ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
rasterman has quit [Quit: Gettin' stinky!]
<Company>
DemiMarie: fun fact: height * stride < buffer_size is technically possible. The last row doesn't need to have the full stride
<Company>
the right formula is (height - 1) * stride + width < buffer_size
Mangix has quit [Read error: Connection reset by peer]
Mangix has joined #dri-devel
pcercuei has quit [Remote host closed the connection]
apinheiro has quit [Quit: Leaving]
iive has quit [Quit: They came for me...]
haaninjo has quit [Quit: Ex-Chat]
u-amarsh04 has quit []
amarsh04 has joined #dri-devel
* pinchartl
is confused
<pinchartl>
I'm trying to import dmabufs to be used as textures in opengl
<pinchartl>
I create an EGL image with the linux dmabuf fd
<pinchartl>
and then use EGLImageTargetTexture2DOES
<pinchartl>
that function is defined in both the OES_EGL_image and OES_EGL_image_external extensions
<pinchartl>
using TEXTURE_2D and TEXTURE_EXTERNAL_OES respectively
<pinchartl>
I don't really understand the difference between the two
<pinchartl>
if anyone could enlighten me, I'd really appreciate it
davispuh has quit [Ping timeout: 480 seconds]
alanc has quit [Remote host closed the connection]
NiGaR has quit [Ping timeout: 480 seconds]
NiGaR has joined #dri-devel
alanc has joined #dri-devel
The_Company has joined #dri-devel
<pinchartl>
I think I got it... TEXTURE_EXTERNAL_OES support multi-planar textures, with a sampler that can directly sample NV12 or YUV420. with TEXTURE_2D, I would need to create a texture for each plane, and write a fragment shader to sample all textures and convert to RGB
<pinchartl>
is there a way to list the formats supported by TEXTURE_EXTERNAL_OES ?
Company has quit [Ping timeout: 480 seconds]
<zamundaaa[m]>
pinchartl: yes, the same way you list the supported dmabuf formats in general
<pinchartl>
zamundaaa[m]: how is that done ? :-)
<zamundaaa[m]>
eglQueryDmaBufModifiersEXT
<pinchartl>
thank you
alane has quit []
alane has joined #dri-devel
kzd has quit [Quit: kzd]
kzd has joined #dri-devel
bluetail has quit []
bluetail has joined #dri-devel
kzd has quit [Ping timeout: 480 seconds]
Calandracas_ has joined #dri-devel
dsimic is now known as Guest10522
dsimic has joined #dri-devel
Guest10522 has quit [Ping timeout: 480 seconds]
Calandracas__ has quit [Ping timeout: 480 seconds]
kzd has joined #dri-devel
The_Company has quit []
Company has joined #dri-devel
<Company>
pinchartl: one thing you should know because it's rarely explained: The stride of the individual planes must be a multiple of 256, or creating the external texture will fail on various GPUs
<eric_engestrom>
and you can also make it optional so that if it's not there you just ignore that check (or print a warning if you want)
paulk-ter has quit []
paulk has joined #dri-devel
rasterman has joined #dri-devel
yrlf has quit [Ping timeout: 480 seconds]
yrlf has joined #dri-devel
phasta has joined #dri-devel
jkrzyszt has quit [Ping timeout: 480 seconds]
Calandracas has joined #dri-devel
vliaskov has joined #dri-devel
jkrzyszt has joined #dri-devel
jkrzyszt_ has joined #dri-devel
jkrzyszt has quit [Ping timeout: 480 seconds]
JRepinc has quit [Remote host closed the connection]
classactatcode has joined #dri-devel
JRepin has joined #dri-devel
classactatcode has quit [Remote host closed the connection]
jkrzyszt_ has quit [Ping timeout: 480 seconds]
nicehabithero has joined #dri-devel
kts has quit [Ping timeout: 480 seconds]
<pinchartl>
is the 256 stride alignment a safe superset of all GPU requirements ?
kts has joined #dri-devel
<nicehabithero>
I personally do not remember what i planned for mesa amber branch r300 isa for control flow emulation, but i remember GCN cards had a comment that soft-clause can replay things, inspecting the clause of memory loads yet deeper might really lead back to command buffer memory done or thread done interrupts, which naturally not occuring on r300 as hoped. But it all was so long time ago
<nicehabithero>
when i delt with ISAs alike.
<nicehabithero>
So as 945 to 950GMA had pretty much same restrictions as r300 such as now flow control and also sw vertex shader, it is complex to get to perform without very modern featuresets.
