#
…

2023-03-13

00:01nick_ has left #asahi-dev [#asahi-dev]
00:44kesslerd_ has quit [Quit: Konversation terminated!]
00:45kesslerd has joined #asahi-dev
00:49stipa is now known as Guest7513
00:49Guest7513 has quit [Read error: Connection reset by peer]
00:49stipa has joined #asahi-dev
00:52kesslerd has quit [Remote host closed the connection]
00:54barrett has quit [Ping timeout: 480 seconds]
01:08barrett has joined #asahi-dev
01:13kesslerd has joined #asahi-dev
01:14kesslerd has quit []
01:17eiln has quit [Remote host closed the connection]
01:23eiln has joined #asahi-dev
02:25systwi has joined #asahi-dev
02:29systwi_ has quit [Ping timeout: 480 seconds]
02:33zef_ has joined #asahi-dev
02:40zef has quit [Ping timeout: 480 seconds]
03:22barrett has quit [Ping timeout: 480 seconds]
03:25barrett has joined #asahi-dev
03:41chip_x has joined #asahi-dev
03:46chipxxx has quit [Ping timeout: 480 seconds]
04:05barrett has quit [Ping timeout: 480 seconds]
05:03marcanoh god I'm testing this on an M2 Air and this is so broken
05:03marcanplasmashell fails to start half the time and stays like that, even if restarted
05:03marcanit segfaults
05:03marcanI can't debug it with gdb because gdb is *also* broken (known upstream bug, fixed in main)
05:04marcanmeanwhile the kwin blur glitches are horrible but not only that, sometimes I get cursor redraw glitches even with blur disabled which tells me there is a deeper problem
05:05marcanI suspect half of this is bugs unrelated to us but I am not patching random KDE packages and we know we can't upstream anything to ALARM
05:06marcanjannau: how practical would it be to get cursor planes into dcp? that would at least alleviate the worst of the kwin issues I think
05:06marcanand I need to track down this segfault mess
05:06* marcancompiles gdb...
05:08marcanat this point I'm half tempted to just throw the towel on ALARM and switch to Fedora and get everything sorted out there, I know I can get this stuff fixed on short notice
05:13marcanthe cursor redraw glitches go away at 100/200% so this is totally yet another kwin fractional scaling bug... sigh.
05:18marcanand gdb still crashes... wtf.
05:19marcanaaand the fix the bug was closed by did not fix it
05:23marcanI guess I'm bisecting gdb now...
05:31marcansegfault is in Qt... sigh... gdb 12.1 works and 13.1 fails.
05:56c10l57 has joined #asahi-dev
05:56c10l5 has quit [Read error: Connection reset by peer]
06:16jannaumarcan: we need a solution to deal with a partially off-screen cursor. I want to try if I can get away with adding 32 pixels to the stride and map it between a transparent page. that should give us 31 pixels transparency horizontally/vertically to work around dcp's limitation
06:18jannauthat depends on unmanaged iommu domains, I don't want to rush that
06:19jannauwe can of course also deal with it by copying the visible part of the cursor to a shadow buffer if it is partially offscreen
06:21jannauugly but not hard to implement
06:40marcanah right, the partial cursor issue. argh.
06:40marcanjannau: it doesn't work using the source x/y box to offset it?
06:48jannauI don't follow. inpractice we have the issue at the bottom and right screen edge so source x/y are always 0
06:51dottedmaga stupid question: if it's RGBA (it's RGBA, right?) can we play with w/h/stride of the buffer to tell DCP that the plane is exactly the size to fit on screen (and thus effectively hide the part of buffer that'd be rendered offscreen)
06:54dottedmagSomething like this (assuming that DCP has a notion of a stride): https://gist.github.com/dottedmag/4f48d81c143d56b93249926ec1d29bd7
06:56jannaumarcan: sigh, the surface limitations are biting us as well surf[2] doesn't overlay on surf[1] (with our current code), surf[1] on top of surf[0] does but uses overlays the background as well
06:56marcanjannau: and cropping w/h at src box doesn't work for that case?
06:57marcanor is this a min plane size thing when w/h drop too low?
