4 ms·
I've often wondered why WOL is so poorly designed. WOL should be a function of the NIC, completely independent of the OS and capable of returning status codes e
by lost_soul 5y ago
I've often wondered why WOL is so poorly designed. WOL should be a function of the NIC, completely independent of the OS and capable of returning status codes even if the computer is "off".
- xyzzy_plugh 5y agoIs... this not how your WoL works? Hitting the NIC is equivalent to pressing the on button, boots the OS, etc. I've been using this for years to wake hosts at home when I'm abroad. It's invaluable.
- derefr 5y agoI think the parent means that you should be able to talk to the NIC about the computer, without the NIC needing to get the CPU powered on so that the CPU can tell the NIC what to do. Like a very-minimally-functional BMC.
- hackmiester 5y agoIntel vPro?
- exikyut 5y agoThat's AMT, provided by Intel ME. As far as I'm concerned ME-equipped NICs should unconditionally be behind VPNs, but the AMT functionality itself is really cool. Even the 10+ year old machines that don't support VNC let you remotely hit CTRL+ALT+DEL, issue an ACPI shutdown, perform a hard reset (equivalent to hitting the reset button on 486-era boxen), etc. So basically you can do either soft or hard shutdowns or reboots. Quite granular. I actually used the hard-reset functionality while playing around with Linux early boot on a laptop a few years ago. Rebuild initramfs (loaded via PXE), use `curl` to hit AMT reset endpoint. Very cool.
- Too 5y agoCan it enter the password required to decrypt my LUKS partition? Lately this has been my biggest barrier to leaving a computer fully remote.
- razemio 5y agoYou could use a Mainboard with an IPMI module and KVM-over-LAN functionality. This way your are able to do everything remote, including mounting isos to install a new system and ofc you could use it to enter your password for the encrypted partition.
- escalt 5y agoYou can embed a minimal ssh server (like dropbear or tinyssh) in your initramfs to remotely enter the passphrase. There are some guides for that online
- justin_oaks 5y agoYes, this works. I have it set up for some servers that have encrypted hard drives.
- londons_explore 5y agoStore the LUKS crypto key in the TPM, and make sure secure boot is used so the TPM will only reveal the key if the systems boot chain is verified.
- mschuster91 5y agoNope. That's dangerous, TPM keys can be sniffed on the bus: https://trmm.net/tpm-sniffing/ https://trmm.net/tpm-sniffing/
- brirec 5y agoIs a LUKS key being sent remotely to Intel ME _not_ vulnerable to being sniffed on the LAN?
- Galanwe 5y agoFor the sake of the argument, I think it's more of a shared capability of the motherboard _and_ the NIC, rather than the sole responsibility of the NIC. I wonder if that could be implemented solely by the motherboard. After all, it could supply a small amount of power to the NIC while shutdown and embed a minimal NIC driver to monitor incoming packets without the need for the NIC to do the monitoring itself.
- hackmiester 5y agoI am not sure how this would meaningfully differ from the current state of affairs, where the NIC runs in a low power mode while the PC is off.
- Galanwe 5y agoBecause currently you need support of WoL from both your NIC (to enter low power WoL monitoring mode) and your motherboard (to listen for WoL events from the NIC and trigger power on). If the motherboard would embed a simplistic NIC/WoL driver then you could have WoL on any NIC, with the motherboard acting like any client. I suppose that would mean using the PCI lane and stuff to communicate with the NIC though which may defeat the purpose of a dedicated low power interface.
- qwerty456127 5y agoNow I'm almost curious if every PCI slot has a pin equivalent to the power button pin on the mainboard and also feed some basic background power to the cards even when turnt off. I doubt it actually is this simple though, the industry rarely chooses KISS.
- anonymfus 5y agoIn PCI Express slot there is indeed a standby power (+3.3 V) pin and WAKE# signal pin, but it works differently than the power button. https://en.wikipedia.org/wiki/PCI_Express#Pinout https://en.wikipedia.org/wiki/PCI_Express#Pinout The main difference is that WAKE# signal only restores power to the system but it does not change the system's power state, after using it the card needs to send an interrupt to the motherboard to properly wake it up. Power button does both things.
- eigen 5y agothere is a WAKE# pin on the PCIe slot so any card should be able to wake the system from sleep (S3). however, an ATX PSU supplies only 5V (5Vsb) when the system is off (S5) which is not available on the PCIe connector which only has 12V and 3.3V. so when the system is off (S5), the PCIe cards are unpowered and thus cannot turn the system on. but if you mean off in the Microsoft "Modern Standby" (S0i) then yes, the PCIe cards are powered and the WAKE# pin can "turn on" the system.
- extrapickles 5y agoThere is a pin that provides access to the 3.3v standby bus, and a wake pin that can be used to wake the CPU.