<namasteandi>
DRM core&drivers uses a lot of idr (integer lookup directories) for mapping userspace IDs to internal objects, and in most places ID=0 means NULL and hence is never used. Switching to idr_init_base() for these would make the idr more efficient.
<namasteandi>
Contact: Daniel Vetter
<namasteandi>
Level: Starter
<namasteandi>
this is the task i am attempting
<namasteandi>
from my understanding of this task, i have to go through the files in the /drivers/drm/gpu subdirectory and replace the instances of ID=0 to idr_init_base()
<namasteandi>
is my understanding of the task correct?
itoral has joined #dri-devel
namasteandi has quit [Ping timeout: 480 seconds]
Duke`` has quit [Ping timeout: 480 seconds]
sdutt has quit [Ping timeout: 480 seconds]
i-garrison has quit []
itoral has quit [Remote host closed the connection]
camus has joined #dri-devel
camus1 has quit [Read error: Connection reset by peer]
mszyprow has joined #dri-devel
mszyprow has quit [Read error: Connection reset by peer]
macromorgan is now known as Guest94
macromorgan has joined #dri-devel
Guest94 has quit [Ping timeout: 480 seconds]
thellstrom has joined #dri-devel
tursulin has joined #dri-devel
tzimmermann has joined #dri-devel
MajorBiscuit has joined #dri-devel
tursulin has quit [Read error: Connection reset by peer]
tursulin has joined #dri-devel
mvlad has joined #dri-devel
itoral has joined #dri-devel
pcercuei has joined #dri-devel
itoral has quit [Remote host closed the connection]
itoral has joined #dri-devel
camus1 has joined #dri-devel
camus has quit [Ping timeout: 480 seconds]
pjakobsson has joined #dri-devel
rasterman has joined #dri-devel
mlankhorst has joined #dri-devel
itoral_ has joined #dri-devel
itoral has quit [Ping timeout: 480 seconds]
JohnnyonFlame has quit [Ping timeout: 480 seconds]
pnowack has joined #dri-devel
thellstrom has quit [Ping timeout: 480 seconds]
MajorBiscuit has quit [Ping timeout: 480 seconds]
anparra[m] has joined #dri-devel
flacks has quit [Quit: Quitter]
flacks has joined #dri-devel
illwieckz has quit [Ping timeout: 480 seconds]
camus has joined #dri-devel
camus1 has quit [Read error: Connection reset by peer]
mclasen has joined #dri-devel
illwieckz has joined #dri-devel
thellstrom has joined #dri-devel
thellstrom1 has joined #dri-devel
thellstrom has quit [Ping timeout: 480 seconds]
dwlsalmeida has joined #dri-devel
vivijim has joined #dri-devel
pcercuei has quit [Quit: brb]
pcercuei has joined #dri-devel
<alyssa>
danvet: ^
itoral_ has quit [Remote host closed the connection]
MajorBiscuit has joined #dri-devel
columbarius has joined #dri-devel
co1umbarius has quit [Ping timeout: 480 seconds]
leandrohrb9 is now known as leandrohrb
fxkamd has joined #dri-devel
jewins has joined #dri-devel
vivijim has quit [Remote host closed the connection]
<danvet>
ishitatsuyuki, thx, but looks like they're gone already :-/
tzimmermann has quit [Read error: Connection reset by peer]
<danvet>
or maybe on the other side of a net split
libv_ has joined #dri-devel
aravind has quit [Ping timeout: 480 seconds]
libv has quit [Ping timeout: 480 seconds]
fxkamd has quit []
jewins has quit [Quit: jewins]
jewins has joined #dri-devel
<marex>
pq: Hi, can I assign a weston MR to you in the FDO gitlab or does someone else have to do that (assuming I know it should be assigned to you, because weston) ?
