#
…

2023-06-02

00:00jonny1996 has quit [Ping timeout: 480 seconds]
00:08Jenzaah has quit [Ping timeout: 480 seconds]
00:15nsklaus has quit [Ping timeout: 480 seconds]
00:17Jenzaah has joined #asahi-dev
00:25Jenzaah has quit [Ping timeout: 480 seconds]
00:34Jenzaah has joined #asahi-dev
00:37Jenzaah_ has joined #asahi-dev
00:37Jenzaah has quit [Read error: Connection reset by peer]
00:37brolin has joined #asahi-dev
00:40jonny1996 has joined #asahi-dev
00:46Jenzaah_ has quit [Ping timeout: 480 seconds]
00:52dylanchapell has quit [Remote host closed the connection]
00:53Jenzaah has joined #asahi-dev
00:55jonny1996_ has joined #asahi-dev
00:55jonny1996 has quit [Read error: Connection reset by peer]
01:02Jenzaah has quit [Ping timeout: 480 seconds]
01:10Jenzaah has joined #asahi-dev
01:13jonny1996_ has quit [Ping timeout: 480 seconds]
01:18Jenzaah has quit [Ping timeout: 480 seconds]
01:26Jenzaah has joined #asahi-dev
01:35Jenzaah has quit [Ping timeout: 480 seconds]
01:36brolin has quit [Ping timeout: 480 seconds]
01:37brolin has joined #asahi-dev
01:42Jenzaah has joined #asahi-dev
01:52Jenzaah has quit [Ping timeout: 480 seconds]
02:00gabuscus has quit []
02:01Jenzaah has joined #asahi-dev
02:04brolin has quit [Ping timeout: 480 seconds]
02:09Jenzaah has quit [Ping timeout: 480 seconds]
02:13brolin has joined #asahi-dev
02:17gabuscus has joined #asahi-dev
02:17Jenzaah has joined #asahi-dev
02:25Jenzaah has quit [Ping timeout: 480 seconds]
02:32jonny1996 has joined #asahi-dev
02:33Jenzaah has joined #asahi-dev
02:34jonny1996 has quit [Remote host closed the connection]
02:34baozich has joined #asahi-dev
02:37sid-linux has joined #asahi-dev
02:41Jenzaah has quit [Ping timeout: 480 seconds]
02:43sid-linux has quit [Quit: User exited]
02:49Jenzaah has joined #asahi-dev
02:57Jenzaah has quit [Ping timeout: 480 seconds]
03:05Jenzaah has joined #asahi-dev
03:12Jenzaah_ has joined #asahi-dev
03:13Jenzaah has quit [Ping timeout: 480 seconds]
03:14brolin has quit [Ping timeout: 480 seconds]
03:21Jenzaah_ has quit [Ping timeout: 480 seconds]
03:29Jenzaah has joined #asahi-dev
03:39Jenzaah has quit [Ping timeout: 480 seconds]
03:42Jenzaah has joined #asahi-dev
03:51Jenzaah has quit [Ping timeout: 480 seconds]
04:00Jenzaah has joined #asahi-dev
04:09Jenzaah has quit [Ping timeout: 480 seconds]
04:09psykose has quit [Remote host closed the connection]
04:10psykose has joined #asahi-dev
04:18Jenzaah has joined #asahi-dev
04:26Jenzaah has quit [Ping timeout: 480 seconds]
04:34Jenzaah has joined #asahi-dev
04:42Jenzaah has quit [Ping timeout: 480 seconds]
04:51Jenzaah has joined #asahi-dev
04:59Jenzaah has quit [Ping timeout: 480 seconds]
05:07Jenzaah has joined #asahi-dev
05:17Jenzaah has quit [Ping timeout: 480 seconds]
05:27Jenzaah has joined #asahi-dev
05:31pthariensflame has joined #asahi-dev
05:32pthariensflame has quit []
05:36Jenzaah has quit [Ping timeout: 480 seconds]
05:43Jenzaah has joined #asahi-dev
05:45Jenzaah_ has joined #asahi-dev
05:45Jenzaah has quit [Read error: Connection reset by peer]
05:53Jenzaah_ has quit [Ping timeout: 480 seconds]
05:58jannaustill a little baffled which difference between M2 macbook air and pro 13" could explain that
06:02Jenzaah has joined #asahi-dev
06:10Jenzaah has quit [Ping timeout: 480 seconds]
06:12Jenzaah has joined #asahi-dev
06:21Jenzaah has quit [Remote host closed the connection]
06:22Jenzaah has joined #asahi-dev
06:35Jenzaah has quit []
07:07MajorBiscuit has joined #asahi-dev
07:19MajorBiscuit has quit [Ping timeout: 480 seconds]
07:24MajorBiscuit has joined #asahi-dev
07:36marcaneric_engestrom: that looks pretty boring though? two extent trees that are oversized sounds like not really corruption, and I think the only reason it says still has errors is the missing /l+f?
