129 | ``authorized`` file to get the new key stored on the device NVM. |
| 130 | |
| 131 | Upgrading NVM on Thunderbolt device or host |
| 132 | ------------------------------------------- |
| 133 | Since most of the functionality is handled in a firmware running on a |
| 134 | host controller or a device, it is important that the firmware can be |
| 135 | upgraded to the latest where possible bugs in it have been fixed. |
| 136 | Typically OEMs provide this firmware from their support site. |
| 137 | |
| 138 | There is also a central site which has links where to download firmwares |
| 139 | for some machines: |
| 140 | |
| 141 | `Thunderbolt Updates <https://thunderbolttechnology.net/updates>`_ |
| 142 | |
| 143 | Before you upgrade firmware on a device or host, please make sure it is |
| 144 | the suitable. Failing to do that may render the device (or host) in a |
| 145 | state where it cannot be used properly anymore without special tools! |
| 146 | |
| 147 | Host NVM upgrade on Apple Macs is not supported. |
| 148 | |
| 149 | Once the NVM image has been downloaded, you need to plug in a |
| 150 | Thunderbolt device so that the host controller appears. It does not |
| 151 | matter which device is connected (unless you are upgrading NVM on a |
| 152 | device - then you need to connect that particular device). |
| 153 | |
| 154 | Note OEM-specific method to power the controller up ("force power") may |
| 155 | be available for your system in which case there is no need to plug in a |
| 156 | Thunderbolt device. |
| 157 | |
| 158 | After that we can write the firmware to the non-active parts of the NVM |
| 159 | of the host or device. As an example here is how Intel NUC6i7KYK (Skull |
| 160 | Canyon) Thunderbolt controller NVM is upgraded:: |
| 161 | |
| 162 | # dd if=KYK_TBT_FW_0018.bin of=/sys/bus/thunderbolt/devices/0-0/nvm_non_active0/nvmem |
| 163 | |
| 164 | Once the operation completes we can trigger NVM authentication and |
| 165 | upgrade process as follows:: |
| 166 | |
| 167 | # echo 1 > /sys/bus/thunderbolt/devices/0-0/nvm_authenticate |
| 168 | |
| 169 | If no errors are returned, the host controller shortly disappears. Once |
| 170 | it comes back the driver notices it and initiates a full power cycle. |
| 171 | After a while the host controller appears again and this time it should |
| 172 | be fully functional. |
| 173 | |
| 174 | We can verify that the new NVM firmware is active by running following |
| 175 | commands:: |
| 176 | |
| 177 | # cat /sys/bus/thunderbolt/devices/0-0/nvm_authenticate |
| 178 | 0x0 |
| 179 | # cat /sys/bus/thunderbolt/devices/0-0/nvm_version |
| 180 | 18.0 |
| 181 | |
| 182 | If ``nvm_authenticate`` contains anything else than 0x0 it is the error |
| 183 | code from the last authentication cycle, which means the authentication |
| 184 | of the NVM image failed. |
| 185 | |
| 186 | Note names of the NVMem devices ``nvm_activeN`` and ``nvm_non_activeN`` |
| 187 | depends on the order they are registered in the NVMem subsystem. N in |
| 188 | the name is the identifier added by the NVMem subsystem. |
| 189 | |
| 190 | Upgrading NVM when host controller is in safe mode |
| 191 | -------------------------------------------------- |
| 192 | If the existing NVM is not properly authenticated (or is missing) the |
| 193 | host controller goes into safe mode which means that only available |
| 194 | functionality is flashing new NVM image. When in this mode the reading |
| 195 | ``nvm_version`` fails with ``ENODATA`` and the device identification |
| 196 | information is missing. |
| 197 | |
| 198 | To recover from this mode, one needs to flash a valid NVM image to the |
| 199 | host host controller in the same way it is done in the previous chapter. |
Amir Levy | e69b6c0 | 2017-10-02 13:38:45 +0300 | [diff] [blame] | 200 | |
| 201 | Networking over Thunderbolt cable |
| 202 | --------------------------------- |
| 203 | Thunderbolt technology allows software communication across two hosts |
| 204 | connected by a Thunderbolt cable. |
| 205 | |
| 206 | It is possible to tunnel any kind of traffic over Thunderbolt link but |
| 207 | currently we only support Apple ThunderboltIP protocol. |
| 208 | |
| 209 | If the other host is running Windows or macOS only thing you need to |
| 210 | do is to connect Thunderbolt cable between the two hosts, the |
| 211 | ``thunderbolt-net`` is loaded automatically. If the other host is also |
| 212 | Linux you should load ``thunderbolt-net`` manually on one host (it does |
| 213 | not matter which one):: |
| 214 | |
| 215 | # modprobe thunderbolt-net |
| 216 | |
| 217 | This triggers module load on the other host automatically. If the driver |
| 218 | is built-in to the kernel image, there is no need to do anything. |
| 219 | |
| 220 | The driver will create one virtual ethernet interface per Thunderbolt |
| 221 | port which are named like ``thunderbolt0`` and so on. From this point |
| 222 | you can either use standard userspace tools like ``ifconfig`` to |
| 223 | configure the interface or let your GUI to handle it automatically. |
Linus Torvalds | 07c455e | 2017-11-18 10:26:57 -0800 | [diff] [blame] | 224 | |
Mario Limonciello | ce6a900 | 2017-09-08 10:23:11 -0500 | [diff] [blame] | 225 | Forcing power |
| 226 | ------------- |
| 227 | Many OEMs include a method that can be used to force the power of a |
| 228 | thunderbolt controller to an "On" state even if nothing is connected. |
| 229 | If supported by your machine this will be exposed by the WMI bus with |
| 230 | a sysfs attribute called "force_power". |
| 231 | |
| 232 | For example the intel-wmi-thunderbolt driver exposes this attribute in: |
Andy Shevchenko | cce1fea | 2017-12-01 15:08:03 +0300 | [diff] [blame] | 233 | /sys/bus/wmi/devices/86CCFD48-205E-4A77-9C48-2021CBEDE341/force_power |
Mario Limonciello | ce6a900 | 2017-09-08 10:23:11 -0500 | [diff] [blame] | 234 | |
| 235 | To force the power to on, write 1 to this attribute file. |
| 236 | To disable force power, write 0 to this attribute file. |
| 237 | |
| 238 | Note: it's currently not possible to query the force power state of a platform. |