00:03 timestwo: as for the commodity equipment i expect uart to still work out, the protocol is fairly simple, but there is some pretty big commune against me, i have looked into ports and something is rippling behind the nat, but even without this port hcdp is capable of signaling something, but it can be turned off i assume, since i was able to photograph the nand chip next to cpu, they had death metal code against my attempts in bitcoining
00:04 timestwo: it is a vulnerability of the amt hw
00:04 timestwo: its content protection code, but i only use open source, though i violate youtube so... whatever
00:05 timestwo: this comes from intel patents and things about project ultraviolet, this theory is also pretty simple
00:05 timestwo: the display leds amplify a ghost signal on networking that is hidden in fw
00:13 timestwo: it is possible to jam signals, of course it is illegal in most countries, i only tried against wifi , and was able to get into the boot logo, but some legal authorities also applied a pre-boot fingerprint technology, which can be beaten with spi programmer maybe with uart too, i have a last look at it, but that ghost signaling was adjusted to satellite direct comm and 3g which i am not interested to jam anymore
00:14 timestwo: death metal can do it over 3g and up masts too
00:14 timestwo: pretty much oob shut off the computer
00:25 timestwo: i was biochiped after i got marked very young, classified as some alien to be abused, i kept playing but felt weird, but i did so much training before that event of abuse, i actually know how to treat the ball in soccer , if that is what they meant, it is physical timing of the performance, you can aim to vertical heights by having the performance slightly adjusted in timing, you bring the shoulders bit infront in the middle of the
00:25 timestwo: swing back with the leg , so that ball takes a vertical height , but the performance requires strong neck and back and hand muscles to be held firm, so that there is no random deviation or error, then the passes go very accurate, in pool i have unorthodox style, the most lethal potting technique for my body is, you stretch the lower leg of the opposite of shot hand to front, so you can straing a bit the back tendonds, and hold the
00:25 timestwo: butt or hip pretty high, again neck is pulled back from the same side and kept strong, then you pull the shot hand hip which has a straight leg instead, and measure something like 30cm distance from it, to get a lethal swing and aim at the shot, they tried to injure me, but did not succeed yet, my tendons are near to recover and neck
00:25 timestwo: my main athletic career was in table tennis, where i played many tournaments, but i did others sports, i suppose i can do it to some degree yet, and no longer fight the impossible
00:36 bl4ckb0ne: timestwo: thats a chan about gpu development, you might be better off with a more relevant channel about embedded dev
00:38 timestwo: you know linux puts the uart interrupts for tx and rx on different IRQs so it can not work this way, to nop the pointer connecting load , it needs a microcontroller nrf way to do it, so on zephyr one would need to throw in an app_config.h with bunch of defines, and intrude into the hw with physical uart protocol NOPing the pointers connecting load, and jumping hence to it's code pointed
00:38 zmike: it's just spam, ignore and wait for someone to ban
00:38 timestwo: bl4ckb0ne: it's actually solved , you want the divine
00:38 timestwo: compiler
00:38 timestwo: and it will perform like a miracle, it's authored for LLVM
00:39 timestwo: gpu is just a more resources parallel units cpu
00:39 timestwo: all that algorithm works well on gpus too, even better, cause they have more parallel units
00:39 bl4ckb0ne: fairly sure ive seen that bloke spamming other channels
00:40 timestwo: failry sure, you neved authored a single sane line too
00:40 timestwo: unlike me, who is a big contributor to theories
00:40 bl4ckb0ne: fair point
00:45 penguin42: thanks
00:50 HdkR: kode54: Nice. Sounds heavily untested
00:51 kode54: the uAPI declaration is nasty too
00:51 kode54: instead of actually having any pointer members anywhere, it just has u64 members and uses some intel_48b_address macro to fill them out
00:53 kode54: maybe I'll look into it if they're not interested right now
00:53 kode54: I mean, i915 only compat's a single function
00:53 kode54: so maybe there's one I need to wrap properly
01:24 HdkR: kode54: Pointers as u64 instead of pointers is recommended in uapi. With 32-bit just eating the zeroing of the upper bits.
