#dri-devel on 2022-12-19 — irc logs at oftc.irclog.whitequark.org

2022-12-16 22:42 ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar

00:25 Haaninjo has quit [Quit: Ex-Chat]

00:26 <Lynne> airlied: you can skip support for VkVideoEncodeH264EmitPictureParametersInfoEXT for now

00:27 <Lynne> as far as I can tell, it's just a convenience thing to let the driver write SPS/PPS/VPS for you

00:27 <Lynne> I can write those myself, not too keen on letting the driver do much

00:28 <Lynne> VkVideoEncodeH264VclFrameInfoEXT is what's needed to write frames

00:30 <Lynne> I saw d3d12 has a nice writer you could reuse for those unfortunate enough to not have CBS

00:36 everfree has quit [Remote host closed the connection]

00:37 tarceri has quit [Remote host closed the connection]

00:38 tarceri has joined #dri-devel

01:00 heat has joined #dri-devel

01:00 heat_ has quit [Read error: Connection reset by peer]

01:30 co1umbarius has joined #dri-devel

01:31 Surkow has quit [Quit: 418 I'm a teapot - NOP NOP NOP]

01:32 columbarius has quit [Ping timeout: 480 seconds]

01:39 OftenTimeConsuming is now known as Guest143

01:39 OftenTimeConsuming has joined #dri-devel

01:41 Guest143 has quit [Ping timeout: 480 seconds]

01:44 oneforall2 has quit [Remote host closed the connection]

01:45 Surkow|laptop has joined #dri-devel

01:45 oneforall2 has joined #dri-devel

02:20 yuq825 has joined #dri-devel

02:28 ArcticAtelier has quit [Read error: Connection reset by peer]

02:35 camus1 has joined #dri-devel

02:36 camus has quit [Remote host closed the connection]

02:50 OftenTimeConsuming has quit [Remote host closed the connection]

02:51 OftenTimeConsuming has joined #dri-devel

03:01 JohnnyonFlame has joined #dri-devel

03:12 Leopold_ has quit [Remote host closed the connection]

03:13 Leopold has joined #dri-devel

03:18 warpme_____ has quit []

03:28 The_Company has quit [Read error: Connection reset by peer]

03:42 JohnnyonFlame has quit [Read error: Connection reset by peer]

03:54 maxzor has quit [Ping timeout: 480 seconds]

03:55 <marcan> kisak: the only requirements to join our channels are that you use SSL/TLS and that you don't use the Matrix gateway (because the trolls and bots all seem to be too dumb to use SSL, and because Matrix is a moderation nightmare)

03:55 <marcan> it's 2022, you should probably be using encryption :p

03:57 <marcan> I might remove the SSL requirement anyway though, that particular mitigation seemed to be effective against the freenode meltdown adjacent spambots, probably not as useful now

03:59 <psykose> requiring encryption is an unthinkable offense sometimes

04:14 kts has joined #dri-devel

04:35 kts has quit [Quit: Leaving]

05:10 bmodem has joined #dri-devel

05:17 YuGiOhJCJ has joined #dri-devel

05:26 heat has quit [Ping timeout: 480 seconds]

05:28 Leopold has quit [Remote host closed the connection]

05:29 Leopold_ has joined #dri-devel

05:42 lemonzest has joined #dri-devel

05:47 Daanct12 has joined #dri-devel

06:14 srslypascal has quit [Remote host closed the connection]

06:15 srslypascal has joined #dri-devel

06:29 bgs has joined #dri-devel

06:42 aravind has joined #dri-devel

06:49 fab has joined #dri-devel

07:02 OftenTimeConsuming has quit [Remote host closed the connection]

07:04 OftenTimeConsuming has joined #dri-devel

07:10 bmodem1 has joined #dri-devel

07:15 kts has joined #dri-devel

07:15 bmodem has quit [Ping timeout: 480 seconds]

07:28 bgs has quit [Remote host closed the connection]

07:32 alanc has quit [Remote host closed the connection]

07:33 alanc has joined #dri-devel

07:36 kts has quit [Quit: Leaving]

07:45 <Lynne> airlied: updated my repo and blog, linked to yours for installation instructions

07:48 dcz_ has joined #dri-devel

07:50 rasterman has joined #dri-devel

07:55 <airlied> Lynne: the slot index for 264 and 265 we made be the same dpb idx

07:55 <airlied> your blog has the old idea

07:56 danvet has joined #dri-devel

08:00 <airlied> Lynne: also didnt intel vaapi require multiplanar to work properly?

