ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
<anarsoul> pushqrdx: just ask
<alyssa> writing mesa-style C after deep diving in kernel is weird
<alyssa> like I can declare variables in the middle of my function? I can do that?!
vocalfan has quit []
<aperezdc> mmh, I am trying to set up 90 degree rotation, and after much fiddling I am stuck in these kernel messages (with /sys/module/drm/paramters/debug = 0x14):
<aperezdc> oct 06 01:57:52 momiji kernel: i915 0000:00:02.0: [drm:skl_plane_check [i915]] Y/Yf tiling required for 90/270!
<aperezdc> oct 06 02:02:41 momiji kernel: i915 0000:00:02.0: [drm:skl_plane_check [i915]] RC support only with 0/180 degree rotation (2)
<aperezdc> first with legacy modesetting, the second with atomic
<pushqrdx> anarsoul for some reason an opengl application (BespokeSynth) is dropping input events if it's render at full framerate, i did some digging and it seems to be related to render batching in mesa as the problem disappears when i either disable DRI or use software rendering. Another way to greatly reduce the issue is if i use INTEL_DEBUG=sync
<pushqrdx> problem is happening for other users as well with different hardware and the workaround that Bespoke implemented was to half the output framerate
alyssa has left #dri-devel [#dri-devel]
<pushqrdx> same problem happens for me in another application called Helio Workstation too, and goes away with using software rendering
<jenatali> Temperature check: If people wanted to add more build system "wrappers" in the style of the Android integration, would people be amenable to that as long as they're small/isolated, or completely opposed on principle?
columbarius has joined #dri-devel
co1umbarius has quit [Ping timeout: 480 seconds]
nchery has quit [Quit: Leaving]
pnowack has quit [Quit: pnowack]
<anarsoul> pushqrdx: (I'm not Intel dev) I'd suggest to open an issue at https://gitlab.freedesktop.org/mesa/mesa/-/issues
<bnieuwenhuizen> jenatali: you mean in style of the Android builds we do using meson?
<jenatali> bnieuwenhuizen: Yeah, a foreign build system that builds by just invoking meson + ninja
<jenatali> The Chromium folks apparently need any dependencies to be able to build using both GN and CMake... and the spirv_to_dxil stuff we've been working on would be one of those dependencies
<jenatali> So I'm trying to convince them to consider using that same Android.mk wrapper approach that was merged a few months back, and I realized I should check if people here would object :)
<bnieuwenhuizen> ah, not familiar with the wrapper, not the right person to answer that one
<jenatali> Basically it just reads the environment, constructs a command line + native file, invokes meson + ninja to build/install, and then imports the resulting binaries as if they were prebuilts
<pushqrdx> anarsoul it does happen with both intel and other gpus
jewins has quit [Quit: jewins]
bluebugs has quit [Ping timeout: 480 seconds]
<anarsoul> pushqrdx: that likely suggests that something's wrong with BespokeSynth, but yet please open an issue at mesa gitlab
Cong has joined #dri-devel
Cong has quit [Remote host closed the connection]
Cong has joined #dri-devel
Cong has quit [Remote host closed the connection]
Cong has joined #dri-devel
Cong has quit [Remote host closed the connection]
moa has joined #dri-devel
boistordu has quit [Ping timeout: 480 seconds]
Company has quit [Quit: Leaving]
guru_ has quit []
oneforall2 has joined #dri-devel
idr has quit [Quit: Leaving]
mattst88 has joined #dri-devel
<pushqrdx> dj-death i think i fixed the issue, can you try https://github.com/BespokeSynth/BespokeSynth/pull/349
flto_ has joined #dri-devel
flto has quit [Ping timeout: 480 seconds]
flto_ has quit []
flto has joined #dri-devel
agd5f has quit [Remote host closed the connection]
agd5f has joined #dri-devel
Duke`` has joined #dri-devel
slattann has joined #dri-devel
elongbug has joined #dri-devel
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
i-garrison has joined #dri-devel
itoral has joined #dri-devel
rpigott has quit [Remote host closed the connection]
rpigott has joined #dri-devel
adjtm has quit [Remote host closed the connection]
adjtm has joined #dri-devel
adjtm has quit [Remote host closed the connection]
adjtm has joined #dri-devel
Hi-Angel has joined #dri-devel
mlankhorst has joined #dri-devel
frieder has joined #dri-devel
danvet has joined #dri-devel
lemonzest has joined #dri-devel
slattann has quit []
unsolo has joined #dri-devel
pnowack has joined #dri-devel
shashank1202 has joined #dri-devel
slattann has joined #dri-devel
jessica_24_ has joined #dri-devel
jessica_24_ has quit []
<pq> company, oh why don't stay online...
lynxeye has joined #dri-devel
<pq> HdkR, since company is not here, I'll tell you: 10-bit is orthogonal to wide gamut, or at least that is what we should assume until we know how specific individual monitor enables WCG or not.