01:26 HdkR: No idea what that macro will do though
01:33 ishitatsuyuki: yeah, using pointers is how you end up with ioctls requiring manual compat variant
02:22 kode54: the macro zeroes the upper 16 bits of a u64 by doing an up then downshift
02:22 kode54: I need to trace why it crashes
02:23 kode54: luckily it's crashing in userspace and not causing a kp
02:23 airlied: that sounds like a gpu addr vs a cpu ptr
03:02 kode54: aha, yes
03:02 kode54: the API is full of u64_to_user_ptr calls
03:02 kode54: looks like I'll have to adapt this if I want it to work
03:04 kode54: or not?
03:22 kode54: this may just be something I need to fix in Mesa
03:22 kode54: since the uAPI is sound
03:22 kode54: uAPI entirely uses u64 for user pointers
03:23 kode54: the only fix the kernel needed was setting compat_ioctl to drm_compat_ioctl, because all the rest of the DRM functions have arch specific pointers in use that need fixing for several functions
03:23 kode54: whereas the Xe uAPI uses u64 for all user pointers
03:24 kode54: and requires the user space code to just deal with that
03:24 kode54: building a debug mesa to test this
03:30 airlied: most modern drivers should have all ioctls being compat from the start
03:30 kode54: which is good
03:30 kode54: something is failing with a mutex
03:30 kode54: steam: ../mesa/src/util/simple_mtx.h:142: simple_mtx_assert_locked: Assertion `mtx->val' failed.
03:30 kode54: I need to somehow build a mesa that lets me have symbols
03:33 airlied: --buildtype=debug should do that
03:36 kode54: the debug files have no symbols
03:36 kode54: I am using buildtype=debug
03:37 kode54: oh, damn
03:37 kode54: how did that get there
03:37 kode54: --strip
03:38 kode54: it shouldn't even be using that option anyway, since pacman can strip and separate debugging symbols
03:47 kode54: okay, this is what's failing, in iris_bo_alloc
03:47 kode54: if (!bufmgr->kmd_backend->gem_vm_bind(bo))
03:47 kode54: goto err_vm_alloc;
03:47 kode54: which is causing it to try to vma_free while the bo is unlocked
03:52 kode54: okay, DRM_IOCTL_XE_VM_BIND is failing
03:53 kode54: oh how incredibly dumb, the error message printer doesn't include errno
03:53 kode54: that's another rebuild
03:56 airlied: just use strace to find the errno :-P
04:02 kode54: running it under strace makes strace crash
04:02 kode54: debug printf it is
04:19 kode54: oh good, dmesg has kernel printk's for this
04:19 kode54: XE_IOCTL_ERR(xe, !obj &&
04:19 kode54: VM_BIND_OP(op) == XE_VM_BIND_OP_MAP) ||
04:19 kode54: that's failing
04:21 kode54: so gem_handle is null
05:55 HdkR: kode54: `DRM_IOCTL_XE_VM_BIND` was one of the ioctls where the struct was different between 32-bit and 64-bit
05:55 HdkR: If it hasn't been fixed in whatever kernel branch you're using
05:57 HdkR: `drm_xe_gem_create`, `drm_xe_vm_bind_op`, `drm_xe_exec`, `drm_xe_wait_user_fence`, and `drm_xe_vm_madvise` all were different.
06:08 kode54: how?
06:08 kode54: both the kernel and the mesa are using the same xe_drm.h
06:09 kode54: I'm using drm-xe-next from gitlab
06:10 HdkR: kode54: u64 members of structs have different alignments so you get holes when it does stuff like `u32 num_syncs; u64 range;`
06:10 kode54: oh, crap
06:11 kode54: is this already fixed in a branch somewhere?
06:13 HdkR: kode54: No idea. I've reported the issue, don't know if Intel peeps fixed it
06:14 kode54: where?
06:14 kode54: mailing list?
