<bslsk05>
imgur.com: The British Kebab Shop - Imgur
<gorgonical>
it's not a proper kebab if you don't get full salad tho
<gorgonical>
this is heresy
<zid>
"heady mix of rat, greyhound and eastern european girls who aren't very good at holding their breath" is one of the best lines commited to.. internet.
<gorgonical>
knoblauch sosse mit KRAUTER
<zid>
knobsauce with cabbage?
<gorgonical>
nailed it
<zid>
german cuisine smh
<gorgonical>
geist do you visit that famous pnw burger place
<gorgonical>
the one that bill gates likes in that famous photo
<geist>
hmmm wich one?
<geist>
now i'm curious
<gorgonical>
dick's
<geist>
oh shit yeah, Dicks
<zid>
I thought they sold sporting goods
<geist>
i figured it might be dicks but had to verify
<geist>
it's like the in n out of seattle
<zid>
Apparently there's a five guys near me
<geist>
yeah those are all over
<geist>
pretty good, not bad
<zid>
not here they're not!
<zid>
american chains are slowly creeping in because there's a lot of brand recognition now from like, memes
<geist>
well, they're not regional in the US anymore, so not surprised they hvae expanded to other countries
alpha2023 has joined #osdev
<zid>
the kids know what five guys is now and would want to try it
xenos1984 has joined #osdev
<zid>
the last big import was domino's
<gorgonical>
I was in the pnw recently and forgot to go and instead went to burgerville
<geist>
next thing you know you'll get a Tim Hortons
<zid>
and papa john's is new
<geist>
i was surprisewd to see one in NYC last time i visited
<zid>
I can name like 30 american chains despite never having set food in the place, it's kind of sad tbh
<gorgonical>
we have nando's now
<zid>
foot*
<gorgonical>
does that count?
<zid>
nando's is south african
<gorgonical>
but it is good
<zid>
it's mediocre, but that's got its own appeal
<zid>
it's a cheap place to get coated chicken
<gorgonical>
and we have the scourge on taste that is jollibee
<bslsk05>
'David Bowie - I'm Afraid of Americans (Official Music Video) [4K Upgrade]' by David Bowie (00:04:27)
<zid>
KFC is wildly inconsistent in quality here and expensive as hell, something to do with how it's franchised
<gorgonical>
welcome to how it is in america, lmao
<zid>
mcdonalds is MUCH more consistent, but corp wants it that way
<zid>
there's still variation, but it's nothing compared to KFC
<gorgonical>
i love the trent reznor cameo in that video geist
<gorgonical>
i have always wanted to go to the uk to try the famed and lauded tesco meal deal
<geist>
yah totally. every time there's some discussion about american culture invading or whatnot, i cant not think of this song
<gorgonical>
and to buy a can of beans for 0.24
<zid>
it's not just tesco
<zid>
everybody jumped on that band wagon
<zid>
even petrol stations and shit do them
<gorgonical>
you have no idea how cheap food is there compared to here
<geist>
and then i get a little sad because david bowie
<zid>
It can be, or it can be stupidly expensive
<zid>
hence everybody going nuts for meal deal action
<gorgonical>
It actually blows my mind that if you are set on it you can get a can of beans for less than a quarter
ManinTheSandbox has joined #osdev
<zid>
do you not know about the BEAN WARS gorgonical
<zid>
They sell them at a loss.
* geist
imagines giant Warhammer 40k style mechas made of a giant bean can
<gorgonical>
The shittiest can of showboat pork n beans here is maybe 0.80 or $1 for a standard ~400cal can
<zid>
It's considered such a vital part of a weekly shop, that the places with expensive baked benas were losing out on £100 shopping refills because the beans were 8p more expensive
<zid>
so there was a race to the bottom
<gorgonical>
u wot m8. Is this for real?
<zid>
they were at one point selling beans for *negative* money
<zid>
100% true, try bean wars in wikipedia
<gorgonical>
geist: isn't the ork killa kan basically that already?
<geist>
see, games workshop is already ahead of us there
<zid>
no wikipedia article :(
<gorgonical>
there was a pork n beans war in maine in 1838
<gorgonical>
which is very on-brand for that part of maine
<geist>
but yteah lots of folks might not know what you mean if you say it
<geist>
so it's one of those, 'use when around fancy and/or educated folks'
<geist>
i do have to admit after having worked at google for so long now i definitely have elevated my speech a bit. too much to be honest
<geist>
i hae to turn it down a bit when talking to non -orkers
<gorgonical>
what
<gorgonical>
"non -orkers"?