<nicehabithero>
but atom processor is something that is manufactured on very nicest convertible tablets in stacks which are also quite affordable.
<jani>
but I claim it's not enough to *make* headers standalone, you have to *keep* them standalone
sguddati has quit [Ping timeout: 480 seconds]
JRepin has quit [Remote host closed the connection]
mehdi-djait3397165695212282475 has quit []
mehdi-djait3397165695212282475 has joined #dri-devel
vliaskov has quit [Ping timeout: 480 seconds]
<linkmauve>
dcbaker, hi, Mesa doesn’t currently build on Rust nightly, probably some meson issue? Here is the error:
<linkmauve>
error: invalid value '1.87.0-nightly' for '--rust-target <RUST_TARGET>': "1.87.0-nightly" is not a valid Rust target, the patch version number must be an unsigned 64-bit integer
<linkmauve>
This is an option being passed to bindgen.
<linkmauve>
bindgen 0.71.1
fab_ has joined #dri-devel
fab_ is now known as Guest10541
sally has quit [Quit: sally]
fab has quit [Ping timeout: 480 seconds]
sally has joined #dri-devel
sally has quit [Quit: sally]
sally has joined #dri-devel
<sally>
It seems there is an issue with mesa 25 on Raspberry Pi 5, some app doesn't work like 'zed editor',
rsalvaterra has quit []
<sally>
When I run this command "vkcube", I got this result "Selected WSI platform: xcb Selected GPU 0: V3D 7.1.7.0, type: IntegratedGpu vkcube: ./cube/cube.c:1486: demo_prepare_buffers: Assertion `!err' failed. Aborted"
rsalvaterra has joined #dri-devel
phasta_ has joined #dri-devel
phasta has quit [Ping timeout: 480 seconds]
vliaskov has joined #dri-devel
pcercuei has joined #dri-devel
<mareko>
karolherbst: note that you can enable ACO for radeonsi+rusticl by setting shader_info::use_aco_amd = true
<karolherbst>
cool
<karolherbst>
but atm I'm testing both. I think the only major regression with aco is int24 performance
JRepin has joined #dri-devel
<mareko>
you can report an issue with attached asm from LLVM and ACO (AMD_DEBUG=cs,asm)
phasta__ has joined #dri-devel
phasta_ has quit [Ping timeout: 480 seconds]
<karolherbst>
mareko: the issue isn't the asm, it's that aco doesn't support imul24 and co. Just needs somebody to wire it all up and test it
ransomware7 has joined #dri-devel
<ransomware7>
You are basing your systems on silence where is not understood whether you extort something, while it is readable that something you want from me. But the systems you do not grasp potentially or have not got time to implement, are not very sophisticated, cause the outrun of number systems had been done for it with centuries in a row of decent work. And the girl some talked about she knew
<ransomware7>
i was the best to her, and how much i loved her to still treat me as freak. So it isn't at all looking anything understandable to me as to what you want until you claim or speak about your demands or requirements at all tbh. Dad had been involved with lawless activity but has not been charged and too big order from my behalf to give him in or out, cause his my dad, same thing with a
<ransomware7>
sister, she is still my sister etc. same with another sister, and same with mom, i want to branch on my own yes, but i do not want to punish them unless there is an unvoidable mass murder reason for it due to narcissism from their behalf or among similar lines. So i am entirely clear my own, your positions i do not know.
<ransomware7>
unavoidable, so case closed
frytaped has joined #dri-devel
frytaped has quit [Remote host closed the connection]
frytaped has joined #dri-devel
frytaped is now known as Guest10546
mehdi-djait3397165695212282475 has quit []
mehdi-djait3397165695212282475 has joined #dri-devel
frytaped has quit [Remote host closed the connection]
ransomware7 has quit [Remote host closed the connection]
jkrzyszt has joined #dri-devel
peiderbundychief has joined #dri-devel
<peiderbundychief>
in that thread or pdf only thing that i ever improvised was fixed point rounding, but the text gives it away, i am reading with high pace, it is such a compact paper too, there is no cardinality or world on arrays lilke at floating point, and i got to the very similar presentation as to for them, it can be considered a stable technique, however yes united states's and china's
<peiderbundychief>
companies are selling the whole mack vechicle burden of socketed 486 chips for thousond dollars, that is true, a lot can be done with such even in ukranian Russian war. But who wants to be in a war with china or united states, i do not know, europe has prepared with war less, and that is human kind thinking but they can push further and prepare cause europe has all the capability and
<peiderbundychief>
know how to do that. But war is still not considered humanity wise necessary at all, it gives no benefits.