<pq> 10-bit framebuffer is totally unrelated to whether a monitor actually displays 10-bit color
tursulin has joined #dri-devel
<pq> framebuffer bit depth <-> KMS pipeline min bit depth <-> monitor cable (signalling) bit depth <-> observed bit depth <-> panel physical bit depth
<pq> they are all fundamentally independent, but at each step defines its minimum possible noise level (precision) that depends on both bit depth and (non-linear) value encoding
<pq> (*max precision)
<pq> the overall visual effect is roughly the minimum precision over everything
rasterman has joined #dri-devel
<pq> whether that end result then maps to sRGB gamut or some WCG, or SDR or HDR, are yet other matters
<pq> then we can start adding assumptions on which depths are tied together in practise - that I don't know
rgallaispou has joined #dri-devel
slattann has quit []
pcercuei has joined #dri-devel
vivijim has joined #dri-devel
<mlankhorst> airlied: will rebase and drop vc4
thellstrom has joined #dri-devel
xexaxo has joined #dri-devel
Ahuj has joined #dri-devel
slattann has joined #dri-devel
lemonzest has quit [Quit: WeeChat 3.2]
shashank1202 has quit [Quit: Connection closed for inactivity]
slattann1 has joined #dri-devel
slattann has quit [Ping timeout: 480 seconds]
gawin has joined #dri-devel
lemonzest has joined #dri-devel
Lucretia-backup has joined #dri-devel
Lucretia has quit [Ping timeout: 480 seconds]
flacks has quit [Quit: Quitter]
flacks has joined #dri-devel
pochu has joined #dri-devel
vivijim has quit [Read error: Connection reset by peer]
vivijim has joined #dri-devel
slattann1 has quit []
iive has joined #dri-devel
shashank1202 has joined #dri-devel
itoral has quit []
Hi-Angel has quit [Ping timeout: 480 seconds]
JohnnyonFlame has quit [Ping timeout: 480 seconds]
Company has joined #dri-devel
Ahuj has quit [Ping timeout: 480 seconds]
Ahuj has joined #dri-devel
Danct12 has quit [Quit: Quitting]
kts has joined #dri-devel
kts has quit []
kts has joined #dri-devel
kts has quit []
kts has joined #dri-devel
kts has quit []
kts has joined #dri-devel
Guest1834 has quit []
enick_789 has joined #dri-devel
Danct12 has joined #dri-devel
kts has quit [Ping timeout: 480 seconds]
Ahuj has quit [Ping timeout: 480 seconds]
Danct12 has quit []
shashank1202 has quit [Quit: Connection closed for inactivity]
pochu has quit [Quit: leaving]
Company has quit [Ping timeout: 480 seconds]
Company has joined #dri-devel
adjtm is now known as Guest1995
adjtm has joined #dri-devel
Guest1995 has quit [Read error: No route to host]
<cwabbott> hakzsam: why does radv_emit_subpass_barrier care about whether the framebuffer is imageless? looks like it's missing a workaround in the imageless case
<cwabbott> also it doesn't have to check if the framebuffer exists bc it's only called in BeginRenderPass/EndRenderPass/NextSubpass where the framebuffer is available
nchery has joined #dri-devel
<cwabbott> you shouldn't be skipping workarounds unless it's actually not needed...
<cwabbott> git blame says you added this
<hakzsam> it was for reducing the number of L2 cache flushes, if we know the image some can be skipped. The fb can be NULL for secondary command buffers ...