06:14 HdkR: here
06:15 HdkR: I don't do kernel development. No mailing list life for me
06:22 kode54: oh
06:22 kode54: I've made a patch against xe_drm.h
06:23 kode54: it inserts u16 and u32 padding members wherever there was non-64 bit alignment caused by interspersed u16 and u32 members
06:23 kode54: I also have a patch for the kernel that enables .compat_ioctl = drm_compat_ioctl, because that's needed for the generic DRM ioctls
06:24 kode54: the uAPI change needs to be patched against both the kernel and Mesa
06:25 kode54: I'll post them to the issue tracker
06:27 kode54: I see why people don't like to use managed git sites for forking the kernel
06:29 kode54: yeah, I've done a few minor kernel patches before
06:29 kode54: kind of a pain
06:33 HdkR: Indeed it is
06:57 kode54: got it working
06:58 kode54: now pushing MRs for review
07:07 kode54: so glad I have the patience to deal with crap like this
07:27 danvet: melissawen, on the amd per-plane color management, personally I'm not fundamentally opposed to per-vendor at first if there's really no way to get this off the ground
07:28 danvet: but I do think we should try pretty hard first at something more agnostic
07:28 danvet: plus I also think there's been plenty of interest for this in the past
07:28 danvet: vsyrjala, pq, emersion ^^ probably as starting point
07:28 danvet: pinchartl, ^^ not sure but iirc you looked at plane color props in the past too?
07:28 danvet: or wast that someone else
07:29 danvet: agd5f, ^^ (hwentland not around it seems)
07:29 pinchartl: danvet: not really. I think the rcar-du could do it, but we haven't looked at that
07:30 danvet: robclark, did cros look at per-plane color mgmt before then?
07:30 danvet: trying to make my brain work on Monday morning before coffee is ... hard
10:19 dolphin: sound-upstream seems to take forever in dim...
11:05 mupuf: dolphin: that;'s because the entire kernel.org infra is mostly-down
13:22 jani: mripard: tzimmermann: mlankhorst: at this point in time, which branch should I apply this to? https://patchwork.freedesktop.org/patch/msgid/20230406134615.1422509-1-jani.nikula@intel.com
13:23 tzimmermann: jani, commit 0575650077ea appears to be in linus' kernel. so drm-misc-fixes
14:21 kolera83: https://forum.lvgl.io/t/stm32f7-disc-hard-fault-on-every-example/7497 , I think making a board micropython binary with mpy-cross, then call host's python mpremote.py blabla would indeed work, but somewhere i read that asl is nowdays equipped to all hardware not sure if that is true, but then it would make sense to compile to use that interpreter , i.e ask for acpi cpuid , and compile the code on the fly like the default is repl, but
14:21 kolera83: fairly little issue overall. Me i am yeah backing off from development , not sure if guys like penguin42 are like my doctor total veggies without taking meds which is outrageous, it's humiliation to me, that such carrots and i do not want to even insult carrots, are really actually terrorizing me.
14:21 kolera83: it really is such a shame on me.
14:29 kolera83: I mean put yourself into my shoes, i am forced to take meds to ruin my own life and health, you are already 25 years with med staff and alien scientists promoting entire tyranny against me, everywhere i go, guy thanks people for banning me, and it's only thing he does
14:30 kolera83: advertising and promoting abuse to a random person
14:32 zmike: moooooooooooooooooooods
14:33 kolera83: i mean look at this ape, what he does, not a single conversation again on my tail when i go have a vaccation, and its a god summer, i am done with code, man if you start all over again
14:33 kolera83: i pulled enough work, i want to get some freedom too
14:38 kolera83: the state is the best we have ever had if to avoid such people and such promotion of terror against totally innocent people, who push hard with mentals to capture sanity for himself, i mena things are good, why you nasty?
14:40 kolera83: maybe you have troubles, but i do not want to see you in my territory harassing me again, or things go very bad i can tell, i warned several times
14:40 kolera83: if you have some trouble, than talk
14:40 kolera83: like a human
14:44 kolera83: I graduated yeah already now, it took some doing, do not harass me, talking how functional people you are, you are pranks to me
14:44 kolera83: cheers.
