ngcortes has quit [Remote host closed the connection]
pendingchaos_ has joined #dri-devel
pendingchaos has quit [Remote host closed the connection]
chadv has joined #dri-devel
litschi has joined #dri-devel
litschi has quit [Remote host closed the connection]
pendingchaos_ has quit [Read error: No route to host]
pendingchaos has joined #dri-devel
GloriousEggroll has quit [Quit: Death is life's way of telling you you've been fired.]
<airlied>
jekstrand: that isl interaction MR will be a bit tricker to work out for crocus I suspect
ella-0 has quit [Remote host closed the connection]
<airlied>
I suppose I can start on gen7 and see how it blowsu p
<airlied>
but the gen4 cubemaps go through that path as well
<jekstrand>
airlied: Yup. But gen4 doesn't do layered rendering so it's fine to just smash to a single 2D surface.
pendingchaos has quit [Remote host closed the connection]
pendingchaos has joined #dri-devel
<alyssa>
tbf neither does mali
<alyssa>
You don't want to know other the blob emulates it.
<imirkin>
jekstrand: what about glFramebufferTexture3D?
<imirkin>
actually hm
<imirkin>
you sorta need glFramebufferTextureLayer, but that came later.
Lightkey has quit [Ping timeout: 480 seconds]
<airlied>
yay I won a v3d ci fail
<airlied>
though I suppose in theory I could be causing it
<airlied>
clicks the retry and sees
<imirkin>
inconceivable
<airlied>
they do use u_blitter so not totally insane
<airlied>
nope just a flakey flake
boistordu_ex has quit [Ping timeout: 480 seconds]
mbrost_ has joined #dri-devel
sdutt has quit [Remote host closed the connection]
mbrost has quit [Remote host closed the connection]
Lightkey has joined #dri-devel
vivijim has quit [Remote host closed the connection]
jewins has quit [Remote host closed the connection]
Namarrgon has quit [Ping timeout: 480 seconds]
<robclark>
so 2 for 2 flake free MRs today (the first being async-flush revert).. I'm sure there are still plenty of other flakes but seems like a significant improvement :-)
<zmike>
ci: continuous improvement
<robclark>
tru dat
<robclark>
we are shipping chromebooks running freedreno with basically random master/main snapshots of mesa (rather than release branches) which would defn not be possible without CI.. the main branch was too unstable before that.. if anyone is curious I can dig up the commits we are using for stable/beta/dev channels (but I guess you can also figure out by looking at CrOS git trees)
luzipher__ has joined #dri-devel
luzipher_ has quit [Ping timeout: 480 seconds]
pendingchaos has quit [Ping timeout: 480 seconds]
<HdkR>
robclark: Have you got Shadow of the Tomb raider traces for turnip?
<robclark>
I've been focusing on the immediate fires which doesn't include vk, tbh.. although I filed a bug today for enabling turnip (so that counts as something, right?)
<HdkR>
Sounds more important
<robclark>
last 6mo has been a lot of cpufreq/schedutil, devfreq, shrinker, etc debug.. so kinda a lot of everything but mesa (which a small smattering of mesa fixes along the way) ;-)
pendingchaos has joined #dri-devel
<HdkR>
Games on ARM aren't real right now anyway :P
mbrost_ has quit [Remote host closed the connection]
<robclark>
there are enough games that are fun to play.. and kinda seems like mobile games are where the $$ is right now.. there are also some bigger issues with how android integrates into CrOS which have been a challenge..
<robclark>
there is a reason that I've been very pro-perfetto.. we need this tooling to debug these bottlenecks ;-)
<HdkR>
mobile F2P is a terrible market :/
khfeng has joined #dri-devel
fweinrich has joined #dri-devel
fweinrich has quit [autokilled: Suspected spammer. Mail support@oftc.net with questions (2021-06-10 03:12:20)]
<robclark>
I'm a stingy gamer, so it wfm.. but in more seriousness there are some interesting challenges for perf that you don't get in the "lets throw water coolers and a giant power supply at the problem" market ;-)
sdutt has joined #dri-devel
<HdkR>
Definitely. I just love both markets. Moving more AAA games on to the ARM platform is a big win. Which is probably why I like emulation so much
<robclark>
yeah, it is defn cool to see steam store on arm with your stuff ;-)
<robclark>
I wish there were more interesting open src games but the economics of that are hard.. so for gpu that is baked into arm SoC, emulation and android games are where there interesting debugging is ;-)
<DrNick>
I'm interested to see whatever the SteamPal is and if it is an x86 GPD-like thing
<HdkR>
robclark: I might have just filled out the Steam Hardware Survey on my ARM board today. That'll be a wacky survey results. Sorry Plagman :P
<robclark>
lol
<jekstrand>
imirkin: It's only a problem if you have layered rendering
sdutt has quit []
shankaru1 has joined #dri-devel
blue__penquin has joined #dri-devel
mbrost_ has joined #dri-devel
<airlied>
mripard: what's the landing for that hdmi codec pull? do I have to care about it?
