#dri-devel on 2021-09-11 — irc logs at oftc.irclog.whitequark.org

2021-07-26 22:56 ChanServ changed the topic of #dri-devel to: <ajax> nothing involved with X should ever be unable to find a bar

00:00 <hanetzer> and I just realized I've been reading /topic this entire time; my mistake was "should be able to find a bar", as in, keep them away from t he booze :)

00:00 <hanetzer> so. that alcoholic screed out of the way; does mesa have a build system (docker/etc) I can use to build HEAD against for dev/testing purposes?

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

00:11 <icecream95> hanetzer: Oh, I always thought that was a reference to BAR (as in the "resizeable BAR" everyone is talking about these days), not an actual bar

00:12 <hanetzer> idk, maybe both lulz.

00:12 <hanetzer> my eyes see it in terms of booze, however :)

00:12 <kisak> the quote is older than the existance of resizable BAR

00:13 <HdkR> I always figured it was a bar of acceptable quality :P

00:13 <anholt_> but yeah, it was in reference to pci bars

00:16 <kisak> here I was under the impression that it was a life lesson to take a swig or two before digging into legacy code from X's era

00:16 nchery has quit [Quit: Leaving]

00:17 * hanetzer drank a fair bit while porting mesa to meson for gentoo :3

00:18 <hanetzer> there was some bug or another, forget who's fault, with buiding with USE=abi_86_32 on gentoo

00:19 jewins has quit [Ping timeout: 480 seconds]

00:26 <HdkR> What is abi 86_32? x32?

00:35 <hanetzer> means 32 bit libs on a 64 bit host.

00:39 <marex> hanetzer: what's the point of that ? :)

00:39 <marex> (cache utilization improvement?)

00:39 <kisak> HdkR: Gentoo portage supports multilib package compiling in the same environment. The end user gets to set ABI_X86="32 64 x32" (or removing any of those archs), which gets converted to a set of USE=abi_x86_xxx

00:40 <kisak> that's across most of the packages in portage as the common mechanism

00:43 <hanetzer> marex: well, some packages do not ship a 64 bit version at all (steam seems to be the primary offender tbh), so we tend to need 32bit versions of a handful of libs on amd64 gentoo machines (same for other distros tbh)

00:49 <hanetzer> like, the games under steam tend to have both 32 and 64 bit builds; but the actual steam client/launcher thingus is only available in 32bit.

00:50 <HdkR> kisak: Ah, so luckily not x32 :)

00:53 robmur01_ has joined #dri-devel

01:00 robmur01 has quit [Ping timeout: 480 seconds]

01:01 columbarius has joined #dri-devel

01:03 co1umbarius has quit [Ping timeout: 480 seconds]

01:03 dfip^ has quit [Ping timeout: 480 seconds]

01:08 mbrost has joined #dri-devel

01:08 dfip^ has joined #dri-devel

01:10 <hanetzer> ah actually the USE flag is abi_x86_{32,64,x32}

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

01:20 Emantor has joined #dri-devel

01:33 jhli has joined #dri-devel

01:43 agx has quit [Read error: Connection reset by peer]

01:43 agx has joined #dri-devel

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

01:49 azorshi has quit [Remote host closed the connection]

01:50 flto has quit [Ping timeout: 480 seconds]

01:54 mattrope has quit [Read error: Connection reset by peer]

01:58 <marex> sravn: could you pick https://patchwork.kernel.org/project/dri-devel/patch/20210713191633.121317-1-marex@denx.de/ ? It has AB/RB, it seems to have fallen through the cracks

02:14 flto has joined #dri-devel

02:17 boistordu has quit [Ping timeout: 480 seconds]

02:28 <clever> https://matrix.org/_matrix/media/r0/download/matrix.org/mqHLsYaujzNqPTbWPkJecidp/20210910_232753_1666962563.jpg

02:29 <clever> this is the latest render coming out of my rpi code

02:30 <clever> each line in the grid is 2 pixels thick, with a spacing of 20 pixels start to start (so 18 pixels of black)

02:31 <clever> the rpi logo, triangle, and text are each a seperate later, unrelated to the current issue, but acting as some references

02:31 <clever> there is then a 200x200 rect (filled with a mix of black, and transparent), that is mapped to /dev/fb0

02:31 <clever> though the transparent half isnt coming thru clearly

02:32 <clever> its top-left corner is located at 200,50

02:33 <clever> now that the grid is on, i can see that the black portion is 50 pixels tall

02:45 dllud has joined #dri-devel

02:45 dllud_ has quit [Read error: Connection reset by peer]

02:59 lemonzest has joined #dri-devel

03:13 <clever> ok, something is definitely screwy with this config

03:13 <clever> i pre-filled the framebuffer with pure red before launching linux

03:13 <clever> its still red now, but /dev/fb0 claims its not

03:38 <marex> clever: time to use the existing kernel driver for the vc4 ?

