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
esv_ has quit [Ping timeout: 246 seconds]
esv_ has joined #fedora-riscv
<thefossguy> Its love
<thefossguy> s/love/live
jednorozec has quit [*.net *.split]
esv_ is now known as esv
jednorozec has joined #fedora-riscv
<davidlt[m]> Finally, but it's for Christmas :D
<davidlt[m]> I don't expect anything large until MW passes.
defolos has quit [Server closed connection]
defolos has joined #fedora-riscv
<davidlt[m]> and v6.3.10 is cooking
<davidlt[m]> I hope Sipeed TH1520 board upstreaming goes fast.
<dtometzki> Good morning at the moment i see only a very slow process
<davidlt[m]> You mean for the upstreaming?
<davidlt[m]> I think, the basic support landed relatively quickly (in v6.5 already IIRC).
<davidlt[m]> The majority of IP blocks in there are well known 3rd party IP (that already has Linux support).
<conchuod> I expect the rest to be a lot slower tbh
<conchuod> Clock drivers that don't have someone with a vested interest to review them take forever
<davidlt[m]> Do we have any indication what happens after that basic support?
<davidlt[m]> I don't recall seeing any new patches come in.
<conchuod> A Gentoo dev sent a clock driver.
<conchuod> dlan I think their name is
<davidlt[m]> Stephen Boyd sboyd at kernel.org
<dtometzki> Yes I mean upstreaming
<davidlt[m]> Ah, this is again that person from Vivo.
<conchuod> Ah, dlan did uboot
<conchuod> Sorry
<davidlt[m]> Ah, this is old stuff. That is from May.
<conchuod> Dudes doing it in his free time, /shrug
<davidlt[m]> I feel I could only trust StarFive these days to understand what's needed to do. Even if they are still learning.
<davidlt[m]> and a lot of that is probably because of BeagleV (i.e. early involved with dev community).
<conchuod> This microchip crowd aren't too bad, albeit not trying to make SBCs etc
<thefossguy> <davidlt[m]> "Finally, but it's for Christmas..." <- I no likey that but it is what it is.
<thefossguy> Do they plan to upstream?
<davidlt[m]> No
<davidlt[m]> Some of that will probably come from TH1520 upstreaming.
<thefossguy> 🥹
<davidlt[m]> Well they do have plans to upstream from what I understand, but I don't know why they aren't doing it now.
<thefossguy> Ah okay
<davidlt[m]> At least some of the drivers could sent out for review, or at least RFC.
<thefossguy> Are you looking to buy it? Seems reasonable to me tbh.  
<conchuod> If they want to maintain their platform upstream, they need to get in ahead of whoever else will get a box
<davidlt[m]> You can find demos (e.g. with Fedora 38 running on it), bug reports on PCB (with levels, P1/2, etc.), some manuals, etc.
<thefossguy> davidlt[m]: Yeah. Maybe they’re still figuring out the HW and therefore the delay in shipping?
<davidlt[m]> They don't have the final PCB. There was v1.0, then v1.1, now they are preparing for v1.2.
<davidlt[m]> There are a few issues with NVMe statbility, overheating near ASMedia PCie switch, etc.
<davidlt[m]> Typically it takes a few iterations before you are feel good enough for mass production.
<thefossguy> idk how it works but maybe thats why they haven’t started sending patches upstream yet? 
<davidlt[m]> For the IP block stuff this should matter a lot.
<thefossguy> Ah so maybe that’s why
<conchuod> Vendor trees are initially much less work than upstreaming
<davidlt[m]> Yeah, but getting hardware and waiting 12-18-24 months before most of the software is upstreamed is highly annoying.
<davidlt[m]> I don't need everything upstreaming, but give me at least some basics.
<thefossguy> With this many cores, I’ll just spin up a VM on top of whatever vendor platform they provide.   
<thefossguy> s/this many/these many/
<thefossguy> *Pratham curses English phonetics as a Sanskrit admirer
<davidlt[m]> There is no hypervisor support :)
<thefossguy> 😬
<davidlt[m]> There are no SoCs available today that would support hypervisor.
<thefossguy> I forgot to check for the H extension 
<davidlt[m]> And hypervisor extensions alone is most likely not enough to provide decent VM performance.
<thefossguy> Ok now I do want support upstreamed asap lol
<davidlt[m]> These are the same T-HEAD cores (the original ones).
<thefossguy> Oh okay
<davidlt[m]> Only C908 (lowest end) is updated.
<davidlt[m]> It claims to be RVA22 + optional V 1.0.
<davidlt[m]> RVA22 does not mandate hypervisor support IIRC.
<davidlt[m]> And you most likely want a new interrupt controller and IOMMU from RVI.
<davidlt[m]> IOMMU is almost here, the next step is approval for ratification process IIRC.
<davidlt[m]> Same with Advanced Interrupt Architecture (AIA).
<conchuod> That things just a fat SBC
<conchuod> Also, distro folks and users are the only ones with leverage in terms of upstream, however limited that leverage may be
jcajka has joined #fedora-riscv
<davidlt[m]> I have a graveyard of ARM64 SBCs (actually, sadly, some unpacked ones) because I got too tired building custom kernels and distribution to keep them running for my own projects.