<emersion>
we don't really assign issues
<emersion>
people work on stuff when they have time
<marex>
emersion: not an issue, a MR for review
<emersion>
yes, same
<emersion>
review work is still work
<marex>
hah, I saw mesa MRs usually got assigned by someone
JohnnyonFlame has joined #dri-devel
<emersion>
there are no guarantees that someone will have tim to look into it
<emersion>
time*
<emersion>
mesa is a bit different, yeah. people usually assign themselves when they are sure they
<emersion>
'll look into it
<marex>
emersion: what I saw there was someone else assigned maintainers of specific driver to the MR
<imirkin>
marex: very rarely are MR's assigned ...
<imirkin>
except to marge
<alyssa>
usually if I assign myself something it's because it's my fault it exists and I get to deal with the consequences
<emersion>
and marge assigns people when she fails
<marex>
imirkin: so how does one CC people who should be CCed ?
<emersion>
in general it's considered a bit rude to assign other people ;)
<imirkin>
marex: you just mention them in a comment
<marex>
oh ok, I already did that
<marex>
thanks
<emersion>
when, cc @foo is totally fine
<emersion>
yeah*
<imirkin>
and the "cc" bit is just so that they know why they were mentioned, it has no effect on them actually being cc'd
sdutt has quit []
sdutt has joined #dri-devel
<alyssa>
in some ways we still use gitlab like a souped up mailing list
<alyssa>
much better than an actual ML though.. I've come around on that.
<marex>
alyssa: if it supported inline comments on commit messages, it would be perfect
<marex>
other than that, indeed, the ML workflow is starting to show its limits
<alyssa>
I comment inline on commit messages though admittedly I have an errrrr unusual way of using gitlab
<marex>
alyssa: the inline commenting is now possible ?
<alyssa>
eh heh heh
<marex>
alyssa: what's the trick ?
<alyssa>
there is no trick
<marex>
alyssa: so, how does one do that -- comment inline into a commit message ?
<alyssa>
jekstrand: perhaps you can explain better than me 😅
<alyssa>
i know it's something you'll need to use a lot at your new position at Tungsten Graphics
<alyssa>
or were you going to SGI?
<zmike>
last news coverage I saw said bose
<alyssa>
zmike: ah right working on the Khronos spec for OpenAL
<emersion>
"news coverage", aka zmike's blog
<emersion>
a very authoritative source :)
<zmike>
look, if it wasn't authoritative then it wouldn't be covered by phoronix
* ccr
nods
camus1 has joined #dri-devel
camus has quit [Remote host closed the connection]
<ccr>
I am certain zmike secretly works at Stargate Command (SGC), "supergoodcode" is just a diversion.
<ccr>
and also a hint, for those who can decode these things
<alyssa>
wait that's what SGC stands for?
<alyssa>
I thought it was for slimey golf clubs
<ccr>
hmm.
<jenatali>
I think I'm just about ready to start upgrading primconvert to support doing converts via compute + indirect. We're really getting bitten for perf in some scenarios by that CPU readback for 8bit index buffers and triangle fans
<alyssa>
jenatali: Why not add 8-bit + fans into D3D12?
<alyssa>
I know they're problematic but hardware supports them ...
<alyssa>
(Does D3D12 have the notion of optional features?)
<jenatali>
I think fans might actually be getting added at some point. 8-bit index buffers we talked about doing but decided against it
<jenatali>
But either way, if we add a feature, we need new drivers to be able to use it, and we want to support GL without requiring new drivers
<jenatali>
And yes, there are optional features
<alyssa>
both would be <10 line patches per driver I imagine
<alyssa>
..if you're allowed to do that
<alyssa>
Is Microsoft seeing the benefits of FOSS drivers now? 😇
<jenatali>
We don't have driver source
<alyssa>
Microsoft /is/ seeing the benefits of FOSS drivers, incredible
Duke`` has joined #dri-devel
<alyssa>
soon: Microsoft ships DXVK atop of anv/radv
<alyssa>
zmike: Did we consider the possibility jekstrand is going to Microsoft to do the Windows port of anv?
