<jrtc27>
it behaves like a fifo but has the resource and timing properties of a memory
freakazoid332 has joined #riscv
frkzoid has quit [Ping timeout: 255 seconds]
<cousteau>
Decoupled is pretty much an AXI Stream interface, right? Bunch of signals plus valid and ready
aerkiaga has joined #riscv
<conchuod>
prabhakarlad: I would say don't bother renaming the file, just add the ARCH_RENESAS symbol. Cleaning up the rest of the file should be done separately to your changes
<cousteau>
And I guess the difference between Decoupled and ReadyValidIO is that the former is explicitly required to be decoupled so eg. there won't be a direct combinational path between valid and ready
<conchuod>
tbh I am not sure that there is really any merit to renaming to .platforms in the first place /shrug
cabal704 has quit [Ping timeout: 244 seconds]
<prabhakarlad>
conchuod: I intended the renaming to be a separate patch. But anyway Ill send v3 for now and we could do the cleanup later.
cabal704 has joined #riscv
<conchuod>
It'd be a new series not a new patch
<conchuod>
I just assume that it'll be me doing it since I am the one that has ~all of the existing hardware
<conchuod>
prabhakarlad: nice that the extension stuff is okay :)
<prabhakarlad>
conchuod: thanks one less thing to worry about;)
<prabhakarlad>
For the extension stuff I still need to figure out a way to plugin vendor callbacks to the existing dma-ops implemented in non-coherent.c
erg_ has joined #riscv
dor has quit [Remote host closed the connection]
<prabhakarlad>
s/ non-coherent.c/dma-nocoherent.c
cabal704 has quit [Read error: Connection reset by peer]
<conchuod>
Is the THEAD not already doing that?
<prabhakarlad>
THEAD has zicbo* extension where as the Andes one doesnt.
<jrtc27>
thead doesn't have zicbo*
<jrtc27>
it has its own similar thing
<prabhakarlad>
cmo?
cabal704 has joined #riscv
jobol has quit [Quit: Leaving]
jjido has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
frkzoid has joined #riscv
frkazoid333 has joined #riscv
freakazoid332 has quit [Ping timeout: 240 seconds]
frkzoid has quit [Ping timeout: 244 seconds]
Noisytoot has quit [Excess Flood]
<conchuod>
cache maint. operations
KombuchaKip has quit [Quit: Leaving.]
<conchuod>
Ah mb, I thought you were asking what it stood for /facepalm
Noisytoot has joined #riscv
vagrantc has joined #riscv
<prabhakarlad>
ahh, no i meant they are using the cmo:)
erg_ has quit [Ping timeout: 252 seconds]
<cousteau>
conchuod: well I didn't know what it meant, so thanks
<cousteau>
Question: do Z<x>* extensions always/usually have to do with the <X> extension (where <X> is any letter)?
<cousteau>
I see Zicmo and Zifencei which are "generic" extensions so they start with Zi, Zfinx which has to do with F (and D), all the Zb* extensions which seem very interesting, and then there's Zk
<cousteau>
(and now I want a K extension) :)
<jrtc27>
> The first letter following the “Z” conventionally indicates the most closely related alphabetical extension category, IMAFDQLCBKJTPV.
<cousteau>
Ah, that's what I suspected
<jrtc27>
27.6 Additional Standard Extension Names
<cousteau>
Wait, so K *is* a thing?
<dramforever__>
Zb *was* like B-parts, but that's gone for now so eh
<jrtc27>
K and B don't exist as extensions but I guess the view is they still exist as categories
<cousteau>
dramforever__: what do you mean gone? Zb* extensions are still a thing, right?
<dramforever__>
i mean the big B is gone
<cousteau>
jrtc27: pretty sure B was a placeholder chapter in the spec, maybe they removed it
<cousteau>
dramforever__: ah
<dramforever__>
wait i think they *actually* removed it, right?
<cousteau>
Well, it's still there in the version the specs page linked last I checked (which is pretty old, 20191213)
<cousteau>
dramforever__: they did remove a lot of stuff from B, which kinda disappointed me because I felt some attachment to parts of it
<cousteau>
(specifically, I "invented" the grev instruction, only to learn that it had been invented, documented and implemented like 2 years before... so I was sad to see it gone)
<cousteau>
I see
<dramforever__>
(checks latest pdf real quick)
<cousteau>
The latest "official" "stable" version of the spec is still 20191213 though, right?
<dramforever__>
huh there's still a B chapter placeholder
<cousteau>
Oh
<dramforever__>
i mean, i checked the lastest draft
<dramforever__>
i thought if it's gone from anywhere it must have been gone from the latest draft
Noisytoot has quit [Excess Flood]
<cousteau>
B appears in italics in the preface
<cousteau>
And so do J and... oh, also V? Thought that was ratified or something
Noisytoot has joined #riscv
<dramforever__>
is it okay to rant about how weirdly everything in riscv-{spec,privileged}.pdf is organized? it's not like there's a better one but, ugggh
<cousteau>
So yeah sections B and J still exist. What I think was gone is N
<cousteau>
dramforever__: I hear you
<dramforever__>
maybe it's just me lacking focus, but i swear every time i read them there's something new to learn (or maybe i'm forgetting things as fast as i'm learning them)
<cousteau>
So yeah, N is the one that is gone. Although maybe because it was moved to Vol2
<dramforever__>
it's definitely gone from vol2
<dramforever__>
> Additionally, the following compatible changes have been made since version 1.11:
<dramforever__>
> - Removed the N extension.
<cousteau>
Interestingly, this version of Vol2 does mention that N is "tentatively reserved for User-level interrupts extension", but B and J are simply "Reserved"
<dramforever__>
'interestingly' is *not* a word you want to hear about specs...
<cousteau>
jrtc27: I wonder which version of the spec you were quoting. The ones I have don't have that exact sentence (missing letter K on both)