07:40marcansid-linux: I've seen that fd leak thing before, it's not a kernel issue
07:41marcanthe dmabuf ones might be a mesa regression
07:42marcanthe pulseaudio one is definitely something else
07:42marcanlina, alyssa: ping for mesa side
07:43linaI think I used plain dup without CLOEXEC in a change recently... that's probably where those come from
07:44cylm has quit [Ping timeout: 480 seconds]
07:53marcaneric_engestrom: but yes the shutdown is the bigger question
07:53marcanI wonder if this has to do with cpuidle...
07:54marcanwhich machine is this?
07:54marcanalso did you update m1n1 and u-boot also?
07:56bps2 has joined #asahi-dev
08:06linasid-linux: Fixed the dmabuf leak ^^. All of this is leaking from plasmashell, so it depends on how you start konsole whether you see the leaks or not. I think the other fd leaks have always been there...
08:17sid-linux has joined #asahi-dev
08:20sid-linuxmarcan lina : thanks both. I installed gnome on Asahi and on wayland in a gnome terminal there are no funny fd leaks of any sort. Even running konsole within gone shows no fd leaks. So the guess that this is related to plasmashell sounds about right.
08:20sid-linux*within gnome
08:23sid-linuxlina: regarding adding CLOEXEC while doing dup -- would this still be the cause of the problem of seeing dmabuf in konsole (started in plasma shell) ? Because then I might have seen this in gnome also -- or was this because gnome-terminal was running via Xwayland ?
08:27_jannau_marcan: m2 macbook air for eric_engestrom and in #152. I see no problems on j493
08:28linasid-linux: The dmabuf leak would affect any app that doesn't explicitly close all FDs
08:29_jannau_cpuidle would be my first bet too but I don't see how it could differ between j413 and j493
08:38MajorBiscuit has quit [Ping timeout: 480 seconds]
08:38MajorBiscuit has joined #asahi-dev
08:49sid-linuxAnother issue unrelated to fd leaks. I noticed some syslog messages arriving _after_ wakeup, rather than before sleep.
08:49sid-linux====== The below message is the first one seen on wakeup from sleep ! ====================================================
08:49sid-linuxJun 02 08:04:23 somehostname kernel: Freezing user space processes
08:49sid-linuxJun 02 08:04:23 somehostname kernel: Freezing user space processes completed (elapsed 1.733 seconds)
08:49sid-linuxJun 02 08:04:23 somehostname kernel: OOM killer disabled.