06:57jannaudottedmag: the issue is that dcp expects an overlay to have at least 32x32 pixels on screen, if not it starts scaling the overlay
06:58marcanurgh okay
07:04dottedmagjannau: So this is also a problem for small cursors, not only partially off-screen ones, right?
07:08jannaudottedmag: default curser fb size is 64x64 independently of the cursor size
07:10jannaulol, I found a solution for the bottom edge. I can display the cursor offscreen as long as src and dst rect have a width of at least 32x32
07:11jannaua, my code is broken, I think that should work for the right edge as well
07:37jannauno, that trick doesn't work at the right edg
07:40jannaualso we can't support a2rgb10 cursors, which kde uses. I was under the impression that the cursor fb is supposed to be argb32
07:41chadmed has quit [Remote host closed the connection]
07:46jannaumarcan: btw, I have see no issues with plasma/wayland on alarm with asahi-dev
07:48marcanjannau: this is random, it's particularly bad on my M2 Air for some reason
07:48marcanplasmashell just dies on half the logins
07:48marcanas for kwin, the non-integer scaling artifacts got a lot worse recently
07:48chadmed has joined #asahi-dev
07:48marcannow cursor redraws leave garbage behind
07:49marcanthis=0x3f sigh
07:49marcanthis is some... heap memory manager... collecting stuff from the javascript stack?
07:49marcanif this is another page size thing...
07:49ChaosPrincessin case it matters - i can also confirm cursor redraw leaving garbage in a square-shaped patter around cursor on plasma. didnt report it at first cause running gentoo, not arch, but apparently its not just my distro being broken.
07:50jannaumaybe related to the pick boxes I see on kde login on m2, but I've never seen a crash
07:50marcannah, the pink boxes are probably the gpu driver still missing sync
07:52jannaucursor leaves square artifacts on the panel but nowhere else
07:52marcanit does that on random surfaces for me, including title bars sometimes
07:52marcanit's not consistent
07:58jannauI see have cursor artifacts pretty consistently on the panel but nowhere else with fractional scaling
07:58marcanmaybe it only happens without plasmashell running (which is the state I'm in right now)
08:00eiln has quit [Quit: Page closed]
08:02chadmedChaosPrincess: i try very hard to not release broken garbage so whatever state gentoo is in is what is going on in asahi
08:02chadmedunless you dont use the overlay in which case youre on your own i guess
08:03chadmedthe only cursor artefacts ive noticed is occasionally a thick black outline appearing around it but it seems completely random
08:04ChaosPrincessi have a smashed together install with some packages from overlay and some from my own weird scripts, but iirc mesa is overlay
08:07marcansigh... valgrind doesn't repro, libasan just deadlocks (??)
08:10jannauhave you tried booting with arm64.nobti / arm64.nopauth to rule out bti/pac issues if you're testing fedora? I suppose valgrind doesn't support that
08:11marcanthis is alarm
08:11marcanand I think it is a race of some sort
08:11marcansomehow pipewire is involved in the deadlock with asan
08:12marcanand gettext
08:12marcanwhy is everything so broken
08:13jannaustrange that I don't hit that on the mbp 13" m2
08:14marcanif it's a race it's probably going to depend on nonsense like how big my taskbar is and what widgets I have on it
08:17marcanhow the hell is gettext deadlocking, who is holding this lock?
08:19dottedmagwhy there is a lock in gettext anyway
08:19dottedmagmaybe something else is screwed up and it's not gettext?
08:21marcanmaybe, or maybe gnulib locks are broken on M1 because reasons
08:21marcanthis is getting ridiculous
08:21psykosewouldn't surprise me re gnulib
08:22psykosecould be the classic "m1 is so o-o-o even gnulib is broken"
08:34marcanok wtaf this is a pthread mutex under the hood and as far as I can tell it is unlocked and locking it deadlocks
08:34marcanis glibc broken now too?
08:45marcan... and I just realized ALARM is shipping glibc 2.35 while their PKGBUILD is at 2.37-2
08:45marcanthis is going great.
08:46psykosedoes it magically fix itself if you build 2.37?