Company has joined #dri-devel
nerdopolis has joined #dri-devel
peiderbundychief has quit [Remote host closed the connection]
sguddati has joined #dri-devel
nerdopolis has quit [Ping timeout: 480 seconds]
mszyprow has quit [Remote host closed the connection]
mszyprow has joined #dri-devel
kzd has joined #dri-devel
nastywastesmell has joined #dri-devel
<nastywastesmell>
and about boundings limits it actually mentions that on the write column of binary entropy paragraph, that it is allowed. so bounding limit is same term as rounding imo. and it even mentiones that both fixed 2 3 4 or 0.xx 0.yy inputs can be given , which also makes sense entirely.
amarsh04 has quit []
<nastywastesmell>
so it is not entirely complex paper either to start harassing me everywhere.
heat has joined #dri-devel
nastywastesmell has quit [Remote host closed the connection]
u-amarsh04 has joined #dri-devel
fab has joined #dri-devel
Guest10541 has quit [Read error: Connection reset by peer]
kts has joined #dri-devel
sguddati1 has joined #dri-devel
sguddati has quit [Ping timeout: 480 seconds]
davispuh has joined #dri-devel
<alyssa>
eric_engestrom: I saw that, but I want the validation to gate the python script running (so the python can assume that it's getting valid XML in). gating on tests doesn't do that - if you have invalid xml you just get inscrutable build fails that validation would've caught with a much more useful error mesage
pcercuei has quit [Quit: brb]
pcercuei has joined #dri-devel
sguddati1 has quit [Ping timeout: 480 seconds]
rsalvaterra_ has joined #dri-devel
rsalvaterra_ is now known as rsalvaterra
eduardreforminai[m] has joined #dri-devel
phasta__ has quit []
phasta__ has joined #dri-devel
phasta__ is now known as phasta
misshorriblesinger has joined #dri-devel
<dianders>
meltq: Hmm, you're right that "panel-novatek-nt36523.c" is a bit wonky. I guess maybe you could make the macro like this:
misshorriblesinger has quit [Remote host closed the connection]
<dianders>
It's not the most amazing solution, but it seems like the most efficient to preserve the original functionality of the code.
tobiasjakobi has joined #dri-devel
pcercuei has quit [Read error: Connection reset by peer]
trancemaniac has joined #dri-devel
bolson has joined #dri-devel
<trancemaniac>
essentialy the bounds of term n in the array/sequence/distribution requirements are rather loose in the original paper, which makes also sense, the pdf shows that (i.e there are many ways to get uniform output distribution cause the limit bounds for nth terms/elements are pretty high for both base2 or base10 logarithm calculated results in the binary distribution), and possibly there was
<trancemaniac>
no human behind that calculated those things, cause for many many years in a row there are solvers that do that. However all the results seem expected to me. I arrived to same conclusions.
tobiasjakobi has quit []
trancemaniac has quit [Remote host closed the connection]
somethinguseful has joined #dri-devel
<Hazematman>
was `ci_run_n_monitor` changes so the target regex has to match the job name entirely? I remember being able to just do `anv` for the job but now it seems have to do stuff like `anv.*`
<zmike>
I'm reasonably sure you always had to use a regex
somethinguseful has quit [Remote host closed the connection]
<lucaceresoli>
Hello, I wonder whether the R-by jani is enough for applying my small DRM bridge debugfs improvement series:
<lucaceresoli>
lumag: maybe you want to review it as well, as you suggested some of that work?
<lucaceresoli>
Sorry for being pushy, I'd love to have this applied soon-ish because it has (trivial) conflicts with my bridge refcounting series, of which I'm probably sending v7 later this week
Duke`` has joined #dri-devel
eduardreforminai[m] has quit [autokilled: This host violated network policy. Contact support@oftc.net for further information and assistance. (2025-03-03 16:20:13)]
pcercuei has joined #dri-devel
<eric_engestrom>
alyssa: ack, makes sense to make it mandatory then 👍
<alyssa>
eric_engestrom: I have a compromise solution for now
<eric_engestrom>
Hazematman: yeah, the first version was a prefix match and we replaced with a full match
<eric_engestrom>
Hazematman: aka `anv` would match `anv-foo` but not `zink-anv-foo`, which was not the most obvious thing, so making it an explicit ful match forces you to think about what you want to match or not
PeterNyance has quit [Remote host closed the connection]
<eric_engestrom>
alyssa: ok; where is that msg printed?