03:38 <clever> marex: nope, that fails due to async external abort

03:38 <clever> marex: you need a magic hangshake before the arm core is allowed to touch the 2d hardware, and i have yet to find it

03:38 <clever> but i did just find a decent clue

03:39 <clever> [root@nixos:~]# ramdumper 0x08000000 64

03:39 <clever> starting at 0x8000000 (128MB)

03:39 <clever> 0x08000000 00 ff 00 ff 00 ff 00 ff 00 ff 00 ff 00 ff 00 ff |................|

03:39 <clever> marex: this is a custom util, that will mmap /dev/mem, and dump a chunk of physical memory

03:39 <clever> via that, i can see the solid green framebuffer, that the firmware had setup

03:39 <clever> [root@nixos:~]# hexdump -C /dev/fb0 | head

03:39 <clever> 00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................|

03:39 <clever> *

03:39 <clever> 00006ac0 55 ff 55 00 00 00 00 00 00 00 00 00 00 00 00 00 |U.U.............|

03:40 <clever> marex: but, the /dev/fb0 created by linux, in response to the DT entry, is something else!

03:43 <marex> clever: well, print the address of the framebuffer in kernel and use print_hex_dump() there ?

03:43 <marex> clever: then you will know exactly what is in the framebuffer on boot

03:44 <clever> but thats also going to be a virtual addr in the kernel

03:44 <clever> and i need to know what phys addr it maps to

03:44 <marex> ioremap it, no ?

03:44 <marex> you know the PA, the PA is the PA of the framebuffer from which the vc4 is scanning data out to the rest of the display pipe, no ?

03:44 <clever> yeah

03:45 <clever> i know that 0x8000000 is the PA of the framebuffer

03:45 <clever> i configured it as such, and i can see the initial boot image in /dev/mem at that offset

03:45 <clever> drivers/video/fbdev/Makefile:obj-$(CONFIG_FB_SIMPLE) += simplefb.o

03:46 <clever> and that is the driver the compatible= i used matches to

03:46 <clever> drivers/gpu/drm/tiny/simpledrm.c: { .compatible = "simple-framebuffer", },

03:46 <clever> ah wait, it shows up in 2 drivers!

03:47 <clever> [root@nixos:/sys/devices/platform/system/8000000.framebuffer/graphics/fb0]# cat name

03:47 <clever> simpledrm

03:47 <clever> marex: and linux is using the other one!

03:47 <marex> clever: tinydrm is for those spi/i2c attached displays

03:48 <clever> CONFIG_DRM_SIMPLEDRM is what stole my node

03:49 idr has quit [Quit: Leaving]

03:51 <clever> marex: krnel rebuilt, and entirely bricked!

03:52 * clever whips out jtag

03:55 <marex> clever: is this drivers/video/fbdev/simplefb.c what you are trying to get operational ?

03:55 <clever> marex: that is the one i thought i was using, until just now

03:55 <clever> ive disabled the conflicting one, but that bricked the system, need to un-brick it first

03:58 <clever> i have dmesg logs once more!

03:58 <clever> not sure why it bricked, maybe a corrupt zImage copy?

03:59 <clever> [root@nixos:/sys/devices/platform/system/8000000.framebuffer]# ls -l

03:59 <clever> marex: that definitely changed the driver, this dir looks entirely different now

04:00 <clever> Documentation/ABI/testing/sysfs-devices-waiting_for_supplier: The /sys/devices/.../waiting_for_supplier attribute is only

04:00 <clever> and this file was not there before, *reads docs*

04:02 <clever> ah, /dev/fb0 doesnt exist anymore, nothing has claimed the DT node

04:03 mbrost has quit [Ping timeout: 480 seconds]

04:05 <clever> yep, CONFIG_FB_SIMPLE was off

04:08 <clever> marex: and i have a blinking cursor once more!!

04:08 macromorgan is now known as Guest6938

04:08 macromorgan has joined #dri-devel

04:08 Guest6938 has quit [Read error: Connection reset by peer]

04:09 <clever> and a login prompt!!

04:10 <clever> it versk!!!