08:49sid-linuxJun 02 08:04:23 somehostname kernel: Freezing remaining freezable tasks
08:49sid-linuxJun 02 08:04:23 somehostname kernel: Freezing remaining freezable tasks completed (elapsed 0.000 seconds)
08:49sid-linuxJun 02 08:04:23 somehostname kernel: printk: Suspending console(s) (use no_console_suspend to debug)
08:49sid-linuxJun 02 08:04:23 somehostname kernel: apple-dcp 231c00000.dcp: RTKit: syslog message: UPPipeDCP_H13P.cpp:7498: set_run_m>
08:49sid-linuxJun 02 08:04:23 somehostname kernel: apple-dcp 231c00000.dcp: RTKit: syslog message: UPPipeDCP_H13P.cpp:7403: virtual I>
08:49sid-linuxJun 02 08:04:23 somehostname kernel: apple-dcp 231c00000.dcp: dcp_poweroff() done
08:49sid-linuxJun 02 08:04:23 somehostname kernel: nvme-apple 27bcc0000.nvme: RTKit: syslog message: cmd.c:7890: NVMe shutdown start >
08:49sid-linuxJun 02 08:04:23 somehostname kernel: nvme-apple 27bcc0000.nvme: RTKit: syslog message: cmd.c:7912: seg->lba 0 saveCtx 1>
08:49sid-linuxJun 02 08:04:23 somehostname kernel: macsmc-rtkit 23e400000.smc: RTKit: syslog message: apComms.cpp:375: SMC HID Event:>
08:49sid-linux======== The ^^ messages should have been logged when the laptop was put to sleep earlier. But these are seen on wakeup ===
08:49sid-linuxJun 02 08:04:23 somehostname kernel: macsmc-hid macsmc-hid: Button wakeup
08:49sid-linuxJun 02 08:04:23 somehostname kernel: apple-pmgr-pwrstate 23b700000.power-management:power-controller@2c0: PS mca0: Fail>
08:49sid-linuxJun 02 08:04:23 somehostname kernel: nvme-apple 27bcc0000.nvme: RTKit: Initializing (protocol version 12)
08:49sid-linuxJun 02 08:04:23 somehostname kernel: nvme-apple 27bcc0000.nvme: RTKit: Unknown system endpoint: 0x0a
08:49sid-linuxJun 02 08:04:23 somehostname kernel: apple-dcp 231c00000.dcp: dcp_poweron() starting
08:51_jannau_sid-linux: please use a pastebin like paste.debian.net when pasting more than 3-5 lines
08:54_jannau_I wouldn't expect anything to be logged after 'printk: Suspending console(s)'
08:54sid-linux_jannau_: yes, that makes sense. will do next time
08:57sid-linuxAn even more important issue that I thought is that all messages _before_ 'printk: Suspending console(s)' should have been logged when putting the latop to sleep ! Yet they are seen on wakeup !
08:58sid-linuxso all the "Freezing use space processes" should have had a timestamp many hours older. Yet they are seen on wakeup which started at 08:04:23
08:59_jannau_how are you looking at syslog? if it involves an user space process it obviously can't displayed after 'Freezing user space processes'
08:59sid-linuxsorry maybe my terminology is wrong
08:59sid-linuxI'm doing `journalctl --since .... --until ....
09:01_jannau_are those timestamps not just the timestamps journald saw the message? please check the timestamps in the kernel with dmesg
09:06sid-linuxJun 02 03:27:26 somehostname systemd[1]: Starting System Suspend...
09:06sid-linuxJun 02 03:27:26 somehostname systemd-sleep[81749]: Entering sleep state 'suspend'...
09:06sid-linuxJun 02 03:27:26 somehostname kernel: PM: suspend entry (s2idle)
09:06sid-linuxJun 02 03:27:26 somehostname kernel: Filesystems sync: 0.001 seconds
09:06sid-linuxJun 02 08:04:23 somehostname kernel: Freezing user space processes
09:07sid-linuxNote that after 03:27 we have 08:04
09:08sid-linuxHow can I check dmesg without journalctl ? What do I look for in /var/log ?
09:08sid-linux(for an older boot)
09:12MajorBiscuit has quit [Ping timeout: 480 seconds]
09:13MajorBiscuit has joined #asahi-dev
09:15mpsdmesg command should work
09:20marcansid-linux: all that is normal for the reasons jannau mentioned
09:20marcanobviously when you put the machine to sleep logs arrive after it wakes up
09:20marcankernel timekeeping stops during sleep so you will not get anything useful out of the kernel timestamps
09:20marcanthere will be no gap
09:21eric_engestromjannau: indeed, looks like the same issue; I'll post there if I manage to get more info, but also starting tomorrow I'll be travelling for a week so I'll prefer downgrading over being able to reproduce the issue ^^
09:22eric_engestrommarcan: re- fsck, I don't know what's normal or not but your assessment sounds right
09:23marcanyeah, so far I'm leaning towards the FS/NVMe behaving as expected and the only real problem we have is the crashes/watchdog reboots
09:23r0ni has joined #asahi-dev
09:23eric_engestrommarcan: re: "what machine" -> m2 air like _jannau_ said, and yes m1n1 & uboot are up to date as well (I know better than doing a partial update on arch :P)
09:24marcanis this with AC connected or disconnected?