luzipher_ has joined #dri-devel
luzipher__ has quit [Ping timeout: 480 seconds]
gpoo has quit [Read error: Connection reset by peer]
lemonzest has joined #dri-devel
Duke`` has joined #dri-devel
luzipher__ has joined #dri-devel
mbrost_ has quit [Remote host closed the connection]
mbrost_ has joined #dri-devel
luzipher_ has quit [Ping timeout: 480 seconds]
imirkin has quit [Quit: Leaving]
leont has joined #dri-devel
leont has quit [Remote host closed the connection]
imirkin has joined #dri-devel
tstm has joined #dri-devel
mbrost_ has quit [Remote host closed the connection]
tstm has quit [Remote host closed the connection]
thellstrom1 has joined #dri-devel
thellstrom has quit [Remote host closed the connection]
ncl3 has joined #dri-devel
ncl3 has quit [Remote host closed the connection]
Ammler13 has joined #dri-devel
Ammler13 has quit [autokilled: This host has violated network policy. Contact support@oftc.net with questions (2021-06-10 05:28:25)]
lohfu has joined #dri-devel
lohfu has quit [Remote host closed the connection]
thellstrom1 has quit [Remote host closed the connection]
thellstrom has joined #dri-devel
cbrill has joined #dri-devel
danvet has joined #dri-devel
ella-0 has joined #dri-devel
luzipher__ has joined #dri-devel
blue__penquin has quit [Remote host closed the connection]
blue__penquin has joined #dri-devel
thellstrom1 has joined #dri-devel
thellstrom1 has quit []
luzipher_ has quit [Ping timeout: 480 seconds]
dri-logger has joined #dri-devel
cbrill has quit []
dri-logger has quit []
cbrill has joined #dri-devel
cbrill has quit []
RobertC has joined #dri-devel
thellstrom has quit [Ping timeout: 480 seconds]
shailangsa has joined #dri-devel
shailangsa has quit [Remote host closed the connection]
dri-logg1r has joined #dri-devel
dri-logg1r has quit []
dri-logger has joined #dri-devel
thellstrom has joined #dri-devel
mtretter has joined #dri-devel
ella-0 has quit [Ping timeout: 480 seconds]
Ultrasauce has quit [Remote host closed the connection]
Namarrgon has joined #dri-devel
Ultrasauce has joined #dri-devel
mlankhorst has joined #dri-devel
jaawerth has joined #dri-devel
jaawerth has quit [Remote host closed the connection]
garrison has joined #dri-devel
i-garrison has quit [Ping timeout: 480 seconds]
pnowack has joined #dri-devel
<danvet>
mripard, I have one more patch for -fixes, so pls don't do the pull just now :-)
ppascher has quit [Quit: Gateway shutdown]
ppascher has joined #dri-devel
andrey-konovalov has joined #dri-devel
luzipher_ has joined #dri-devel
luzipher__ has quit [Ping timeout: 480 seconds]
<mripard>
airlied: it should be drm-misc-next, so you don't have to worry about it directly :)
<mripard>
but I need to send another version, the first one had a bisection issue
NamanGupta[m] has joined #dri-devel
<mripard>
danvet: ack
<mripard>
let me know when it's ok :)
NamanGupta[m] has quit [Remote host closed the connection]
agx__ has joined #dri-devel
pcercuei has joined #dri-devel
agx__ has quit []
agx_ has joined #dri-devel
rasterman has joined #dri-devel
agx__ has joined #dri-devel
agx__ has quit []
agx has joined #dri-devel
agx has quit []
agx_ has quit []
TanoMarcelo has joined #dri-devel
agx_ has joined #dri-devel
TanoMarcelo has quit [Remote host closed the connection]
agx_ has quit []
agx_ has joined #dri-devel
Thymo has quit []
elongbug has joined #dri-devel
arthuredelstein25 has joined #dri-devel
<danvet>
mripard, constantly meetings that overrun :-(
arthuredelstein25 has quit [Remote host closed the connection]
CorePoint11 has joined #dri-devel
boistordu_ex has joined #dri-devel
CorePoint11 has quit [Remote host closed the connection]
MrCooper has quit [Remote host closed the connection]
MrCooper has joined #dri-devel
randomher0 has joined #dri-devel
luzipher__ has joined #dri-devel
Thymo has joined #dri-devel
