ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Chrome OS Trogdor Devices - Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630 - Lenovo ThinkPad X13s - and various other snapdragon laptops) - https://oftc.irclog.whitequark.org/aarch64-laptops
<HdkR>
Oh yea, the Arch image is old
<HdkR>
Doesn't have X1E bits
<HdkR>
I actually /just/ rebuilt it locally to un-bitrot it
<HdkR>
And I can't test it because the EC freaked out on my Yoga 7x again, so it discharged and I'm not at my desk to reconnect it!
<icecream95>
maz: Thanks for pointing that out, qemu -accel kvm does work fine with a newer kernel... or at least as well as it seems to for others
<icecream95>
NetBSD manages to panic about the architectural timer going backwards somehow...
<HdkR>
Oh wait, I can test the updated Arch image since my Orin board is still alive
<craftyguy>
steev: ya wifi is looking better for me too after removing the GIC thing for pcie4
<icecream95>
Well now time to recompile the kernel with hid-bpf enabled, so I can run my "bpflet" to turn off this horrible RGB keyboard backlight and get get Fn-keys working
<craftyguy>
at least you _can_ turn off the RBG lighting. I have a mouse that seems to be impossible to turn it off
<craftyguy>
it just cycles through a bunch of colors, quickly, non-stop. I want to run it over with a car
<icecream95>
The vivobook does cycling as well, but it goes between low and high brightness so wouldn't be useful if you actually needed it
<steev>
for a while, my x13s kept turning the backlight on, and i still dunno what it was because i default it to off
<HdkR>
icecream95: New Arch image with Mesa 24.3.3 is live as soon as the CDN flushes its cache
<craftyguy>
strange, I haven't seen that
<HdkR>
8
<steev>
9
hipboi has joined #aarch64-laptops
<icecream95>
a
<HdkR>
Sometimes I have a bit of latency. I'm like two VMs and three SSHs sessions over wifi deep, it can't be helped :)
<HdkR>
There we go, all of FEX rootfs images updated with mesa 24.3.3
hipboi has quit [Quit: hipboi]
ungeskriptet has quit [Ping timeout: 480 seconds]
<steev>
craftyguy: yeah, i'm not seeing it here either, but i don't know a good way to do this since it seems like it might be x13s specific
<craftyguy>
maybe Johan has some idea 🤷
<craftyguy>
wifi still going OK though after that dts change though :D
<dgilmore>
hrrm, now I need to look at my usb stick install. grub with the memory limiting patch allos cutmem to work. and touchpat and keyboard to work
tobhe_ has joined #aarch64-laptops
tobhe has quit [Ping timeout: 480 seconds]
hipboi has joined #aarch64-laptops
martiert has quit [Ping timeout: 480 seconds]
martiert has joined #aarch64-laptops
nothorseface has joined #aarch64-laptops
hexdump01 has quit [Ping timeout: 480 seconds]
hipboi has quit [Quit: hipboi]
nothorseface has quit [Ping timeout: 480 seconds]
hipboi has joined #aarch64-laptops
jglathe_volterra has quit [Quit: Leaving]
<_mike>
i cant compile box86 on arch
nothorseface has joined #aarch64-laptops
<_mike>
nothorseface: on the flOptus
<macc24>
you're not going to be able to run it on x1e
<macc24>
it can't run arm32 programs
<_mike>
macc24: bummer
<macc24>
*points at fex*
<_mike>
fex is giving me errors in steam
<_mike>
missing vulkan files etc
<_mike>
amongst others this one stands out
<_mike>
pressure-vessel-wrap[36875]: Internal error: /usr/share/vulkan/explicit_layer.d/VkLayer_MESA_overlay.json is not in /usr/lib/pressure-vessel/overrides/share/vulkan/explicit_layer.d
alfredo has joined #aarch64-laptops
<JensGlathe[m]>
@steev craftyguy I have disabled msi-map on pcie4 for test, its using one irq for all of them then:
<jhovold>
this isn't caused by gic its enablement, but we started getting more reports about it around the time it was enabled
<jhovold>
but the underlying fw (or driver) bug has been there since long before, as I also mentioned in the bugzilla entry
<jhovold>
it's likely that random changes in timing can mask this bug, which may explain why some kernel builds (i.e. not necessarily kernel versions) are more affected than others
<jhovold>
I've only hit this is once in the last year, while you guys have machines and networks that now trigger this daily it sounds like
<steev>
Not just daily, multiple times daily, and when it’s spitting out the kernel spam, the system locks up for a second or 5, then usable for 5 then locks up for 5
<JensGlathe[m]>
I will test both variants, forcing back to one physical interrupt (by renaming irq_names in the dtsi) and only disabling msi-map. The EL2 box was more prone to this, so we'll see
<jhovold>
indeed sounds like something qcom should be prioritising...