09:24marcanand what compile workload?
09:24eric_engestromagreed, I think we can consider the fs side "normal"
09:24eric_engestromac disconnected, and the workload was compiling mesa
09:24marcancool, easy enough to repro
09:24eric_engestromlet me try with ac connected
09:24marcanninja -jwhat?
09:25eric_engestromplain `ninja` but with `alias ninja='nice -n20 ninja'` in my .bashrc
09:27_jannau_I only tested with AC connected
09:30eric_engestromjust mesa compiled with ac connected, and it went fine
09:30marcanstarting to sound like a power delivery problem...
09:31marcanany idea what % your battery was at, roughly?
09:33eric_engestromabove 30% because I have a visual warning when it gets below that, and below 80% because I never let it charge higher than that
09:33eric_engestromI don't know more than that though
09:37marcan(you can enable the 80% charge threshold now btw, it works in sleep even)
09:40eric_engestrombut not powered off, right? I think powered off I've seen it continue to charge to 100%
09:41eric_engestrombtw I unplugged the power cord and re-ran the compilation, and it reset again pretty much right away, so it's definitely that
09:42marcanI haven't tested powered off, it may reset SMC yeah
09:42marcanok cool, I'll try to repro
09:42marcanis the reset instant, or does it hang for a while?
09:44eric_engestromI wasn't paying attention this time, but last time it hung for about 30 sec
09:44eric_engestromlet me do it again
09:44marcaninteresting
09:45marcanI'm wondering if this is something just hanging linux due to a power control interaction we don't understand (and then we watchdog timeout), or if it's outright SMC panicking (but then why would it hang?)
09:46marcanThough if SMC is in charge of voltage regulator management it could be in the hot path for powering up clusters and that could explain a hang (linux tries to wake up some cores and gets stuck)
09:47eric_engestromI'm running `dmesg -w` through ssh to try to see if there's anything; any other log I should watch?
09:47marcannah, that's good
09:48marcanif SMC panics hopefully you'd get a crashdump
09:48eric_engestromor perhaps a /sys knob to make it more verbose?
09:48marcanshould already be verbose on crashes
09:48sid-linux_ has joined #asahi-dev
09:48eric_engestrom(btw if most of the compilation gets caught by ccache, it doesn't reproduce)
09:49marcanyeah that makes sense
09:49marcanoh yeah, there's one more trick you can try
09:49marcanwhen the crash happens, immediately hold down the power button and go into the boot picker, and boot macOS
09:50marcansometimes it can catch panic logs from magic places
09:50sid-linux_marcan: eric_engestrom: it is a puzzling issue. regarding the reset. So I was plugged in. (Though my battery was a 100%. I was on a 80/75% limit).
09:50marcanthen you weren't plugged in
09:50sid-linux_I was able to get a successful compile of `cargo install alacritty`
09:50marcanwhen the 80/75% limit is engaged, if the battery is >=81%, it forces discharge
09:50marcanso it's as good as not being plugged in
09:50Major_Biscuit has joined #asahi-dev
09:51sid-linux_Then when I tried again with NO CHANGE absolutely to the setup/plug, it rebooted
09:51sid-linux_tried again = tried to compile again
09:51marcanmight be borderline for you
09:52MajorBiscuit has quit [Ping timeout: 480 seconds]
09:52sid-linux_My battery is at 98% now and I'm plugged in. So if the 80/75 limit is engaged -- does it not mean that the battery does not get charged and I get power directly from the power adapter?
09:53sid-linux_Why should the battery be discharging?
09:54sid-linux_I confused as to why discharge should be forced when I'm plugged in. I would assume the manufacturer would just bypass battery and supply directly to the circuits
09:54marcanthe charge limit behavior is to explicitly force discharge to bring down the battery to 80%
09:54sid-linux_(once battery is above the 80% threshold)
09:54marcanthat is what the firmware does
09:54sid-linux has quit [Ping timeout: 480 seconds]
09:56sid-linux_"Battery at 100% charging"
09:56sid-linux_trying a compile again
09:56marcanare you sure the charge limit is enabled?