<zmike>
I didn't personally consider that, no
<zmike>
my sources were pushing bose last time I talked tohim
<zmike>
this is great, I won't have to keep updating/rebasing your patches in my branch anymore :)
macromorgan has quit [Read error: Connection reset by peer]
macromorgan has joined #dri-devel
lstrano has joined #dri-devel
<jekstrand>
alyssa: Stop giving reasonable suggestions! You need to be more imaginative, like zmike. :-P
<zmike>
IT'S A HINT
<jekstrand>
If ever someone was going to teach a class on bullshitting....
<Ristovski>
ccr: I am confident zmike feeds on driver bugs, which is also what allows him to come up with all those jokes.. its the only explanation
<ccr>
it's all becoming clear now!
<Ristovski>
in short: don't stop introducing bugs in mesa if you enjoy his humor!
<zmike>
teaching...imaginitive...bullshitting...
<zmike>
HE'S GOING TO PIXAR
<jekstrand>
Pixar could be fun
<ccr>
why not the classic: he's going to Valve .. "Half-Life 3 confirmed."
<zmike>
trying not to get fired for leaking our next project
<zmike>
ffffuuuuuuuuuuuuuuuuuuuuuuuuu
<ccr>
:P
nchery has joined #dri-devel
<Ristovski>
zmike: Btw, did you ever look into what Valve exactly updates when they added -vulkan support to Portal2/HL2 et al? I assume they also update their SDL initialization code to setup a vulkan instance instead of opengl?
<zmike>
it adds support for running the games on dxvk
<zmike>
JoshuaAshton did all the work to make this possible
<zmike>
💪💪💪
<Ristovski>
well yes, they replace libtogl.so with dxvk, but the reason I wonder is that it would be funny to get TF2 working with dxvk as well. However, patching the binary to make SDL setup vulkan instead of opengl would be tricky (I assume they /do/ do that in their engine initialization code)
* Ristovski
should perhaps look into his 200% totally legitimate and purely hypothetical "Valve" folder
<zmike>
it's possible that we might be doing this at some point anyway
<zmike>
I'm not sure what the roadmap is and everyone's very busy now
<Ristovski>
I see
ybogdano has joined #dri-devel
rgallaispou has joined #dri-devel
<alyssa>
jekstrand: Pixar sounds entirely too reasonable.
<alyssa>
It's not Pixar, you're going to /Disney/ directly, specifically to work on the graphics for a new Pirates of the Carribean theme park ride, exclusively opening in Disneyland Paris as soon as we get past Omega
dreda has quit [Read error: Connection reset by peer]
dreda_ has joined #dri-devel
mbrost has joined #dri-devel
dreda_ is now known as dreda
<jekstrand>
Didn't you all know? Valve already released HL3. It's embedded in Portal2 near the cake. It's just that no one's figured out how to get to it yet.
mbrost has quit [Read error: Connection reset by peer]
<alyssa>
this was a triumph
mbrost has joined #dri-devel
<ccr>
a huge success
* alyssa
makes note
<ccr>
open source .. we do what we must because we can
<jekstrand>
For the good of all of us
<bnieuwenhuizen>
emersion: in radv we totally assign explicit reviewers for MRs
<emersion>
bnieuwenhuizen: yeah, the exception is that you all work for the same company?
<bnieuwenhuizen>
I don't
<bnieuwenhuizen>
but yeah we have actual discussions when review is too slow
<jekstrand>
On ANV, they all just assign me. :-P
JohnnyonFlame has quit [Ping timeout: 480 seconds]
<bnieuwenhuizen>
jekstrand: not anymore, I hope?
rgallaispou has quit [Ping timeout: 480 seconds]
<emersion>
bnieuwenhuizen: ah, i thought there was a cross-company contract or something
<bnieuwenhuizen>
emersion: I started the practice because I've had good experiences within Google for tracking which of the open MRs need my attention
lumag__ has joined #dri-devel
<emersion>
there's also the "request review" button nowadays
mbrost has quit [Ping timeout: 480 seconds]
<zmike>
I think that's just there as a joke though? 🤔
lumag_ has quit [Write error: connection closed]
<bnieuwenhuizen>
isn't that just the reviewer assignment?
