#dri-devel on 2022-05-21 — irc logs at oftc.irclog.whitequark.org

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

00:01 nchery is now known as Guest1042

00:01 nchery has joined #dri-devel

00:06 simon-perretta-img_ has quit [Ping timeout: 480 seconds]

00:07 Guest1042 has quit [Ping timeout: 480 seconds]

00:21 sdutt has joined #dri-devel

00:21 alyssa has quit [Quit: leaving]

00:24 mclasen has joined #dri-devel

00:29 nchery has quit [Read error: Connection reset by peer]

00:30 pcercuei has quit [Quit: dodo]

00:40 rkanwal1 has quit [Ping timeout: 480 seconds]

00:50 Daanct12 has joined #dri-devel

00:56 columbarius has joined #dri-devel

00:57 co1umbarius has quit [Ping timeout: 480 seconds]

01:20 Emantor has quit [Quit: ZNC - http://znc.in]

01:20 Emantor has joined #dri-devel

01:30 apinheiro has quit [Ping timeout: 480 seconds]

01:51 e|oon has quit [Ping timeout: 480 seconds]

02:01 mclasen has quit [Ping timeout: 480 seconds]

02:11 mclasen has joined #dri-devel

02:19 JohnnyonF has joined #dri-devel

02:23 JohnnyonFlame has quit [Ping timeout: 480 seconds]

02:59 mclasen has quit [Ping timeout: 480 seconds]

03:03 sumits has quit [Quit: ZNC - http://znc.in]

03:03 andrey-konovalov has quit []

03:09 heat has quit [Read error: Connection reset by peer]

03:10 heat has joined #dri-devel

03:10 sumits has joined #dri-devel

03:11 andrey-konovalov has joined #dri-devel

03:18 heat_ has joined #dri-devel

03:18 heat has quit [Read error: Connection reset by peer]

03:34 heat has joined #dri-devel

03:34 heat_ has quit [Read error: Connection reset by peer]

03:56 oneforall2 has quit [Remote host closed the connection]

03:58 oneforall2 has joined #dri-devel

04:09 oneforall2 has quit [Remote host closed the connection]

04:09 eukara has quit [Remote host closed the connection]

04:10 Daanct12 has quit [Ping timeout: 480 seconds]

04:15 oneforall2 has joined #dri-devel

04:19 eukara has joined #dri-devel

04:47 Daanct12 has joined #dri-devel

04:56 Duke`` has joined #dri-devel

05:02 digetx has quit [Ping timeout: 480 seconds]

05:03 digetx has joined #dri-devel

05:16 <HdkR> Is there a DRM ioctl that allocates memory with CPU VA hint for placement?

05:20 OftenTimeConsuming is now known as Guest1059

05:20 OftenTimeConsuming has joined #dri-devel

05:24 Guest1059 has quit [Remote host closed the connection]

05:37 Company has quit [Quit: Leaving]

05:42 icecream95 has joined #dri-devel

05:54 heat has quit [Ping timeout: 480 seconds]

06:03 hch12907 has joined #dri-devel

06:21 lumag_ has quit [Remote host closed the connection]

06:21 lumag_ has joined #dri-devel

06:42 ppascher has quit [Ping timeout: 480 seconds]

06:47 X512 has joined #dri-devel

06:47 <X512> I think you can call mmap on buffer fd and specify virtual address.

06:48 <HdkR> Cute

06:55 hch12907 has quit [Ping timeout: 480 seconds]

06:57 <HdkR> Mostly coming from a passing thought about driver only facing buffers being allocated outside of the lower 32-bit VA space for 32-bit applications. But I doubt anyone wants to deal with the discipline that requires.

06:58 <HdkR> Resolve some of the 32-bit VA problems

07:01 mszyprow has joined #dri-devel

07:03 <airlied> HdkR: we shouldnt be allocating va space at all then should we?

07:04 <HdkR> airlied: Not allocating memory is definitely better than allocating :P

07:05 <HdkR> But any internal staging buffers that never get shown to the application could resolve some of the 32-bit OOM issues

07:07 <HdkR> I don't recall which 32-bit games have OOM problems though

07:09 <HdkR> Or if getting rid of internal buffers from the 32-bit VA space would be enough to resolve that OOM :)

07:17 <airlied> HdkR: but doesnt a 32 bit mesa still only have 32bit va?

07:17 <HdkR> Not under FEX :)

07:18 <airlied> ah

07:18 <airlied> so can fex rum

07:18 <airlied> run 32 bit x86 on 64 bit x86? :-)

07:18 <HdkR> aye. Not optimized for it like the x86->aarch64 path, but still quite fast

07:19 <HdkR> Since we use that path for debugging things

07:21 <HdkR> Once we wire up 32-bit thunking then that could be an avenue for fixing 32-bit OOM problems. If anyone cares enough to categorize mesa allocations to implement the punchthrough.

