2021-06-22 12:29
ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
00:08
camus1 has joined #dri-devel
00:09
camus has quit [Read error: Connection reset by peer]
00:18
mbrost has quit [Ping timeout: 480 seconds]
00:18
nchery has quit [Quit: Leaving]
00:18
pnowack has quit [Quit: pnowack]
00:19
Lucretia has quit []
00:45
alyssa has left #dri-devel [#dri-devel]
00:49
rasterman has quit [Quit: Gettin' stinky!]
00:50
alatiera has quit [Ping timeout: 480 seconds]
00:55
alatiera has joined #dri-devel
00:55
Lightkey has quit [Ping timeout: 480 seconds]
00:55
alatiera is now known as Guest1826
00:58
alatiera3 has joined #dri-devel
01:03
Guest1826 has quit [Ping timeout: 480 seconds]
01:05
Lightkey has joined #dri-devel
02:01
sdutt_ has joined #dri-devel
02:07
sdutt has quit [Ping timeout: 480 seconds]
02:10
boistordu_old has joined #dri-devel
02:17
sdutt_ has quit [Ping timeout: 480 seconds]
02:17
boistordu has quit [Ping timeout: 480 seconds]
02:21
alatiera3 is now known as alatiera
02:22
xexaxo_ has quit [Ping timeout: 480 seconds]
02:34
Bennett has quit [Remote host closed the connection]
02:40
boistordu_old has quit [Remote host closed the connection]
02:47
boistordu_ex has joined #dri-devel
03:00
camus has joined #dri-devel
03:05
camus1 has quit [Ping timeout: 480 seconds]
03:17
camus1 has joined #dri-devel
03:17
camus has quit [Remote host closed the connection]
03:17
cphealy has quit [Remote host closed the connection]
03:29
lemonzest has quit [Quit: Quitting]
03:32
boistordu_ex has quit [Remote host closed the connection]
03:37
craftyguy has quit [Remote host closed the connection]
03:38
craftyguy has joined #dri-devel
03:41
boistordu_ex has joined #dri-devel
03:41
sdutt has joined #dri-devel
04:00
mbrost has joined #dri-devel
04:06
NiksDev has quit [Ping timeout: 480 seconds]
04:34
Company has quit [Quit: Leaving]
04:52
camus1 has quit [Remote host closed the connection]
04:52
camus has joined #dri-devel
05:06
meatloaf has joined #dri-devel
05:08
Duke`` has joined #dri-devel
05:12
macromorgan has joined #dri-devel
05:12
macromorgan_ has quit [Read error: Connection reset by peer]
05:16
Duke`` has joined #dri-devel
05:22
<
airlied >
uggh they added bunch of GLES3.2 test to GL4.5 cts, but they end up mismatching shader versions between 3.1 and 3.2
05:24
mbrost has quit [Ping timeout: 480 seconds]
05:25
<
HdkR >
airlied: Is that a problem with ES 3.1 shaders being forward compatible with ES 3.2 contexts?
05:25
<
airlied >
HdkR: the ES rules are you have to have all the shader versions the same
05:25
<
HdkR >
oh, mismatched between stages
05:25
<
HdkR >
That's a big oof
05:26
<
airlied >
the desktop rules are the same, but ignored enough they changed the spec,
05:26
<
HdkR >
Did they only test these changes on the Windows drivers? :D
05:26
<
HdkR >
Nvidia blob or bust?
05:28
itoral has joined #dri-devel
05:29
<
airlied >
looks like it :-P
05:30
<
airlied >
now I suppose the question is whether GLES shaders on Desktop stack matching matters
05:46
* jekstrand
really wishes he could stop writing WSI patches....
05:49
mlankhorst has joined #dri-devel
05:50
<
airlied >
okay 3 internal vk-gl-cts issues filed for the new tests
05:54
pnowack has joined #dri-devel
05:58
<
airlied >
oh one bug looks to be a mesa bug
05:59
sdutt has quit [Ping timeout: 480 seconds]
06:03
mattrope has quit [Read error: Connection reset by peer]
06:09
camus has quit [Remote host closed the connection]
06:10
camus has joined #dri-devel
06:14
danvet has joined #dri-devel
06:20
Duke`` has quit [Ping timeout: 480 seconds]
06:39
alanc has quit [Remote host closed the connection]
06:39
alanc has joined #dri-devel
06:41
meatloaf|2 has joined #dri-devel
06:48
meatloaf has quit [Ping timeout: 480 seconds]
06:48
sumits has quit [Remote host closed the connection]
06:49
cbaylis has joined #dri-devel
06:54
sumits has joined #dri-devel
07:27
camus1 has joined #dri-devel
07:30
camus has quit [Ping timeout: 480 seconds]
07:58
pcercuei has joined #dri-devel
08:00
bcarvalho has joined #dri-devel
08:01
lynxeye has joined #dri-devel
08:07
tursulin has joined #dri-devel
08:11
Lucretia has joined #dri-devel
08:14
itoral has quit [Remote host closed the connection]
08:24
rasterman has joined #dri-devel
08:31
bcarvalho has quit [Read error: Connection reset by peer]
08:33
bcarvalho has joined #dri-devel
08:52
<
pq >
emersion, what do you mean you can't ask for pageflip event on an atomic commit turning a CRTC on->off? I believe Weston does exactly that and it works.
