#
…

2022-02-09

00:07nyx_o has quit [Quit: WeeChat 3.3]
00:14user982492 has joined #asahi-gpu
00:26yuyichao has joined #asahi-gpu
00:37NightsOnly[m]hmmm, i managed to get it compiled and installed with ninja install. Although i am somewhat confused on how to get it working?
00:37NightsOnly[m]like, how do i attempt to use it and have something crash so i can figure out what needs implementing?
01:21jbowen has joined #asahi-gpu
01:30jbowen has quit [Ping timeout: 480 seconds]
01:30chadmedDYLD_INSERT_LIBRARIES=/path/to/your/compiled/libGL.1.dyld then run whatever opengl program you want to profile
01:32chadmedyou should keep in mind that virtually all opengl software for macos is going to be built for and linked against amd64 libraries, so they will not accept your arm64 mesa libGL. alyssa has been using the dEQP conformance tests since you can build them yourself.
01:37chadmedthe Metal shader examples are just used to get memory dumps of data structures to see how the hardware expects things like textures to be presented to it for processing, since we can assume that Apple's Metal libraries do this the "correct" way for the hardware. that information is then used to teach mesa how to do the same thing for GLES shader programs
01:51rowang077[m] has joined #asahi-gpu
02:03jbowen has joined #asahi-gpu
02:20JuniorJPDJ has joined #asahi-gpu
02:30phireright, I'm writing up an infodump for you marcan
02:31phireyou still interested at looking into gpu stuff today?
02:32perigoso[m] has joined #asahi-gpu
02:41The_DarkFire_[m] has joined #asahi-gpu
03:54nyx_o has joined #asahi-gpu
03:55phiologe has joined #asahi-gpu
03:59PhilippvK has quit [Ping timeout: 480 seconds]
04:33user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
04:38sheepgoose has joined #asahi-gpu
04:39user982492 has joined #asahi-gpu
05:24petermlyon[m] has quit [Server closed connection]
05:24petermlyon[m] has joined #asahi-gpu
05:25davay[m] has joined #asahi-gpu
05:27DanielHuisman[m] has quit [Server closed connection]
05:27DanielHuisman[m] has joined #asahi-gpu
05:28Liam[m] has joined #asahi-gpu
05:29fried_dede[m] has joined #asahi-gpu
05:32HaoYanQi[m] has joined #asahi-gpu
05:33peerp[m] has joined #asahi-gpu
05:33RasmusEneman[m] has joined #asahi-gpu
05:33pulpy_orange2[m] has joined #asahi-gpu
05:34retonlage[m] has joined #asahi-gpu
05:36Synth[m] has joined #asahi-gpu
05:36Rakshit[m] has quit [Server closed connection]
05:36Rakshit[m] has joined #asahi-gpu
05:37YichaoYu[m] has quit [Server closed connection]
05:37YichaoYu[m] has joined #asahi-gpu
07:09fbd222[m] has joined #asahi-gpu
07:16user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
07:39MajorBiscuit has joined #asahi-gpu
07:46Major_Biscuit has joined #asahi-gpu
07:48MajorBiscuit has quit [Ping timeout: 480 seconds]
08:28svenendpoint 0x1 is probably the normal rtkit crashlog
08:36phirethanks. Prehaps I should be trying harder to crash it
08:46svenjust send 0x40 or 0x22 to it and it should crash
08:48svenweird that gfx-asc can essentially get access to every page
08:48phireyes, it's very privliaged
08:57phireoh... thank you crashlog and your very creative endpoint naming scheme of "user01" and "user02"
08:58phireRTKit Task List: rtk_ep_work, power, agx_background, agx_recovery, agx_interrupt, agx_power, agx_sampler
10:11phireforcing crashes hasn't told me that much
10:13phireep 0x21 crashes when you send it message types it doesn't exist, but doesn't seem to do any checking on the args to the 0x83 message
10:13phireEP 0x20 doesn't seem to crash no matter what you send it
10:15phirewell, unless you do something stupid like senting a null pointer on it's init data
10:16phire GFX PANIC - Unable to grab init data from host - agx_background(2)
11:27marcanphire: the kick via ASC plus shared memory ring buffer/work queue pattern is used for DCP, so finding one of those somewhere wouldn't surprise me as a work submission mechanism
11:27marcan*via mbox
11:30phireyeah, sounds right
11:30phireand way more sane than my other theories
11:30phirethough the DCP ASC looks nothing kike the GFX ASC. I doubt they are related
11:31marcanI mean the general idea
11:31marcanobviously GFX does not use EPIC nor that bizarro dcpep thing
11:31marcanbut e.g. the kick stuff is also kinda how syslog works
11:32phireyeah, the general idea
11:39phireI'm current mapping out this "init data" struct
11:40phireif there is an address to a shared memory queue, it should be there
11:43marcanI would expect at least a few global queues and then possibly some per-context stuff or similar
11:43marcanthe latter could be dynamically allocated via global messages
11:44phireyeah, I haven't even tried to look for the per-context stuff, because I assume there needs to be a global queue to set the context
11:44phireGFX PANIC - Host-mapped FW allocations are disabled, but FW only supports enabled (init_data_addr=0xffffffa0000b0000)! - agx_background(2)
11:45phirethat's an interesting panic
11:45marcanit's also possible the GPU allocates the queues btw, and updates the structure or something
11:46phiremonitoring doesn't show any updates to this structure
12:49alyssaphire: Keep in mind that WindowServer renders everything via Metal internally
12:50alyssaso unless you're doing all this headless, the "is Metal in use?" piece is questionable
12:50alyssa(Maybe you *are* doing it headless, though?)