luzipher_ has quit [Ping timeout: 480 seconds]
<dolphin>
airlied, danvet: PR sent for -gt-next, after it gets merged there is the plan to create topic branch based on drm-next for the TTM changes in core and merge both to drm-misc-next and drm-intel-gt-next
gouchi has joined #dri-devel
gouchi has quit []
luzipher_ has joined #dri-devel
MrCooper_ has joined #dri-devel
MrCooper has quit [Ping timeout: 480 seconds]
luzipher__ has quit [Ping timeout: 480 seconds]
Lucretia has quit [Remote host closed the connection]
Lucretia has joined #dri-devel
MrCooper_ is now known as MrCooper
karolherbst_ has joined #dri-devel
karolherbst has quit [Ping timeout: 480 seconds]
karolherbst has joined #dri-devel
itoral has joined #dri-devel
karolherbst_ has quit [Ping timeout: 480 seconds]
hch12907 has quit [Remote host closed the connection]
Sachiel has quit [Ping timeout: 480 seconds]
Sachiel has joined #dri-devel
pochu has quit [Quit: leaving]
pochu has joined #dri-devel
randomher0 has quit []
khfeng has quit [Remote host closed the connection]
khfeng has joined #dri-devel
garrison has quit [Ping timeout: 480 seconds]
ntz has joined #dri-devel
ntz has quit [Remote host closed the connection]
i-garrison has joined #dri-devel
pjakobsson has quit []
luzipher__ has joined #dri-devel
jubalh has joined #dri-devel
jubalh has quit [Remote host closed the connection]
luzipher_ has quit [Ping timeout: 480 seconds]
itoral has quit [Remote host closed the connection]
vivijim has joined #dri-devel
dv_ has joined #dri-devel
blue__penquin has quit [Remote host closed the connection]
luzipher_ has joined #dri-devel
luzipher__ has quit [Ping timeout: 480 seconds]
<alyssa>
what's the rule of thumb for CI pipeline time these days?
<alyssa>
(if dEQP takes 10 minutes, I guess that's too long because compiling + booting + etc adds a few minutes of overhead and then you're way over the 10min golden rule)
<MrCooper>
10 minutes max per job is the rule of thumb
<alyssa>
Ack
<alyssa>
So that leaves... 8 minutes for dEQP?
<danvet>
mripard, ok it's pushed
<danvet>
sry forgot to tell you, got distrated
aiddamse has joined #dri-devel
pcercuei_ has joined #dri-devel
pcercuei has quit [Read error: Connection reset by peer]
aravind has quit [Remote host closed the connection]
MrCooper has quit [Remote host closed the connection]
MrCooper has joined #dri-devel
aravind has joined #dri-devel
aiddamse has quit [Remote host closed the connection]
mnabid has joined #dri-devel
mnabid has quit [Remote host closed the connection]
xp4ns3 has joined #dri-devel
blue__penquin has joined #dri-devel
<mripard>
danvet: np, I was doing something else
<robclark>
alyssa: compiling is it's own job, but try to get booting + deqp run into 10min
Daanct12 has joined #dri-devel
aravind has quit [Remote host closed the connection]
Danct12 has quit [Ping timeout: 480 seconds]
<alyssa>
robclark: ack
RobertChiras has joined #dri-devel
RobertC has quit [Ping timeout: 480 seconds]
blue__penquin has quit []
mbrost has joined #dri-devel
Duke`` has joined #dri-devel
awygle has joined #dri-devel
awygle has quit [Remote host closed the connection]
NiksDev has quit [Ping timeout: 480 seconds]
Daaanct12 has joined #dri-devel
Daaanct12 has quit []
Daanct12 has quit [Ping timeout: 480 seconds]
Danct12 has joined #dri-devel
tvon has joined #dri-devel
tvon has quit [Remote host closed the connection]
hch12907 has joined #dri-devel
xp4ns3 has quit []
purpleidea8 has joined #dri-devel
purpleidea8 has quit [Remote host closed the connection]
xp4ns3 has joined #dri-devel
DasSkele- has joined #dri-devel
DasSkele- has quit [autokilled: Suspected spammer. Mail support@oftc.net with questions (2021-06-10 15:51:54)]
RobertChiras has quit [Ping timeout: 480 seconds]
<alyssa>
daniels: Any idea why dEQP-GLES31 is taking nearly 2x longer on the rk3399 kevin lava than on my own machine?