<davidlt[m]> Hardware alone means very little for me.
<conchuod> Building custom kernels is fine for me, having 500 different ways to flash things is much more annoying
<conchuod> Having to replace vendor u-boots that are not even close to upstream u-boot is painful too
<davidlt[m]> At some point you get tired. If you run X different SBCs, and new distro version / kernel shows up. So much work.
<conchuod> Although, custom out of tree kernels I cannot be bothered with :)
<davidlt[m]> Most of these will be out-of-tree vendor kernel.
<davidlt[m]> If you also need to rebase those, etc. that's even more annoying.
<conchuod> I already rebase one vendor tree, and I get paid for that. No more pls.
<davidlt[m]> Just so much time wasted. I could just saved money (at a time) just buying x86_64 Intel Atom stuff.
davidlt[m] has quit [Server closed connection]
davidlt[m] has joined #fedora-riscv
<davidlt[m]> Seems to be 6.4-rc6 based.
<davidlt[m]> This should track Linus master branch.
<davidlt[m]> Based on their latest issues there definitely is a problem with PCIe stuff.
<davidlt[m]> Like if a GPU is plugged in into PCIe slot 1, and you dd between two SSDs the system crash.
<davidlt[m]> And similar issues, e.g. system crashes while playing supertuxkart.
<davidlt[m]> They have plenty of time to discover and fix software or/and PCB issues, or discover erratas is various chips :)
_whitelogger has joined #fedora-riscv
Ferix_afk has joined #fedora-riscv
<conchuod> Being based on ~mainline gives me hope davidlt[m]
fuwei has joined #fedora-riscv
jcajka has quit [Quit: Leaving]
jcajka has joined #fedora-riscv
fuwei has quit [Ping timeout: 240 seconds]
fuwei has joined #fedora-riscv
somlo_ has quit [Remote host closed the connection]
fuwei has quit [Quit: Konversation terminated!]
<davidlt[m]> Holy Moly, I am back!
<davidlt[m]> What happens when doing Fedora 37 -> 38 upgrade your computer powersoff just 10-15% in the upgrade? A lot of fun 🤡
somlo has joined #fedora-riscv
<davidlt[m]> All that work on SBCs gives you a lot of skills to recover from this :)
<davidlt[m]> I am still impressed.
<davidlt[m]> and kudos to my aarch64 laptop that helped :D
* davidlt[m] 💃💃🎆
<somlo> davidlt[m]: what sort of aarch64 laptop do you have ?
<davidlt[m]> PineBook Pro with a Rockchip.
<somlo> only $220, interesting!
<davidlt[m]> Nah, it was 300+ a long time ago because all the customs, etc.
<somlo> yeah, it's 219.99 plus $40 shipping from where I'm looking at it (mid-west USA). I *want* one, not sure I *need* one, so I'm conflicted :)
<davidlt[m]> Get macbook with M1 instead. There are sales for that thing in US and it will be a way better investment.
<somlo> oh I have one of those from work :) Don't love it, it's checking with the mother ship each time I try to reinstall the OS, hate its guts for doing that :D
<davidlt[m]> Perf/USD there is nothing to compare here.
<davidlt[m]> You can run Linux on it.
<somlo> but Fedora works great as a guest VM under UTM
<somlo> as for asahi linux, I'm not sure I'm *that* early of an adopter :D
<somlo> once Fedora supports m1 out of the box, I might stop hating it...
<somlo> but who knows how many years that will take...
<davidlt[m]> Conan Kudo: is working on that, there is Fedora Asahi SIG.
jcajka has quit [Quit: Leaving]
<Eighth_Doctor> indeed
<Eighth_Doctor> though it'll be a while that it'll reach mainline
<somlo> interesting, I'll check out the SIG. For now, all I can try is being a guinea pig, if there's any need for that sort of thing :)
<somlo> Eighth_Doctor: so the mailing list has no traffic, and there's mention of a matrix channel -- my boomer question is, can I monitor that from my IRC client? :D (I know it works the other way around, but monitoring stuff using my web browser doesn't come naturally to me...)
<Eighth_Doctor> nope
<Eighth_Doctor> It's only Matrix
<somlo> oh well :(
<Eighth_Doctor> the mailing list asahi@ is not heavily used much because we don't have a release yet
<Eighth_Doctor> but there's preview stuff you can find and try
<somlo> yeah, I was mainly going to open the channel in my irc client and peek if/when there's any activity, to get a feel for where things are and where they're headed
<somlo> logging into something over the web for just that kind of thing is a bridge too far :)
alexsaezm has quit [Server closed connection]
alexsaezm has joined #fedora-riscv
somlo[m] has joined #fedora-riscv
somlo[m] has quit [Remote host closed the connection]
somlo[m] has joined #fedora-riscv
<somlo[m]> guess I'll give this a try, see if s/hexchat/thunderbird/ will be workable...
<somlo[m]> not sure what was with the quit and re-join, and whether that's something that happens often (and spams the "real" irc window for others) :)
esv has quit [Remote host closed the connection]
esv has joined #fedora-riscv
esv has quit [Ping timeout: 245 seconds]
tg has quit [Quit: tg]