14:55 pinchartl: several years ago I would have wondered if this was a bot. nowadays we seem to have reversed the turing test: you can differentiate some humans from machines by how incoherent the humans are
14:57 bl4ckb0ne: "i do not want to even insult carrots, they are actually terrorizing me"
15:03 tzimmermann: ask any 3-yrs old about cooked vegetables. they will agree
15:07 javierm: cooked or raw
15:12 tzimmermann: both, i guess :)
15:12 javierm: yeah :)
15:15 kolera83: pinchartl: you are garbage stop chasing me and talking about me, do your porn with other garbage bags without single sane thought nor acheivement, you are thrash to me, you do not want to appear around me, i said one more time for estonian lampleg or jack or charl or any of your fuckers to stalk me, i kill them
15:15 kolera83: cause my father owns that hotel on the island
15:15 kolera83: and i have enough friends to kill them
15:16 kolera83: you are malfunctioning parrot to me, i do not care about your coherency
15:16 kolera83: get lost take your buggs bunnies away from my areas giving me lessons about gangsters or wankers
15:18 kolera83: we will promptly kill them off, cause i counted over 20 or more assaults at me
15:19 kolera83: i let others to know that such bad jokes stalk me
15:19 kolera83: and assaulted behind the back talking entire nonsense
15:19 kolera83: everyone is fine with the kill off , scum you are
15:20 psykose: take ur meds
15:20 kolera83: psykose: shut your mouth abortion leftover
15:21 psykose: i wish i was aborted
15:21 kolera83: get lost from my territory
15:21 kolera83: take your bugs bunny simon and shit with you
15:21 kolera83: i know, but harassing me is not allowed
15:21 kolera83: that is not my problem that you have unstable moods
15:22 kolera83: i sat 2years in mental insitution cause someone terrorised me, and blew up the domededov airport
15:22 kolera83: you know what shit i have been through or what?
15:23 bl4ckb0ne: are you the one bloke who's missing part of his skull/brain?
15:24 pinchartl: bl4ckb0ne: the best strategy is to not engage. /ignore is your friend
15:25 bl4ckb0ne: yeah, just being curious there
15:25 bl4ckb0ne: ive seen this kind of rants in multiple chans and i wonder if its the same person behind
15:26 kolera83: yet i do not stalk people, i do not act improperly , take your gangster shit out of my territory
15:26 kolera83: and sluts needing for gangsters along with you
15:26 kolera83: i've been forced to clear too many murders and in that shit for long
15:28 kolera83: i do not stop being nice, cause i am a nice and sane person, cause some slut requests people to be gangsters and bad boys
15:28 kolera83: find another tank driver i say, i know what those mud brainless crews are doing anyways
15:32 kolera83: you are not accountable humans, when you open your mouth it's clear that you should not be talking to my dad, cause he might be smaller, but his brain is hell of a more capable one
15:32 kolera83: and hence you lose there too for sure, and you lose period, do not drag me into this
15:33 kolera83: me i put no limits to your porn outside of my areas at all, bang your retarded slut how much you want, me i signed not up with her, retarded human
15:34 kolera83: all my friends despise your words, and terror
15:34 kolera83: you are not ok people
15:38 kolera83: And my therapist is not a doctor, she is a scammer, she looks like a pigonoid, and makes never any sense, she wants to munch chicken, and but does not want to do any work
15:39 kolera83: like what do you expect from such person, miracles?
15:39 kolera83: they do not happen
15:39 kolera83: i just hoped that you can find some enjoyment from programming and theories and science
15:39 kolera83: which would calm you down or give some satisfaction
15:40 kolera83: obviously you do not feel like it, and only keep thinking your porn is really my problem
15:40 kolera83: or your sexual issues with pretty retarded women indeed
15:42 kolera83: me i get calmer when i see, science being the topest during my lifeline
15:42 kolera83: it's at the highest peak
15:43 kolera83: 3d metal printers, bioprinters, such chips, all science figured out
15:43 kolera83: but this is not enough for you , i do not know what you search from the world
15:43 kolera83: only to despise and disguise and complain and ban and actions like conflicts
15:43 kolera83: i do not understand such behavior, but everyone says it is my fault
15:44 kolera83: but i disagree, i do not want to understand such people
15:45 kolera83: and that record as linux was broken too, linux used to be pain in the ass
15:45 kolera83: but now it's not, everything is skyrocketed package management are accurate
15:45 kolera83: such lovely experience
15:47 kolera83: it's that i do not care if vulkan is not too modern api to me
15:47 kolera83: overall everything has been made available, and people think different
15:48 kolera83: my work was to offer alternative to those who need
15:48 kolera83: it was my duty
15:49 kolera83: i do not say that h265 is so bad either
15:49 kolera83: same wise mono and digilentengine and all 3d engines, it's not bad work you know
15:50 kolera83: what i considered my work was, everyone who want to get more modernized ways
15:50 kolera83: that this option would be available without me interference
15:50 kolera83: cause you hate me as i see!