<steev>
It would be nice, but no one from qcom has even responded about the bug
<jhovold>
but good to hear that swtiching to the internal msi controller at least works around the bug
<jhovold>
is the difference as noticable for you as it seems to be for craftyguy, steev and JensGlathe[m]?
hipboi has quit [Quit: hipboi]
<steev>
For my side, it has happened since I switched to AT&T (fiber), when I was on Spectrum (cable) I never ran into it - my sister had Spectrum but shortly before Christmas, they switched to AT&T as well, so now it occurs at their place as well
<jhovold>
same router, just different connection?
<steev>
I’ll check Thursday if we do family dinner
<steev>
I’m pretty sure we do though, sadly
<steev>
Or I can text them much later today as it’s currently 3am here
<jhovold>
no rush, qcom has been sitting on their hands for a year and half soon anyway...
<jhovold>
but would be good to know, if just increasing the speed (presumably) affects this
<steev>
The AT&T router I have is a model BGW320-500
<jhovold>
not as surprising that switcing router may affect this
<jhovold>
craftyguy: thanks for testing the mhi patch anyway, you could try disabling the modem in dt as well (e.g. in case the patch is not working as intended)
alfredo1 has joined #aarch64-laptops
<jhovold>
losing the rtc offset when you hard reset the machine is a known issue
alfredo has quit [Ping timeout: 480 seconds]
alfredo1 is now known as alfredo
hipboi has joined #aarch64-laptops
<JensGlathe[m]>
Time is too short yet. An indication at the weekend I think...
<jhovold>
at least it sounds like you're not hitting the issue as frequently as steev (multiple times a day)
<jhovold>
steev: when you wake up, could you try dropping the msi-mask parameter and see if that makes a difference in your setup since you seem you can reproduce this as easily as craftyguy?
<JensGlathe[m]>
One additional data point. My HP X14 has the same chip Id on the WCN6855, its a FastConnect 6900. Never had this HTC Rx issue.
hogliux has joined #aarch64-laptops
<JensGlathe[m]>
Same firmware
<jhovold>
interesting
<jhovold>
same board_id as well?
<JensGlathe[m]>
no
nothorseface has quit [Ping timeout: 480 seconds]
alfredo has quit [Quit: alfredo]
<hogliux>
kuruczgy: sorry for misspelling your name (I typed that on my phone yesterday)
srinik has joined #aarch64-laptops
hipboi has quit [Quit: hipboi]
hipboi has joined #aarch64-laptops
anthony25 has quit [Quit: Quit]
anthony25 has joined #aarch64-laptops
<_mike>
HdkR: yeah i get the usual printout in the console after loading steam, but nothing that points to why it's failing with 'steamwebhelper' is not responding
<_mike>
the steam ui will not be usable
<_mike>
i'm running this in a ubuntu 24.04 podman distrobox on arch with fex in it
<_mike>
basically the same result as if i use FEX 'natively'
<_mike>
VDPAU is a broadcom thing right? it can't find the driver search path for it but that shouldn't matter...