04:12 <clever> marex: thanks to your questions, i realized there are 2 simple-framebuffer drivers, and found the one that works

04:12 <clever> https://matrix.org/_matrix/media/r0/download/matrix.org/pGBfGKVrJnKwUuecpimEZdpX/20210911_011204_935977913.jpg

04:13 <marex> clever: you should probably have that LCD checked, it doesn't look too healthy ;-)

04:13 <clever> marex: its a CRT tv

04:14 <marex> ;-)

04:14 <clever> because of the previously mentioned issues, the linux vc4 driver cant run

04:14 <clever> and i havent RE'd the hdmi init enough yet, to bring hdmi online

04:14 <clever> so i'm stuck with just ntsc out

04:15 <clever> https://i.imgur.com/Yeh6wEr.jpg

04:15 <clever> marex: but it does look way way better, when i feed that ntsc into a proper lcd based screen

04:17 <clever> --- a/drivers/video/fbdev/core/bitblit.c

04:17 <clever> + printk("rendering %c to %d,%d\n", s[0], xx, yy);

04:17 <clever> marex: oh, thats a very bad idea, now that its working, infinite recursion!

04:28 mbrost has joined #dri-devel

04:29 shadeslayer has quit [Quit: Ping timeout (120 seconds)]

04:29 shadeslayer has joined #dri-devel

04:31 dllud_ has joined #dri-devel

04:31 dllud has quit [Read error: Connection reset by peer]

04:51 thellstrom1 has joined #dri-devel

04:53 dllud has joined #dri-devel

04:53 dllud_ has quit [Read error: Connection reset by peer]

04:55 dllud_ has joined #dri-devel

04:55 dllud has quit [Read error: Connection reset by peer]

04:57 thellstrom has quit [Ping timeout: 480 seconds]

05:03 Company has quit [Quit: Leaving]

05:23 i-garrison has joined #dri-devel

05:29 lemonzest has quit [Quit: WeeChat 3.2]

05:30 lemonzest has joined #dri-devel

05:44 jhli has quit [Remote host closed the connection]

05:51 mlankhorst has joined #dri-devel

05:53 Duke`` has joined #dri-devel

05:58 mbrost has quit [Remote host closed the connection]

06:00 danvet has joined #dri-devel

06:31 mmenzyns has quit []

06:31 sdutt has quit [Remote host closed the connection]

06:31 mmenzyns has joined #dri-devel

06:32 Duke`` has quit [Ping timeout: 480 seconds]

06:38 dllud has joined #dri-devel

06:38 dllud_ has quit [Read error: Connection reset by peer]

06:39 dllud_ has joined #dri-devel

06:39 dllud has quit [Read error: Connection reset by peer]

06:40 Duke`` has joined #dri-devel

06:41 alanc has quit [Remote host closed the connection]

06:42 alanc has joined #dri-devel

06:58 quasselcore has joined #dri-devel

08:03 quasselcore has quit [Read error: Connection reset by peer]

08:13 <pq> marex, sorry, I've no idea about Mesa internals.

08:14 quasselcore has joined #dri-devel

08:27 quasselcore has quit [Remote host closed the connection]

08:40 gouchi has joined #dri-devel

08:46 pcercuei has joined #dri-devel

09:01 mlankhorst has quit [Ping timeout: 480 seconds]

09:16 mlankhorst has joined #dri-devel

09:43 boistordu_ex has joined #dri-devel

09:56 Hi-Angel has joined #dri-devel

10:06 mmenzyns has quit []

10:07 mmenzyns has joined #dri-devel

10:11 yogesh_mohan has joined #dri-devel

11:01 dllud has joined #dri-devel

11:01 dllud_ has quit [Read error: Connection reset by peer]

11:09 camus has quit []

11:44 camus has joined #dri-devel

11:54 mlankhorst has quit [Ping timeout: 480 seconds]

11:55 camus1 has joined #dri-devel

12:00 camus has quit [Ping timeout: 480 seconds]

12:27 dllud_ has joined #dri-devel

12:27 dllud has quit [Remote host closed the connection]

12:29 dllud has joined #dri-devel

12:29 dllud_ has quit [Read error: Connection reset by peer]

12:31 dllud_ has joined #dri-devel

12:31 dllud has quit [Read error: Connection reset by peer]

12:33 shashanks has quit [Ping timeout: 480 seconds]

12:33 macromorgan is now known as Guest6957

12:33 Guest6957 has quit [Read error: Connection reset by peer]

12:33 macromorgan has joined #dri-devel

12:36 shashanks has joined #dri-devel

12:40 mlankhorst has joined #dri-devel

12:56 Company has joined #dri-devel

13:43 perr has joined #dri-devel

13:50 macromorgan is now known as Guest6958

13:50 Guest6958 has quit [Read error: Connection reset by peer]