<alyssa>
Is this the ulimit thing biting me?
GloriousEggroll has joined #dri-devel
<daniels>
alyssa: is it crashing all the time?
<alyssa>
Not all the time but lots.
<daniels>
yeah that
<alyssa>
Should I roll in a ulimit -c 0 into this then..?
<daniels>
I'll make that HEAD+2
<daniels>
yeah, itmt just whack that into .gitlab-ci/lava.yml.jinja2 under - modprobe amdgpu
<alyssa>
ack
<alyssa>
erm is that going to create a new container
NiksDev has joined #dri-devel
niiko has joined #dri-devel
niiko has quit [Remote host closed the connection]
<MrCooper>
shouldn't
<alyssa>
ty
<kisak>
hopefuly an easy question for someone who regularly debugs the kernel... If I bump dwarves from 1.15 to 1.17 on Ubuntu Focal so that I can use the Ubuntu default CONFIG_DEBUG_INFO_BPF, is dwarves 1.17 a build-time or runtime/debug symbol lookup time dependency?
<daniels>
YAML: not even once
<danvet>
MrCooper, there was some thread with robclark about adding boosting
<danvet>
well just clock boosting for msm
<danvet>
but priority boosting in the drm/scheduler isn't that far
<danvet>
ofc since no preempt you're still stuck behind everything that's in the hw already
<danvet>
but that's about as good as it would get
<danvet>
and I think for atomic flips we could queue up the boosting automatically
<danvet>
so compositors don't need to fiddle around with "do I need to be root to gain elevated priority levels"
<alyssa>
"atomic flip" sounds super cyberpunk
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
benedikt93 has joined #dri-devel
benedikt93 has quit [Remote host closed the connection]
<danvet>
alyssa, actually called it nuclear flip for a while to distinguish from atomic modeset
<danvet>
very 50s vibe on that
gouchi has joined #dri-devel
xp4ns3 has quit []
<danvet>
emersion, I thought you know email?
<danvet>
(threading butchered)
<MrCooper>
danvet: yeah, boosting still won't help without preemption
xp4ns3 has joined #dri-devel
<emersion>
i've tried to fix it, but it's complicated
<danvet>
MrCooper, it should if the app is doing at least a few CS
<emersion>
tl;dr my email provider is shit
<danvet>
you'll still get in earlier than without any boosting at all
<MrCooper>
well I'm actually testing with hacked-up boosting, doesn't seem to help at all for GpuTest plot3d at least
<MrCooper>
but I'm not satisfied with "helps a little" anyway :)
<MrCooper>
danvet: BTW, surely a DRM master should be able to get a high priority context without root; mutter certainly does on Fedora at least
<danvet>
MrCooper, ah but amdgpu folks want to disallow rendering on the master
<danvet>
or at least könig said he'll do that
<danvet>
not sure it happened
<danvet>
but yeah on i915 you should be able to get higher priority on master
vivijim_ has joined #dri-devel
vivijim has quit [Ping timeout: 480 seconds]
pendingchaos has quit [Ping timeout: 480 seconds]
mboddu[m] has joined #dri-devel
mboddu[m] has quit [Remote host closed the connection]
pendingchaos has joined #dri-devel
xp4ns3 has quit []
aswar002 has joined #dri-devel
thellstrom has quit [Quit: thellstrom]
WARBIRD199 has joined #dri-devel
WARBIRD199 has quit [Remote host closed the connection]
shankaru1 has quit []
khfeng has quit [Ping timeout: 480 seconds]
Kayden has quit [Quit: off to....lunch. lunch? what is this concept]
sagar_ has quit [Ping timeout: 480 seconds]
pekkari has quit []
iive has joined #dri-devel
Nightmare7 has joined #dri-devel
Nightmare7 has quit [autokilled: Suspected spammer. Mail support@oftc.net with questions (2021-06-10 18:34:53)]
tott has joined #dri-devel
tott has quit [Remote host closed the connection]
<alyssa>
robclark: OOI what's stopping freedreno from doing GL4.x?