08:05 mvlad has joined #dri-devel

08:14 tursulin has joined #dri-devel

08:14 fab has quit [Quit: fab]

08:14 fab has joined #dri-devel

08:14 fab has quit []

08:14 fab has joined #dri-devel

08:15 maxzor has joined #dri-devel

08:17 jfalempe has joined #dri-devel

08:20 tzimmermann has joined #dri-devel

08:22 maxzor has quit [Remote host closed the connection]

08:23 maxzor has joined #dri-devel

08:30 frieder has joined #dri-devel

08:31 kts has joined #dri-devel

08:33 oneforall2 has quit [Remote host closed the connection]

08:35 oneforall2 has joined #dri-devel

08:38 ahajda_ has joined #dri-devel

08:39 mwalle has joined #dri-devel

08:43 kts has quit [Ping timeout: 480 seconds]

08:54 warpme_____ has joined #dri-devel

08:55 rgallaispou has joined #dri-devel

08:56 frieder has quit [Remote host closed the connection]

08:59 lynxeye has joined #dri-devel

08:59 frieder has joined #dri-devel

09:00 apinheiro has joined #dri-devel

09:22 guru_ has joined #dri-devel

09:27 oneforall2 has quit [Ping timeout: 480 seconds]

09:27 frieder has quit [Ping timeout: 480 seconds]

09:29 kts has joined #dri-devel

09:36 frieder has joined #dri-devel

09:44 Mangix has quit [Read error: Connection reset by peer]

09:44 Mangix has joined #dri-devel

09:45 tzimmermann has quit [Quit: Leaving]

09:45 tzimmermann has joined #dri-devel

09:51 <tzimmermann> airlied, i came across ae1287865f53 ("fbcon: don't lose the console font across generic->chip driver switch") from long ago. is this still necessary? because system appears to have fixed this in userspace: https://cgit.freedesktop.org/systemd/systemd/commit/?id=f6ba8671d83f9fce9a00045d8fa399a1c07ba7fc

09:51 <tzimmermann> i'd like to remove the rsp kernel code

09:54 <javierm> tzimmermann: have you seen https://bugzilla.kernel.org/show_bug.cgi?id=216475 ?

09:56 <tzimmermann> javierm, no i don't follow the kernel bugzilla

09:59 <javierm> tzimmermann: about your previous question, what's the uevent sent during firmware-provided framebuffer driver -> native DRM driver ?

10:00 <javierm> because IIUC the systemd patch, that's when console drivers are switched (i.e: vgacon -> fbcon) but not when fbdev devices bound to fbcon are changed (i.e: efifb -> DRM native emulated fbdev)

10:02 <tzimmermann> javierm, the rule has meanwhile been updated: https://cgit.freedesktop.org/systemd/systemd/commit/src/vconsole/90-vconsole.rules.in?id=a52750d1483ff139df33149afc0b675531e9cd79

10:02 <tzimmermann> how do i find out about the uevent?

10:03 <javierm> tzimmermann: with `udevadm monitor` maybe ?

10:03 <javierm> and load your DRM driver that will cause the simpledrm to be removed

10:03 <tzimmermann> i need to do that during the handover

10:04 <tzimmermann> ok, good idea

10:04 <javierm> tzimmermann: hmm, it seems you are correct that the new rule should handle that case

10:05 <tzimmermann> i never tried

10:05 <tzimmermann> i just noticed that the rule changed

10:05 <javierm> tzimmermann: but I'm thinking that just reading Documentation/driver-api/console.rst, didn't try it either

10:05 <tzimmermann> we currently don't keep fonts with generic drm drivers

10:06 <tzimmermann> only the old fbdev drivers support this

10:06 <javierm> tzimmermann: but fbcon=font:<name> works with simpledrm IIRC

10:07 <javierm> but also, systemd would change the font according to /etc/vconsole.conf anyways I think

10:08 <tzimmermann> i guess i'll configure a font and test is it's still there after handover

10:11 <javierm> tzimmermann: Ok

10:11 <javierm> tzimmermann: but yes, if user-space can handle that and we can drop ae1287865f53, that would be much better IMO

10:13 <tzimmermann> i agree

10:14 <tzimmermann> especially as the fix doesn't with with transitions from vgacon or textmode, AFAIU

