2022-03-22 11:57
ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar
00:04
kts has joined #dri-devel
00:05
jewins has quit [Ping timeout: 480 seconds]
00:14
jewins has joined #dri-devel
00:22
kts has quit [Ping timeout: 480 seconds]
00:42
codingkoopa has joined #dri-devel
00:46
pcercuei has quit [Quit: dodo]
00:57
co1umbarius has joined #dri-devel
00:59
columbarius has quit [Ping timeout: 480 seconds]
01:11
kts has joined #dri-devel
01:32
kts has quit [Ping timeout: 480 seconds]
01:38
camus has quit [Ping timeout: 480 seconds]
01:39
jewins has quit [Ping timeout: 480 seconds]
01:46
gnuiyl has joined #dri-devel
01:51
camus has joined #dri-devel
01:52
camus has joined #dri-devel
01:57
mclasen has joined #dri-devel
02:06
Danct12 has joined #dri-devel
02:19
bmodem has joined #dri-devel
02:29
mclasen has quit [Ping timeout: 480 seconds]
02:41
bmodem has quit [Ping timeout: 480 seconds]
02:55
bmodem has joined #dri-devel
02:58
bmodem1 has joined #dri-devel
02:59
heat has quit [Ping timeout: 480 seconds]
03:00
kts has joined #dri-devel
03:03
bmodem has quit [Ping timeout: 480 seconds]
03:06
bmodem1 has quit [Ping timeout: 480 seconds]
03:11
ppascher has joined #dri-devel
03:25
camus has quit [Remote host closed the connection]
03:26
camus has joined #dri-devel
03:31
YuGiOhJCJ has joined #dri-devel
03:39
i-garrison has joined #dri-devel
03:39
garrison has quit [Read error: Connection reset by peer]
03:53
JohnnyonFlame has joined #dri-devel
03:54
rsalvaterra_ has joined #dri-devel
03:54
rsalvaterra is now known as Guest3367
03:54
rsalvaterra_ is now known as rsalvaterra
03:58
Guest3368 has quit [Ping timeout: 480 seconds]
04:03
YuGiOhJCJ has quit [Remote host closed the connection]
04:04
YuGiOhJCJ has joined #dri-devel
04:06
mbrost has joined #dri-devel
04:13
Daaanct12 has joined #dri-devel
04:13
Danct12 has quit [Read error: Connection reset by peer]
04:28
YuGiOhJCJ has quit [Remote host closed the connection]
04:28
YuGiOhJCJ has joined #dri-devel
04:35
shankaru has joined #dri-devel
04:37
bmodem has joined #dri-devel
04:39
toolchains has joined #dri-devel
04:49
YuGiOhJCJ has quit [Remote host closed the connection]
04:49
YuGiOhJCJ has joined #dri-devel
04:58
Duke`` has joined #dri-devel
05:08
bmodem has joined #dri-devel
05:22
saurabhg has joined #dri-devel
05:36
itoral has joined #dri-devel
05:38
Duke`` has quit [Ping timeout: 480 seconds]
05:59
saurabhg has quit [Ping timeout: 480 seconds]
06:00
saurabhg has joined #dri-devel
06:08
TMM has joined #dri-devel
06:10
shankaru has quit [Quit: Leaving.]
