18:41:55 If I do decide to try this PCIe passthrough thing is there a known to work usb serial dongle that reliably works as a console port? 19:28:26 wonko: why would you need a phisical serial port then you not even forward the USB? 19:29:38 I need a console for the NUC itself if I'm going to pass the GPU through 19:31:31 ah, I see. 19:32:04 THere are a couple in the OpenIndiana HCL, those will work for smartos as well https://docs.openindiana.org/community-hcl/components/#serial-controllers 20:58:49 wonko: I'm not sure the USB console stuff works at all to be honest 21:17:49 jclulow: it would surprise me greatly if it did to be honest. 21:18:58 I mean it's theoretically possible for it to work for some purposes 21:19:05 I just don't think it does right now haha 21:19:23 You would not be able to interact with the bootloader for example, or at least early kmdb 21:19:34 Console ttya? 21:19:45 But with fixes it would likely work for the actual system ttymon 21:21:12 With temporary boot environment activation available now though I suspect the cases where you'd need to interact with the boot loader to fix boot are fewer 21:22:20 isn't serial console ttya? 21:22:37 shouldn't it work the same as bhyve 21:22:46 or libvirt? 23:23:16 ttya is, on x86 systems, specifically the PC-compatible COM1 23:23:33 It's a specific kind of UART chip that is attached to specific I/O ports on the CPU (effectively) 23:23:45 a USB serial console is vastly different, and requires a lot more machinery (e.g., a USB stack) to wokr 23:23:47 *work 23:23:50 They're definitely not equivalent 23:39:26 spi uart usb pick something....sparc had nice ttyA ttyB 23:42:16 handshaking, device detection, auto speed negotiation yeah usb going to be more complicated