<eric_engestrom>
I have a feeling it might end up in the meson log and not somewhere visible by default
<alyssa>
eric_engestrom: exactly
<Hazematman>
Thanks for confirming eric!
<alyssa>
it ends up in the log and silent.. unless the script ValueErrors or whatever and then it shows up in stderr
NiGaR has quit []
<eric_engestrom>
alyssa: ah wait, /me missed the grayed out comment, reading it now ^^
NiGaR has joined #dri-devel
<eric_engestrom>
ok yeah I see, so if it works you don't need the dep, but if you have a problem you need it to know what's wrong
<alyssa>
in theory
<eric_engestrom>
sounds like a really good compromise actuall
<alyssa>
i'd love for everyone to have the dep but
<eric_engestrom>
*actually
<alyssa>
I don't exactly have the political capital to spare right now for "add more deps to mesa"
<alyssa>
:p
frytaped has joined #dri-devel
<eric_engestrom>
yeah I get that
frytaped is now known as Guest10553
<eric_engestrom>
you _could_ suggest the dep and if anyone pushes back, not resist and directly suggest this compromise
<dcbaker>
I'm hoping to get some of that pulled back to 1.7.1
tyalie has quit []
tyalie has joined #dri-devel
<alyssa>
trying to push my first DRM patch, can someone sanity check my steps of:
<alyssa>
1. dim checkout drm-misc-next
<alyssa>
2. dim b4-shazam -s <message-id>
<alyssa>
3. ... I guess dim push-branch now?
<alyssa>
I guess I should probably build test the applied thing first just in case?
phasta has quit [Quit: Leaving]
frieder has quit [Remote host closed the connection]
warpme has quit []
vliaskov has quit [Read error: Connection reset by peer]
vliaskov has joined #dri-devel
fab_ has joined #dri-devel
fab has quit [Read error: Connection reset by peer]
fab_ is now known as Guest10555
jsa1 has quit [Ping timeout: 480 seconds]
tzimmermann has quit [Quit: Leaving]
jkrzyszt has quit [Quit: Konversation terminated!]
fomys_ has joined #dri-devel
jsa1 has joined #dri-devel
<alyssa>
i'm glad i did build test becuase there's a conflict, yay!
<alyssa>
eric_engestrom: btw, I'm thinking some drivers will want u16's to save space
<alyssa>
dunno whether that should be a different type="" or just add a new attrib for min/max and infer the correct integer type
<eric_engestrom>
infering sounds risky to me, better to make it an explicit `size=16` or something like that
<linkmauve>
“12:30:58 jadahl> can't be off screen IIRC was the apple hw limitation, or something along those lines. that just makes it impossible to use as a cursor plane”, in Weston we switch to GL cursor everytime the cursor is even partially off-screen, including in-between two outputs. I don’t know which hardware originally caused that.
<alyssa>
eric_engestrom: hmm. at that point might as well just have types `uint16` or something
<eric_engestrom>
alyssa: sure, but your type_map will grow a lot
<alyssa>
it has to anyway to map to uint16_t concrete types?
<alyssa>
so.. meh?
<eric_engestrom>
yeah, maybe that's the simplest solution
<eric_engestrom>
so instead of uint|int|float|bool you'd have uint16|uint32|uint64|int16|int32|int64|float16|float32|float64|bool
<alyssa>
well float is just float32
<alyssa>
well just float
<alyssa>
i think only panfrost uses float stats and fp32 is fine
<eric_engestrom>
but if you can want u16 you might also want f16
<eric_engestrom>
anyway, if it's all in the `type` field, if doesn't matter which ones we add at first, we can always add others later
<alyssa>
dim is screaming at me uhhh
<eric_engestrom>
so long as we don't have to add another field on both sides later
<alyssa>
"Branch setup for the integration repo is borked'
<alyssa>
sima: ^?
* eric_engestrom
has to go, good luck with dim :]
movies_and_games has joined #dri-devel
mehdi-djait3397165695212282475 has quit []
mehdi-djait3397165695212282475 has joined #dri-devel
<alyssa>
indeed .src/drm-tip is not on a branch, what? dim bug? user error?
<movies_and_games>
best joke is that everything is now in the cloud but is fuzzed for me to read only. Man how strong man I am, it impresses even me. It's unbelievable how accurately i managed to do calculations, i am shocked i did this.