08:52
<
pq >
emersion, or are you hitting a CRTC off->off transition?
08:52
<
emersion >
pq, i'm getting [drm:drm_atomic_check_only [drm]] [CRTC:70:crtc-1] requesting event but off
08:52
<
pq >
...which Weston specifically avoids adding to a atomic req because it'd... fail
08:53
<
emersion >
ah, that may be off->off yeah
08:53
<
pq >
which is annoying code
08:53
<
emersion >
hm. interesting.
08:53
<
emersion >
that's a very good thing to know, thanks
08:53
<
pq >
off->off transition just cannot create an event
08:54
<
emersion >
but on->off can?
08:54
<
emersion >
ah, i guess for buffer release
08:55
<
emersion >
should really document these flags…
08:55
<
emersion >
oh well.
08:56
<
pq >
yeah, on->off seems to work just fine.
09:02
<
daniels >
I honestly don't mind on->off not generating a timestamp
09:03
<
emersion >
instead of inventing a timestamp, making it clear it's non-sensical lgtm
09:03
Namarrgon has quit [Ping timeout: 480 seconds]
09:04
<
emersion >
but yeah, ideally we'd have two callbacks, one for "the buffer has been presented at $timestamp", the other for "the buffer is released"
09:04
<
emersion >
OUT_FENCE_PTR kind of addresses this
09:04
<
MrCooper >
at the least, if something wants a timestamp for that, first it needs to be clearly defined in the kernel what the timestamp corresponds to
09:05
Namarrgon has joined #dri-devel
09:20
xexaxo_ has joined #dri-devel
09:30
vivijim has joined #dri-devel
09:51
<
airlied >
mripard: ack
09:52
<
airlied >
evadot: not sure, just i915 added some newer apis over the years that 965 has optional support for but crocus depends on
09:52
<
airlied >
evadot: would have to dig in a bit if that might be in, esp if Linux work on the same hw, my gm45 at least works as well as 965 does
09:53
<
evadot >
airlied: ok thanks, if you know those api I'll have a look on our side too
09:58
iive has joined #dri-devel
10:08
jkrzyszt has joined #dri-devel
10:09
itoral has joined #dri-devel
10:29
K`den has joined #dri-devel
10:29
Kayden has quit [Remote host closed the connection]
10:40
flacks has quit [Quit: Quitter]
10:44
flacks has joined #dri-devel
11:09
phomes has joined #dri-devel
11:14
degasus has quit [Ping timeout: 480 seconds]
11:18
tobiasjakobi has joined #dri-devel
11:18
tobiasjakobi has quit [Remote host closed the connection]
11:26
degasus has joined #dri-devel
11:27
degasus is now known as Guest1889
11:48
Company has joined #dri-devel
11:51
rsalvaterra_ has joined #dri-devel
11:51
<
mripard >
airlied: thanks :)
11:57
rsalvaterra has quit [Ping timeout: 480 seconds]
11:59
rsalvaterra_ has quit []
11:59
rsalvaterra has joined #dri-devel
11:59
camus has joined #dri-devel
12:03
camus1 has quit [Ping timeout: 480 seconds]
12:24
vivek has joined #dri-devel
13:12
Ristovski has quit [Quit: 0]
13:14
Ristovski has joined #dri-devel
13:22
xexaxo_ has quit [Read error: Connection reset by peer]
13:22
xexaxo_ has joined #dri-devel
13:45
Peste_Bubonica has joined #dri-devel
14:06
camus1 has joined #dri-devel
14:07
cphealy has joined #dri-devel
14:11
camus has quit [Ping timeout: 480 seconds]
14:22
thellstrom has joined #dri-devel
14:32
sdutt has joined #dri-devel
14:34
mattrope has joined #dri-devel
14:34
alyssa has joined #dri-devel
14:34
<
alyssa >
Does anyone have experience with mesa + crappy android vendor kernels without root?
14:34
<
alyssa >
anholt_: maybe?