08:47marcangood question
08:49___nick___ has joined #asahi-dev
09:03jannaukwin ctm night color fixes are merged and should be in the upcoming 5.27.3
09:03marcannow it's starting to look like someone unlocks an already unlocked lock
09:03marcangettext bug?
09:04marcanwait, not sure
09:08marcanokay no it's just asan
09:08marcanthe asan hooks end up with it doing a rlock -> (asan) -> wlock -> wunlock -> (out of asan) -> runlock sequence I think
09:08marcanwhich is illegal
09:13marcanand this all happens because it tries to dlsym... swift_demangle??
09:14marcan... which should only happen for things that look like Swift symbols??
09:15marcanwhich is things that start with _T
09:15marcan... and there is a _TLS_MODULE_BASE symbol
09:25skipwich has quit [Remote host closed the connection]
09:26skipwich has joined #asahi-dev
09:35___nick___ has quit []
09:35bisko has joined #asahi-dev
09:37___nick___ has joined #asahi-dev
09:37___nick___ has quit []
09:39___nick___ has joined #asahi-dev
10:06marcanjannau: QV4_MM_AGGRESSIVE_GC=1 *may* help repro the issue
10:06marcanthis is some messed up javascript type confusion thing or something...
10:13marcansigh, and now hardware watchpoints don't work properly?
10:13ChaosPrincess has quit [Quit: WeeChat 3.8]
10:14marcan(and software is too slow here...)
10:14ChaosPrincess has joined #asahi-dev
10:14Mary6 has quit []
10:15jannaumarcan: reproduced with QV4_MM_AGGRESSIVE_GC=1
10:15Mary6 has joined #asahi-dev
10:16marcanand the bad value is in the JS stack
10:16marcanI might just punt this to upstream
10:16marcanjannau: consistently?
10:21jannauyes, with startplasma-wayland from a tty, plasmashell crashes consistently, starts as expected without QV4_MM_AGGRESSIVE_GC=1
10:21marcanok yeah, repro'd on M1 Pro
10:21marcanthis is just broken everywhere depending on when GC runs
10:21marcanI'll file a KDE bug
10:21marcan(though it might be a Qt issue)
10:22marcanlet me see if I can repro this on a random Arch amd64 system
10:26nyilas has joined #asahi-dev
10:31jannaureproduced on fedora 37 on m2
10:31marcandoes not repro on amd64 :/
10:32marcanwhat bugs me here is this *could* be a random memory corruption issue
10:32marcanbut I just can't get the gdb watchpoint stuff to not suck
10:37marcanI think this is another gdb bug...
10:37marcanI think if you set a watchpoint and it gets hit by "dc zva", it hangs
10:59chadmed_ has joined #asahi-dev
10:59chadmed has quit [Read error: Connection reset by peer]
11:00marcanand "stp" also...
11:05jannauhave you tried lldb?
11:09marcanok, it's a JIT bug
11:09marcanI worked around it with a ridiculous number of automated breakpoints to skip over the problem areas
11:10marcanthe write of the bad value occurs in what pretty clearly is JIT code
11:10marcanlet me see if I can do this again in a less stupid way...
11:22marcanok, WTF, this JIT code is calling a function then immediately storing a caller-preserved register into the stack
11:22marcandid they get the ABI wrong?!
11:35marcanok, I think we are shipping QV4_FORCE_INTERPRETER=1
11:36marcanthat makes it work and it's not that slow
11:36gladiac is now known as Guest7557
11:36gladiac has joined #asahi-dev
11:37jannauABI allows to return more 8 byte directly so x1 could be part of the return value. the rules for function arguments apply to return types as well
11:39marcanjannau: the problem is x9 getting clobbered
11:39marcansee toot
11:40jannauyes, I overlooked that, the x1 looked unusual as well
11:42Guest7557 has quit [Ping timeout: 480 seconds]
12:11marcanjannau: the fedora repro, was that on qtdeclarative-5.15.8?
