ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
OveRSeeR has joined #dri-devel
OveRSeeR has quit []
<jekstrand>
daniels: Hah! Yeah, that's not happening. :P
<jekstrand>
daniels: It'd require me to be able to learn how to build/test Android which I plan never to do.
<jekstrand>
Too many good people have been lost down that rabit hole
<bnieuwenhuizen>
I wonder how hard it would be to mock that stuff on non-Android. Especially if we can implement minigbm on top of gbm and then use cros_gralloc
orbea has quit [Quit: You defeated orbea! 2383232 XP gained!]
orbea has joined #dri-devel
<Kayden>
minigbm-on-top-of-gbm just sounds bizarre, if you don't know what it is
Haaninjo has quit [Quit: Ex-Chat]
columbarius has joined #dri-devel
<bnieuwenhuizen>
well, everyone seems to play the "it is kinda a completely separate library with its own API" card so if I can make it not completely clash to have both in the same process it actually sounds decent to me
<bnieuwenhuizen>
something we might want even since we have minigbm-on-dri and all the usages of dri seem to be slowly going the gbm way
co1umbarius has quit [Ping timeout: 480 seconds]
<jekstrand>
Or we could just stop supporting ancient Android versions...
* jekstrand
runs
<bnieuwenhuizen>
wait, what has this to do with ancient android versions?
<jekstrand>
According to the comments in libsync.h, recent android versions can just use the Mesa version because they have "real" sync_file. At least, that's the way I understood it. I could be totally wrong, though.
rasterman has quit [Quit: Gettin' stinky!]
<daniels>
correct
<bnieuwenhuizen>
uh I was just talking about developing for android
<bnieuwenhuizen>
AFAIU even with the new ahb stuff we gotta call into gralloc for image allocation sometimes
<daniels>
bnieuwenhuizen: it's not mandatory, no
<jekstrand>
Right. It'd be nice if we could test Android paths on !Android but, then again, that setup will likely have a distinct set of bugs and it'll just be a disaster.
<daniels>
bnieuwenhuizen: we periodically refresh Android-on-generic-Linux to make sure it keeps working; when we last did that a few weeks ago, it was using minigbm but only because it meant dealing with less build system friction, as the backend just called into the usual gem alloc ioctls
<daniels>
minigbm used to have a meaningful difference to Mesa-GBM in that it had a couple of new entrypoints, but tomeu brought those back to Mesa-GBM so they're on level terms now
<bnieuwenhuizen>
what is not mandatory?
<bnieuwenhuizen>
AFAIU the problem is cros_gralloc looks a bit at minigbm internals?
<bnieuwenhuizen>
if we're talking about directly using gbm
<daniels>
hmm, I didn't think that was the case, but I might have missed it
<daniels>
(the 'it' that I didn't think was mandatory was both calling into gralloc + using minigbm)
<bnieuwenhuizen>
maybe technically not, but I'm not sure how we get a gralloc handle for export without calling into gralloc
<bnieuwenhuizen>
(for the vulkan ahb export case)
<daniels>
gbm_gralloc, no?
<bnieuwenhuizen>
oh you have an alternative gralloc impl
<daniels>
it's in-tree in AOSP now afaik
<daniels>
(I don't actually know this first-hand; the last time I touched it personally was ~5y ago and ever since then I've managed to avoid having to deal with it personally ...)
<bnieuwenhuizen>
interesting
lemonzest has quit [Quit: WeeChat 3.3]
<bnieuwenhuizen>
AFAIU we do a bunch of weird stuff in minigbm because e.g. we have to report the map-stride on image creation time (or I guess queryable from that point) instead of at map time
* daniels
blinks
<daniels>
define 'image creation'? I assume if you're talking about minigbm then you mean a gbm_bo ... ?
<bnieuwenhuizen>
and of course that isn't a great match for gallium especially around drivers that copy into a temp image
<bnieuwenhuizen>
yeah gbm_bo, AFAIU the app can query from gralloc before mapping or something like that
<daniels>
why would you not know the stride(s) of a gbm_bo which is created but not yet mapped?
gawin has joined #dri-devel
<jekstrand>
Why not? As soon as the size has been calculated, so has the stride.
<jekstrand>
So you need to know the size by the time you've allocated memory
<bnieuwenhuizen>
because if the driver copies into a temp linear image for mapping a tiled image the stride might be different for that temp image
<jekstrand>
right...
<bnieuwenhuizen>
in particular for AMD the stride requirement for linear images ends up higher than for tiled images
<bnieuwenhuizen>
(or rather the same in bytes, but for tiled the alignment is for the tile which means lower horizontal alignment)
<daniels>
uh.
