Crosscompile or not?

I see posts on this forum and it seems most of the people are using native compile on the board instead of crosscompile on some linux box. I decided to time kernel compilation (just kernel, from starfive github repo) and got the following:

  • On the board with SSD: 39 minutes
  • On the box with AMD 5950X (16 cores): 1 minute
  • On the box with AMD 3990X (64 cores): 28 seconds

So I figure there must be a very good reason for doing native compile. But what is it?


have you tried an incremental build on the board?

i did compile the kernel onboard because i didn’t want to setup the environment here and there, and then copy the build and all.
but in my case i changed for cross because the native compile didn’t work well: systemd on the image that i use is crosscompiled (on ubuntu 20.04 or 22.04), and there are enough differences between the gcc’s that it hung during the boot (during the systemd execution).

I always native compile because I’m really love to use LLVM for kernel building, and I can’t find cross-LLVM yet. :sweat_smile:

BTW, the reason I buy 8G RAM version because I want to do native compile on the board.


very easy answer: no 5950x or 3990x around :slight_smile:


I tried incremental build. I just touched few header files. It took 18 minutes, not that speedy.

Dont care about the board really; just configure it and type make -j4 is enough to produce results the next day, mostly successful. So why not to compile natively? It does not eat a dozen watts so quite affordable. I run my board idling most time for example.


Kernel is fully support cross-compile.
If you talk about kernel only, this is OK.

But other software not sure supported.
eg. java1.6 to 1.8 . it is hard to cross build.

So. If you are not build kernel only, native compile is a easy way.
native built depent lesser than cross-compile.


Thanks for pointing it out, I was not aware of this.

I have a PD cable which shows on a small display the Watts used by my Vision Five 2. I have plugged in a USB Hub, with a third Gigabit Ethernet (which is currently not supported as the recent upstream kernel lacks USB subsystem support) and a 2TB NVMe and it draws in idle mode (running gnome desktop) around 5W. I am also running it 24/7. Even with expensive electricity prices in my home country Austria since Ukraine problems started, that device costs in a whole year around 30 Euro max for its power consumption. I guess running a 5950x or 3990x draws a tiny little bit more power…


You are correct. My 5950x system draws about 300W on average (including monitors), but 3990x starts at something like 400W and up, and that does not include power drawn by GPUs which I use a lot for machine learning. Electricity rates in California probably not as bad as in your country, but it’s by no means cheap. I run all my SBCs non stop, but all other systems I suspend when not in use.

I have just switched off one Raspberry Pi 4B in favour of my new StarFive Vision Five 2. Though i am a little bit angry about the non-working nftables and the missing USB subsystem, i believe that the platform has potential as a solid foundation for completely free open-source computing. If the ZSBL ROM boot step is not spoiled with spyware, Trojan horses and other trapdoors and backdoors, the Vision Five 2 can really be successful where the Raspberry Pi 4 is failing (ip4/ip6 packet forwarding and routing not working, no good 64-bit support, too many custom distros). What is lacking in the Raspberry Pi 4 is that basically every specific application needs a specially tailored linux distribution. Since a few years Debian is infiltrated by enemies of open-source and free software, who break one packet after another (with Debian 11 around 10.000 packets were thrown out of the distro) and looses more and more ground to be the only distro necessary for doing all the jobs the Raspberry Pi 4 can be used for. And now the Vision Five 2 has a chance to do it better and maybe establish a solid foundation for free and open-source software and hardware giving people really free computing as a kind of new human right.


The ZSBL rom being closed source atm is what puzzling me most now. Even big player, MilkV decided to release their ZSBL upon release of actual hardware.

There were rumors that SF had made a big deal with some smartmeter company and lots of 7110’s going there securing resources consumption. Researchers already speculated the chip contains form of secureboot hardwired in it, but it is disabled for VF2 dev.boards. Hence why (initially proprietary) spl_tool was needed. The simple one contributed by me is now their property and it’s opensource, but original binary only one before mine contains additional cryptosigning code inside.

I am on almost finish line to disasm zsbl properly, a week probably and I could upload it restricted somewhere and open access to those interested. Since SF already violated few terms of GNU GPL already and shamelessly borrowed drivers from U-Boot into bootrom directly, my hope SF will not try interfere with us, peasants just for our natural curiosity and desire for having completely open yet still capable and versatile platform.


Though i feel what you say about the ROM might be true, you still sound like someone, who is not on the side of free and open-source software and hardware users. I am not wanting to get paranoid here, but encrypting the boot procedure and starting the custom kernel in a kind of fake Supervisor mode and pretending to a custom firmware to be sitting on an untampered board, that is what would be the natural way in our days to infiltrate private users home networks and VPN’s and whatever else by simply selling them hardware with Trojan horses, backdoors and trapdoors included. I guess that is where we are since some years. That is the reality that we have to live with. And the other part of this reality is, that systems attempting such a high level of control and access to devices, where they have no business whatsoever by human rights law (privacy law, secrecy and privacy of communication,…) and maybe also by the terms of business under which such devices were sold to customers, will inevitably enter the realm of crime: rape, murder and theft is their natural habitat be it state departments of intelligence or private maniacs, who try to have total control and power over some users hardware and software.

Thus, in order to solve such questions, it needs: legally binding terms of business for trading and selling and producing hardware, e.g. legal safe use standards for hardware, signed by producers and guranteed by the state. Then maybe the community should also start to think, what is actually the purpose of such high infiltration of our computational devices by spyware, malware,…? Not everyone in this world is abusing your device for cryptomining (some companies selling you hardware might even do it legally without you knowing by the terms of business your hardware was sold to you).