<movies_and_games>
I do not actually know what went wrong in my life at some point things turned over, i became very good.
mehdi-djait3397165695212282475 has quit []
<movies_and_games>
even purposely i am not arrogant either, even though my self-determinism kind of forcing me to take me seriously might suggest, just i know i am correct on those things, and i am using such style, i know that cause i am sufficiently old, and logically known each last step or history of your doings, leaves only the correct ones on the table for very long history of taking mental notes.
<movies_and_games>
if i know the adjective arrogant , well let's say i do, then actually that is mis interpretatable, i have more like determined feed of info, rather than being arrogant i am more like so good in my heart that i want you to have that intelligence.
<movies_and_games>
and that intel i give.
<Ermine>
I'm trying to run piglit and getting this: FileNotFoundError: [Errno 2] No such file or directory: '/home/peter/code/piglit/tests/spec/amd_shader_trinary_minmax/execution/built-in-functions/cs-max3-uvec4-uvec4-uvec4.shader_test' --- did i forget to download something?
movies_and_games has quit [Remote host closed the connection]
<sima>
alyssa, might be very old dim setup and need to start over fresh perhaps?
<sima>
we've moved branches around a lot and I think the transition period wasn't entirely perfect
<sima>
so just setup again probably easiest
<alyssa>
sima: I thought I setup like 2 weeks ago..
<sima>
hm ...
<sima>
this should work I think?
<sima>
demarchi, ^^ dim seems to be unhappy with alyssa
jsa1 has quit [Ping timeout: 480 seconds]
<alyssa>
~~feeling's mutual~~ delet
<alyssa>
:P
<alyssa>
(i'm joking, bug aside this was surprisingly painless as far as email-driven workflows go. thank you demarchi and sima ^^ )
<alyssa>
i cd'd into drm-tip and did `git branch --track drm-tip drm-tip/drm-tip && git checkout drm-tip`
<alyssa>
and like.. that seems to work now but also
<alyssa>
wtf
<alyssa>
and the rebuild worked. so i guess i can fix my error now?
<sima>
yeah
<sima>
could be that dim setup broke, it's probably the least used part
<sima>
I currently don't see where/how it creates the drm-tip tracking branch, and without that dim create-worktree indeed doesn't work
<sima>
could be due to ee96f65ba846d but feels a bit too funny if this has been broken for almost a year
<sima>
demarchi, ^^ ideas?
<sima>
javierm, just realized that you've put dim cherry-pick into the getting started section of dim, maybe that's the reason people use it instead of maintainers?
gouchi has quit [Remote host closed the connection]
Calandracas has quit [Ping timeout: 480 seconds]
<javierm>
sima: IIRC I did that because was asked to cherry-pick a commit that was in drm-misc-next to drm-misc-fixes and didn't use cherry-pick, so was asked to extend the docs
<sima>
javierm, yeah the other part of the patch look good, but this one was maybe a bit too enthusiastic
<sima>
I think I'll ping mripard
<sima>
but not around now
<alyssa>
sima: his kernel.org email bounced, is everything ok?
<sima>
I think so, would be news to me
Kayden has quit [Quit: -> JF]
<javierm>
alyssa: his == mripard? Yeah, I talked with him earlier today :)
<javierm>
sima: could be possible that since 2022 (when I posted that patch) it was decided that we should ask drm-misc maintainers to do cherry-picking ?
<sima>
javierm, yeah we've tried to clarify that
<sima>
was kinda always the rule though, I guess just missed that when landing your patch
<sima>
and now it's still there
<sima>
iirc I've chatted with mripard about improving the docs and we concluded it's all good, but missed that one
DodoGTA has quit [Quit: DodoGTA]
<javierm>
sima: got it, sorry for the misleading doc change then :/
<sima>
for the starting guide maybe just a general rule that you have to use the dim tool if it exists, not raw git, is better
<sima>
and then a note that you shouldn't use the ones in the maintainer section but ping maintainers instead
DodoGTA has joined #dri-devel
<sima>
javierm, nah not on you, just noticed and got confused
<sima>
good docs are hard
<javierm>
sima: right and I see that change was before the commiter / maintainer split
<alyssa>
javierm: ok, guess kernel.org acting up then
<sima>
alyssa, I've heard from other cases that kernel.org is struggling with forwarding because it ended up on a spam list or something
<sima>
so it's not k.o bouncing, but the next hop rejecting stuff from k.org