dgilmore changed the topic of #fedora-riscv to: Fedora on RISC-V https://fedoraproject.org/wiki/Architectures/RISC-V || Logs: https://libera.irclog.whitequark.org/fedora-riscv || Alt Arch discussions are welcome in #fedora-alt-arches
rwmjones_ has joined #fedora-riscv
rwmjones has quit [Ping timeout: 260 seconds]
tg has quit [Quit: tg]
jcajka has joined #fedora-riscv
<davidlt[m]> Yeah, that's something to consider for the build farm.
<davidlt[m]> I kinda expect this to get upstream support, but most likely via community effort. Maybe from Alibaba, or partial. They have been contributing a lot.
<davidlt[m]> I do want to see company pushing the effort. This is a reason I somewhat happy with JH7110 and StarFive Tech.
<davidlt[m]> After 1-2 years of talking they seem to understand that. There are learning pains, but that's something you cannot avoid.
<davidlt[m]> They posted a number of patchsets already, they have more WIP and also preparing to send patches for other things, not only the kernel.
<davidlt[m]> So that makes me happy (so far).
<davidlt[m]> If the hardware is stuck with some not that maintained BSP dump I have little interest.
<davidlt[m]> There is a graveyard of ARMv7/ARMv8 SBCs in the box that I am too tired to deal with.
<davidlt[m]> Time to review twitter feed for #RISCVSummit
<davidlt[m]> So Android now officially supports RISC-V
<davidlt[m]> Okay, that probably means there will be unplanned smartphone update at some point :)
<davidlt[m]> I hope videos are posted on YouTube sooner than later.
Kevinsadminaccou has quit [Quit: You have been kicked for being idle]
<varlad[m]> davidlt: Is there a translation layer to translate ARM binaries to RISCV? :v
<varlad[m]> I think that'd be essential for producing consumer Android devices in the short run
zsun has joined #fedora-riscv
masami has joined #fedora-riscv
masami has quit [Quit: Leaving]
zsun has quit [Ping timeout: 256 seconds]
<davidlt[m]> Don't know. Are apps still pre-compiled these days? I assumed that Apple and Android eco-system doesn't use compiled apps these days.
zsun has joined #fedora-riscv
<davidlt[m]> So StarFive did announce JH8100
<Eighth_Doctor> <davidlt[m]> "Don't know. Are apps still pre-..." <- They are not. They're compiled to LLVMIR and the AppStore compiles them to machine code
<Eighth_Doctor> That's what the "optimizing applications" thing is on Android, and on iOS it's done server side
<Eighth_Doctor> <davidlt[m]> "Okay, that probably means..." <- You should probably ask jforbes to enable binder on riscv in the Fedora kernel then :)
<davidlt[m]> Conan Kudo: wasn't there an alternative to binderfs discussed like a couple of months ago?
<Eighth_Doctor> nope
<Eighth_Doctor> that's ashmem you're thinking of
<Eighth_Doctor> we don't have ashmem enabled on the Fedora kernel for ARM
<Eighth_Doctor> and x86
<Eighth_Doctor> we have Binder enabled though
<davidlt[m]> Well Qualcomm has invested into SiFive
<davidlt[m]> and Patrick Little (SiFive CEO) used to work at Qualcomm
<Eighth_Doctor> yeah no surprise really
<Eighth_Doctor> but no matter how much Qualcomm says they want it, we're going to have a screwy ecosystem with RISC-V
<Eighth_Doctor> there's no incentives by anyone to avoid the mistakes of MIPS and ARM
<davidlt[m]> What I can tell is that there is a large interest into RISC-V, but what companies want is the future of RISC-V.
<Eighth_Doctor> * says they don't want it,
<Eighth_Doctor> well, a future of RISC-V
<davidlt[m]> Well, we copied some mistakes from MIPS and ARM ;)
<davidlt[m]> The beginning is always a mess, there is no way around it, I guess.
<Eighth_Doctor> well, sure, but it depends on whether they realize it's a mess or not :)
<Eighth_Doctor> and the problem is, they don't
<davidlt[m]> There wasn't much of interest into that for the 1st ~10 years of RISC-V.
<davidlt[m]> It's all custom, build exactly what you want, etc. Basically freedom.
<davidlt[m]> There is a lot of effort now to work on non-ISA specifications. Have common implementation for specific things, restrain that freedom with profiles, platforms, and other specs.
<davidlt[m]> I hope that today all the boards, etc. are just e-waste and will not require support once all the specs are done.
<davidlt[m]> ISA wise RVA23 is considered "major" one, which is not significantly larger than RVA22.
<davidlt[m]> The expectation was that a new profile gets defined every 2 years, but things are moving a bit faster thus RVA23 will happen.
<davidlt[m]> I think RVI do not expect distros to pick up every "minor" profile, but major one probably.
<davidlt[m]> Another example of non-ISA important spec is RISC-V IOMMU (Rivos Inc, SiFive, T-HEAD, Thales and probably more donated initial specs).
<davidlt[m]> We just got the initial psABI ratified :)
<davidlt[m]> I did expect to hear more about new stuff being ratified at RISC-V Summit :/
zsun has quit [Quit: Leaving.]
jcajka has quit [Quit: Leaving]
tg has joined #fedora-riscv
<varlad[m]> <Eighth_Doctor> "They are not. They're compiled..." <- No idea but, tools like Waydroid still largely need the translation layer to run Android apps.