Although Google has shown significant progress in recent weeks in improving RISC-V support in Android, it seems that we’re still quite a bit away from seeing RISC-V hardware running certified builds of Android. Earlier today, a Senior Staff Software Engineer at Google who, according to their LinkedIn, leads the Android Systems Team and works on Android’s Linux kernel fork, submitted a series of patches to AOSP that “remove ACK’s support for riscv64.” The description of these patches states that “support for risc64 GKI kernels is discontinued.”
↫ Mishaal Rahman
Google provided Android Authority with a statement, claiming that Android will continue to support RISC-V. What these patches do, however, is remove support for the architecture from the Generic Kernel Image, which is the only type of kernel Google certifies for Android, which means that it is now no longer possible to ship a certified Android device that uses RISC-V. Any OEM shipping a RISC-V Android device will have to create and maintain its own kernel fork with the required patches. This doesn’t seem to align with Google’s statement.
So, unless Google intends to add RISC-V support back into GKI, there won’t be any officially certified Android devices running on RISC-V. Definitely an odd chain of events here.
With cpu arch support, its easy to make assumptions based on small sample size of implementations of the arch. I think Google found one that broke their assumptions and caused them to remove what they had. In software that happens at a smaller scale often enough. Usually behind doors, so no one sees the first attempt at support. Once they re-introduce it someone better versed in it can shed light on why they did it. But I’m guessing it has a lot to do with non-public information that can’t be shared.
Bill Shooter of Bul,
Generic Kernel Image pushes for a stable ABI, which naturally requires the compiler to be stable on all supported architectures. If the compiler doesn’t output stable structure types and calling conventions on RISC-V, then everything in the article makes sense why GKI can’t be supported. But I don’t know the state of android compilers on RISC-V and I’d definitely like to see more details to confirm this.
Alternatively I guess there could be something broken with the hardware itself such that refuse to support it at all. All major superscalar CPUs have gone through a spectre style reckoning where hardware shortcuts can lead to statistical data leaks: intel x86, amd x86, apple ARM, etc. I have no evidence for this, but it’s possible that google found similar weaknesses in RISC-V cpus. Rather than trying to patch their software with suboptimal performance harming workarounds they might just conclude they don’t need to support RISC-V right now.
The employees involved could be be under non disclosure agreements and bureaucratic red tape.
This is a good point. I think the compiler might be GCC. I don’t feel like digging through the Andoid repo to find the bazel file in order to figure it out.
RISC-V is saying support has been upstreamed in Debian, Fedora, and OpenSUSE. so I’m not sure.
My hypothesis is still accountants got a hold of it.
https://risc-v-getting-started-guide.readthedocs.io/en/latest/linux-introduction.html
Or unemployment.
It could have something to do with politics in the US:
https://www.reuters.com/technology/us-china-tech-war-risc-v-chip-technology-emerges-new-battleground-2023-10-06/
Morgan,
Interesting theory, Maybe there’s nothing wrong with android RISC-V and it’s just political. It does seem plausible that our vindictive politicians would push against open architectures. Openness naturally hampers their ability to control the technology and keep it out of the hands of “enemies”. This quote from your link is very telling…
Another more devious (tin foil hat) reason could be that open architectures have fewer opportunities to hide back doors.
It will be such a shame if politics ends up killing RISC-V because I really want open architectures to flourish.
We say that, but it’s not like we have a way to verify the hardware and firmware we get is as advertised. It’s something we take on faith.
Flatland_Spider,
Couldn’t very high resolution die photographs using microscopes help expose alterations?
I imagine there are labs that have the ability to photograph the dies using very high resolution microscopes.
https://www.reddit.com/r/Damnthatsinteresting/comments/iqnuhs/destroyed_cpu_in_electron_microscope_intel_d320_i/
If so, researchers could compare samples from different fabs and investigate differences from known good lithographic patterns. None of this seems easy, but maybe not impossible either. I suspect it would be much easier to get access to the correct lithographic patterns when the architecture is open. With a proprietary architecture everything may be a black box, even if you manage to obtain pictures of the die, you wouldn’t know if they were altered.
Possibly. More likely it was killed by the accountants. I don’t see Sundar Pichai as the kind of person who lets things like morals get in the way of making money.
Google keeps firing people for protesting a contract with Israel.
If it is political, I don’t think it’s about morals. Google is a US based company subject to US laws. If the US government decides to ban import of RISC-V based devices, Google would be throwing money away trying to develop for a platform they can’t support in their home market.
Another angle… Intel, AMD, and Arm are using China as cover to bribe some politicians to effectively ban RISC-V chips. “People who use RISC-V are terrerrists. If you’re not with us, you’re against us. Fredom Fries!” to paraphrase a former US president.
First thing to came to my mind too.
P.s.: We need upvoting system in the comments section!
Two things on this one.
1) There isn’t any shipping RISC-V Android devices, so removing support from the golden GKI probably shrinks the hardware they have to test on.
2) Google just cut 3 programming language teams (Dart, Flutter, Python). Google is not interested in going cool sh*% anymore. They’re in their Ballmer phase, or is it their IBM phase, where they coast on past glory while milking the AdWords cash cow. Gutting the Fuschia team was probably the first sign. They’re going to double down on established tech because because “no one got in trouble for buying Microsoft.” and Sundar needs the line to keep going up to keep Wall Street happy.
Well said F_S. Completely agree. “Balmer Phase” haha, love it. Coined here first!