<alyssa>
from mesamatrix it looks like ES3.2 + softfp64 gets you to GL4.3
thellstrom has joined #dri-devel
<robclark>
hmm, does look like there are a lot of GL4.x stuff that we could check off if we bothered to update features.txt.. I guess softfp64 (and probably similar for int?) would be needed.. other than that I'd have to check what things various gl extensions want that gles counterparts don't require
sdutt has joined #dri-devel
<alyssa>
good news is that intel already does softfp64 piping
<robclark>
are there any open src games that use higher GL? Or is this all for HdkR? :-P
<alyssa>
you better have a darn good register allocator, though
<alyssa>
say... SSA-based with efficient spilling to pvtmem.... hmmmm
h0m1 has joined #dri-devel
<robclark>
there are probably a lot of things we (or anyone with interest) could enable on desktop gl side.. but -ENOTIME
<robclark>
probably if I'd had time I would have wired up clover by now
h0m1 has quit [Remote host closed the connection]
<HdkR>
haha :)
<alyssa>
nod
<robclark>
although I suppose clover might force the issue of wiring up int64/fp64 emulation
<HdkR>
Theoretically you can just implement all the Vulkan features and zink will take care of the rest ;)
<robclark>
hmm, zink on tu only expose gl3.2 core profile..
<robclark>
but I think tu itself has implemented a bunch of things we'd probably like to have for higher gl incl the ir3 parts, so probably not too hard to port things to gallium
<zmike>
3.3 prob missing... timestamps?
<robclark>
timestamps are a wonky thing on tilers
<robclark>
(it's easy enough to support.. but what you get is probably not what apps expect)
<alyssa>
yep
<bnieuwenhuizen>
now I'm curious, what do you actually get?
<zmike>
an ice cream sandwich
<alyssa>
bnieuwenhuizen: on mali, we can get timestamps accurately for vertex/tiling/compute shaders
<alyssa>
but for fragment, we can only get the timestamps before any fragment shaders have run or after all fragmen thsaders have run
<alyssa>
whereas big GL apps would expect timestamps on a per-draw granularity
<alyssa>
which doesn't make sense when draws are interleaved across different tiles
NiksDev has quit [Ping timeout: 480 seconds]
thellstrom has quit [Remote host closed the connection]
thellstrom has joined #dri-devel
thellstrom has quit []
thellstrom has joined #dri-devel
<robclark>
bnieuwenhuizen: on adreno you get the time in *one* of the tiles/bins.. IIRC it will end up being the time for the last tile (ie. each tile will just overwrite the previous timestamp)
elongbug has quit [Remote host closed the connection]
badsnappa has joined #dri-devel
badsnappa has quit [Remote host closed the connection]
<kusma>
Yuck, timestamps
shankaru1 has joined #dri-devel
<kusma>
Maybe we should just do something terrible hacky there. I mean, it's not going to go away, but it's also not really going to work on tilers. And it's hard to prove that we're doing something whack anyway...
<kusma>
So maybe just claim the cap and write jibberish?
<kusma>
YOLO all the way to the bank.
<alyssa>
kusma: yeah i mean
<ccr>
aren't you afraid at all that the graphics police unit (GPU) will come after you?
<alyssa>
ccr: the GPU? no, not so much
<alyssa>
but apparently the fashion police have a warrant for my arrest because of my daytime pajamas
<ccr>
FPU?
<kisak>
alyssa: fashion police don't visit this neighborhood, too disgusted
<kusma>
ccr: They'll never make the charges stick!
<kusma>
I mean, what's even time? I'm sure we can use some general relativity hokum to cast doubt on the whole thing!