15:52 kolera83: i am graduated in that terms i do not have those tasks anymore
15:52 kolera83: i need to switch to new work, cause those things are done perfectly
15:53 kolera83: Or do you at all understand what divine actually does in it's mstring implementation?
15:54 kolera83: i am deeply happy that czechs did that job, exactly the same as the best solution is
15:54 kolera83: i had exactly the same algorithm
15:54 kolera83: ANd i know, it is the very best, and so do they
15:55 kolera83: stop being such a dicks, i will exactly explain what they do instead...
15:56 kolera83: it's completely amazing compiler, just one hundred percent exactly like i visioned
15:56 kolera83: uart is very simple protocol
15:57 kolera83: but divine is indeed paradise wizdom
16:02 gfxstrand: Does anyone know what bit of spec text requires that flag be applied at vec4 granularity
16:03 gfxstrand: i.e., why I can't have a flat vec2 and a non-flat vec2 sharing the same slot?
16:03 gfxstrand: The Intel compiler assumes this but I can't find spec text
16:07 gfxstrand: Intel hardware seems to assume this as well
16:09 gfxstrand: Kayden, maybe?
16:09 danylo: gfxstrand: I think "Component aliasing" covers this
16:10 danylo: glsl spec has "Further, when location aliasing, the aliases sharing the location must have the same underlying numerical type and bit width and the same auxiliary storage and interpolation qualification."
16:10 gfxstrand: danylo: Yup. Found it! Thanks.
16:22 kisak: bl4ckb0ne: even if you're curious, the problem is that any reaction completing the communication loop is incentive to continue this unchecked behavior.
16:23 kisak: I've had my wrists slapped for the same curiousity.
16:42 soreau: now they're on #radeon
16:45 penguin42: they're doing their part in making AIs look smarter than humanity
16:54 ccr: I suspect it's the same person who plagued this and many other FLOSS channels back in Freenode. all the signs point to it, e.g. vague/possible Estonian origin, using VPNs or TOR exit nodes for new hosts/accounts etc.
16:55 bl4ckb0ne: yuo, thats what i thought
16:56 bl4ckb0ne: might not be linked to the one i thought about in #alpine-devel
17:56 kchibisov: Is there some way to have printf or debug values in glsl shaders with mesa? I know that Vulkan has extensions for that.
17:57 kchibisov: Using software rendering and patching mesa is also fine.
17:58 penguin42:tends just to add an extra array parameter to the shader
18:00 kchibisov: I want to debug something I'm computing in shader.
18:00 anholt: kchibisov: renderdoc is what I see people use for that. Or, in llvm, there's lp_debug_dump_value if you want to instrument the backend compiler.
18:00 anholt: llvmpipe/lavapipe, I mean
18:01 jenatali: What's the current upper bound target for CI jobs? I'm increasing coverage for dzn but want to make sure I stay well under that limit since the Windows runner speeds seem to fluctuate so much
18:01 kchibisov: anholt: is the LLVM stuff documented somewhere, I have more experience with it than with renderdoc?
18:01 anholt: That function name is the information, you just call it afaik.
18:02 Kayden: there's nir_lower_printf in rusticl, clover, and microsoft clc, and some GLSL extensions for it IIRC, someone could possibly wire it up
18:03 kchibisov: I know that there's a GL_EXT_debug_printf.
18:27 tjaalton: why is git.fd.o so slow? gives ~11 KiB/s here
18:35 daniels: tjaalton: some kind of weird network issue - machine is unloaded and NFS is fine because a local clone of mesa succeeds in milliseconds
18:36 tjaalton: ah ok
18:36 tjaalton: well, I'll let it fetch overnight ;)
19:26 jenatali: At what point do I get to remove "-experimental" from dzn's meson enable flag?
19:27 zmike:shakes magic 8 ball
19:27 zmike: ask again later
19:27 kisak:recommends when passing baseline conformance
19:27 jenatali: Yeah I'm close
19:36 ccr: groovy bassline comformance
23:03 anholt: if someone wants to ack the workaround at the top of https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22495/commits we should be ready to uprev deqp
23:11 zmike: anholt: ack
23:49 kisak:ponders if Dozen counts as Microsoft or Software in the Public Interest when submitting for Vulkan conformance.
23:52 jenatali: kisak: Pretty sure Microsoft
23:55 kisak: Ah right, RADV is Software in the Public Interest, but ANV is Intel on their submissions