<jekstrand>
bnieuwenhuizen: I hope so too!
<emersion>
"approve review" buttons are unambiguously a joke
<emersion>
i often see contributors click it, then the MR shows up as "approved :tick::tick::tick:" in the list
<jenatali>
And vbuf can't do the save since it only has vertex state, hm
<jenatali>
I'll need to think about that more
<zmike>
it's a tricky problem
ybogdano has joined #dri-devel
<jasuarez>
anholt: restarted the failing jobs
<jasuarez>
now it should go fine
<jenatali>
Yeah... I mean I could always just put it in the d3d12 driver but I figure this might be useful for others so I want to put it in common code if I can
<anholt>
jasuarez: weird. those sure don't seem like that MR should matter to them
<zmike>
jenatali: for sure
<zmike>
might be worth creating a ticket for discussion
<jenatali>
Anyways, I'll noodle on it some more. I'm still polishing up my indirect support before I can even handle that
<zmike>
cool
<zmike>
I'm just blogging about jekstrand's latest job leaks
<jenatali>
Hm... I'm hitting a test failure in arb_indirect_parameters/tf-count-arrays that I'm not convinced is my fault. The GLSL linker seems to be assigning locations to the VS input vars in the opposite order that the test's expecting
<zmike>
passes for me?
<jenatali>
:(
<imirkin>
jenatali: i wrote it, so a solid 97% chance the test is busted in some way
<pq>
marex, like emersion said, we don't really assign MRs. Assigning wouldn't change that person's priorities anyway, and everyone is welcome to review too. I'm also still on holidays.
<imirkin>
jenatali: looks like i assume that the inputs are laid out sequentially. not sure if that's a legal assumption to make.
<jenatali>
I don't see how it could pass for zmike though
<imirkin>
there's a "nir" linker now, perhaps it does things differently?
<zmike>
good news
<imirkin>
the test _did_ pass back when i wrote it ;)
<jenatali>
I'm pretty sure that's what I'm using?
<zmike>
it doesn't just pass for me
<zmike>
it passes for Kayden too
<zmike>
and radeonsi
<jenatali>
Hmmm... something's odd
* zmike
is starting to suspect that imirkin successfully wrote a test this once
<jenatali>
Yeah I checked the CI fails and didn't see any references, so I assumed others were passing it, but I don't get how just yet :)
<imirkin>
jenatali: so you're seeing vtx as "attr 1" and color as "attr 0"?
<jenatali>
Yeah
<imirkin>
er, make that in_color
<imirkin>
only thing i can think of is the nir linker is doing something different
agx has quit [Read error: Connection reset by peer]
<jenatali>
Seems it's the logic in assign_attribute_or_color_invocations() in compiler/glsl/linker.cpp, the temp_attr::compare sorts by size
<jenatali>
Er, nvm, it's by slots, they're both 1
<jenatali>
Oh but qsort's not stable... wonder if this is an MSVC thing O.o
<imirkin>
huh. well, i don't know if it's safe to assume the order in a pre-explicit-by-location world
<imirkin>
so ultimately i think the test may be busted
<imirkin>
even if it often works ;)
mlankhorst has quit [Ping timeout: 480 seconds]
<imirkin>
i guess it should be calling glBindAttribLocation()
<jenatali>
Yeah, looks like the qsort is inverting the order of the inputs for me. I bet it's a C standard library difference. Amazing
<zmike>
have you tried installing linux
<imirkin>
jenatali: can you try adding glBindAttribLocation(draw_prog, 0, "vtx"); glBindAttribLocation(draw_prog, 1, "in_color");
<imirkin>
jenatali: somewhere in piglit_init
<jenatali>
Yeah, trying it now
<imirkin>
(preferably after the draw_prog variable is initialized)
<alyssa>
jenatali: Trying to install Linux?