13:50 macromorgan has joined #dri-devel

13:51 perr has quit [Quit: Leaving]

13:51 perr has joined #dri-devel

13:54 perr has quit []

14:04 smpl has joined #dri-devel

14:20 camus1 has quit []

14:30 co1umbarius has joined #dri-devel

14:32 columbarius has quit [Ping timeout: 480 seconds]

15:25 cphealy has quit [Remote host closed the connection]

15:40 <danvet> emersion, thx a lot for the pull, applied :-)

15:43 <danvet> also realized I had a commit here locally to remove the g+ links ...

16:11 <marex> pq: oh ... I thought you were one of the experts :)

16:22 <hanetzer> yeah man. the shorter the name the greater the expertise, this is a fact :P

16:29 azorshi has joined #dri-devel

16:31 <FLHerne> marex: pq is a windowing/Wayland/kernel ABI expert, just not a mesa one

16:34 <hanetzer> fair enough lol

16:37 <marex> FLHerne: ACK

16:43 iive has joined #dri-devel

16:59 Lucretia has quit []

17:00 sarnex_ has quit []

17:04 Lucretia has joined #dri-devel

17:08 sarnex has joined #dri-devel

17:34 hanetzer1 has joined #dri-devel

17:35 hanetzer has quit [Ping timeout: 480 seconds]

17:47 <daniels> hanetzer1: you can just look at any pipeline that's run in GitLab CI - those all execute in publicly-available Docker images

17:48 <daniels> hanetzer1: e.g. https://gitlab.freedesktop.org/mesa/mesa/-/jobs/13578327 tells you the container name in the first couple of lines of the job output; you can just pull that and execute the same stuff inside it

17:57 <sravn> marex: backlight patches go through Lee Jones <lee.jones@linaro.org> - backlight is not part of drm-misc although that would make total sense if it was

18:01 mlankhorst has quit [Ping timeout: 480 seconds]

18:12 sdutt has joined #dri-devel

18:38 azorshi has quit []

19:03 dllud has joined #dri-devel

19:03 dllud_ has quit [Remote host closed the connection]

19:37 boistordu has joined #dri-devel

19:42 boistordu_ex has quit [Ping timeout: 480 seconds]

19:43 flto has quit [Remote host closed the connection]

19:44 flto has joined #dri-devel

19:45 flto has quit [Remote host closed the connection]

19:45 flto has joined #dri-devel

19:47 illwieckz has quit [Ping timeout: 480 seconds]

19:54 lemonzest has quit [Quit: WeeChat 3.2]

20:01 illwieckz has joined #dri-devel

20:18 Daanct12 has joined #dri-devel

20:25 Danct12 has quit [Ping timeout: 480 seconds]

20:31 sdutt has quit []

20:44 cphealy has joined #dri-devel

20:53 thellstrom has joined #dri-devel

20:53 thellstrom1 has quit [Read error: Connection reset by peer]

21:17 dllud_ has joined #dri-devel

21:17 dllud has quit [Read error: Connection reset by peer]

21:18 dllud has joined #dri-devel

21:18 dllud_ has quit [Read error: Connection reset by peer]

21:25 quasselcore has joined #dri-devel

21:36 danvet has quit [Ping timeout: 480 seconds]

21:46 Duke`` has quit [Ping timeout: 480 seconds]

21:48 gouchi has quit [Remote host closed the connection]

22:05 ivyl has quit [Ping timeout: 480 seconds]

22:06 ivyl has joined #dri-devel

22:08 rossy has quit [Ping timeout: 480 seconds]

22:08 rossy has joined #dri-devel

22:18 bl4ckb0ne has quit [Ping timeout: 480 seconds]

22:18 bl4ckb0ne has joined #dri-devel

22:20 vyivel has quit [Ping timeout: 480 seconds]

22:20 vyivel has joined #dri-devel

22:42 imre is now known as Guest6984

22:42 imre has joined #dri-devel

22:43 vsyrjala has quit [Ping timeout: 480 seconds]

22:43 Guest6984 has quit [Ping timeout: 480 seconds]

22:47 vsyrjala has joined #dri-devel

22:50 dllud_ has joined #dri-devel

22:50 dllud has quit [Read error: Connection reset by peer]

23:24 quasselcore has quit [Read error: Connection reset by peer]

23:25 quasselcore has joined #dri-devel

23:27 Hi-Angel has quit [Ping timeout: 480 seconds]

23:54 iive has quit []