<geist>
chiken butt?
<geist>
oh, cow-orkers
<zid>
amongst is the other -st word I was trying to think of
<zid>
cat ass trophy
<Hammdist>
hi all I have made a tiny bit of progress on my device driver for ethernet. I compared the mii register values from my os to those of linux (obtained with mii-tool). one thing that sticks out like a sore thumb is the "power on" bit which refuses to switch from power off -> power on in my os but is in the power on state in linux. any ideas what the
<gorgonical>
I'm having issues with secure/non-secure mode and I'm bitter
<geist>
that's a 3588
<geist>
yah honestly i didn't look close enough at the rockchip stuff to see if a real TRM is available
<kof123>
if you mix fancy words with non-fancy words then you piss them both off
<gorgonical>
there's a trm for the 3399, but certain parts of the board i have, the rockpro64, are not in the trm
<gorgonical>
e.g. my earlier comments about secure/non-secure mode and the exact semantics of res0
<gorgonical>
It is unexpected that res0 is seemingly just "please behave and don't write to these bits, or else badt things will happen"
<Hammdist>
ah there is a schematic after all - looking through it now
<zid>
Language thing americans do that keeps pissing me off while reading my trashy light novels: Not know the difference between venemous and poisonous, and fewer and less.
<zid>
the latter is universal
<zid>
the former is unconscionable
<gorgonical>
I have just re-re-read the manual and now it makes even less sense. "Hardware must ignore writes to this field" for res0
<gorgonical>
Wtf yo
<zid>
That sounds like a complicated way to write "read only"
<gorgonical>
There's extra language that makes it worse
<zid>
oh god, is it a telnet MUST
<zid>
IAC MUST ignore writes to res0 IAC
<gorgonical>
"If software has read the field since the PE implementing the field was last reset and initialized, it must preserve the value of the field by writing the value that it previously read from the field. Otherwise, it must write the field as all 0s"
<gorgonical>
Which means what exactly?
<zid>
PE?
<gorgonical>
processing element, I think
<gorgonical>
yes it's that
<zid>
so you must do a writeback if it has a value you care about
<zid>
otherwise you must clear it after reading it
<zid>
reading corrupts it, basically, it seems
<zid>
so either set it back to what you read, or set it to 0, but don't just leave it
<gorgonical>
but then doesn't that implicitly mean the modification pattern is: read, reconstruct, write-back?
<zid>
reconstruct?
<gorgonical>
If you want to change a bit from 0 to 1
<gorgonical>
You can't
<zid>
I mean technically by the letter of the law
<gorgonical>
Because the only valid modification is to zero
<zid>
you would do a = *p; *p = a; a |= 3; *p = a;
<zid>
but I imagine you can skip the middle write in practice
<zid>
it doesn't say you can *only* write to it after reading it in what you pasted.
<zid>
Just that if you read it, you must write it.
<gorgonical>
It seems like you actually have to do a = *p; *p = 0; a |= whatever; *p = a
<zid>
yes, I just said that
<zid>
well no, *p = a; not *p = 0
<zid>
unless you're doing it off a reset
<gorgonical>
I guess for res0 values you aren't allowed to be writing them to other values, anyway
<zid>
after a reset it wants you to read garbage, then write a 0
<zid>
otherwise it wants you to put back what you found
<gorgonical>
yes
<gorgonical>
I was getting hung up on how this is for individual bits in a register, not the whole register
<gorgonical>
Well if hardware *must* ignore writes to these fields, either my hardware is non-conforming or Linux is performing a secure access
<gorgonical>
Or maybe a third option where linux can read all sorts of wacky nonsense from the register and because those fields are res0 that's "allowed"
<zid>
Couldn't possibly comment, as I have no fucking idea what this device is.
<gorgonical>
I have a rockpro64 that has a rk3399 soc on it, fwiw
<geist>
gorgonical: res0 generally means exactly that
<geist>
as in, 'in the future these may be used but if you write 0s to it it probably will fall back to whatever default it is'
<geist>
that's specifically different from say 'doesn't matter' or 'ignored'
<geist>
these bits may be used in some future version of a thing, or they're undocumented features that they dont want you to use
<gorgonical>
can you share your thoughts on this gic thing? My secure kernel Kitten configures the gicd to set up a timer on core 1, then linux comes up on core 0, reconfigures the gicd which causes me to lose my timer. But the timer is configured as a secure interrupt so Linux shouldn't be able to see the bits to turn it off
<geist>
is linux explicitly running not in secure mode?