07:31 jkrzyszt has joined #dri-devel

07:36 danvet has joined #dri-devel

07:37 eukara has quit []

07:43 lumag_ has quit [Ping timeout: 480 seconds]

07:43 hch12907 has joined #dri-devel

07:44 <pq> feaneron, eh heh, well, good that you figured it out - you never checked if open() succeeded? :-p

07:54 rasterman has joined #dri-devel

08:02 glennk has quit [Remote host closed the connection]

08:03 glennk has joined #dri-devel

08:10 mszyprow has quit [Remote host closed the connection]

08:10 mszyprow has joined #dri-devel

08:13 Haaninjo has joined #dri-devel

08:20 mszyprow has quit [Ping timeout: 480 seconds]

08:28 sdutt has quit [Ping timeout: 480 seconds]

08:35 Daaanct12 has joined #dri-devel

08:50 jkrzyszt has quit [Ping timeout: 480 seconds]

08:51 hch12907 has quit [Ping timeout: 480 seconds]

08:59 ppascher has joined #dri-devel

09:16 Daaanct12 has quit [Remote host closed the connection]

09:19 hch12907 has joined #dri-devel

09:26 Haaninjo has quit [Quit: Ex-Chat]

09:27 i-garrison has joined #dri-devel

09:27 garrison has quit [Read error: Connection reset by peer]

09:35 pcercuei has joined #dri-devel

09:44 Daanct12 has quit [Quit: Leaving]

09:44 lemonzest has joined #dri-devel

10:00 hch12907 has quit [Ping timeout: 480 seconds]

10:31 Danct12 has quit [Remote host closed the connection]

10:32 flacks has quit [Quit: Quitter]

10:32 soreau has quit [Read error: Connection reset by peer]

10:32 Danct12 has joined #dri-devel

10:33 soreau has joined #dri-devel

10:33 flacks has joined #dri-devel

10:36 mclasen has joined #dri-devel

10:42 rkanwal has joined #dri-devel

10:45 mclasen has quit []

10:46 mclasen has joined #dri-devel

10:47 mclasen has quit []

10:53 bbrezillon has quit [Ping timeout: 480 seconds]

10:58 bbrezillon has joined #dri-devel

11:07 <feaneron> pq: i did, and it did succeed - the problem is that FD_CLOEXEC's value is 1, matches O_WRONLY, so it *was* valid, but things fell apart on the first attempt to write() to it

11:11 digetx has quit [Ping timeout: 480 seconds]

11:12 digetx has joined #dri-devel

11:15 icecream95 has quit [Ping timeout: 480 seconds]

11:18 alanc has quit [Remote host closed the connection]

11:18 alanc has joined #dri-devel

11:36 Duke`` has quit []

11:47 Duke`` has joined #dri-devel

11:49 sadlerap has quit [Ping timeout: 480 seconds]

11:53 hch12907 has joined #dri-devel

12:17 Net147 has quit [Quit: Quit]

12:18 Net147 has joined #dri-devel

12:34 camus has joined #dri-devel

12:37 pcercuei has quit [Quit: brb]

12:40 pcercuei has joined #dri-devel

12:57 lumag_ has joined #dri-devel

13:36 tobiasjakobi has joined #dri-devel

13:36 tobiasjakobi has quit [Remote host closed the connection]

13:46 MajorBiscuit has joined #dri-devel

13:48 Company has joined #dri-devel

14:00 mi6x3m has joined #dri-devel

14:01 <mi6x3m> hey, using OSMesa and GL from the same process, LLVM gives me "CommandLine Error: Option 'x86-use-base-pointer' registered more than once!"

14:01 <mi6x3m> does anyone have an idea?

14:01 <mi6x3m> i think its' because i have 2 llvm instances next to each other statically but how do they interfere exactly

14:03 <mi6x3m> this is latest llvm

14:10 lumag_ has quit [Remote host closed the connection]

14:10 lumag_ has joined #dri-devel

14:16 <HdkR> mi6x3m: Should be safe to ignore and also !16587 will supposedly resolve it

14:18 ella-0 has joined #dri-devel

14:21 ella-0_ has quit [Read error: Connection reset by peer]

14:23 <mi6x3m> HdkR, it crashes my app :)

14:23 <mi6x3m> what is 16587?

14:26 <mi6x3m> the merge request?

14:27 <HdkR> mi6x3m: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16587 aye

14:29 <mi6x3m> can you give me a brief summary of what's going on?