10:15 <javierm> right

10:31 repetitivestrain has joined #dri-devel

10:32 srslypascal is now known as Guest171

10:32 srslypascal has joined #dri-devel

10:36 Guest171 has quit [Ping timeout: 480 seconds]

10:38 vv221 has left #dri-devel [WeeChat 3.7.1]

10:43 Daanct12 has quit [Remote host closed the connection]

10:44 junaid has joined #dri-devel

10:44 devilhorns has joined #dri-devel

10:45 rgallaispou has quit []

10:49 nuh^ has joined #dri-devel

10:49 junaid has quit [Remote host closed the connection]

10:53 <tzimmermann> javierm, just tested with simpledrm->radeon. the configured font got survived the handover

11:02 <javierm> tzimmermann: cool

11:02 aravind has quit [Ping timeout: 480 seconds]

11:16 pjakobsson_ has joined #dri-devel

11:17 ayaka_ has joined #dri-devel

11:18 pjakobsson has quit [Ping timeout: 480 seconds]

11:19 Haaninjo has joined #dri-devel

11:19 <ayaka_> are enum drm_color_encoding and enum drm_color_range not a part of userspace api?

11:21 <emersion> you're supposed to get the values dynamically from the KMS API, but i just hardcode them

11:22 <ayaka_> I know I should get the supported enum values from the plane properties

11:23 <ayaka_> but I think the value of bt709, bt601, linear, non linear would have a different value

11:24 <emersion> what do you mean?

11:26 <ayaka_> for example, I want to know what transfer functions are available to a plane

11:27 <ayaka_> then a get the enum counts and enum from COLOR_ENCODING

11:28 <ayaka_> but I don't know which is the value for bt709 in the enum

11:28 <emersion> in drmModePropertyRes, it's in the values array iirc

11:28 <emersion> ah, no, it's not

11:29 <emersion> it's prop->enums[i].value

11:29 <ayaka_> or you mean I should parse the enum name?

11:29 <emersion> https://gitlab.freedesktop.org/emersion/drm_info/-/blob/master/json.c#L466

11:29 <emersion> you're supposed to find the entry in the array with the name you want

11:30 <emersion> yeah, it's not great :/

11:30 apinheiro has quit [Ping timeout: 480 seconds]

11:30 <emersion> i'd recommend just giving up on the whole idea and hardcode the values

11:31 <ayaka_> what I was worrying is that parsing string could make unmatch with a single char missing

11:31 <ayaka_> if hardcode is acceptable, I would did that in gstreamer kms atomic

11:31 <ayaka_> s/did/do/

11:39 apinheiro has joined #dri-devel

11:43 <emersion> yeah, a typo would mess things up without giving any clear error

11:59 maxzor has quit [Ping timeout: 480 seconds]

12:03 maxzor has joined #dri-devel

12:29 devilhorns has quit []

12:46 tobiasjakobi has joined #dri-devel

12:46 tobiasjakobi has quit []

12:47 sinatosk has joined #dri-devel

12:53 maxzor has quit [Ping timeout: 480 seconds]

12:56 camus has joined #dri-devel

12:58 pcercuei has joined #dri-devel

13:00 Company has joined #dri-devel

13:01 camus1 has quit [Ping timeout: 480 seconds]

13:07 aravind has joined #dri-devel

13:08 JohnnyonFlame has joined #dri-devel

13:09 sinatosk has quit [Quit: Leaving]

13:24 srslypascal has quit [Remote host closed the connection]

13:24 heat has joined #dri-devel

13:25 maxzor has joined #dri-devel

13:27 srslypascal has joined #dri-devel

13:28 maxzor has quit [Remote host closed the connection]

13:28 maxzor has joined #dri-devel

13:30 frieder has quit [Ping timeout: 480 seconds]

13:31 kts has quit [Quit: Leaving]

13:40 frieder has joined #dri-devel

13:47 apinheiro has quit [Ping timeout: 480 seconds]

13:47 JohnnyonFlame has quit [Read error: Connection reset by peer]

13:47 JohnnyonFlame has joined #dri-devel

13:47 Akari has joined #dri-devel

14:01 frieder has quit [Ping timeout: 480 seconds]

14:04 yuq825 has left #dri-devel [#dri-devel]

14:08 camus has quit [Remote host closed the connection]

14:09 camus has joined #dri-devel

14:10 frieder has joined #dri-devel

14:13 Akari has quit [Quit: segmentation fault (core dumped)]

14:20 bgs has joined #dri-devel

14:35 maxzor has quit [Ping timeout: 480 seconds]