09:56sid-linux_now it is at 100%/99%
09:56marcanit does not persist across reboots if you use KDE to set it
09:57sid-linux_Yes Ive noticed that
09:57sid-linux_How should I set it so that it is always persistent?
09:57eric_engestrom(i've been doign a full-mesa build with ccache cleared, it's like 75% through and still no issue...)
09:58marcanyou could create a udev rule or something like that
09:58_jannau_set it from a systemd single shot service
10:00sid-linux has joined #asahi-dev
10:00mpsmarcan: is the battery charge with latest asahi-wip controlled by charge_control_end_threshold and charge_control_start_threshold?
10:00marcanyes, but your only options are 80 and 100
10:01mpsaha, thanks
10:01marcaneric_engestrom: fwiw, modulo the peculiarities of what thresholds are supported, this is all a completely standard kernel sysfs interface and I would expect it to be supported by generic code, not anything apple-specific
10:02sid-linux__ has joined #asahi-dev
10:02eric_engestromargh, I got distracted by another chat window and I didn't pay attention when it crashed again, so I still don't know if it hung for a while before resetting... guess I'm doing it again xD
10:03sid-linux__marcan: eric_engestrom: boom ! I rebooted again. In all my time I've had ~10 reboots by now. Through some luck, I had a single successful compile of alacritty
10:03eric_engestromre: thresholds sure, but I think TLP doesn't enumerate /sys/class/power_supply/*, which is why it needs to be told about each one
10:04sid-linux__I was plugged in with battery this time with 0:07 minutes to fully being charged
10:04marcanit probably should, writing plugins for every single device seems silly :p
10:04sid-linux__sorry I meant plugged into power supply
10:05eric_engestrommarcan: agreed, it probably made sense in the early days of each vendor having its own custom interface, but not anymore
10:05dubiousness has quit [Remote host closed the connection]
10:05cds has quit [Remote host closed the connection]
10:05amada95 has quit [Remote host closed the connection]
10:05dk_ has quit [Remote host closed the connection]
10:05akspecs has quit [Remote host closed the connection]
10:05rosefromthedead has quit [Remote host closed the connection]
10:05okt has quit [Remote host closed the connection]
10:05probie has quit [Remote host closed the connection]
10:05tsujp has quit [Remote host closed the connection]
10:05d4ve has quit [Remote host closed the connection]
10:05coder_kalyan has quit [Remote host closed the connection]
10:05handlerug has quit [Remote host closed the connection]
10:05alethkit has quit [Remote host closed the connection]
10:05nsklaus has joined #asahi-dev
10:05sid-linux_ has quit [Ping timeout: 480 seconds]
10:05eric_engestrombtw I did have the dmesg output logged through ssh, and there was nothing, so if it did output something it didn't have time to make it through the network
10:06rosefromthedead has joined #asahi-dev
10:06cds has joined #asahi-dev
10:06d4ve has joined #asahi-dev
10:06alethkit has joined #asahi-dev
10:06handlerug has joined #asahi-dev
10:06tsujp has joined #asahi-dev
10:06dubiousness has joined #asahi-dev
10:06probie has joined #asahi-dev
10:06dk_ has joined #asahi-dev
10:06amada95 has joined #asahi-dev
10:06okt has joined #asahi-dev
10:06akspecs has joined #asahi-dev
10:06coder_kalyan has joined #asahi-dev
10:08sid-linux has quit [Ping timeout: 480 seconds]
10:09sid-linux__marcan: im wondering if it makes sense to downgrade back to 6.2 -- hopefully these reboots should be harmless to the nvme? I don't mind ext4 doing an fsck everytime but I'm worried if there could be some implications. In your experience with thousands of such reboots by now, I know you can't gaurantee anything, but how does apple hardware perform generally with these kind of hard reboots?
10:10chadmedive been using 6.3 for quite some time now (long before it was released) and have hard rebooted it many many many times and have not suffered any data corruption whatsoever
10:10sid-linux__chadmed: +1 -- thanks !