14:41
uis_ has joined #dri-devel
14:43
thellstrom has quit [Remote host closed the connection]
14:43
thellstrom has joined #dri-devel
14:43
sneil has quit [Quit: Leaving]
14:49
<
bl4ckb0ne >
is there something special to link when using GL_GLEXT_PROTOTYPES?
14:49
<
bl4ckb0ne >
ive been hitting undefined references during link, arent the symbols in the libGL im using?
14:51
<
emersion >
GL_GLEXT_PROTOTYPES just exposes the functions as symbols in the header
14:52
<
emersion >
but the GL library also needs to expose these, or you'll hit link errors
14:52
<
bl4ckb0ne >
so I still have to load them by hand
14:52
<
emersion >
tl;dr don't use GL_GLEXT_PROTOTYPES :P
14:52
<
bl4ckb0ne >
makes sense
14:57
thellstrom has quit [Read error: Connection reset by peer]
15:01
<
daniels >
or use epoxy if you want to avoid the GetProcAddress dance
15:07
flto has quit [Read error: No route to host]
15:07
Peste_Bubonica has quit [Quit: Leaving]
15:07
flto has joined #dri-devel
15:07
sneil has joined #dri-devel
15:12
Duke`` has joined #dri-devel
15:13
<
karolherbst >
why is gdbserver so slow? :(
15:17
nielsdg has joined #dri-devel
15:24
<
danvet >
siqueira, you're also going to push the doc fix you just replied to?
15:24
* danvet
was about to get busy, but I prefer not to
15:25
Bennett has joined #dri-devel
15:25
<
siqueira >
danvet I can, but can I drop that Fixes tag?
15:25
<
danvet >
siqueira, imo it's ok
15:25
<
danvet >
it does fix an issue in my patch
15:25
<
danvet >
if it motivates stable team to pick that up for backporting ... not my problem :-)
15:27
<
emersion >
seems like
*something* is disabling CRTCs behind my back
15:27
<
emersion >
90% sure it's not me, checked with strace
15:28
<
emersion >
could that happen if i don't correctly do VT switching things?
15:31
<
siqueira >
danvet ok, I'm going to keep that.
15:36
<
danvet >
emersion, as long as your drm master, you own the display, and vt switching won't nuke crtc behind your back
15:36
<
danvet >
ofc your code might fail to restore for some reason
15:36
<
emersion >
hrm. really wondering what's causing this CRTC disable right after i modeset then
15:36
<
danvet >
emersion, the more likely is the RMFB trap
15:37
<
emersion >
yeah, entirely possible
15:37
<
danvet >
I think all the other traps we've nuked where drivers where killing your CRTC on hotunplug, dp link training issues or anything like that
15:37
<
emersion >
i see the CRTC disable as an atomic commit in the kernel logs
15:38
<
danvet >
emersion, sadly no drm_dbg for that case, I think we should fix that
15:38
<
emersion >
yeah doing that right now :)
15:38
soreau has quit [Read error: Connection reset by peer]
15:39
<
danvet >
drm_mode_rmfb_work_fn probably the case to augment with logging
15:39
soreau has joined #dri-devel
15:39
<
danvet >
drm_framebuffer_remove is also used for other cases (fbdev cleanup for drivers not yet using the generic ones, sadly)
16:05
sdutt has joined #dri-devel
16:17
nchery has joined #dri-devel
16:21
cedric has joined #dri-devel
16:22
cedric is now known as bluebugs
16:28
ngcortes has joined #dri-devel
17:04
gouchi has joined #dri-devel
17:16
thellstrom has joined #dri-devel
17:31
K`den is now known as Kayden
17:32
jkrzyszt has quit [Remote host closed the connection]
17:55
tzimmermann has joined #dri-devel
17:56
Kayden has quit [Quit: Leaving]
17:56
Kayden has joined #dri-devel
18:10
mbrost has joined #dri-devel
18:27
thellstrom has quit [Ping timeout: 480 seconds]
18:27
rasterman has quit [Quit: Gettin' stinky!]
18:31
nchery has quit [Ping timeout: 480 seconds]
18:31
danvet has quit [Read error: Connection reset by peer]
18:38
ngcortes has quit [Remote host closed the connection]
19:01
lynxeye has quit [Quit: Leaving.]
19:13
<
bl4ckb0ne >
is there a draft somewhere to port glBlitFramebuffer to gles2? i see there's an ANGLE and NV ext already
19:13
tzimmermann has quit [Quit: Leaving]
19:15
<
bl4ckb0ne >
there is! EXT_framebuffer_blit
19:15
<
HdkR >
That's the desktop GL one
19:15
<
bl4ckb0ne >
han yes
19:16
<
HdkR >
You'll need to do your own blit shader for ES 2.0 land
19:16
<
bl4ckb0ne >
so it could be ported to gles2
19:17
<
bl4ckb0ne >
I have a glCopyTexImage2D but it's very slow
19:17
<
HdkR >
Does anyone even care about ES 2.0 these days? Sort of doubt anyone would do the effort
19:18
<
bl4ckb0ne >
i could move to ES 3, i guess
19:19
<
HdkR >
Depends on if you want to care about Mali Utgard, Adreno 2xx, and Tegra 2/3/4 :P
19:28
mlankhorst has quit [Ping timeout: 480 seconds]
19:30
danvet has joined #dri-devel
19:35
<
bl4ckb0ne >
gles3 has no pkg config?