<jenatali>
:P
<imirkin>
my knowledge of GL was and continues to be rather incomplete, when it comes to understanding how to write complete programs that actually display things
<imirkin>
i can tell you exactly what each command does, but not what set of commands you need to run to get it to actually do stuff ;)
<jenatali>
Hm... doesn't seem to be changing anything
<jenatali>
My knowledge of GL is probably far less complete :P
<imirkin>
jenatali: oh ... i think you need to relink
<imirkin>
jenatali: add a glLinkProgram(draw_prog) after that
<jenatali>
Yeah
<imirkin>
"Attribute bindings do not go into effect until glLinkProgram is called"
mbrost has joined #dri-devel
<jenatali>
Yep, makes sense since the location are assigned during linking
<imirkin>
this stuff always makes sense in hindsight
<jenatali>
Could also just change "vtx" to "piglit_vertex" I think to force it to be 0 since piglit_link_simple_program assigns that location
<imirkin>
sure
<imirkin>
errrr
<imirkin>
maybe.
<imirkin>
one way to find out
<imirkin>
i think i avoided that name since it's variously hooked up to other things
<imirkin>
and i didn't want to fight
<jenatali>
PIGLIT: {"result": "pass" }
<imirkin>
same deal in tf-count-elements of course
<jenatali>
Yep. Of course this pass is somewhat moot since the test also uses triangle fans meaning I'm running primconvert, and therefore doing a CPU readback of the count buffer instead of an indirect
<imirkin>
lol
<imirkin>
feel free to change it
<imirkin>
i probably just copied something
<imirkin>
nvidia supports it all
<jenatali>
Yeah, I had to hack up a different test from fans to strips to actually validate my changes. But there's another indirect_parameters test that actually hits the indirect path for me which validates it
<zmike>
seems suspicious that every other driver passes it
<jenatali>
zmike: Every other driver runs on linux where qsort is apparently stable
<jenatali>
Or at least for a 2-element sort where the sort parameters are ==, it doesn't invert their order
<zmike>
doesn't that imply that any usage of qsort in mesa components would be broken for you?
<jenatali>
Any usage of qsort that expects stable results would be broken, sure, since qsort isn't spec'd to be stable
<zmike>
seems like the more optimal fix would be to provide a functional qsort so that you don't have to try rewriting every GL app to work according to your needs
<jenatali>
This only took me like 20 minutes to get to the bottom of, could've been worse
nchery is now known as Guest150
nchery has joined #dri-devel
nchery is now known as Guest152
nchery has joined #dri-devel
Guest150 has quit [Ping timeout: 480 seconds]
nchery is now known as Guest155
nchery has joined #dri-devel
Guest152 has quit [Ping timeout: 480 seconds]
tzimmermann_ has quit []
mbrost has quit [Read error: Connection reset by peer]
_rgallaispou has joined #dri-devel
Guest155 has quit [Ping timeout: 480 seconds]
rgallaispou has quit [Ping timeout: 480 seconds]
nchery is now known as Guest157
nchery has joined #dri-devel
<marex>
pq: it wasn't about changing your priorities, more like you seemed to be the go-to person for anything weston, hence my question ... anyway, enjoy your holiday
<pq>
marex, ah, yeah, but nowadays I'd like to be the last resort -kinda person. :-)
<marex>
pq: oh :)
<pq>
I have notifications about everything happening in Weston Gitlab.