12:14marcan5.15.8-1.fc37 I assume
12:16marcanjannau: the really nasty kwin stuff goes away with plasmashell running, probably by accident
12:16marcanso we can prooobably punt on that as long as we at least disable blur by default because that is very broken
12:18zef has joined #asahi-dev
12:20jannaumarcan: yes, 5.15.8-1.fc37
12:23zef_ has quit [Ping timeout: 480 seconds]
12:25MajorBiscuit has joined #asahi-dev
12:40c10l57 has quit []
12:40c10l has joined #asahi-dev
13:35jannaumarcan: plasmashell is slow with QV4_MM_AGGRESSIVE_GC=1 but the patch seems to fix the segfaults
13:39tobhe_ has joined #asahi-dev
13:39kettenisall the world's x86 linux ;)
13:44tobhe has quit [Ping timeout: 480 seconds]
13:49marcanjannau: yes, it's supposed to be slow as heck with that
13:53marcanpushed another asahi-configs bump to disable blur by default in kwin
13:56jannaukettenis: u-boot patches sent, should I have cc-ed Tom?
14:03kettenisjannau: see it; does this diff actually fix booting linux or does it merely prevent a (very annoying) warning?
14:06kettenis(the apple pcie diff I mean)
14:21jannaukettenis: it avoids the warning and prevents the disabled root ports from being discovered. i.e. it restores the observed pcie setup to having the ports deleted from the dt
14:22jannauthe lmb max region change fixes booting, the issue was that there was no lmb region for the uefi binary
14:25kettenisThe 2023.4 release is around the corner, and already has the number of lmb regions bumped to 16
14:25kettenisI'm basically asking whether we should push for these changes to be included in 2023.4 or not
14:27jannau16 is should be ok for now, pcie is mostly cosmetical and releng has pcie disabled anyway. no need to rush any of the two patches
14:30gladiac is now known as Guest7571
14:30gladiac has joined #asahi-dev
14:34kettenisthanks; hopefully this will be merged into the next branch or merged onto master early in the 2023.7 cycle
14:36Guest7571 has quit [Ping timeout: 480 seconds]
14:51veeyee has joined #asahi-dev
15:13bisko has quit [Read error: Connection reset by peer]
15:15bisko has joined #asahi-dev
15:25leftas has joined #asahi-dev
15:29Leftas7 has quit [Ping timeout: 480 seconds]
15:53pranav has joined #asahi-dev
16:17bisko has quit [Read error: Connection reset by peer]
16:47amarioguy has joined #asahi-dev
17:27nyilas has quit [Remote host closed the connection]
17:28barrett has joined #asahi-dev
17:29nyilas has joined #asahi-dev
18:02pranav has quit [Quit: Connection closed for inactivity]
18:06MajorBiscuit has quit [Ping timeout: 480 seconds]
18:06barrett_ has joined #asahi-dev
18:07barrett has quit [Remote host closed the connection]
18:12amarioguymaz: heads up running those pmuv3 tests now
18:20amarioguy has quit [Remote host closed the connection]
18:22veeyee has quit [Quit: Leaving.]
18:28amarioguy has joined #asahi-dev
18:31mazamarioguy: great, thanks. let me know when you have something!
19:26winter has quit [Quit: The Lounge - https://thelounge.chat]
19:38veeyee has joined #asahi-dev
19:45gladiac has quit [Quit: k thx bye]
19:50amarioguymaz: ok got results for MDCR, trapped regs and effects on Apple PMU writes
19:50amarioguyanything else you need me to test?
20:14veeyee has quit [Quit: Leaving.]
20:14veeyee has joined #asahi-dev
20:15veeyee has quit []
20:28eiln has joined #asahi-dev
21:08___nick___ has quit [Ping timeout: 480 seconds]
21:36stickytoffee has joined #asahi-dev
21:39WindowPa- has joined #asahi-dev
21:39WindowPain has quit [Read error: Connection reset by peer]
22:09nyilas has quit [Remote host closed the connection]
22:30skipwich has quit [Quit: DISCONNECT]
22:31skipwich has joined #asahi-dev
22:40barrett_ has quit [Ping timeout: 480 seconds]
22:52barrett has joined #asahi-dev
23:54chadmed has joined #asahi-dev
23:59chadmed_ has quit [Ping timeout: 480 seconds]
…