14:40 eletrotupi_ has left #dri-devel [#dri-devel]

14:41 maxzor has joined #dri-devel

14:43 kts has joined #dri-devel

14:54 fxkamd has joined #dri-devel

15:03 frieder has quit [Remote host closed the connection]

15:27 maxzor has quit [Ping timeout: 480 seconds]

15:40 fxkamd has quit []

15:59 aravind has quit [Ping timeout: 480 seconds]

16:09 Duke`` has joined #dri-devel

16:10 Leopold_ has quit [Remote host closed the connection]

16:11 Leopold_ has joined #dri-devel

16:33 FireBurn has quit [Quit: Konversation terminated!]

16:47 tzimmermann has quit [Quit: Leaving]

16:52 rsjw has joined #dri-devel

16:54 tursulin has quit [Ping timeout: 480 seconds]

16:56 <Lynne> airlied: apparently the dg2 preproduction cards have no built-in firmware

16:57 <Lynne> as far as I was told, you have to boot under windows, install intel's drivers, let windows boot up with the card, and let it upload the firmware for the first boot

16:57 <Lynne> so the card I've got access to is currently in brick form

16:58 <Lynne> (the drivers intel linked were apparently no good)

16:59 <Lynne> modern hardware sure is nice, implementing VGA FB output is too low-class for them

17:03 nuh^ has quit [Remote host closed the connection]

17:09 gouchi has joined #dri-devel

17:13 <Lynne> well, they're not wasting any time, the headers repo s/EXT/KHR'd

17:14 <Lynne> I'll fix my repo in 2 hours

17:14 <mlankhorst> Not sure you if you've read the VGA spec, but most vga modes are hardly sane. :)

17:15 maxzor has joined #dri-devel

17:17 <Lynne> no, but they could've implemented a 640x4eight0 fallback, it's been present on all systems since time immemorial

17:18 <Lynne> how else would you run DOS on your brand new system :P

17:28 robin_ has joined #dri-devel

17:44 <airlied> Lynne: I'll update the driver branches in a while with the latest headers

17:48 YuGiOhJCJ has quit [Quit: YuGiOhJCJ]

18:07 <gekret005> I have verified on the 7900xtx on windows 10 11, and gentoo ~amd64 with mesa-9999 that the card cannot run the valve index at 144hz

18:08 <gekret005> it ran run it at 89, 90, and 120

18:09 <gekret005> i see reports on reddit of windows users with the problem too, could this be a hardware issue and unfixable in updates?

18:09 <gekret005> *other windows users

18:11 bmodem1 has quit [Ping timeout: 480 seconds]

18:25 jbarnes_ has left #dri-devel [#dri-devel]

18:25 jbarnes has joined #dri-devel

18:30 epoll has quit [Ping timeout: 480 seconds]

18:38 alyssa has joined #dri-devel

18:38 <alyssa> Mmm, these tests are sometimes passing, sometimes failing (with no fault), and sometimes faulting

18:38 <alyssa> what delight

18:39 epoll has joined #dri-devel

18:41 guru_ has quit []

18:41 oneforall2 has joined #dri-devel

18:43 <alyssa> dEQP-GLES3.functional.texture.specification.teximage2d_pbo.*

18:46 ybogdano has joined #dri-devel

18:46 <alyssa> Fault from some random address..

18:47 <alyssa> not an actual address

18:57 maxzor has quit [Ping timeout: 480 seconds]

18:57 <alyssa> oh boy i see a bug

18:57 <alyssa> not my bug tho

18:58 <alyssa> :

19:05 lynxeye has quit [Quit: Leaving.]

19:05 <alyssa> yeah no idea what's causing this.

19:14 ngcortes has joined #dri-devel

19:26 <Lynne> airlied: repo updated

19:31 alyssa has quit [Quit: leaving]

19:37 <italove> my DRM_IOCTL_MODE_CREATE_DUMB drmIoctl is returning permission denied even though I have root privileges, anyone know what might be the issue?

19:44 * airlied is fighting the mesa vulkan registry integration

19:53 camus has quit [Remote host closed the connection]

19:53 camus has joined #dri-devel

19:59 rsjw has quit [Quit: rsjw]

20:03 <airlied> jekstrand: maybe take another quick look at 20384 for the updated python

20:03 kts has quit [Quit: Leaving]

20:15 <jekstrand> airlied: What things are now guarded and by what?