19:54
<
imirkin_ >
seems straightforward enough to implement GL_NV_framebuffer_blit
19:55
<
imirkin_ >
if someone cared.
19:55
<
bl4ckb0ne >
would it be much bikeshed to present the ext for gles
19:55
<
imirkin_ >
it's a registered gles ext already
19:57
<
bl4ckb0ne >
i think migrate my stuff to es3 would be easier
19:59
<
HdkR >
NV_framebuffer_blit was explicitly made for Tegra 2/3/4 stuff, since that didn't have ES 3.0 :D
20:20
nchery has joined #dri-devel
20:22
<
ajax >
i mostly have it written
20:22
<
ajax >
^ re ANGLE_framebuffer_blit
20:23
<
ajax >
if anyone wanted to push that across the finish line go for it, i'm not spending any real time on it atm
20:24
* bl4ckb0ne
has his shits lagging as well with glBlitFramebuffer
20:24
<
bl4ckb0ne >
can find some time eventually i guess
20:24
<
ajax >
it should essentially work. i think all that was left was making it reject more cases that big-gl allows
20:30
<
HdkR >
bl4ckb0ne: Could be that glblit isn't the source of your problem, just where the symptom shows up
20:31
<
bl4ckb0ne >
problem was blitting too much into too little
20:31
<
bl4ckb0ne >
instant gain of perf
20:31
<
bl4ckb0ne >
src size was way bigger than dst
20:32
<
HdkR >
Sounds reasonable if you're using a blit to down sample to me :)
20:33
<
Lightkey >
HdkR: ..and PowerVR SGX, I've just heard the DragonBox Pyra pre-pre[sic!]-orders have been completed, 200 of them.
20:33
<
HdkR >
Oh right, that has the blit extension as well
20:33
<
Lightkey >
From six years ago.
20:34
<
ajax >
HdkR: i'm kind of keeping gles2 as a target since there's a
_lot_ of dx9 hardware out there that still works
20:35
<
ajax >
slowly, much of it, but a fast r500 is still a decently fast card
20:35
<
ajax >
and the i915 sold for ages
20:36
<
airlied >
I wonder does my x1950 still work
20:37
<
ajax >
hm. fastest r500's memory bandwidth is somewhere between that of skylake and icelake.
20:38
<
HdkR >
Hm. I guess something like a DE supporting ES2?
20:38
<
ajax >
both glamor and mutter require only gles2
20:39
<
HdkR >
Guess that makes sense for that software
20:39
<
ajax >
"don't require much more than" more than "require only". i think there's a few extensions they demand too but they're basically software features
20:40
<
bl4ckb0ne >
wlroots only requires gles2 too
20:42
nchery has quit [Ping timeout: 480 seconds]
20:46
nchery has joined #dri-devel
20:58
nchery has quit [Ping timeout: 480 seconds]
21:09
Viciouss has quit [Ping timeout: 480 seconds]
21:09
sdutt has quit [Ping timeout: 480 seconds]
21:09
<
ajax >
bl4ckb0ne: gles3 has no pkgconfig because it's defined to use the same soname as gles2
21:17
gouchi has quit [Remote host closed the connection]
21:19
nchery has joined #dri-devel
21:25
Viciouss has joined #dri-devel
21:28
<
daniels >
you forgot Weston!
21:43
<
ajax >
weston doesn't even require gles2 tbf. but point taken about its gl renderer
21:47
sdutt has joined #dri-devel
21:47
sdutt has quit [Remote host closed the connection]
22:17
rasterman has joined #dri-devel
22:21
Duke`` has quit [Ping timeout: 480 seconds]
22:42
cbaylis has quit [Ping timeout: 480 seconds]
22:44
vivijim has quit [Ping timeout: 480 seconds]
22:44
danvet has quit [Ping timeout: 480 seconds]
22:50
idr has quit [Remote host closed the connection]
22:52
emersion has quit [Ping timeout: 480 seconds]
22:52
emersion has joined #dri-devel
22:52
idr has joined #dri-devel
23:28
pcercuei has quit [Quit: dodo]
23:53
rasterman has quit [Quit: Gettin' stinky!]