<gorgonical>
I *should be*
<geist>
double check that
<gorgonical>
I'm trying to but I can't fucking figure out who handles the smc calls when Linux is running
<geist>
probably ATF
<gorgonical>
I have uboot and arm tf-a running but I can't figure out which handler linux gets into
<geist>
TF-A for almost certain
<geist>
it leaves around the EL3 thing that traps it
<geist>
and if linux is in secure mode then it has full access to the GIC
<gorgonical>
It should be, right. Kitten clearly uses it when it does PSCI discovery as the TF-A printfs show up
<gorgonical>
But I never see TF-A issuing any printfs
<gorgonical>
for linux on core 0
<geist>
what do you mean?
<geist>
why would it?
<gorgonical>
Because I added debug printfs to make sure certain things were happening
<geist>
ah
<geist>
what calls do you expect it to make?
<gorgonical>
And I see the top-level smc handler (or at least what I *think* is the top level) printfs showing up when Kitten makes PSCI smc calls
<gorgonical>
But when Linux makes PSCI smc calls I do not see that same handler's printfs
<zid>
oh *that's* what res0 was, that makes 100% sense as a requirement now.
<zid>
atm it's all 0s, so reset it to 0s, but in future the high bit might be an interrupt acknowledge that clears when read or something, so write it back if you read it without understanding wtf you're doing please.
<gorgonical>
zid: right. it's like a "just leave this the way it is if it isn't 0, please"
<zid>
why you would read it in the first place is the mystery
<gorgonical>
well if you read a double word from gicd_ctlr and like 55 of the bits are res0
<zid>
oh it's various bits? fair
<gorgonical>
yeah
<zid>
Now there are 0 mysteries.
<geist>
yah in general ARM does that pretty consistently. they'll define unused bits in a register as res0, and then future versions of the spec will strt to use the bits
<gorgonical>
slightly different from intel. Usually bits for x86 are just "reserved"
<geist>
but they prett much always default the 0 value to mean 'feature disabled' or 'the way it used to be' such that if old software uses it and wriets zeros to it it is safe
<zid>
tell is telling you not to do a = *p; flag1 = a & FLAG1; flag2 = a & FLAG2; ... *p = flag1 | flag2;
<zid>
but ... *p = a;
<geist>
right, it's ARM being much mroe precise aobout what resserved means
<zid>
s/tell is/
<geist>
there are a few places of res1
<gorgonical>
I wonder if maybe tf-a knows to go quiet on the output under certain circumstances?
<gorgonical>
the thing that makes me sad is that this whole arrangement worked really well on qemu. I had an interkernel channel that userspace programs were using to communicate across the trustzone/secure boundary
<gorgonical>
But then the actual hardware fucked everything up
DrPatater has quit [Quit: Explodes into a thousand pieces]
DrPatater has joined #osdev
qbasiq has quit [Ping timeout: 248 seconds]
qbasiq has joined #osdev
<geist>
ah i was gonna suggest trying it all out on qemu but sounds like you're already well ahead of me
<geist>
i dont really know how TF-A does all the handoff between secure and nonsecure mode
<geist>
i suspect you're not really supposed to futz aournd with the gic from secure mode, so it's just blasting over your changes
<gorgonical>
I am coming up to a lot of architectural difficulties that arise from arm's notion of this "globalplatform" api and the underlying system that supports it, yes
<gorgonical>
an availability probem is one thing and if I have to patch linux to make it cooperative then that can go in the paper, but I have to understand wtf is happening first, you know?
<gorgonical>
from QEMU to hardware the tf-a/uboot configuration is a *little* different so probably something there is to blame (or qemu being sloppy, not unheard of)
qbasiq has quit [Ping timeout: 248 seconds]
qbasiq has joined #osdev
qbasiq has quit [Ping timeout: 248 seconds]
sham1 has joined #osdev
mhall has quit [Quit: Connection closed for inactivity]
gabi-250 has quit [Remote host closed the connection]
gabi-250 has joined #osdev
elastic_dog has quit [Ping timeout: 246 seconds]
elastic_dog has joined #osdev
lg has joined #osdev
Vercas has quit [Remote host closed the connection]
Vercas has joined #osdev
gabi-250 has quit [Remote host closed the connection]
gabi-250 has joined #osdev
benlyn has quit [Remote host closed the connection]
foudfou has quit [Remote host closed the connection]
foudfou has joined #osdev
zxrom has quit [Quit: Leaving]
gxt has quit [Remote host closed the connection]
gxt has joined #osdev
Brnocrist has quit [Ping timeout: 255 seconds]
Brnocrist has joined #osdev
danilogondolfo has joined #osdev
Burgundy has joined #osdev
mhall has joined #osdev
nyah has joined #osdev
zxrom has joined #osdev
GeDaMo has joined #osdev
Jari-- has joined #osdev
zhiayang has quit [Quit: oof.]