06:17
Daaanct12 has quit [Read error: Connection reset by peer]
06:18
Danct12 has joined #dri-devel
06:18
kts has quit [Ping timeout: 480 seconds]
06:22
WinLinux1028 has joined #dri-devel
06:25
kts has joined #dri-devel
06:28
alanc has quit [Remote host closed the connection]
06:29
alanc has joined #dri-devel
06:30
toolchains has quit [Remote host closed the connection]
06:31
toolchains has joined #dri-devel
06:39
toolchains has quit [Ping timeout: 480 seconds]
06:42
jkrzyszt has joined #dri-devel
06:50
frieder has joined #dri-devel
06:59
tzimmermann has joined #dri-devel
06:59
jfalempe has joined #dri-devel
07:00
itoral has quit [Remote host closed the connection]
07:00
itoral has joined #dri-devel
07:01
lygstate has joined #dri-devel
07:13
tango_ has joined #dri-devel
07:20
kts has quit [Ping timeout: 480 seconds]
07:23
wvanhauwaert has joined #dri-devel
07:24
lplc has quit [Remote host closed the connection]
07:24
itoral has quit [Remote host closed the connection]
07:25
itoral has joined #dri-devel
07:29
itoral has quit [Remote host closed the connection]
07:30
itoral has joined #dri-devel
07:32
danvet has joined #dri-devel
07:36
lygstate has quit [Remote host closed the connection]
07:37
JohnnyonFlame has quit [Ping timeout: 480 seconds]
07:42
lplc has joined #dri-devel
07:42
bmodem has quit [Ping timeout: 480 seconds]
07:42
kts has joined #dri-devel
07:47
MajorBiscuit has joined #dri-devel
07:58
Company has quit [Quit: Leaving]
08:00
lynxeye has joined #dri-devel
08:00
bmodem has joined #dri-devel
08:08
bmodem has quit [Remote host closed the connection]
08:17
tursulin has joined #dri-devel
08:19
bmodem has joined #dri-devel
08:23
Danct12 has quit [Ping timeout: 480 seconds]
08:26
bmodem has quit [Remote host closed the connection]
08:27
lplc has quit [Remote host closed the connection]
08:30
lplc has joined #dri-devel
08:30
Danct12 has joined #dri-devel
08:36
lplc has quit [Remote host closed the connection]
08:38
itoral has quit [Remote host closed the connection]
08:39
itoral has joined #dri-devel
08:39
ahajda has joined #dri-devel
08:39
lplc has joined #dri-devel
08:43
aravind has joined #dri-devel
08:49
itoral has quit [Remote host closed the connection]
08:50
itoral has joined #dri-devel
08:51
mvlad has joined #dri-devel
08:52
pcercuei has joined #dri-devel
08:54
itoral has quit [Remote host closed the connection]
08:55
itoral has joined #dri-devel
08:58
itoral has quit [Remote host closed the connection]
08:59
itoral has joined #dri-devel
08:59
frankbinns has joined #dri-devel
09:28
saurabhg has quit [Ping timeout: 480 seconds]
09:28
rkanwal has joined #dri-devel
09:30
pixelcluster has joined #dri-devel
09:30
Danct12 has quit [Remote host closed the connection]
09:39
tursulin has quit [Read error: Connection reset by peer]
09:42
tursulin has joined #dri-devel
09:44
srslypascal has quit [Ping timeout: 480 seconds]
09:46
srslypascal has joined #dri-devel
09:50
<
dj-death >
../src/intel/vulkan/grl/gpu/intrinsics.h:39:34: error: implicit declaration of function 'get_sub_group_local_id' is invalid in OpenCL
09:55
saurabhg has joined #dri-devel
10:08
srslypascal has quit [Ping timeout: 480 seconds]
10:09
srslypascal has joined #dri-devel
10:11
<
dj-death >
ah llvm14 special handling in mesa/clc
10:14
gawin has joined #dri-devel
10:15
JohnnyonFlame has joined #dri-devel
10:16
devilhorns has joined #dri-devel
10:16
mclasen has joined #dri-devel
10:17
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
10:35
pixelcluster has quit [Remote host closed the connection]
10:38
sul has quit [Ping timeout: 480 seconds]
10:38
sul has joined #dri-devel
10:38
icecream95 has quit [Ping timeout: 480 seconds]
11:09
JohnnyonFlame has quit [Ping timeout: 480 seconds]
11:10
aravind has quit [Remote host closed the connection]
11:10
aravind has joined #dri-devel
11:12
wvanhauwaert has quit [Ping timeout: 480 seconds]
11:13
gawin has quit [Ping timeout: 480 seconds]
11:15
<
mripard >
José Expósito: It is yeah
11:15
<
mripard >
it's a new thing, so there's nothing to fix in the first place :)
11:23
saurabh_1 has joined #dri-devel
11:23
saurabhg has quit [Read error: Connection reset by peer]
11:37
rsalvaterra has quit []
11:38
rsalvaterra has joined #dri-devel
11:42