<daniels>
is it the fact that the image is mapped _at all_ that decides linear vs. tiled, or something else?
<bnieuwenhuizen>
I think on Android everything can be mapped?
<daniels>
right
<bnieuwenhuizen>
and the app can request linear but since above it ends up kinda meaningless
<daniels>
so do you unconditionally morph/resolve to linear when it's mapped, or ... ?
<bnieuwenhuizen>
yeah, no other choice
<daniels>
well
<daniels>
I hear there's some kind of API which lets you explicitly declare acceptable layouts
<bnieuwenhuizen>
everyone expects a linear view AFAIU
<daniels>
well, gbm_bo_map() always exposes a row-linear+uncompressed mapping, however it does also provide stride as an out param
<bnieuwenhuizen>
yeah so the gbm interface is fine, but then cros_gralloc kinda ignores because it has nothing useful it can do with it at that point
<jekstrand>
gawin: I just tagged danvet. He can add others as needed.
<gawin>
thanks
moa has joined #dri-devel
bluebugs has quit [Ping timeout: 480 seconds]
gawin has quit [Ping timeout: 480 seconds]
lstrano has joined #dri-devel
aravind has joined #dri-devel
LexSfX has quit []
ceyusa has joined #dri-devel
Walter_ has joined #dri-devel
Walter_ has quit [Killed (NickServ (Too many failed password attempts.))]
aravind has quit []
aravind has joined #dri-devel
Wally has joined #dri-devel
<Wally>
What do Vulkan drivers provide to the Vulkan loader? Looking through the Vulkan-Loader documentation for drivers it seems that it just needs an entrypoint and a few extentions to the drivers, is that true? Not trying to make a Vulkan driver or anything(at the moment) just curious.
lemonzest has joined #dri-devel
<Wally>
Also how much does the loader handle itself?
Wally has quit [Remote host closed the connection]
LexSfX has joined #dri-devel
Wally has joined #dri-devel
<Wally>
Sorry, I was briefly logged off, if anyone answered my question above could you please copy paste it bellow?
YuGiOhJCJ has joined #dri-devel
<HdkR>
Wally: Driver should only need to provide a couple of entry points and then the loader pulls all entry points through the vkGetDeviceProcAddr entrypoint
<HdkR>
Driver also provides a json so the loader can find the library
<Wally>
Thats all it does to interface with the loader?
<Wally>
why do drivers need to be updated for Vulkan 1.0,....., 1.x
<Wally>
Ive read through the whole document yesterday, I was extremly confused
<Wally>
(at the simplicity)
<HdkR>
But also yes, a driver needs to be updated to support a new version because just supported all the new feature bits isn't enough to claim support for a new Vulkan version :)
<HdkR>
just supporting*
<Wally>
so all the necessary extensions would have to be provided by the driver for a version(not the loader0
Wally has quit [Quit: Page closed]
<HdkR>
Definitely all the required extensions need to be provided by the driver
Duke`` has joined #dri-devel
Duke`` has quit []
Duke`` has joined #dri-devel
itoral has joined #dri-devel
ewrvvvvvvvvvvvvvvvvvvvvvvvvvvd has quit [Remote host closed the connection]
aravind has quit [Read error: Connection reset by peer]
aravind has joined #dri-devel
itoral_ has joined #dri-devel
itoral has quit [Ping timeout: 480 seconds]
pcercuei has joined #dri-devel
<MrCooper>
hakzsam: could the radv-fossils job be changed to clone the fossils-db repo only once?
