PoC 都嘛能動,放到實機就各種摸不著頭緒的黑盒子… #tianocore #edk2 #uefi https://github.com/hiroshiyui/netbsd-labs/tree/main/labs/0-build_edk2

@lopta @clf

I can likewise confirm, albeit that I'm using a wired network, that #NetBSD aarch64 also works on a #RaspberryPi 4, bootstrapped from #TianoCore.

https://github.com/tianocore/edk2-platforms/tree/master/Platform/RaspberryPi/RPi4

edk2-platforms/Platform/RaspberryPi/RPi4 at master · tianocore/edk2-platforms

EDK II sample platform branches and tags. Contribute to tianocore/edk2-platforms development by creating an account on GitHub.

GitHub

I'm possibly being unfair to CP/M there. Its user interface was on occasion better than u-boot's.

But in the early 1980s IBM came along with this program named SETUP for its PC. It was menu driven and full-screen.

Although originally loadable from floppy disc, by the 1990s it was generally in ROM, and had colours and used function keys and had list boxes and had little help messages that were even sometimes helpful.

By the 1990s, IBM had invented its Boot Manager, originally with OS/2 and PC-DOS but later cloned by others, that sat in a type 0A partition and gave a menu of bootable things, complete with human-readable partition descriptions that one could edit with FDISK and LVM.

And now it's 2025, and for machines that post-date the 1990s by a decade, I have a choice between something that sits in boot virus areas and names partitions "(hd0,msdos2)" and something that gives one flashbacks to PIP and ERA.

Or Microsoft Windows.

#EFI #UEFI #bootstrap #DUET #TianoCore

It appears that there was a golden age when #DUET existed, during which if one had an old PC98 machine one could blam DUET into (say) an EFI System Partition, mark that as bootable and either using one of the world's many Master Boot Records for MBR-style partitions or my code for EFI partition tables, boot straight into what was mostly #TianoCore and thence in turn into a modern EFI-compatible operating system, via its EFI boot loader.

https://superuser.com/a/619386/38062

But all of the hyperlinks to DUET by Rod Smith et al. have rotted, and DUET has quietly been discontinued, some time in the late 2010s.

So if you have a PC98 machine nowadays, it's either back to fixed-location stages and map files with GRUB et al., or back to the user interface of CP/M with u-boot.

#EFI #UEFI #bootstrap #TianoCore

What can TianoCore's DUET be used for in a PC?

I've recently become interested in DUET, Intel's UEFI emulator for developers. I'm reading that it can be installed to a hard disk, but what does this make it? Some kind of boot monitor? A virtual

Super User

@ptribble

And yet the downloads page explicitly states "All #OmniOS releases are for the x86-64 architecture only.".

* https://omnios.org/download.html

They could do with letting the world know that this experiment exists, instead of hiding it in a tree of its own on a downloads site that isn't linked-to.

Then people like me casting about for some #RaspberryPi diversity could find it. (-:

As it is, the operating system's own site doesn't mention it, and the search engines turn up dead end stuff like this as top results:

* https://solarisdesktop.blogspot.com/2013/02/illumos-on-raspberrypi.html

I don't expect Wikipedia to be accurate when it comes to this stuff, but the operating system's own WWW site could at least mention that there's a port even if it adds a lot of caveats about unsupported, best-effort, tier 2, and so forth.

That looks like an EFI bootloader. I wonder whether I could use it from #TianoCore.

Gah!

I'm going to have to resist. I'm *supposed* to be resurrecting my Debian amd64 build machine.

#Illumos

Downloads

illumos based server OS with ZFS, Bhyve, DTrace, Crossbow, SMF and Linux zone support

@jamesoff

I don't know whether you can get #TianoCore onto a 2. I have it on a #RaspberryPi 4.

It just drops straight in as an armstub in the Pi boot volume. Unlike OpenBSD, the NetBSD install image (for Pi 4, anyway) has enough free space for it.

@jmcwhatever is looking into the #NetBSD boot loader spinners, should you spot that.

@jamesoff

If you can get #TianoCore on that, it gives the kernel a pretend real-time clock. It's very noddy, doesn't get updated by the kernel at any point, and only ticks when you're in the SETUP utility, which means that it loses time like nobody's business.

But you get a clock. (-:

@kraxel

https://github.com/tianocore/edk2/blob/master/MdeModulePkg/Universal/Console/GraphicsConsoleDxe/GraphicsConsole.c#L1013

Who do you know that works on #TianoCore?

BSD bootloaders emit the BS character quite a lot, as they use it for progress indications during load, and this is an inefficient way of handling it.

A helper function akin to FlushCursor that just blanks the current cell would be a lot more efficient, and also eliminate the need for what that extra FlushCursor is doing.

edk2/MdeModulePkg/Universal/Console/GraphicsConsoleDxe/GraphicsConsole.c at master · tianocore/edk2

EDK II. Contribute to tianocore/edk2 development by creating an account on GitHub.

GitHub

You're welcome. I cottoned on when there was a very brief flicker of a character at one point, and went looking.

#TianoCore doing that self-recursion for backspace and #NetBSD calling OutputString twice makes enough of a window that the character not being erased yet, can synchronize with the display refresh on rare occasion.

So I expect that some sort of putchar_twiddle() that calls OutputString then SetCursorPosition would likely also cut down twiddling overhead.

#FreeBSD #DragonFlyBSD