20:16 <airlied> jekstrand: some stuff was always guarded, we just didn't handle the combination

20:16 nuh^ has joined #dri-devel

20:17 <jekstrand> Yeah, but why would a vkCmd thing be guarded?

20:18 <jekstrand> Guards are typically for platform defines and vkCmd* should be platform-agnostic

20:20 <airlied> jekstrand: I've pushed an updated commit msg

20:20 <airlied> jekstrand: VkVideoCodingControlInfoKHR

20:20 <airlied> is extended by VkVideoEncodeRateControlInfoKHR

20:20 <airlied> the extended struct is protected

20:20 <airlied> but the cmd queue generator always generated unguarded pNext handlers for extended structs

20:22 <jekstrand> But what's guarding it? BETA?

20:23 <airlied> yeah beta

20:23 <airlied> beta was originally guarding the first struct and the entrypoint using it

20:23 <airlied> but that is removed now for decode

20:26 <jekstrand> airlied: I'm debating if we just want to throw #define VK_ENABLE_BETA_EXTENSIONS at the top of the file instead

20:26 <jekstrand> I think probably no

20:27 <jekstrand> Yeah, ok, that's probably not a great idea.

20:28 maxzor has joined #dri-devel

20:34 mvlad has quit [Remote host closed the connection]

20:47 nchery has quit [Ping timeout: 480 seconds]

20:48 <Lynne> why not?

20:49 <Lynne> I'm considering enabling it by default in our public API headers

20:52 <airlied> I'm not sure that would be a good plan

20:52 <airlied> since nothing underneath it remains at all stable

20:53 <airlied> okay https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20388 contains the radv work out of beta in an MR

20:57 nchery has joined #dri-devel

20:57 nchery has quit [Remote host closed the connection]

20:57 nchery has joined #dri-devel

20:59 ngcortes has quit [Ping timeout: 480 seconds]

20:59 nchery has quit []

20:59 nchery has joined #dri-devel

21:03 nchery has quit []

21:10 bgs has quit [Remote host closed the connection]

21:11 <jekstrand> Also, you're not supposed to ship anything beta in production code

21:11 <jekstrand> Better to have it entirely compile out.

21:17 Duke`` has quit [Ping timeout: 480 seconds]

21:25 rasterman has quit [Quit: Gettin' stinky!]

21:41 cphealy has quit [Remote host closed the connection]

21:41 Haaninjo has quit [Quit: Ex-Chat]

21:41 cphealy has joined #dri-devel

21:43 alyssa has joined #dri-devel

21:43 * alyssa tries to write silly nir pass

21:52 MajorBiscuit has joined #dri-devel

22:07 gouchi has quit [Remote host closed the connection]

22:08 apinheiro has joined #dri-devel

22:08 <alyssa> nir_cf_reinsert to the rescue

22:18 <Lynne> tested, radv video PR still works

22:19 <Lynne> good idea to keep it behind a flag, it's likely machine-crashing if supplied with a bad file

22:20 <Lynne> then again, I have a corrupt sample which crashes the GPU with vaapi

22:23 <airlied> Lynne: I'd like to have it pass CTS at least before unflagging it, the hw encoders are going to fall over no matter what, they usually recover, sometimes the driver recovery fails in other places

22:27 dcz_ has quit [Ping timeout: 480 seconds]

22:29 K0bin has joined #dri-devel

22:29 K0bin has quit []

22:29 robin_ has quit []

22:30 K0bin has joined #dri-devel

22:30 apinheiro has quit [Ping timeout: 480 seconds]

22:40 djbw has joined #dri-devel

22:41 alyssa has quit [Quit: leaving]

22:48 agd5f has joined #dri-devel

22:50 AndroUser has joined #dri-devel

22:54 ahajda_ has quit []

23:00 <emersion> italove: you need to be DRM master

23:10 fab has quit [Quit: fab]

23:16 ngcortes has joined #dri-devel

23:30 AndroUser has quit []

23:32 danvet has quit [Ping timeout: 480 seconds]

23:40 the_sea_peoples has quit [Remote host closed the connection]

23:42 the_sea_peoples has joined #dri-devel

23:53 <airlied> Lynne: my MR branch now uses RADV_PERFTEST=video_decode as the enable to align it with others, I've updated my blogpost

23:54 <Lynne> thanks, I'll add it to my env

23:54 <Lynne> 12th flag in there now, not sure why there are 12 of them, I've never had any faults with any