14:29 <mi6x3m> I can try the merge request for you in 22.0.3 if you want

14:30 MajorBiscuit has quit [Quit: WeeChat 3.4]

14:31 MajorBiscuit has joined #dri-devel

14:31 <HdkR> mi6x3m: LLVM is getting initialized twice and their global initializers get upset about it. Usually it's just a warning and it continues while ignoring duplicated arguments

14:32 <HdkR> That ends up calling `llvm::cl::ResetCommandLineParser();` to reset the internal command line options so on the second initialization it doesn't warn about it

14:32 <mi6x3m> HdkR I'm getting an assertion error in Wine after it :)

14:35 <HdkR> uh oh :)

14:51 lumag_ has quit [Quit: Leaving]

14:55 MajorBiscuit has quit [Ping timeout: 480 seconds]

15:06 heat has joined #dri-devel

15:30 MajorBiscuit has joined #dri-devel

15:35 <marex> airlied: daniels: danvet: hi, can either of you help me with dim a bit ?

15:36 <mi6x3m> HdkR, in case that's the reason, could this make it into 22.0.5 potentially?

15:36 <marex> I am following https://drm.pages.freedesktop.org/maintainer-tools/committer-drm-misc.html#where-do-i-apply-my-patch to apply patches which are on top of https://cgit.freedesktop.org/drm/drm-misc/log/?h=drm-misc-next , but drm-misc-next is in feature-freeze

15:36 <marex> so they should go into drm-misc-next-fixes, right ?

15:36 <marex> but drm-misc-next-fixes is still missing the commits which I need to fix in drm-misc-next, so dim complains about Fixes: tag commit SHA1 not being an ancestor

15:37 <HdkR> mi6x3m: No idea. I'm also just a hapless user :P

15:38 <mi6x3m> wow, but how did you know of the issue

15:38 <HdkR> Someone yesterday brought it up so I recalled it

15:39 <mi6x3m> interesting

15:39 <mi6x3m> what a coincidence

15:45 <mi6x3m> HdkR can you dig out the messages and PM me if you have time?

15:46 <marex> https://oftc.irclog.whitequark.org/dri-devel/2022-05-20

15:47 <HdkR> ^ Then just search for 16587 in that

15:49 <danvet> marex, hm might need a backmerge?

15:49 <mi6x3m> thanks

15:50 <danvet> marex, pls ping mripard tzimmermann mlankhorst (maybe next week) to sort this out

15:50 <danvet> marex, also if the patches you're fixing are not in drm-next already, then they missed the merge window and just pushing them to drm-misc-next is the right branch

15:50 <marex> danvet: I was afraid of that

15:51 mi6x3m has quit [Quit: Leaving]

15:51 <danvet> but if they are in drm-next then I guess you need a backmerge

15:51 <marex> danvet: which one is drm-next again ?

15:51 <danvet> should be quick to sort out if you ping them + why (i.e. which sha1) you need backmerged

15:51 <danvet> marex, the sha1 you're fixing up

15:52 <marex> danvet: I mean, which tree/branch is drm-next ?

15:52 <danvet> marex, the one airlied&me maintain

15:52 <danvet> https://cgit.freedesktop.org/drm/drm

15:52 <marex> I am fixing https://cgit.freedesktop.org/drm/drm-misc/commit/?id=3080c21a043abc83e56da9bb5942eb3f4b92c90e this one

15:53 <danvet> marex, per dim tag-contains that's only in drm-misc-next

15:53 <marex> yes, it is

15:53 <danvet> so drm-misc-next would be the right branch to push this to

15:54 <danvet> marex, I guess we should have one more question in the flowchart to check that the patch you're fixing actually made it to drm-next

15:54 <marex> even though "is drm in feature freeze (occurs after rc6)" ?

15:54 <danvet> if not, then drm-misc-next is still the right branch

15:54 <danvet> marex, yeah

15:54 <marex> OK, then yes, the flow chart needs updating

15:54 <danvet> that question is for when the broken commit has escaped

15:54 <marex> I have been following it rigorously

15:55 <marex> danvet: OK, so ... can you give me a quick hint /wrt dim ?

15:56 <marex> danvet: I am really worried I might mess something up with dim now

15:56 <danvet> marex, reset your drm-misc-next-fixes branch to upstream

15:56 <marex> danvet: so ... I did 'dim checkout drm-misc-next-fixes ; cat /tmp/patches/*patch | dim apply-branch drm-misc-next-fixes'

15:56 <danvet> git reset --hard @{u}

15:57 <danvet> to make sure the patch is gone from there

15:57 <danvet> then do the usual patch apply flow to drm-misc-next branch

15:57 <marex> just like that ? cd drm-misc ; git reset --hard drm-misc/drm-misc-next-fixes ?