<kettenis>
I noticed the usb_2_dwc3 node in the x1e80100.dtsi file is missing a dma-coherent property
<maz>
icecream95: with the CNTVOFF_EL2 workaround, I don't see any screaming with NetBSD (just installed 10.0)
<icecream95>
So probably 10.0 wouldn't have panicked in the same situation
anarsoul has quit [Read error: Connection reset by peer]
anarsoul has joined #aarch64-laptops
<icecream95>
_mike: You'd need my qemu patch to fix that
hipboi has quit [Quit: hipboi]
hipboi has joined #aarch64-laptops
SpieringsAE has joined #aarch64-laptops
<SpieringsAE>
kettenis: the sd-card reader only shows up once a card has been inserted. Are you seeing the usb root hub? dont know if bsd has something like lsusb
<SpieringsAE>
anthony25: yeah I'm pretty sure usb_2 is not used on the slim7x
<maz>
but I can't tell whether this is build with diags enabled or not.
<icecream95>
maz: Is that with the patch? I haven't tried it with the patch myself yet.
<icecream95>
But I think that build will have the check enabled.
<maz>
icecream95: the CNTVOFF_EL2 workaround? yes, that's the only way I have to make this machine usable.
<kettenis>
SpieringsAE: oj, interesting
<kettenis>
I didn't try with an sd-card inserted
<JensGlathe[m]>
take a UHS-II card, mine doesn't dork with UHS-I (although it says it registers as it) or cheaper. With UHS-II it runs fine.
<SpieringsAE>
I'm guessing the chip they use to read the sd-card simulates a usb plug/unplug event when an sd-card is inserted or extracted
<SpieringsAE>
I do still kind of wonder if that chip has any vregs or so attached to it
anthony25 has quit [Quit: Quit]
anthony25 has joined #aarch64-laptops
<SpieringsAE>
side project question, could running a 6.12 kernel on debian mess with mesa? I am working on an imx8mp system with the vivante gpu, I built a 6.12 kernel for it, debootstrapped a rootfs. But kmscube is showing llvmpipe instead of the vivante gpu
<SpieringsAE>
debian bookworm btw, its mesa version does seem to come with vivante in there
<SpieringsAE>
on alarm it just worked :(
<SpieringsAE>
nvm
<SpieringsAE>
something funky with that display or something
<SpieringsAE>
yay gpu acceleration!
hipboi has quit [Quit: hipboi]
alfredo has joined #aarch64-laptops
hogliux has quit [Quit: Leaving]
alfredo has quit [Ping timeout: 480 seconds]
alfredo has joined #aarch64-laptops
iivanov_ has joined #aarch64-laptops
iivanov has quit [Ping timeout: 480 seconds]
agl has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
agl has joined #aarch64-laptops
shoragan has quit [Ping timeout: 480 seconds]
agl has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
agl has joined #aarch64-laptops
icecream95 has quit [Quit: rcirc on GNU Emacs 29.1]
alfredo1 has joined #aarch64-laptops
alfredo has quit [Ping timeout: 480 seconds]
alfredo1 is now known as alfredo
alfredo1 has joined #aarch64-laptops
iivanov_ has quit [Remote host closed the connection]
iivanov has joined #aarch64-laptops
alfredo has quit [Ping timeout: 480 seconds]
alfredo1 has quit [Ping timeout: 480 seconds]
nothorseface has joined #aarch64-laptops
nothorseface_ has joined #aarch64-laptops
nothorseface has quit [Ping timeout: 480 seconds]
nothorseface has joined #aarch64-laptops
<steev>
jhovold: sorry, i'm dumb, what msi-mask parameter?
<jhovold>
and you still hit the issue multiple times a day?
<steev>
no
<steev>
that's what we're testing :D
<jhovold>
got it
<steev>
with that in, i hit it multiple times a day, however, i don't know what actually causes it to be hit
nothorseface_ has quit [Ping timeout: 480 seconds]
<steev>
i just checked, and overnight i did not hit anything or at least, not in the ~6 hours
<jhovold>
using the gic will increase parallelism and change timings, both which could make it easier to hit the underlying bug
nothorseface has joined #aarch64-laptops
<steev>
often times though, the system is idle (for the most part) maybe a terminal running
<jhovold>
sounds promising
<jhovold>
that when I hit it too the other week
<craftyguy>
Ya can also confirm that making that change to disable GIC on pcie4 has greatly improved wifi stability for me too
nothorseface has quit []
<craftyguy>
jhovold: also, that mhi recovery patch didn't help with my issue resumung from s2idle, I think maybe you were offline earlier when I mentioned this. Not even magic sysrq works when it's wedged, so maybe it's stuck very early in resume, or ?