wvanhauwaert has joined #dri-devel
12:01
saurabh_1 has quit [Ping timeout: 480 seconds]
12:06
aravind has quit []
12:18
mbrost has joined #dri-devel
12:20
saurabh_1 has joined #dri-devel
12:20
RSpliet has quit [Read error: Connection reset by peer]
12:20
RSpliet has joined #dri-devel
12:28
RSpliet has quit [Quit: Bye bye man, bye bye]
12:29
RSpliet has joined #dri-devel
12:31
itoral has quit [Remote host closed the connection]
12:32
maxzor_ has joined #dri-devel
12:34
pixelcluster has joined #dri-devel
12:49
lygstate has joined #dri-devel
13:00
fxkamd has joined #dri-devel
13:02
jewins has joined #dri-devel
13:03
mbrost has quit [Ping timeout: 480 seconds]
13:20
ella-0 has joined #dri-devel
13:24
ella-0_ has quit [Read error: Connection reset by peer]
13:30
mbrost has joined #dri-devel
13:35
nchery has joined #dri-devel
13:36
Daanct12 is now known as Danct12
13:53
yogesh_m1 has joined #dri-devel
13:55
yogesh_mohan has quit [Ping timeout: 480 seconds]
14:06
kts has quit [Ping timeout: 480 seconds]
14:13
kts has joined #dri-devel
14:35
JohnnyonFlame has joined #dri-devel
14:40
mbrost has quit [Ping timeout: 480 seconds]
14:46
mbrost has joined #dri-devel
14:49
Company has joined #dri-devel
14:50
wvanhauwaert has quit [Quit: Leaving]
14:50
wvanhauwaert has joined #dri-devel
14:57
mbrost has quit [Ping timeout: 480 seconds]
15:05
nchery has quit [Ping timeout: 480 seconds]
15:06
frieder has quit [Remote host closed the connection]
15:15
nchery has joined #dri-devel
15:15
saurabh_1 has quit [Ping timeout: 480 seconds]
15:23
frankbinns1 has joined #dri-devel
15:24
frankbinns has quit [Remote host closed the connection]
15:25
tursulin has quit [Ping timeout: 480 seconds]
15:26
saurabhg has joined #dri-devel
15:26
frankbinns2 has joined #dri-devel
15:32
frankbinns1 has quit [Ping timeout: 480 seconds]
15:38
wvanhauwaert has quit [Ping timeout: 480 seconds]
15:42
nchery is now known as Guest3426
15:42
nchery has joined #dri-devel
15:46
Guest3426 has quit [Ping timeout: 480 seconds]
15:48
agd5f has joined #dri-devel
15:51
<
agd5f >
danvet, seems that way unfortunately. Not sure why the old amdgpu fbdev code didn't require the vt switch
15:51
<
agd5f >
setting mode_config.prefer_shadow = 0 or doing a VT switch "fixes" it
15:59
<
danvet >
agd5f, hm what's the issue?
15:59
<
danvet >
like frozen fbcon or not even showing?
15:59
<
danvet >
the shadow thing is kinda funky
16:02
<
agd5f >
danvet, hibernation fails with a warning in blkdev_flush_mapping()
16:07
<
danvet >
agd5f, there's a bugfix for the shadow stuff in flight to handle some overflow due to busted math
16:07
<
danvet >
javierm, ^^
16:08
<
danvet >
agd5f, also that blkdev thing is most likely not the cause, since just a warning
16:08
<
danvet >
we shut down consoles plus much is done under console_lock held anyway, so without hacks you wont see much
16:08
<
danvet >
no_console_suspend might help
16:08
<
danvet >
otherwise maybe a general console_lock busting, not sure we have that
16:08
<
danvet >
(why is fbcon so terrible)
16:11
ybogdano has joined #dri-devel
16:12
mbrost has joined #dri-devel
16:14
jkrzyszt has quit [Ping timeout: 480 seconds]
16:17
<
danvet >
hm I cant find the patch, was I dreaming ...
16:18
dliviu has joined #dri-devel
16:22
<
danvet >
agd5f, might also be worth to retest with kasan or so enabled
16:22
<
danvet >
in case we scribble over stuff with the shadow fb implementation
16:23
<
danvet >
what I don't get is how both disabling shadow fb or forced vt_switch switches anything here
16:24
tobiasjakobi has joined #dri-devel
16:29
tobiasjakobi has quit []
16:30
devilhorns has quit [Remote host closed the connection]
16:34
ybogdano has quit [Ping timeout: 480 seconds]
16:37
<
agd5f >
danvet, yeah, agreed
16:39
Danct12 has quit [Remote host closed the connection]
16:39
<
agd5f >
also weird that only hibernation is affected
16:43
Danct12 has joined #dri-devel
16:47
ybogdano has joined #dri-devel
16:53
toolchains has joined #dri-devel
16:58
frankbinns1 has joined #dri-devel
16:59
technopoirot has quit [Ping timeout: 480 seconds]
17:01
frankbinns2 has quit [Remote host closed the connection]
17:03
Duke`` has joined #dri-devel
17:11
toolchains has quit [Ping timeout: 480 seconds]
17:12
lynxeye has quit [Quit: Leaving.]