<zmike>
forcing me to adhere to your time standard infringes on my freedoms
<kusma>
"Look, your GPU must have been very close to a supermassive black hole, time doesn't work the way you think"
luzipher__ has joined #dri-devel
<ccr>
"well, it's an accelerator .. just like the LHC .. the structure and nature of time is not absolute in these matters."
<kusma>
Yeah, that's the one!
<Prf_Jakob>
Silly question, does hardware in general support sRGB conversion on R10G10B10A2 formats?
luzipher_ has quit [Ping timeout: 480 seconds]
<zmike>
r
<alyssa>
Prf_Jakob: Probably no categorical yes or no
<alyssa>
but I expect some does (I think Mali included but need to check)
<pendingchaos>
I don't think AMD hardware supports it
<pendingchaos>
only R8, R8G8 and R8G8B8A8
<pendingchaos>
(AMD hardware: GCN and newer)
<Prf_Jakob>
alyssa, pendingchaos: Thanks, and typical.
* alyssa
looks for mali-g77 hardware
<Prf_Jakob>
Intel lists ISL_FORMAT_R10G10B10A2_UNORM_SRGB so that seems to indicate that the hardware supports it.
mlankhorst has quit [Ping timeout: 480 seconds]
<jekstrand>
Prf_Jakob: Yup. Supported for sampling.
<jekstrand>
Prf_Jakob: Critically, though, not supported for render, I'm afraid.
<jekstrand>
So doing GL with that as a framebuffer format would be tricky at best.
<jekstrand>
I'm assuming that's what you're really getting at
<Prf_Jakob>
Sampling only would work fine, I think.
<jekstrand>
Prf_Jakob: We do, however, support all the fun stuff on B10G10R10A2_UNORM_SRGB
<jekstrand>
Prf_Jakob: isl_format.c has the full table of HW caps, FYI.
<Prf_Jakob>
Hmm annoying :)
<jekstrand>
We could emulate R10G10B10A2 with a BGRA and a swizzle if needed, though.
<Prf_Jakob>
Long story short, Unreal really wants to submit sRGB space pixels. I think it writes them with a compute shader tho. We need to sample them in our Monado compositor. OpenGL support is optional.
<jekstrand>
At least on HSW+
<alyssa>
Prf_Jakob: Writing out with a compute shader? That is unreal.
<jekstrand>
Right. So you probably really only care about sampling, in which case we support either.
<Prf_Jakob>
Rather avoid having to add sRGB->linear stuff on sample.
<Prf_Jakob>
Yeah
<Prf_Jakob>
They *might* want GFX render which would be annoying.
<jekstrand>
Better to use the actual format or you'll get interpolation subtly wrong. :)
<jekstrand>
If they want GFX render, we can do that; we just need to smash the ANV format table.
<Prf_Jakob>
Hehe
<jekstrand>
No idea where things with with RADV.
<jekstrand>
It's a 1-line patch to support render on RGBA
<Prf_Jakob>
Cool
Daanct12 has joined #dri-devel
<jekstrand>
Uh... Does Vulkan have a format for that?
<Prf_Jakob>
Nope :)
<Prf_Jakob>
So would need to write a extension for that.
<jekstrand>
Ok, so it's a Vulkan extension, a small bit of boilerplate, and then we can do it. :)
<Prf_Jakob>
RADV seems to split bit layout and bit meaning, so technically can probably smash the 10,10,10,2 layout in there with the sRGB meaning.
<jekstrand>
But extensions to add formats are the easiest of extensions.
<jekstrand>
Prf_Jakob: Yeah, I would expect AMD and nvidia could both do it.
* jekstrand
asks some AMD and Nvidia people.
<pendingchaos>
Prf_Jakob: not on Navi/GFX10+, images switched to a single enum
thellstrom has quit [Ping timeout: 480 seconds]
<Prf_Jakob>
pendingchaos: Oh
<Prf_Jakob>
pendingchaos: In which file are that enum listed?
<Prf_Jakob>
I would assume gfx10-rsrc.json
Danct12 has quit [Ping timeout: 480 seconds]
<Prf_Jakob>
Dang it no 10bit sRGB format.