agl has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
nothorseface has joined #aarch64-laptops
agl has joined #aarch64-laptops
<steev>
also, fwiw, it's not specific to any patches you or i have jhovold; with debian's 6.12.8 kernel i hit it consistently
SpieringsAE has quit [Quit: Leaving]
<steev>
though, i think you only carry the venus patches at this point, unless there are regressions
whattt has joined #aarch64-laptops
whattt has left #aarch64-laptops [#aarch64-laptops]
<JensGlathe[m]>
disabled msi-map is up now on the EL2 box too. It was the one most sensitive to these HTC Rx errors. Stressing it with VM load, let's see what happens
<steev>
i suppose i can checkout our 735 repos, that should be a decent test
<steev>
into the e's and so far all i've gotten is 1 "msdu len not valid"
<JensGlathe[m]>
nothing yet.
<JensGlathe[m]>
Interrupts are all on core 0
<maz>
the joys of a single SPI for all the queues...
<JensGlathe[m]>
looks like it. The dtsi has 4 defined, for the port, but I guess the card is usingonly one.
nothorseface_ has joined #aarch64-laptops
nothorseface has quit [Ping timeout: 480 seconds]
<maz>
no. without the ITS, all the MSIs are funnelled to the RC widget can mux 32 MSIs onto a single SPI.
<maz>
so still multiple queues, but they only bring overhead.
nothorseface_ has quit [Ping timeout: 480 seconds]
iivanov has quit [Quit: Leaving...]
SpieringsAE has joined #aarch64-laptops
<_mike>
yay i have steam running and downloading CS2 to try it out
SpieringsAE has quit [Remote host closed the connection]
SpieringsAE has joined #aarch64-laptops
<SpieringsAE>
kettenis: the dma-coherent attribute does not seem to fix my issue with the sd-card reader, but it doesn't seem to cause any negative effects either
<konradybcio>
SpieringsAE: did you pick's Abel interrupt fixes?
<SpieringsAE>
yep, otherwise it wouldn't work at all
<SpieringsAE>
I had been waiting for those patches for quite a while
<SpieringsAE>
but jeah, /dev/sda, but no partitions
<SpieringsAE>
I want to send in a v7 if that thingy needs to added
<konradybcio>
sgerhold: jhovold ^ I think this is the one case where we can't really sanely describe it.. unless we maybe add a new supply to the wifi chip bindings which would be like soldered-m2-input-thats-stepped-down-later-internally-supply
nothorseface has joined #aarch64-laptops
<JensGlathe[m]>
I have it working on the SnapDragon Dev Kit, with UHS-II only, though
<_mike>
hmm
<_mike>
counter strike 2 is stuck on loading screen
<kettenis>
SpieringsAE: umass0 at uhub0 port 1 configuration 1 interface 0 "Genesys Logic USB Storage" rev 2.01/24.02 addr 2
<kettenis>
and it actually works just fine it seems
<kettenis>
at least I can mount a partition on that sd-card
hogliux has joined #aarch64-laptops
<_mike>
damn it gets so close, i can connect to a game then it closes..