<cwabbott> you can't call NextSubpass without calling BeginRenderPass in the same cmdbuf
<hakzsam> I'm not sure what workaround you are talking about though
<hakzsam> pretty sure the fb check was needed by that time
<cwabbott> it can only be null inside a single subpass
Danct12 has joined #dri-devel
<cwabbott> hakzsam: it's for for 4c99d6f ("radv: flush L2 for images affected by the pipe misaligned issue on GFX10+")
<cwabbott> although yeah, I guess it does flush conservatively
<cwabbott> but you should always know the framebuffer at that point
<cwabbott> or rather you should always know the attachments
fxkamd has joined #dri-devel
JohnnyonFlame has joined #dri-devel
<hakzsam> cwabbott: I will double check
<hakzsam> thanks for pointing this out
<cwabbott> hakzsam: CmdExecuteCommands says "If vkCmdExecuteCommands is being called within a render pass instance, each element of pCommandBuffers must have been recorded with the VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT" and "If vkCmdExecuteCommands is being called within a render pass instance, each element of pCommandBuffers must have been recorded with VkCommandBufferInheritanceInfo::subpass set to the index of the subpass
<cwabbott> which the given command buffer will be executed in"
ppascher has quit [Ping timeout: 480 seconds]
<cwabbott> I guess the implication being that a secondary can't span more than one subpass at a time so can't call NextSubpass or EndRenderPass
fxkamd has quit []
ppascher has joined #dri-devel
shashank1202 has joined #dri-devel
Anorelsan has joined #dri-devel
gouchi has joined #dri-devel
ybogdano has joined #dri-devel
rgallaispou has quit [Ping timeout: 480 seconds]
fxkamd has joined #dri-devel
slattann has joined #dri-devel
<karolherbst> tagr: btw.. still in Hamburg? :D if you keep having issues with setting up UEFI and fedora I can actually help you in person :D
<karolherbst> but normally this should'nt be that problematic, sometimes you just need to do it another time
JohnnyonFlame has quit [Ping timeout: 480 seconds]
ngcortes has joined #dri-devel
RauLi has joined #dri-devel
slattann has quit []
RauLi has quit []
mlankhorst has quit [Ping timeout: 480 seconds]
Danct12 has quit [Quit: Quitting]
shashank1202 has quit [Quit: Connection closed for inactivity]
gouchi has quit [Remote host closed the connection]
Anorelsan has quit [Quit: Leaving]
i-garrison has quit []
i-garrison has joined #dri-devel
frieder has quit [Remote host closed the connection]
xexaxo has quit [Read error: Connection reset by peer]
JohnnyonFlame has joined #dri-devel
xexaxo has joined #dri-devel
<hikiko> jekstrand, is it ok to merge MR 13162 with my review only? all the added code is between ifdef __somebsd__ so it can't affect Linux but without it you can't create a Vulkan device on FreeBSD when the GPU is Gen < 7. I don't know who to tag to review a BSD/Gen6 problem. :/
<jekstrand> hikiko: Let me dig through some IRC logs. I think mlankhorst explained it to me once.
<hikiko> ok, thank you!
gouchi has joined #dri-devel
<jekstrand> hikiko: My IRC logs are not informative. :(
<jekstrand> hikiko: If it doesn't cause HSW to blow up, it's probably fine.
<hikiko> jbeich who wrote the patch had actually tested it on HSW, I use an IVB and verified it works there too
<jekstrand> hikiko: Do you need userptr for iris/crocus?
<hikiko> only for clover (experimental)
<jekstrand> For ANV on IVB/HSW, it's a hard requirement.
<jekstrand> It's needed for clover?!?
<hikiko> this is what jbeich said actually
* jekstrand doesn't remember why that would be
<jekstrand> Ok.
<hikiko> I didn't test
<hikiko> but I can remove the crocus/iris part
<hikiko> I was able to run opengl without it
ngcortes has quit [Remote host closed the connection]
kallisti6 has quit [Remote host closed the connection]
JTL has quit [Remote host closed the connection]
gpiccoli has quit [Quit: Bears...Beets...Battlestar Galactica]
<jekstrand> hikiko: Ok, I think I've done enough reading to give myself a very sketchy understanding and I left a few comments.
<jekstrand> hikiko: I think we can at least merge something that gets ANV going again.
<hikiko> ok, then I'll remove the iris/crocus changes as you can use the driver without them
<hikiko> jekstrand, thank you!
Lucretia-backup has quit [Remote host closed the connection]
Lucretia has joined #dri-devel
<jekstrand> hikiko: YW. Sorry it took so long.
kallisti5 has joined #dri-devel
gpiccoli has joined #dri-devel
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
JTL has joined #dri-devel
pushqrdx has quit [Ping timeout: 480 seconds]
<jekstrand> hikiko: gen < 7? You're not running on Sandy-bridge somehow, are you? :P
<jekstrand> That'd be pretty impressive. hehe.