zhiayang has joined #osdev
zhiayang has quit [Client Quit]
zhiayang has joined #osdev
zhiayang has quit [Client Quit]
zhiayang has joined #osdev
Left_Turn has joined #osdev
gog has joined #osdev
mahk has quit [Quit: WeeChat 3.0]
mahk has joined #osdev
Burgundy has quit [Ping timeout: 250 seconds]
goliath has joined #osdev
ManinTheSandbox has joined #osdev
ManinTheSandbox has quit [Max SendQ exceeded]
ManinTheSandbox has joined #osdev
ManinTheSandbox has quit [Remote host closed the connection]
ManinTheSandbox has joined #osdev
terminalpusher has joined #osdev
Burgundy has joined #osdev
ManinTheSandbox1 has joined #osdev
terminalpusher has quit [Remote host closed the connection]
ManinTheSandbox has quit [Ping timeout: 250 seconds]
Turn_Left has joined #osdev
terminalpusher has joined #osdev
Left_Turn has quit [Ping timeout: 246 seconds]
<Hammdist>
so I'm trying now to get i2c working, in order to communicate with the pmic. I noticed the clock divider has a wrong value. however, it is resistant to attempts to set it. I noticed one of the i2c clocks was off so I enabled it in the gate register, but still the clock divider cannot be set
Jari-- has quit [Read error: Connection reset by peer]
<Hammdist>
so I ended up simply bulk clearing all the clock gates and managed to set the divider
eddof13 has joined #osdev
eddof13 has quit [Client Quit]
alexander has quit [Quit: ZNC 1.8.2+deb3.1 - https://znc.in]
alexander has joined #osdev
eddof13 has joined #osdev
eddof13 has quit [Client Quit]
eddof13 has joined #osdev
terminalpusher has quit [Remote host closed the connection]
eddof13 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
yoyofreeman has joined #osdev
<Hammdist>
well I managed to read some of pmic registers through i2c. however according to the pmic everything is on already by default, so I hit a bit of a wall there
<Hammdist>
the pmic is an rk805
<Hammdist>
am I correct in assuming that if I can talk to the pmic via i2c, it is not in the so-called "sleep" mode
yoyofreeman has quit [Remote host closed the connection]
heat has joined #osdev
<Ermine>
heat: please remind me: do I need minimal sysroot for onyx?
<heat>
Depends
<heat>
what are you doing'
<heat>
you need the minimal sysroots if you're building a toolchain, else no
<gog>
hi
<Ermine>
Ah, I'm going to use prebuilt toolchain
<heat>
(you can also build the minimal sysroots yourself through a bespoke process that involves building another toolchain with --no-libc)
<bslsk05>
'Laugh Track' by HollywoodLaughTracks (00:00:13)
<gog>
i wanta laugh track where the audience starts chuckling a little, then gradually goes to uproarious laughter, then into coughing and sneezing fits, then being violently sick
<gog>
and then the sounds of madness as reality crumbles into meaninglessness
<heat>
u ok gog
<kof123>
there was kind of an ancient duckman where he joins the cast of friends and then they end up choking eachother, something like that
<kof123>
which probably was not that great, but any sitcom is just begging for that
<kof123>
*they are all ancient
<Ermine>
heat: I guess I can specify toolchain dir as $SYSROOT ?
<heat>
Ermine: bad idea, make clean clears that
<heat>
install it to a separate location and add it to PATH
<Ermine>
Ok
<heat>
(except for clang, clang doesn't need to be added to PATH for REASONS)
<zid>
Ring ring, ring ring.
<Ermine>
So, if toolchain is prebuilt, is it ok for sysroot to be empty?
<Ermine>
Also, kernel.config.example where
<Ermine>
Well, anyway, will just do the thing from gh workflow file