10:10eric_engestromok so it only hangs for about 5 sec before resetting
10:10chadmedthe hardware itself is fine, and our driver stack is as resilient as one could expect it to be so its really all down to the fs code and what youre doing when it hangs/you reboot
10:11marcansid-linux__: I've been hard rebooting machines dozens of times daily during testing
10:12eric_engestromchadmed: yeah I think we came to the same conclusion, the fs side is solid, the power side is the only issue
10:12sid-linux__yes that is my experience also, the hang is barely 4-5 seconds and then you hear the (designed to be pleasing) infuriating sound !
10:12chadmedi use btrfs too so according to the talking heads my laptop should have melted and all data on any drive ive ever used should be ciphered
10:14eric_engestrom(oh, and no dmesg output, even locally)
10:15marcaneric_engestrom: can you paste me `grep -a . /proc/device-tree/chosen/asahi*` somewhere?
10:16_jannau_the power issue might be m2 macbook air specific. I ran 6.3 based asahi-wip for weeks on j314c without issues
10:17marcanI just survived two mesa compiles on j413 on battery...
10:17marcaneric_engestrom: maybe also your meson configuration?
10:18marcanit just crashed
10:20marcanI would say it's about 15 seconds before the reboot
10:22eric_engestrombtw sid-linux__ if you boot on macos, press mute and shutdown, you won't have that power-on sounds anymore :)
10:23_jannau_or disable boot chime in system settings
10:23marcan24 seconds this time
10:23marcanhow often does the kernel kick the watchdog? could still be that
10:24_jannau_watchdog timeout is 30s, not sure how often it is fed
10:25_jannau_seems to update a timeout / 2
10:26marcanyeah, so then it's still consistent with wdt timing out
10:28marcanwatchdog_stop_on_unregister doesn't seem to be working, I unbound the device and it rebooted me
10:29_jannau_I'd start looking if that works with WDOG_HW_RUNNING
10:33jonny1996 has joined #asahi-dev
10:33marcancrashed again while plugged in, I don't think it's power delivery
10:33marcanI think we are deadlocking on something, probably related to cpuidle
10:33marcanand then wdt reboots us
10:43marcanrepro'd under the hypervisor... and the hypervisor died too :(
10:43marcanno reboot though, it just hangs
10:44marcanso it's definitely the watchdog, and it's not SMC thinking the world is on fire
10:47bps2 has quit [Ping timeout: 480 seconds]
10:48marcanReturning from exception
10:48marcan^CTTY> HV: Failed to rendezvous, missing CPUs: 0x76
10:48marcanheey a message
10:49marcanokay so definitely something going wrong there
10:56sid-linux has joined #asahi-dev
10:58sid-linux has quit []
11:00sid-linux__ has quit [Ping timeout: 480 seconds]
11:18bps2 has joined #asahi-dev
11:27nickchan has joined #asahi-dev
11:31MajorBiscuit has joined #asahi-dev
11:32jannauburned through from 80% to 30% battery with kernel and mesa compiles without reboot on j493
11:32Major_Biscuit has quit [Ping timeout: 480 seconds]
11:35marcanstarting to wonder if it's thermal
11:41bps3 has joined #asahi-dev
11:43baozich has quit [Ping timeout: 480 seconds]
11:47bps2 has quit [Ping timeout: 480 seconds]
11:55sid-linux has joined #asahi-dev
11:56sid-linuxthermal issues could be a possibility... For instance `cargo install alacritty --jobs=1` did not cause a reboot for me.
11:57sid-linuxbut `cargo install alacritty` almost always does
11:57sid-linuxload on cpu is high -> thermal goes off the charts ?? -> cpu reboots
12:00marcanI managed to get a debug shell in the broken state, but I don't know what's up
12:00marcanit's like a bunch of CPUs are just dead
12:00marcanon both clusters
12:01marcanmaybe I should go back to macos and see if there are any magic pokes we missed...
12:02marcanI should also just try it under newer firmware, because it's entirely possible this is an iBoot bug. we're running the first version M2 shipped with after all.