12:51phirewhat I mean is that I've only seen in while running the metal spinning cube example, not during login or on the desktop
12:52alyssaok.. my point is that login and the desktop are both using the GPU heavily
12:52alyssa(and that there will be kicks and work submitted from those too)
12:53phireyes... I agree it is odd that only that example would cause them
12:54alyssaphire: It might be worth it to run a metal compute kernel test over SSH with no display connected (lid closed for the laptop, I guess)
12:54alyssasignal:noise ratio might be better for finding the kick
12:55phirehmm, lid closed might work
13:01phirewell... m1n1 crashes if I boot when the lid closed
13:03phiredid stop most of the kicks. I think it still drew the white apple logo
13:08marcanphire: that's because it goes into suspend
13:08marcanyou need to disable that
13:08* alyssasupposes this part would be easier on a mini
13:08phirein macos settings?
13:08marcanyes, I think that should work
13:08marcanyou could also just turn off the screen
13:08marcanI think setting the brightness to 0 should stop display updates, but it shouldn't suspend
13:18phiresetting the "prevent sleep when display off" option didn't stop it from sleeping
13:19phireand setting zero brightness doesn't prevent rendering
13:22phirethis apparently works: ssh in, then close the lind
13:22phires/lind/lid/
14:45yuyichao has quit [Ping timeout: 480 seconds]
15:03yuyichao has joined #asahi-gpu
15:11phireAlright, I've found a set of circular buffers
15:28Andre[m]1 has joined #asahi-gpu
15:31Jamie[m]1 has joined #asahi-gpu
15:45Hinata[m] has joined #asahi-gpu
16:07mort_8 is now known as mort_
16:12spokv[m] has joined #asahi-gpu
16:21blassphemy[m] has quit [Server closed connection]
16:21blassphemy[m] has joined #asahi-gpu
16:24rusty-nail[m] has joined #asahi-gpu
16:34NicolasOury[m] has joined #asahi-gpu
16:55jbowen has quit [Ping timeout: 480 seconds]
17:07jbowen has joined #asahi-gpu
17:18Major_Biscuit has quit [Ping timeout: 480 seconds]
17:18jbowen_ has joined #asahi-gpu
17:22jbowen has quit [Ping timeout: 480 seconds]
17:25Genbuchan[m] has joined #asahi-gpu
17:27Nspace has joined #asahi-gpu
17:29jbowen_ has quit [Ping timeout: 480 seconds]
18:02anparra[m] has quit [Server closed connection]
18:02anparra[m] has joined #asahi-gpu
18:05jbowen has joined #asahi-gpu
18:16marcan has quit [Server closed connection]
18:16marcan has joined #asahi-gpu
18:20lonjil has joined #asahi-gpu
18:34zelig_[m] has quit [Server closed connection]
18:34zelig_[m] has joined #asahi-gpu
18:42user982492 has joined #asahi-gpu
19:15jbowen has quit [Quit: leaving]
19:18feeleep[m] has quit [Server closed connection]
19:18feeleep[m] has joined #asahi-gpu
19:24minecrell has quit [Server closed connection]
19:24minecrell has joined #asahi-gpu
19:35DmitrySboychakov[m] has quit [Server closed connection]
19:35DmitrySboychakov[m] has joined #asahi-gpu
19:40DragoonAethis has quit [Quit: hej-hej!]
19:41DragoonAethis has joined #asahi-gpu
19:46jbowen has joined #asahi-gpu
20:05LilleCarl[m] has quit [Server closed connection]
20:05LilleCarl[m] has joined #asahi-gpu
20:05user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
20:29user982492 has joined #asahi-gpu
20:43grange_c has joined #asahi-gpu
20:48pg12 has quit [Server closed connection]
20:48pg12 has joined #asahi-gpu
21:15TheLink has quit [Server closed connection]
21:15TheLink has joined #asahi-gpu
21:17aleasto has quit [Server closed connection]
21:17aleasto has joined #asahi-gpu
21:21BenPetterborg[m] has quit [Server closed connection]
21:21BenPetterborg[m] has joined #asahi-gpu
21:26vimsos[m] has quit [Server closed connection]
21:26vimsos[m] has joined #asahi-gpu
21:34FireFox317 has quit [Server closed connection]
21:36FireFox317 has joined #asahi-gpu
21:42user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
21:45M1bn3mar[m] has quit [Server closed connection]
21:45M1bn3mar[m] has joined #asahi-gpu
22:27user982492 has joined #asahi-gpu
22:43eragon has quit [Server closed connection]
22:43eragon has joined #asahi-gpu
23:48commandoline_ has quit [Quit: Bye!]
23:49commandoline has joined #asahi-gpu
…