17:15
tzimmermann has quit [Quit: Leaving]
17:20
lygstate has quit [Remote host closed the connection]
17:22
nchery has quit [Ping timeout: 480 seconds]
17:23
heat has joined #dri-devel
17:26
Peste_Bubonica has joined #dri-devel
17:33
nchery has joined #dri-devel
17:42
JohnnyonFlame has quit [Ping timeout: 480 seconds]
17:55
flacks has quit [Ping timeout: 480 seconds]
17:58
Rayyan_ has joined #dri-devel
17:59
flacks has joined #dri-devel
18:03
nchery is now known as Guest3438
18:03
nchery has joined #dri-devel
18:08
Guest3438 has quit [Ping timeout: 480 seconds]
18:23
<
javierm >
danvet: I'm not aware of such a patch but IIRC Thomas shared some patches ?
18:24
<
javierm >
danvet, agd5f: while I have your attention, did you see the warning due BOOTFB mem resource not released when amdgpu takes over ?
18:24
<
javierm >
danvet, agd5f: what Linus reported
18:25
* airlied
saw it, is there a plan? :-)
18:30
<
danvet >
javierm, thx, found the right patch
18:31
<
danvet >
agd5f, for lack of better ideas
18:31
<
javierm >
airlied: I don't have an amdgpu machine but jfalempe tried on his amd machine and wasn't able to reproduce it
18:32
<
javierm >
airlied: I'm trying to understand why the simple-framebuffer isn't unregistered and the resource released on Linus' setup
18:36
<
airlied >
is there any possibility of a race?
18:36
<
airlied >
he is on a threadripper
18:39
JohnnyonFlame has joined #dri-devel
18:40
<
javierm >
airlied: I'm still trying to figure that out
18:40
<
airlied >
has userspace already opened it?
18:41
heat has quit [Ping timeout: 480 seconds]
18:45
heat has joined #dri-devel
18:49
technopoirot has joined #dri-devel
18:51
rkanwal has quit [Ping timeout: 480 seconds]
18:54
MajorBiscuit has quit [Ping timeout: 480 seconds]
18:58
iive has joined #dri-devel
19:06
maxzor_ has quit [Ping timeout: 480 seconds]
19:12
mbrost has quit [Read error: Connection reset by peer]
19:14
mclasen has quit []
19:14
mclasen has joined #dri-devel
19:17
technopo1rot has joined #dri-devel
19:18
technopoirot has quit [Ping timeout: 480 seconds]
19:21
Haaninjo has joined #dri-devel
19:21
alyssa has joined #dri-devel
19:21
<
alyssa >
Are subgroup barriers in non-uniform convergent control flow legal?
19:22
<
alyssa >
i.e. `if (subgroupAny(not_uniform)) { subgroupBarrier(); }`
19:23
<
javierm >
airlied: that's indeed a possibility since the release_mem_region() happens in the .fb_destroy callback and not in .remove
19:23
<
javierm >
for {efi,simple}fb
19:24
<
pendingchaos >
"non-uniform convergent control flow"? wasn't "convergent" synonymous with "uniform"
19:24
<
pendingchaos >
that example looks legal
19:26
<
jenatali >
pendingchaos: I'd interpret "uniform" as never diverged
19:26
<
alyssa >
pendingchaos: convergent within the subgroup but not the workgroup
19:26
<
alyssa >
what I'm getting at is "is it legal to replace subgroup barriers with workgroup barriers?"
19:26
<
jenatali >
Oh, or that I guess
19:27
<
alyssa >
if the snippet is legal, then "no, it's not"
19:36
<
javierm >
danvet: IIUC from your explanation last time release_mem_region() should be done in .remove rather than .fb_destroy right ?
19:37
<
javierm >
danvet: that is, you mentioned that all the HW cleanup (i.e: iounmap, release_mem_region, etc) should be done in .remove and only the SW cleanup (i.e: fb_dealloc_cmap, framebuffer_release, etc) in .fb_destroy
19:40
mbrost has joined #dri-devel
19:40
kts has quit [Quit: Konversation terminated!]
19:41
<
alyssa >
Do we have a NIR pass that deletes barriers at the end of a shader?
19:41
<
alyssa >
If not, would any other backends benefit from one?