patrickm has joined #aarch64-laptops
patrickm has quit [Remote host closed the connection]
patrickm has joined #aarch64-laptops
jglathe_volterra has joined #aarch64-laptops
<SpieringsAE>
kettenis: okay that is good to hear, I really think I've just missed something in the kernel config but just cannot find why, I'll just label the patch as working but I'm an idiot
<SpieringsAE>
coming to a mailing list near you this weekend
<SpieringsAE>
I should also try more with display altmode, hopefully it will be a three part patch set, usb-a, sd-card, usb dp altmode
nothorseface has quit []
<kettenis>
at this point OpenBSD dosn't really do any clock, reset, power domain and interconnect management
<kettenis>
so it could be related to that
<SpieringsAE>
lol
<kettenis>
at least we know the firmware turns on everything that matters
<SpieringsAE>
but jeah that sd-card is not going to help, different interface
srinik has quit [Ping timeout: 480 seconds]
SpieringsAE has quit [Quit: SpieringsAE]
nothorseface has joined #aarch64-laptops
kettenis has quit [Remote host closed the connection]
nothorseface_ has joined #aarch64-laptops
nothorseface_ has quit [Remote host closed the connection]
<dgilmore>
Dumb question, could the address space of a 64GiB laptop overlap with the address space of the i2c hid devices?
hogliux has quit [Quit: Leaving]
nothorseface has quit [Ping timeout: 480 seconds]
<maz>
dgilmore: i2c devices are not memory mapped.
<dgilmore>
Thanks, that's what I expected. Was a stretch theory
<tobhe_>
dgilmore: the reason for the bluescreen crashes is buried somewhere in the firmware, possibly a bug in the smmu
<maz>
I'd be curious to find out if anyone saw these crashes on 64GB machines when booted at EL2...
kettenis has joined #aarch64-laptops
hogliux has joined #aarch64-laptops
<hogliux>
robclark: anthony25: interesting! when booting in el2 on my slim 7x, the iris driver suddenly rejects the vpu30_p4.mbn firmware with the same error message I posted yesterday. Neither the linux-firmware nor the Windows version work.
hogliux has quit []
nothorseface has joined #aarch64-laptops
<robclark>
I coulda told you that ;-)
<robclark>
well, I think the thing that validates and loads the fw doesn't exist if you boot in el2
<robclark>
(otoh, for chromebooks on sc7180, which boot in el2, venus somehow manages to load the fw and boot the coproc)
nothorseface has quit []
jglathe_volterra has quit [Remote host closed the connection]
nothorseface has joined #aarch64-laptops
nothorseface has quit []
<steev>
jhovold: yes, BIL just confirmed that they have the same router from AT&T that I have
<anthony25>
robclark: And Windows boots in EL2?
<Jasper[m]>
Nope
<robclark>
no, windows boots EL1
jhovold has quit [Ping timeout: 480 seconds]
nothorseface has joined #aarch64-laptops
nothorseface has quit [Remote host closed the connection]
nothorseface has joined #aarch64-laptops
<anthony25>
I need to read more about what does each mode, I thought that windows was in EL2 to be able to run hyperv
nothorseface has quit []
<robclark>
the windows setup is maybe more like zen dom0... there is a secure-launch thing that traps unto gunya (qc hyp) to launch a vm.. slbounce kinda reuses this mechanism
<anonymix007[m]>
Is this not the case? In fact, running `strings` on some drivers suggests that these are actually in control of image validation and stuff, it's just that no one with access to the documentation actually cared enough to implement that.
nothorseface has quit []
<robclark>
hmm, I assumed that windows only goes thru that when a VM is started, but I could be wrong
<robclark>
if windows wanted to be el2 all the time then why doesn't uefi just boot the kernel in el2
<robclark>
I'm pretty sure that wouldn't be allows for protected content, as only the hypervizor is trusted... the kernel is not
<Jasper[m]>
iirc it's only tcblaunch that sits in el2 no?
<JensGlathe[m]>
tcblaunch does the switch to EL2... iow, slbounce initializes the switch to EL2 with tcblaunch as payload which fails, but returns on EL2, if I understood this correctly
<JensGlathe[m]>
The crucial part is to have a signed executabe that switches to EL2. And there aren't that many out in the wild
nothorseface has joined #aarch64-laptops
<tobhe_>
do we know of more than that one?
nothorseface_ has joined #aarch64-laptops
nothorseface has quit [Read error: Connection reset by peer]
nothorseface has joined #aarch64-laptops
nothorseface has quit [Remote host closed the connection]
nothorseface_ has quit [Ping timeout: 480 seconds]