Akari` has quit [Remote host closed the connection]
Akari has joined #dri-devel
<pq>
if you think you really need me, then talk to me, in Gitlab or irc
<jenatali>
You may have looked but I don't think you said anything
<zmike>
maybe I saw it when I was on vacation
mbrost has joined #dri-devel
nchery has quit [Remote host closed the connection]
nchery has joined #dri-devel
Guest158 has quit [Ping timeout: 480 seconds]
lyudess has quit []
Lyude has joined #dri-devel
gouchi has joined #dri-devel
ybogdano has quit [Ping timeout: 480 seconds]
ybogdano has joined #dri-devel
_rgallaispou has quit [Ping timeout: 480 seconds]
camus has joined #dri-devel
camus1 has quit [Ping timeout: 480 seconds]
Haaninjo has joined #dri-devel
mbrost has quit [Ping timeout: 480 seconds]
mlankhorst has joined #dri-devel
* ccr
boogies.
<alyssa>
zmike: what's vacation
<imirkin>
is that time when you can actually get some work done
<imirkin>
it's*
<zmike>
alyssa: it's the time when you disconnect from irc and discord and slack and all the other chat services and don't reply on gitlab no matter how many times people ping you to tell you they can't live without your expert opinions and/or fixing all the stuff you broke until you can't remember how triangles work anymore
<anarsoul>
alyssa: that's when you work on your personal projects
<ccr>
disconnect .. from .. irc? but how? @__@
karolherbst has quit [Remote host closed the connection]
karolherbst has joined #dri-devel
nchery is now known as Guest179
nchery has joined #dri-devel
<alyssa>
zmike: hmm.. sounds hard, not hanks
<zmike>
the key is you have to do 12 months of work in 11 months
nchery is now known as Guest181
nchery has joined #dri-devel
<alyssa>
instructions unclear, did 11 months of work in 12 months
<zmike>
sounds like you took a vacation
nchery is now known as Guest182
nchery has joined #dri-devel
rasterman has quit [Quit: Gettin' stinky!]
Guest179 has quit [Ping timeout: 480 seconds]
Guest181 has quit [Ping timeout: 480 seconds]
Guest182 has quit [Ping timeout: 480 seconds]
boistordu has joined #dri-devel
<jekstrand>
zmike: But what if you did 12 months of work in 2 months but it only feels like you did 2 weeks of work?
* jekstrand
has no sense of how much work he does.
<imirkin>
jekstrand: esp while unemployed :p
<zmike>
jekstrand: call me after you start at pixar, I need that job
JohnnyonFlame has joined #dri-devel
<jekstrand>
zmike: Oh, I don't mean that it only feels like two weeks of work in terms of exhaustion. You're still exhausted. I mean that you have all the guilt of only having done 2 weeks of work in 2 months.
<zmike>
why would you feel guilty for that
<zmike>
you did work
mbrost has joined #dri-devel
<Sachiel>
after ten years of that you get numb to it, don't worry
<zmike>
it was probably really solid
<jekstrand>
Sachiel: Good! Then I should be numb soon!
<alyssa>
or you could just not do work...
<zmike>
that's called vacation
<alyssa>
Hmmm
<alyssa>
It seems I take a lot of vacation after all
<alyssa>
🙃
* alyssa
is taking a vacation from school right now!
gouchi has quit [Remote host closed the connection]
nchery is now known as Guest185
nchery has joined #dri-devel
Duke`` has quit [Ping timeout: 480 seconds]
mvlad has quit [Remote host closed the connection]
Guest185 has quit [Ping timeout: 480 seconds]
<danvet>
agd5f_, hwentlan__ the joys of Linus using your driver on his main dev machine ...
<danvet>
cc'ed you + amd-gfx list
rauji_ has joined #dri-devel
agd5f_ has quit []
agd5f has joined #dri-devel
<agd5f>
danvet, yeah, just saw that
ngcortes has joined #dri-devel
<agd5f>
is anyone else having problems with updating dim? When I do dim update-branches, it just sits forever on Updating maintainer-tools...
opotin23 has left #dri-devel [#dri-devel]
opotin has joined #dri-devel
<jenatali>
Woo, PR posted to flip 6 more extensions on :)
_rgallaispou has joined #dri-devel
mbrost has quit [Ping timeout: 480 seconds]
join_subline has quit [Remote host closed the connection]