<pendingchaos>
yes, gfx10-rsrc.json
<Prf_Jakob>
Thanks for the help
<pendingchaos>
that combination isn't listed in the Vega ISA doc (8.4.4. Data Formats)
<pendingchaos>
even if you could give the hardware with split bit layout and bit meaning a combination of 10,10,10,2 and sRGB, I'm not sure if it would work
<Prf_Jakob>
*nods*
rasterman has quit []
macromorgan has quit [Quit: Leaving]
danvet has quit [Ping timeout: 480 seconds]
<anholt>
I accidentally found a "sudo reboot" in my bash history on fd-farm. apologies to owners of runs that just got killed, we should be back shortly
<jekstrand>
This is why I make my sudo require a password. :)
<anholt>
like half of the commands on that system are sudo whatever gitlab-runner/config.toml, so not really happening.
<jekstrand>
Yeah....
danvet has joined #dri-devel
<anholt>
the good news is the reason I was on there is I think I've got a workaround for a530 fastboot prompt missing, and it should fix the recovery in other intermittent cases too if all goes well.
<airlied>
jekstrand: I think I'm mostly finished with major reworks in crocus now before landing it
<jekstrand>
airlied: Ok. I'm not looking to hold you up. Just trying to make sure someone has at least attempted a review.
<jekstrand>
airlied: I did find a bunch of little to moderate things in the files I did look at.
<jekstrand>
airlied: But if you'd rather ack-and-land and fix stuff up later, we can do that.
<jekstrand>
It's just super-convenient to make comments on an MR.
pnowack has quit [Quit: pnowack]
<airlied>
jekstrand: I fixed up most of the ones you called out, I probably just need to make sure the precursor patches get reviewed properly
<jekstrand>
airlied: I only really read two files. :P
<danvet>
mripard, shots fired
<danvet>
did you honestly think you could keep my r-b on something that officially sanctions vendor uapi when we'er trying to standardize this all :-)
<jekstrand>
daniels, emersion, pq: Any way I could get some pitty from a window-system person to review all but the last patch of !4037?
<jekstrand>
and bnieuwenhuizen may have an opinion too, though I don't think RADV gets anything from it given amdgpu's "creative" interpretation of fences.
<jekstrand>
The patches on dri-devel for the ioctl are all reviewed but we should have userspace reviewed before we go landing any new uapi.
Duke`` has quit [Ping timeout: 480 seconds]
danvet has quit [Ping timeout: 480 seconds]
<emersion>
jekstrand: is it worth it to move this forward if it's going to get replaced with Something Else?
<jekstrand>
emersion: I don't think sync_file is getting replaced. Not quickly, anyway.
<emersion>
drm_syncobj would be better
<jekstrand>
There's some discussion between Christian and I to that effect on dri-devel on the patch which adds the ioctl.
<emersion>
but not if drm_syncobj will get replaced by something else
gouchi has quit [Remote host closed the connection]
* emersion
needs to go to sleep
<emersion>
alright, i'll have a look tomorrow
<emersion>
sorry for the confusion
<jekstrand>
no worries
<jekstrand>
And FTR, I'm very much in team "don't re-plumb the entire stack more than once, please"
<zmike>
more than once per year, right?
<jekstrand>
Bad zmike!
* zmike
runs off to hide in nir MRs
<bnieuwenhuizen>
jekstrand: wrt radv you guessed correctly. I currently see no way to use it in radv and actually improve anything, until we improve amdgpu fencing
<alyssa>
rip
<jekstrand>
bnieuwenhuizen: That's ok. We've got like 5 Vulkan drivers these days. The rest of us can use it. :P
<jekstrand>
bnieuwenhuizen: It would be good for you to look at it and ensure we're not going to break radv, though.
<airlied>
jekstrand: I think I cleaned up the decoder patch in parallel with your comment on it
<jekstrand>
airlied: Oh, ok. Let me look again.
tobiasjakobi has quit [Remote host closed the connection]
shankaru1 has quit []
newdimension has joined #dri-devel
newdimension has quit [Remote host closed the connection]
aravind has joined #dri-devel
sagar_ has joined #dri-devel
pcercuei_ has quit []
equake[m|gr] has joined #dri-devel
equake[m|gr] has quit [Remote host closed the connection]
iive has quit []
andrey-konovalov has quit [Ping timeout: 480 seconds]