lstrano has quit [Ping timeout: 480 seconds]
Haaninjo has joined #dri-devel
thellstrom has quit [Remote host closed the connection]
thellstrom has joined #dri-devel
<hakzsam>
MrCooper: I don't know offhand how to do that but yes maybe. It could also be purely removed (it never triggered anything useful). The initial plan was to run our private fossils db repository but it never happened
agners has joined #dri-devel
pochu has joined #dri-devel
ewryyyyyyyyyyyyyyyyyyyyyyyyyyd has joined #dri-devel
join_subline has joined #dri-devel
mvlad has joined #dri-devel
<tomeu>
eric_engestrom: was wondering what has brought you to the tracie dashboard
<tomeu>
I'm currently thinking that it's not bringing much value over what is in gitlab already
aravind has quit [Remote host closed the connection]
aravind has joined #dri-devel
qyliss has quit [Quit: bye]
itoral_ has quit [Remote host closed the connection]
qyliss has joined #dri-devel
qyliss has quit []
qyliss has joined #dri-devel
Company has joined #dri-devel
Company has quit []
txenoo has joined #dri-devel
gawin has joined #dri-devel
OftenTimeConsuming has quit [Remote host closed the connection]
frieder has joined #dri-devel
frieder has quit [Remote host closed the connection]
<eric_engestrom>
tomeu: I was catching up with a bunch of gitlab notifications, and someone asked something about tracie so I got curious and looked at the repo, and I noticed an old issue that I thought wouldn't take me long to fix, so I did
<eric_engestrom>
but I'm not using it or anything
<tomeu>
ah, phew :)
<eric_engestrom>
haha
hasebastian[m] has joined #dri-devel
OftenTimeConsuming has joined #dri-devel
<ccr>
blarp
gawin has quit [Ping timeout: 480 seconds]
sneil has quit [Quit: Leaving]
Kamikaze has joined #dri-devel
<Kamikaze>
Hi. I'm trying to get panfrost working on my firefly RK3399, which has a Mali T860MP4. I'm using Arch Linux ARM with a custom compiled 5.16.0-rc7 with panfrost compiled as a module. On userspace I have latest available mesa (21.3.3) & X (1.21.1.2) just from the package manager. My understanding was that I shouldn't need to do much else and the modeset driver would then somehow know to use panfrost for DRI, but it always uses llvmpipe.
sneil has joined #dri-devel
ivyl has quit [Quit: issued !quit command]
ivyl has joined #dri-devel
karolherbst has joined #dri-devel
ella-0 has joined #dri-devel
ella-0_ has quit [Read error: Connection reset by peer]
gawin has joined #dri-devel
cef is now known as Guest10149
cef has joined #dri-devel
Guest10149 has quit [Ping timeout: 480 seconds]
anujp has joined #dri-devel
Kamikaze has quit [Quit: Page closed]
kem has quit [Ping timeout: 480 seconds]
pendingchaos has quit [Ping timeout: 480 seconds]
kem has joined #dri-devel
Duke`` has joined #dri-devel
pendingchaos has joined #dri-devel
cef is now known as Guest10152
cef has joined #dri-devel
aravind has quit [Ping timeout: 480 seconds]
gawin has quit [Ping timeout: 480 seconds]
Guest10152 has quit [Ping timeout: 480 seconds]
rasterman has quit [Quit: Gettin' stinky!]
macromorgan has joined #dri-devel
Surkow|laptop has quit [Quit: 418 I'm a teapot - NOP NOP NOP]
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
ybogdano has joined #dri-devel
Namarrgon has joined #dri-devel
moa has quit [Ping timeout: 480 seconds]
angerctl has quit [Ping timeout: 480 seconds]
mbrost has joined #dri-devel
bluebugs has joined #dri-devel
rasterman has joined #dri-devel
sdutt has joined #dri-devel
mbrost has quit [Remote host closed the connection]
pochu has quit [Quit: leaving]
jhli has joined #dri-devel
gouchi has joined #dri-devel
iive has joined #dri-devel
fxkamd has joined #dri-devel
thellstrom has quit [Ping timeout: 480 seconds]
gawin has joined #dri-devel
ewryyyyyyyyyyyyyyyyyyyyyyyyyyd has quit [Remote host closed the connection]
LexSfX has quit []
LexSfX has joined #dri-devel
boistordu has joined #dri-devel
pnowack has quit [Quit: pnowack]
fxkamd has quit []
heat has joined #dri-devel
jhli has quit []
<jekstrand>
"Is this because Intels upcoming GPUs are based off AMDs designs?"
<jekstrand>
This is why we love phoronix...
<zmike>
I saw a top graphics news site reporting on your upcoming move into the network space just a moment ago
gawin has quit [Ping timeout: 480 seconds]
boistordu has quit [Remote host closed the connection]
<jekstrand>
FAKE NEWS!!!
<zmike>
it's being reported, so there has to be something to it
<jekstrand>
It's something, all right. :P
<ccr>
"I saw it on the Internet, thus it must be true."
<anholt>
mareko: how are you at r600 shaders? I've got a couple of things noted in the nir-to-tgsi regressions for it at this point, and someone that knows r600 might be able to identify the bugs much faster.
sdutt has quit [Read error: Connection reset by peer]
mlankhorst has quit [Ping timeout: 480 seconds]
<jenatali>
Finally bit the bullet and started implementing tessellation to get some GL4.0 support
<jenatali>
Up to 31 patches so far, and not quite done yet :D
<kisak>
only 45 to go! good luck!
<jenatali>
I think I'm almost done, just gotta hook up barrier
<jenatali>
Lots of fun though since DXIL expresses tess ctrl / hull shaders as 2 functions as opposed to GL's 1, and our entire bitcode emitter didn't support multiple functions