15:57 <danvet> yup

15:57 <danvet> that should do the trick

15:59 <marex> danvet: all right, whew

16:00 <marex> OK, drm-misc-next pushed with the three fixes

16:01 <marex> danvet: thank you

16:09 <danvet> marex, btw just noticed that for me that bridge driver doesn't even compile right now in drm-misc-next :-/

16:19 <marex> danvet: which bridge driver ?

16:20 <danvet> Anx7625

16:21 <marex> danvet: that's not something I maintain, I did like 1 patch to it, was it applied ?

16:21 <danvet> ah ok I just thought that's the patch you're going to push

16:21 <marex> danvet: no, but I can take a look

16:21 <danvet> I pinged already on dri-devel

16:21 <danvet> nah only if you're really bored :-)

16:21 <marex> I pushed three small tc358767 patches

16:23 <marex> danvet: it does compile on linux-next/master at least

16:23 <danvet> marex, yeah but it should also compile on that branch

16:23 <danvet> probably should have been applied to another branch or needed a topic branch first

16:24 <danvet> this randomized patch-set split-up is occasionally just too much :-)

16:24 <marex> danvet: what build error do you get, can you tell or point me to an email ?

16:24 <danvet> V4L2_FWNODE_BUS_TYPE_DPI doesn't exist yet

16:24 <danvet> in drm-misc-next

16:24 <danvet> maybe backmerge is enough to fix this

16:25 * danvet didn't look closer, it's w/e

16:26 <marex> 5e052a4d2a47e ("media: media/v4l2-core: Add enum V4L2_FWNODE_BUS_TYPE_DPI")

16:26 <marex> that went through linux-media it seems

16:26 <danvet> yeah just split stuff up too much

16:26 <danvet> instead of merging it all through one tree

16:26 <danvet> with the "foreign" patches acked for merging through that

16:27 <marex> maybe drm and v4l should somehow be ... kind-of ... merged together

16:27 MajorBiscuit has quit [Ping timeout: 480 seconds]

16:28 <marex> they seem to address a lot of the same stuff, except going in different direction

16:28 <marex> it starts to seem like one is the mirror image of the other, except incompatible

16:28 <marex> oh well

16:28 <danvet> marex, this is a very ... complicated topic :-)

16:29 <marex> I will stop at this point

16:29 <marex> let's discuss next topic

16:34 MajorBiscuit has joined #dri-devel

16:51 mszyprow has joined #dri-devel

17:00 mszyprow has quit [Ping timeout: 480 seconds]

17:07 MajorBiscuit has quit [Quit: WeeChat 3.4]

17:14 illwieckz has quit [Quit: I'll be back!]

17:56 MajorBiscuit has joined #dri-devel

18:18 MajorBiscuit has quit [Ping timeout: 480 seconds]

18:20 iive has joined #dri-devel

18:40 Haaninjo has joined #dri-devel

19:06 mhenning has joined #dri-devel

19:16 Duke`` has quit [Ping timeout: 480 seconds]

19:24 Duke`` has joined #dri-devel

19:25 X512 has quit [Quit: Vision[]: i've been blurred!]

19:33 pcercuei has quit [Ping timeout: 480 seconds]

19:43 MajorBiscuit has joined #dri-devel

19:48 eukara has joined #dri-devel

19:51 pcercuei has joined #dri-devel

19:51 Anorelsan has joined #dri-devel

19:57 hch12907 has quit [Ping timeout: 480 seconds]

20:03 MajorBiscuit has quit [Ping timeout: 480 seconds]

20:45 lemonzest has quit [Quit: WeeChat 3.4]

21:00 Daanct12 has joined #dri-devel

21:06 Danct12 has quit [Ping timeout: 480 seconds]

21:29 Haaninjo has quit [Quit: Ex-Chat]

21:44 mhenning has quit [Quit: mhenning]

21:45 apinheiro has joined #dri-devel

21:45 mwk has quit [Remote host closed the connection]

21:46 mwk has joined #dri-devel

21:50 mhenning has joined #dri-devel

22:01 Anorelsan has quit [Quit: Leaving]

22:14 sdutt has joined #dri-devel

22:19 Duke`` has quit [Ping timeout: 480 seconds]

22:20 mhenning has quit [Quit: mhenning]

22:20 mhenning has joined #dri-devel

22:24 rkanwal has quit [Quit: rkanwal]

22:34 illwieckz has joined #dri-devel

22:54 danvet has quit [Ping timeout: 480 seconds]

23:11 flto has quit [Ping timeout: 480 seconds]

23:13 morphis has quit [Ping timeout: 480 seconds]

23:13 flto has joined #dri-devel

23:13 morphis has joined #dri-devel

23:17 iive has quit []

23:22 pcercuei has quit [Quit: dodo]

23:27 heat has quit [Remote host closed the connection]

23:41 apinheiro has quit [Quit: Leaving]