<jekstrand> hikiko: I'm not sure how I feel about enabling things that require you to be running as root. :-/
pushqrdx has joined #dri-devel
<jekstrand> But I'm not sure how I feel about the other options, either. :-/
<jekstrand> The userptr hack is admittedly terrible.
<jekstrand> It'd be better if we had a "real" BO growing API in the kernel.
<airlied> jekstrand: no way on things have to be run as root
<jekstrand> ^^
<airlied> that's a rabbit hole we don't go down again, suggesting users run any app using GL or Vulkan as root isn't a good idea
<jekstrand> airlied: !13162 if you want to weigh in there.
gawin has quit [Ping timeout: 480 seconds]
ybogdano has quit [Ping timeout: 480 seconds]
JohnnyonFlame has quit [Ping timeout: 480 seconds]
<jekstrand> hikiko: Another option would be to use softpin even on Gen7. We can't use it for everything but we can probably use it for the state buffers.
<jekstrand> danvet: We always have PPGTT on gen7, right?
* jekstrand should know this
<danvet> no
<danvet> currently disabled on all gen7
<jekstrand> *sigh*
<jekstrand> There goes that clever plan
<clever> ?
<clever> :D
<jekstrand> :P
<APic> B-)
<danvet> jekstrand, we're throwing out unsynchronized userptr
<danvet> pls don't merge or that nice plan goes down the drain because "regression!"
<danvet> s/are throwing out/have thrown out/
<danvet> build your kernel with proper mmu notifiers and get over it
<danvet> airlied, ^^
<airlied> I think it's the freebsd kernel
<jekstrand> danvet: Yeah, it's bsd
<danvet> well then make it bsd only at least
<jekstrand> And I think the Mesa patch has been thoroughly NAKd at this point
<danvet> it's still terrible
ybogdano has joined #dri-devel
<airlied> yeah I don't think we want to reintroduce that upstream
<danvet> I added a comment to nack it harder :-)
<jekstrand> I think it's been well-nacked now. :)
danvet has quit [Ping timeout: 480 seconds]
mattrope has joined #dri-devel
lemonzest has quit [Quit: WeeChat 3.2]
<hikiko> sorry I read this now after I replied :)
<hikiko> ok I won't merge it
<jekstrand> hikiko: Yeah, sorry this is such a mess. :(
<hikiko> I thought with ifdef __FreeBSD__ it would be ok but it makes sense to prefer a proper fix from running a hack as root
ngcortes has joined #dri-devel
enick_789 has left #dri-devel [#dri-devel]
Duke`` has quit [Ping timeout: 480 seconds]
DrNick has joined #dri-devel
ybogdano has quit [Ping timeout: 480 seconds]
gouchi has quit [Remote host closed the connection]
gawin has joined #dri-devel
ybogdano has joined #dri-devel
lynxeye has quit []
<gawin> airlied: sorry for that lazy MR for crocus in last week, now I think it should be just an issue
<gawin> I mean issue how to handle &...->... stuff overall
ybogdano has quit [Ping timeout: 480 seconds]
ybogdano has joined #dri-devel
Bennett has joined #dri-devel
ybogdano has quit [Ping timeout: 480 seconds]
moa is now known as bluebugs
thellstrom1 has joined #dri-devel
thellstrom has quit [Read error: Connection reset by peer]
gawin has quit [Quit: Konversation terminated!]
alyssa has joined #dri-devel
<alyssa> Drive-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
<alyssa> ...lol
<alyssa> zmike: ^ found that looking through gitlab
<zmike> alyssa: look if you say it's a drive-by, that's what it is
adjtm has quit [Remote host closed the connection]
adjtm has joined #dri-devel
<alyssa> :d
<alyssa> + bool clamp_to_border = sampler->min_img_filter != PIPE_TEX_FILTER_NEAREST &&
<alyssa> + sampler->mag_img_filter != PIPE_TEX_FILTER_NEAREST;
<alyssa> shouldn't that be ||?
<alyssa> also, if both min and mag are NEAREST, can't we translate to clamp_to_edge and skip the expensive shader lowering?
JohnnyonFlame has joined #dri-devel
yoslin_ has joined #dri-devel
pcercuei has quit [Quit: dodo]
idr has joined #dri-devel
yoslin has quit [Ping timeout: 480 seconds]
pnowack has quit [Quit: pnowack]
Bennett has quit [Remote host closed the connection]
adjtm has quit [Remote host closed the connection]
adjtm has joined #dri-devel
tursulin has quit [Read error: Connection reset by peer]
iive has quit []