<sorear>
SDK = software development kit = generally means a compiler, linker, and sometimes OS components; a SDK will _not_ have HDL
<lumberjack123>
It says XuanTie C906 RISC-V in this doc which IIRC is Apache 2.0 licensed online
<sorear>
ah, that finally got published?
<lumberjack123>
But this is from Allwinner not Alibaba so maybe they added more code and did not release it, it is not copylefted so they can do this legally.
<sorear>
that's going to happen regardless
<lumberjack123>
sorear: What do you mean regardless?
<sorear>
any physical chip you're holding has a whole bunch of components from a whole bunch of companies, even if they use the core unmodified there are things there other than the core
<lumberjack123>
What if it were copylefted?
<lumberjack123>
Or would they just not be able to use the core becayse they can't get the other components from other companies under the same license?
<sorear>
this is a linking situation so copyleft doesn't straigtforwardly apply
<sorear>
not that there's enough legal precedent to know
Sofia has quit [Remote host closed the connection]
Sofia has joined #riscv
jack_lsw has quit [Ping timeout: 260 seconds]
\dev\ice_ has quit [Quit: Reconnecting]
\dev\ice has joined #riscv
\dev\ice has quit [Changing host]
\dev\ice has joined #riscv
abelvesa_ has joined #riscv
abelvesa_ has quit [Client Quit]
davidlt_ has joined #riscv
davidlt has quit [Ping timeout: 252 seconds]
seninha has joined #riscv
elastic_dog has quit [Ping timeout: 240 seconds]
elastic_dog has joined #riscv
davidlt__ has joined #riscv
davidlt_ has quit [Ping timeout: 252 seconds]
jellydonut has quit [Quit: jellydonut]
tgamblin has joined #riscv
BOKALDO_ has joined #riscv
BOKALDO has quit [Read error: Connection reset by peer]
BOKALDO_ has quit [Client Quit]
jellydonut has joined #riscv
jack_lsw has joined #riscv
jack_lsw has quit [Client Quit]
handsome_feng has quit [Quit: Connection closed for inactivity]
zjason` is now known as zjason
<la_mettrie>
trying to understand how these systems work... VisionFive's pages tell you how to compile kernel but no word about making initramfs. actually update-initramfs and update-grub seem both to be absent on the Debian-build i have
<la_mettrie>
at the same time i get errors like this which may suggest that initramfs should be updated: videobuf2_memops: version magic '5.4.61 preempt mod_unload riscv' should be '5.17.0-rc5-starfive-rc5 SMP preempt mod_unload riscv'
EchelonX has joined #riscv
BOKALDO has joined #riscv
<courmisch>
do I understand right that LLVM implements 8- and 16-bit atomics by rounding down the address to 32-bit and then masking the other bits?
Sofia has quit [Ping timeout: 240 seconds]
<courmisch>
so the actual storage size of the atomics is 8- or 16-bit, not 32-bit
Sofia has joined #riscv
compnerd has joined #riscv
Sofia has quit [Ping timeout: 240 seconds]
shoragan has quit [Ping timeout: 256 seconds]
shoragan has joined #riscv
prabhakarlad has quit [Quit: Client closed]
Sofia has joined #riscv
kaph has quit [Ping timeout: 252 seconds]
cousteau has joined #riscv
aerkiaga has joined #riscv
<cousteau>
I'm curious about where did the N extension go. It seems that volume I of the standard moved the N extension to volume II, and volume II just dropped it... Did it get absorbed into another extension, was it temporarily abandoned, what happened?
<cousteau>
sorear: oh OK, so "remove it for now until we've decided what it'll actually look like"
<cousteau>
I thought interrupts were on a more mature state
<dh`>
the problem with N is that as it was written the OS _must_ handle it or it becomes a security problem
<dh`>
or a problem
<cousteau>
I see
<dh`>
though I don't see why it needs anything more than an enable bit somewhere that usermode can't see
<dh`>
(or maybe that it can see but not touch)
<cousteau>
So it was more of a "hey, the way we initially approached this is extremely insecure, we will need to rewrite it from scratch" than a "this way to do things is much more convenient in practice"?
<cousteau>
dh`: maybe they haven't decided yet, and it might end up being just what you said, but until they've thought about it in depth they have preferred to just remove it
mahmutov has joined #riscv
<jimwilson>
lumberjack123, and it isn't V, it is an draft version of V incompatible with the actual V standard
jacklsw has quit [Read error: Connection reset by peer]
lumberjack123 has quit [Ping timeout: 240 seconds]
The_Decryptor3 has joined #riscv
The_Decryptor has quit [Read error: Connection reset by peer]
The_Decryptor3 is now known as The_Decryptor
prabhakarlad has joined #riscv
geranim0 has joined #riscv
cwebber has joined #riscv
<dh`>
having a "V" for "RISC-V" always causes a momentary double-take
<sorear>
cousteau: also N was redundant for its intended purpose. there was always the option to implement S mode and *not* implement paging (this is what bare S means in the commit message), which gives you exactly the same results as the N extension in a M/U system except for two more bits in mstatus
<dh`>
I always figured the goal was more efficient interrupt delivery to microkernel drivers
<sorear>
doesn't really help you for that purpose because if you're intentionally taking interrupts, it means you were doing something else, which means you have to go into the kernel to do a context switch
<sorear>
it _would_ help for things like hardware assisted GC write barriers
<cousteau>
sorear: interesting. I'm not familiar with the details, but I expected paging to be pretty much what a supervisor does..
<cousteau>
("paging" is virtual memory, right?)
<sorear>
"pretty much" but the individual page table formats are all optional, and there's nothing actually in the spec to prevent you from implementing none of them
<sorear>
(yes)
seninha has quit [Quit: Leaving]
___nick___ has joined #riscv
Andre_H has joined #riscv
cousteau has quit [Read error: Connection reset by peer]
aerkiaga has quit [Remote host closed the connection]
davidlt__ has quit [Ping timeout: 250 seconds]
radu242 has quit [Ping timeout: 256 seconds]
kaph has joined #riscv
BOKALDO has quit [Quit: Leaving]
radu242 has joined #riscv
sobkas has joined #riscv
radu242 has quit [Ping timeout: 240 seconds]
radu242 has joined #riscv
mahmutov has quit [Ping timeout: 268 seconds]
vagrantc has joined #riscv
zjason` has joined #riscv
zjason has quit [Ping timeout: 256 seconds]
radu242 has quit [Ping timeout: 256 seconds]
___nick___ has quit [Ping timeout: 252 seconds]
radu242 has joined #riscv
mahmutov has joined #riscv
FL4SHK has joined #riscv
radu242 has quit [Ping timeout: 252 seconds]
radu242 has joined #riscv
seninha has joined #riscv
radu242 has quit [Ping timeout: 256 seconds]
FL4SHK has quit [Ping timeout: 256 seconds]
mahmutov has quit [Ping timeout: 256 seconds]
radu242 has joined #riscv
radu242 has quit [Client Quit]
JanC has quit [Read error: Connection reset by peer]
JanC has joined #riscv
radu242 has joined #riscv
cwebber has quit [Ping timeout: 240 seconds]
aerkiaga has joined #riscv
<jrtc27>
courmisch: yes, and gcc just calls into libatomic which does exactly the same thing internally; note it's an outstanding issue that gcc doesn't just do the same as llvm
EchelonX has quit [Ping timeout: 256 seconds]
aerkiaga has quit [Remote host closed the connection]