Personally, i made my experiences with spooky and improbable things happening in my personal home network, which would everyone make to feel as if being in a horror movie, where it becomes clear that some entity is there, which wants to demonstrate to you its all-powerfulness over your network and computation devices. Let us assume, you actually are completely infiltrated and your every move on your network and computer is protocoled, logged and observed by some outsiders. Then: Who could that be? Is it state entities, who have pissed off the geniuses of their country and who want to find ways to do some NLP and social hacking in order to FORCE them against their human and social rights to work for them, because through all their crimes and violations of human rights they have been uncapable of simply writing a letter, where they make a nice offer to their geniusses according to their real standing (despite what communism and savage capitalism think, there IS actually some objective value of knowledge, understanding and the human possessing it)? Or private entities, who do not want to bother to respect anti-slavery and anti-torture human rights?

If we look into the problem of establishing a good (uninfiltrated) basis for our computing devices on a deeper level, it is maybe our own subconscious, which we are struggling against, when we try to get rid of spyware, malware, Trojan horses, backdoors, trapdoors,… Maybe it is a matter of integrating this dark omnipresent part of us into our way of computing, such that we can actually be happy in a world like ours where DRM and police states with a thousand eyes are getting too strong.

From my personal experience i can share one remedy against people abusing their power over such infiltration systems and who simply conspire to destroy an individual by putting up a wrong case against this individual in order to be able to “justify” infiltrating the individuals network and computer systems by backdoors and trapdoors and other Trojan horses restricted for government use only and only under circumstances clearly listed in the exceptions for human rights policy (war, plague or other situations endangering the state as a whole or certain crimes,…). As Jacob was struggling against god in the book of Genesis in the bible and he won over him, in the same way it is possible to overcome ones infiltrators by wrestling them down and dominate them instead of being dominated by them: Control your controllers, destroy your destroyers, wipe out the enemies of god so to speak. This is the more so true, as people, who establish total control over a persons home network and computers have the burden of seeing everything and this can easily mean “seeing too much, to be left alive”. Even attempts to distribute the burden by shift changes or dividing up competences of the people monitoring a person does not really solve the problem, as the only way to really get rid of the burden is to let loose on the total control over a person. Thus, even if you are under complete control and you are watching child porn all the time, then the reaction of your controllers maybe would not be to turn you over to the authorities, but to restrict your access to such sites (even in networks like Tor) as they actually do not care about you breaking the law (they are themselves violating YOUR inalienable human rights massively by monitoring you), but just about such things appearing in their logs and protocols monitoring you as they want to go around with that logs and protocols to show to others for achieving certain things, e.g. getting legal permissions and justification after the fact for things they have been illegally doing already before…

We remember all how during Corona and now during the Ukraine case our human rights are stretched, state properties all around the world have been systematically raided and abused leaving a large part of a states people deprived of necessary state protection. And now slowly the tide is turning and the failures of the states will bounce back on those, who have tried to uproot the states primary functions to protect its weak members.

It is also possible to put up arguments on the level of security and cleanliness from infiltration one can achieve by cryptographic protocols commonly applied in our computer (networking) protocols as e.g. Bruce Schneier explains in his famous book “Applied Cryptography” (Applied Cryptography ( Bruce Schneier) : Free Download, Borrow, and Streaming : Internet Archive). The general concept in such issues seems to be that there are too extremes: 1) taking no care at all about security of your communication and 2) securing your communication so much, that there is nothing left to talk about to anyone. It is also a little bit like in the uncertainty principle in quantum mechanics, but just for information technology: either you get complete safety of communication or meaningful messages transmitted over your communication channels, but not both at once. Thus, a trade-off has to be made. Look e.g. at Telegram messenger vs. WhatsApp. Live is a give and a take as my beloved father used to say always, while he was still alive…


You brought up way too much for such a simple topic here, please, don’t overestimate.

I nowhere said or gave an impression that I am against free and open source software. I just gave a little facts that we’re on now.

Look, what you have described is already implemented widely. Say, an average ARM A64 modern phone is literally riddled with hardware security walled garden blobs which disallow you to own them like they’re truly your devices (getting root on average modern Android? Sorry, Google has deep distrust with you covered by bullshit like rooted phone is an open gateway for banking thieves). Boycott ARM simply, buy certified Libre devices. It is expensive and specs are not within even average, but doable.

Or take modern x86_64: UEFI, blobs, all around, SMM / Ring minus two running Minix monitoring firmware inside. It’s well known but nobody cares.

Raspberry Pi was from the very beginning a properietary walled garden. Nobody cared.

Things that in past like Clipper chip were condemned already happened, multiple times. Nobody cared. Those who prefer convenience over freedom don’t get both in very end.

But what are you going to do about it now? Everybody got their Google/Apple chained device, and happy computing amateur times are long gone.

You think RISC-V is a magic escape to freedom? Nope, look at the news: because secureboot BS is trendy, even SiFive does it. Game over. Start hacking.



I known you hate secureboot.
But please remember: user and owner are not have to be the same one.

There always must be a defined way to disable it when it’s not needed, and behave since then like it never existed.

1 Like

But the owner also has only limited rights to his property. If it were really about the protection of the owner, then secureboot should only be an option that the owner can only switch on if necessary.

1 Like

Did you mean “user” can disable it ?

Better solution is thewe have 2-plans can choose: A is can not disable for owner/user is not some one. B is user can disable it.
RISC-V is not have common user porduct. And secureboot is a enterprise feature. plan-A is necessary.

How to determined “unlock” is owner’s decision && is he realy known the result?