19:41
<
alyssa >
(I just wrote a backend IR pass for it but maybe it'd be useful in common)
19:42
toolchains has joined #dri-devel
19:56
rkanwal has joined #dri-devel
19:58
Haaninjo has quit [Quit: Ex-Chat]
20:02
maxzor_ has joined #dri-devel
20:19
nchery is now known as Guest3445
20:19
nchery has joined #dri-devel
20:20
Guest3445 has quit [Ping timeout: 480 seconds]
20:21
toolchains has quit [Remote host closed the connection]
20:27
toolchains has joined #dri-devel
20:28
neonking_ has joined #dri-devel
20:30
maxzor_ has quit [Ping timeout: 480 seconds]
20:34
neonking has quit [Ping timeout: 480 seconds]
20:40
toolchains has quit [Ping timeout: 480 seconds]
20:43
<
agd5f >
javierm, I saw the thread. I haven't seen anything similar internally
20:45
neonking__ has joined #dri-devel
20:49
mvlad has quit [Remote host closed the connection]
20:51
neonking_ has quit [Ping timeout: 480 seconds]
20:54
neonking_ has joined #dri-devel
20:56
alyssa has quit [Quit: leaving]
21:01
neonking__ has quit [Ping timeout: 480 seconds]
21:05
technopo1rot has quit [Ping timeout: 480 seconds]
21:06
ngcortes has joined #dri-devel
21:08
Duke`` has quit [Ping timeout: 480 seconds]
21:11
<
danvet >
javierm, yeah, but also simpledrm.c is already using devm, which should dtrt
21:11
<
danvet >
at least if we nuke the platform_dev and correctly unbind the driver
21:11
<
danvet >
if we don't unbind the driver then devm might not kick in and things wont work correctly
21:12
<
danvet >
javierm, maybe linus needs one of the sysfb shufflings we've been doing?
21:12
<
danvet >
javierm, for debugging I'd install a printk devm action and see when that fires :-)
21:14
heat has quit [Quit: Leaving]
21:15
heat has joined #dri-devel
21:22
technopoirot has joined #dri-devel
21:37
saurabhg has quit [Remote host closed the connection]
21:37
srslypascal has quit [Quit: Leaving]
21:39
srslypascal has joined #dri-devel
21:43
danvet has quit [Ping timeout: 480 seconds]
21:49
technopoirot has quit [Ping timeout: 480 seconds]
22:05
rasterman has joined #dri-devel
22:06
sul has quit [Ping timeout: 480 seconds]
22:06
sul has joined #dri-devel
22:33
rkanwal has quit [Ping timeout: 480 seconds]
22:37
icecream95 has joined #dri-devel
22:40
mbrost has quit [Remote host closed the connection]
22:41
mbrost has joined #dri-devel
22:43
pixelcluster has quit [Quit: Bye-bye!]
22:48
* jekstrand
hates graphics pipeline libraries. There. I said it.
22:49
mbrost has quit [Ping timeout: 480 seconds]
22:49
<
zmike >
how dare you.
22:52
<
jekstrand >
Yeah...
22:52
<
jekstrand >
I'm gonna copy+paste all the radv_blah_info stuff into common.
22:53
<
jekstrand >
Because while you can, in theory, actually create your pipeline piecemeal (might actually work for AMD/NV), it's just not tractable most of the time. Not really.
22:53
<
zmike >
just means you haven't tried hard enough
22:54
<
jekstrand >
You have to just stash absolutely everything from the pCreateInfo and combine it all together later.
22:54
<
jekstrand >
Which is fine, really. As long as you don't actually compile shaders at the final link step.
22:55
<
zmike >
take this as a preemptive a-b for when you copy/paste the lavapipe implementation around
22:57
<
airlied >
jekstrand: 16919 is still out there :-)
22:58
<
airlied >
whether you come up with a better plan for anv, I think the common code should be refactored out of the 3 other drivers
23:03
nchery has quit [Ping timeout: 480 seconds]
23:03
<
jekstrand >
airlied: Yeah...
23:05
technopoirot has joined #dri-devel
23:13
rasterman has quit [Quit: Gettin' stinky!]
23:23
agx has quit [Ping timeout: 480 seconds]
23:35
pcercuei has quit [Quit: dodo]
23:41
agx has joined #dri-devel
23:57
ybogdano has quit [Ping timeout: 480 seconds]
23:57
mclasen has quit []