<dramforever__>
cousteau: i think it's just sort of infinity. it's the natural result for some hardware implementations of division
<dramforever__>
imagine doing long division, but in binary you only need to subtract the divisor once or not subtract at all at every bit position, so you really only ever check if remaining > divisor, and subtract if it is greater
<dramforever__>
>=, rather
<dramforever__>
but if divisor is zero, you'd just decide to subtract at every position, so it's all ones
<dramforever__>
i don't know if fancier division algorithms do this too
<cousteau>
dramforever__: yeah I was just concerned because the math didn't add up
<cousteau>
but the math did add up, I was wrong
<dramforever__>
la_mettrie: No there isn't any more datasheet to be found, at least publicly. I got this answer when I asked around here a while ago
ghee has joined #riscv
<cousteau>
for some reason, I thought that if the result of div was anything other than 0, then the result of (x/y)*y + (x%y) would be (something other than 0) + (x%y)
<cousteau>
so if x%y = x when y = 0 then you wouldn't get x
<dramforever__>
It's kind of a bummer, if their idea of 'supporting open-source community' really is 'we sent money to canonical and tell their engineers to build kernels', I'm going to be so disappointed
<cousteau>
but, as courmish pointed out, that's not true... no matter what you consider x/0 to be, (x/0) * 0 is going to be 0, and (x/0)*0 + (x%0) is going to be (x%0)
<cousteau>
so (x/0) can literally be any value
<dramforever__>
cousteau: yeah you can also see that if you follow the long division
<dramforever__>
(wrt the starfive kernel again) i don't know if this is just a china thing or what but it's kinda really bothering me? like what the hell
eroux_ has joined #riscv
<dramforever__>
ended up developing a sorta unhealthy relationship with the jh7100 and decided to keep digging and i've put some? most? of my findings here: https://hackmd.io/@dramforever/jh7100-ideas
aerkiaga has quit [Remote host closed the connection]
eroux has joined #riscv
compscipunk has quit [Ping timeout: 256 seconds]
compscipunk has joined #riscv
eroux_ has joined #riscv
eroux has quit [Ping timeout: 268 seconds]
compscipunk has quit [Ping timeout: 256 seconds]
compscipunk has joined #riscv
compscipunk has quit [Ping timeout: 268 seconds]
dramforever_ has joined #riscv
dramforever__ has quit [Read error: Connection reset by peer]
BootLayer has quit [Quit: Leaving]
cousteau has quit [Quit: ♫ I can't forget the day I shot that network down ♫]
crabbedhaloablut has quit [Write error: Connection reset by peer]
crabbedhaloablut has joined #riscv
fedorafan has joined #riscv
raym has quit [Remote host closed the connection]
raym has joined #riscv
crabbedhaloablut has quit [Remote host closed the connection]
KombuchaKip has quit [Quit: Leaving.]
crabbedhaloablut has joined #riscv
crabbedhaloablut has quit [Remote host closed the connection]
crabbedhaloablut has joined #riscv
crabbedhaloablut has quit [Remote host closed the connection]
crabbedhaloablut has joined #riscv
<conchuod>
palmer: are you seeing sparse fails at the moment because of some rcu stuff?
<palmer>
I'm running without errors, but not seeing any other failures (though I didn't look over the weekend)
<conchuod>
I am seeing them with allmodconfig
<conchuod>
But I am trying this Svnapot stuff and I got it there too
* jrtc27
doesn't understand how svnapot is worth it when you can have 2m and 1g pages
<jrtc27>
sure, you don't need 4k pages for the ends, but, those are generally well-aligned too and you really don't need a hardware extension just for that
* conchuod
shrugs
<conchuod>
I just wanna make sure nothing breaks for me :)
<jrtc27>
sure
<jrtc27>
I'm just not rushing to make use of the feature in freebsd...
<jrtc27>
(and am a little sad the precious PTE bits are wasted on this...)
<conchuod>
I now can't repro this on this machine, but sparse is failing 100% of the time with allmodconfig for me in our CI
<conchuod>
(with the minimal config + svnapot0
vagrantc has quit [Quit: leaving]
crabbedhaloablut has quit [Remote host closed the connection]
crabbedhaloablut has joined #riscv
<conchuod>
palmer: it's not the C=1 it is the W=1 that is the problem, nvm
<palmer>
OK, no worries. I'm just merging in rc2 now, everything looks OK so far on my end
<conchuod>
Are other archs able to run W=1 for allmodconfig :?
<conchuod>
palmer: btw, not sure if I got the intent right, but since it's hitting that zicbom warning I folded some stuff into that patch you sent to clear the warning..
<palmer>
arnd would know better than I would on the W=1 bits, since RISC-V has so many different configurations we seem to run into a lot of generic build bugs for the first time
<palmer>
generally there's a lot of testing we don't do and we probably should, I don't usually have the time to fix whatever's broken but I'm happy to add the test to my pre-merge stuff once it's fixed so we don't regress
<atishp[m][m]>
conchuod: Are you going to attend plumbers in person ?
<atishp[m][m]>
We are trying to finalize the RISC-V MC timetable. Just wanted to know if you have any preference for timeslot
* muurkha
imagines a guy in a leather jacket with "RISC-V" and "1%" patches on the back
<conchuod>
atishp[m][m]: yup, I'll be there
<conchuod>
I have zero preference tbh
<atishp[m][m]>
ok cool.
<conchuod>
I live in Dublin atishp so if I couldn't make it...
<atishp[m][m]>
ohh you are local!
<conchuod>
palmer: I've been going through the warnings that pop up in the configs I do run, but there's a lot on the fs/mm side of things and I am just lost there haha
<palmer>
conchuod: ya, that's kind of where I gave up
<conchuod>
atishp[m][m]: 35 mins on a bike from my house to plumbers..
<palmer>
at some point I was trying to chase them all down, it's probably the right thing to do as sometimes it's arch/riscv issues but there's just more every few weeks
matt__ has joined #riscv
<conchuod>
palmer: ye, idk. I just don't know enough to fix em. At least for my small config I know what's expected so I can "complain"/send fixes if there are new ones.
<conchuod>
beyond that, idk. i dont care about sending the wrong patches, at least I learn something - but it takes ages to even get my head around half of it
<palmer>
you can always just post the failures to LKML, someone will probably point out how to deal with it
frkzoid has quit [Ping timeout: 255 seconds]
<palmer>
if you don't want to fix them all no big deal, most folks don't
<conchuod>
Ye that too. Will prob just end up doing that a bunch I suppose.
<conchuod>
Eh, my OCD or w/e will keep me at it whether I want to fix them or not.
<conchuod>
I just like being able to turn on things like Werror..
<dh`>
-Werror isn't on by default?
<nathanchance>
It was for a brief moment but now it is only enabled by default when CONFIG_COMPILE_TEST is set (i.e., allmodconfig and allyesconfig)
<muurkha>
-Werror isn't a good default for distributing software for other people to compile later
<muurkha>
because it stops compiling when they upgrade their compilers
<muurkha>
it's great for development
<conchuod>
The one I don't get is -Winitializer-overrides
<conchuod>
there's like 700 of those on riscv and arm64 builds b/c of compat