12:02marcananyway, I recommend playing with /sys/devices/system/cpu/cpu?/cpuidle/state1/disable
12:02marcanI bet writing 1 to that on some/all CPUs will fix it
12:02marcanif you just want it to work
12:03marcanI'll have to take a deeper look at this later, too tired to do to go hunting for it
12:04sid-linuxsorry -- maybe I missed the earlier part of is /sys/devices/... discussion -- setting this in advance of running the cpu intensive compile?
12:05jannauI'd set it in general on at least one core and then look if other cores get stuck in cpuidle
12:05sid-linuxim a bit concerned about the whole dmabuf / pulseaudio / various pipes etc leaking into all programs derived from plasma shell. Sounds like its something unrelated to 6.3 and the problem has always existed?
12:05sid-linux(different topic)
12:07sid-linuxThe alternative is to use gnome of course which does not leak any fds. How would I set the battery 80/75% from the command line manually? Any link I could refer to so that you don't have to explain it to me...?
12:11j`eysid-linux: is the leaked fds actually causing you issues?
12:12marcanthe dmabuf leaks are fixed and could cause a memory leak. the pulseaudio/inotify leaks are probably harmless.
12:13sid-linuxj`ey: It feels wrong for every program derived from plasma shell to have the ability to write to pulse audio, dma buff, some kind of pipes, memfds
12:13marcanthe pulseaudio leak is probably the pulseaudio lib's fault
12:14sid-linuxprobably harmless but everything is harmless until it becomes a problem
12:14marcanif you're so concerned feel free to track it down and fix it ;)
12:14sid-linuxgood to hear the dmabuf is fixed
12:15jonny1996 has quit [Read error: Connection reset by peer]
12:15sid-linuxmarcan: true. The itch I scratched recently was the PMU not working while investigating why rr was not working -- thanks to maz's (who did the real work) that is now fixed
12:15jonny1996 has joined #asahi-dev
12:16sid-linuxI wonder if I have the desire to go down hunting for fds in the kde codebase. I'm feeling more inclined to just ignore the problem or use gnome if I can figure out the 80/75% thing from the command line
12:16baozich has joined #asahi-dev
12:17sid-linuxmarcan: oh and thanks for the kde bug link. reassuring
12:17jjanzic has quit [Ping timeout: 480 seconds]
12:18sid-linuxmarcan: SHGTFM (Should have Googled that for Myself)
12:20marcanseems the PA one is already fixed
12:21baozich has quit [Remote host closed the connection]
12:22marcanthere hasn't been a pulse release since then though
12:22marcanso *shrug*
12:22baozich has joined #asahi-dev
12:23sid-linuxj`ey: actually a correction -- i remember that an fd leak did cause a problem -- which I forgot. A CI test in the rr-debugger was failing. It was assuming a pristine environment of stdin/stdout/sderr. It was expecting fd 50 or something to exist but that was taken up by konsole
12:23marcannot sure who owns the inotify one
12:24sid-linuxwhich is what caused me to look at the /proc/<pid>/fd while doing a gdb of the test. To my shock when I saw all the un-CLOEXCed fds there :-) !
12:24jonny1996 has quit [Read error: Connection reset by peer]
12:25sid-linux(while dup-ing fd 49 -- which should have given fd 50)
12:25marcanexpecting a clean fd environment without explicitly cleaning up seems fragile to me, tbh
12:25marcanfd leaks are nothing new
12:26sid-linuxin rr's case we can do that
12:26sid-linuxsorry we "can't"
12:26marcanif it's a CI issue I'm sure you can wrap the CI scripts in an fd cleaner ;)
12:26sid-linuxhmm... I take that back. We can
12:27sid-linuxin the record phase of the rr debugger, we simply close all the fds to create a clean environement. During replay the same thing we will be OK
12:27sid-linuxbecause it will go through the same steps
12:28sid-linuxsorry my bad. It is possible to clean up. The tricky thing is that the clean also is recorded !
12:28sid-linuxAnd the same cleanup will be replayed
12:29jonny1996 has joined #asahi-dev
12:32sid-linux(recording of any program tree by rr happens from inception to death. never an arbitrary point.)
12:38jonny1996 has quit [Remote host closed the connection]
12:38jonny1996 has joined #asahi-dev
12:47jonny1996 has quit [Ping timeout: 480 seconds]
12:54sid-linux has quit [Quit: User exited]
13:23chadmedsid-linux: re pulse bear in mind that once speakers support is ready to go we will only be officially supporting pipewire going forward
13:24chadmedof course you can still use PA but you wont get any of the DSP required to make the speakers work properly and this will probably result in the safety daemon kneecapping your volume semi-frequently
13:41sid-linux has joined #asahi-dev
13:43sid-linuxeric_engestrom: thanks for the link to TLP earlier !
13:44sid-linux has quit []
13:51brolin has joined #asahi-dev
13:53nehsou^ has quit [Remote host closed the connection]
14:00sid-linux has joined #asahi-dev
14:01sid-linuxlina : can you tell me which repo+branch you have pushed out the dmabuf fix? I want to try to compile that locally ...
14:02sid-linux has quit []
14:02Major_Biscuit has joined #asahi-dev
14:04MajorBiscuit has quit [Ping timeout: 480 seconds]
14:09MajorBiscuit has joined #asahi-dev
14:10nsklaus /engestrom
14:11Major_Biscuit has quit [Ping timeout: 480 seconds]
14:34marcanchadmed: this is libpulse which is what everything uses with pipewire
14:34marcanit's the client lib
14:47veloek has quit [Remote host closed the connection]
14:50dylanchapell has joined #asahi-dev
14:52ellyq has joined #asahi-dev
15:01MajorBiscuit has quit [Ping timeout: 480 seconds]
15:02MajorBiscuit has joined #asahi-dev
15:12MajorBiscuit has quit [Ping timeout: 480 seconds]
15:15MajorBiscuit has joined #asahi-dev
15:24nela has quit [Quit: bye!]
15:25nela has joined #asahi-dev
15:26billak has joined #asahi-dev
15:35billak has quit [Quit: Konversation terminated!]
15:38MajorBiscuit has quit [Ping timeout: 480 seconds]
15:42brolin has quit [Ping timeout: 480 seconds]
16:00sid-linux has joined #asahi-dev
16:02sid-linux has quit []
16:08baozich has quit [Ping timeout: 480 seconds]
16:13cylm has joined #asahi-dev
16:18hightower4 has joined #asahi-dev
16:18dylanchapell has quit [Quit: Page closed]
16:20baozich has joined #asahi-dev
16:24hightower3 has quit [Ping timeout: 480 seconds]
16:31brolin has joined #asahi-dev
16:37veloek has joined #asahi-dev
16:38baozich has quit [Ping timeout: 480 seconds]
16:52MajorBiscuit has joined #asahi-dev
16:53MajorBiscuit has quit []
16:54hightower4 has quit [Ping timeout: 480 seconds]
16:54bps3 has quit [Ping timeout: 480 seconds]
17:27Guest1832 has quit [Quit: Bridge terminating on SIGTERM]
17:27rhysmdnz has quit [Quit: Bridge terminating on SIGTERM]
17:59brolin has quit [Ping timeout: 480 seconds]
18:07brolin has joined #asahi-dev
18:37brolin has quit [Ping timeout: 480 seconds]
18:44brolin has joined #asahi-dev
19:37brolin has quit [Ping timeout: 480 seconds]
19:38brolin has joined #asahi-dev
20:05maximbaz has quit [Quit: bye]
20:07maximbaz has joined #asahi-dev
20:24StupidYui has joined #asahi-dev
20:27brolin has quit [Ping timeout: 480 seconds]
20:37brolin has joined #asahi-dev
20:40Dementor has quit [Ping timeout: 480 seconds]
20:44Dementor has joined #asahi-dev
20:50Dementor3 has joined #asahi-dev
20:56Dementor has quit [Ping timeout: 480 seconds]
20:56Dementor3 is now known as Dementor
20:58roxfan2 has joined #asahi-dev
21:03roxfan has quit [Ping timeout: 480 seconds]
21:03brolin has quit [Quit: Lost terminal]
23:10chadmed_ has joined #asahi-dev
23:10chadmed has quit [Read error: Connection reset by peer]
23:57DVD has joined #asahi-dev
23:57codingkoopa3 has quit [Ping timeout: 480 seconds]
…