galliumos-distro
galliumos-distro copied to clipboard
GalliumOS
Trackpad pointer freezes on Samsung Chromebook 3 (CELES) XE500C13, two different units
I have installed GalliumOS 2.1 with the beta full ROM firmware (via @MattDevo) and the 4.14.6 galliumOS beta kernel on two separate Samsung Chromebook 3 systems (CELES, XE500C13).
I have experienced the same problem on both machines: everything is fine at first, but eventually the pointer will freeze.
With the first machine I assumed it was because I did a clumsy job removing the write protect screw from behind the motherboard and messed something up. However, on the second machine everything went very smoothly (antistatic strap, removed the battery as step one, etc) and it seems unlikely I'd mess it up the same way twice.
At one point I wrote a script to unload and reload the atmel_mxt_ts driver; I'd alt-tab over to a terminal window and run this on each freeze. It worked, but for varying periods of time, and would always freeze again.
I don't see anything interesting in /var/log/syslog when the pointer freezes. However, at boot I see this:
Jan 27 12:33:38 iloverocks kernel: [ 11.641900] atmel_mxt_ts i2c-ATML0000:00: Direct firmware load for maxtouch.cfg failed with error -2 Jan 27 12:33:38 iloverocks kernel: [ 11.642424] atmel_mxt_ts i2c-ATML0000:00: Resetting device
There appear to be official drivers here:
https://github.com/atmel-maxtouch/linux
I'm not sure if the latest version of this code is included in the kernel or not.
Any ideas?
Thanks!
The 4.14.14 kernel makes this issue much less frequent, but it still occurs. Seems liked to high load, but not necessarily.
Confirming that I get the same issue on 4.14.14 and the full ROM firmware. I also notice stuttering video and audio during this time that the trackpad pointer freezes. Possibly related is that I notice that sometimes keyboard input (characters displaying on screen) sometimes lags around the time that the trackpad pointer freezes.
I had been running the RW_LEGACY firmware for about a week and never had any of these issues.
For those needing a temporary fix, something like:
#/usr/bin/env bash
sudo modprobe -r atmel_mxt_ts
sudo modprobe atmel_mxt_ts
in an executable script will unload and re-load the trackpad driver which seems to temporarily fix the issue as @boutell noticed in his first message.
Yes, I noticed the stuttering yesterday.
On Feb 18, 2018 4:58 PM, "jeff tecca" [email protected] wrote:
Confirming that I get the same issue on 4.14.14 and the full ROM firmware. I also notice stuttering video and audio during this time that the trackpad pointer freezes.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-366551586, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fesEKLhh1SXOhA15gFntRXtTwhipks5tWJzzgaJpZM4Rvb5Q .
idk if its a coincidence, but, i've created a 2 GB swap file on a micro sd and mysteriously the problem with trackpad was fixed!
Interesting! Just a swap file, or a swap partition? I could do either, but a swap file is lazier (:
On Wed, Feb 21, 2018 at 11:36 PM, Evandro Begati [email protected] wrote:
idk if is a coincidence, but, i've created an 2 GB swap file on a micro sd and mysteriously the problem with trackpad was fixed!
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-367565059, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fSUDHeIZbFYy1W27yci36sn-msZYks5tXO7lgaJpZM4Rvb5Q .
--
THOMAS BOUTELL, CHIEF SOFTWARE ARCHITECT P'UNK AVENUE | (215) 755-1330 | punkave.com
give a try on swap file, if it works, try creating a swap partition! I’ve tested both.
I do already have a swap partition, but I'll have to check how large. Yours is really large.
On Thu, Feb 22, 2018 at 10:33 AM, Evandro Begati [email protected] wrote:
give a try on swap file, if it works, try creating a swap partition! I’ve tested both.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-367718710, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fdHIGKEi37HIKnIDdkTfgtnc2rXtks5tXYjkgaJpZM4Rvb5Q .
--
THOMAS BOUTELL, CHIEF SOFTWARE ARCHITECT P'UNK AVENUE | (215) 755-1330 | punkave.com
Not sure if this is related but I'm running RW_LEGACY with testing enabled and have recently started running into issues with the trackpad pointer freezing.
At boot, get atmel_mxt_ts i2c-ATML0000:01: Direct firmware load for maxtouch.cfg failed with error -2 too.
But I'm also seeing a lot of errors in /var/log/syslog:
Feb 21 18:42:33 fluorine kernel: [28025.196806] pipe A vblank wait timed out
Feb 21 18:42:33 fluorine kernel: [28025.196882] ------------[ cut here ]------------
Feb 21 18:42:33 fluorine kernel: [28025.196941] WARNING: CPU: 0 PID: 26987 at drivers/gpu/drm/i915/intel_display.c:12195 intel_atomic_commit_tail+0xe5a/0xe80 [i915]
Feb 21 18:42:33 fluorine kernel: [28025.196943] Modules linked in: iwlmvm mac80211 iwlwifi cfg80211 xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack libcrc32c ipt_REJECT nf_reject_ipv4 rfcomm ccm xt_tcpudp bridge stp llc ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter ip_tables x_tables msr bnep uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 btusb videobuf2_core btrtl btbcm btintel videodev lz4 bluetooth lz4_compress media ecdh_generic zram binfmt_misc arc4 snd_soc_sst_cht_bsw_rt5645 intel_rapl intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel joydev pcbc snd_hda_codec_hdmi snd_hda_intel snd_hda_codec kvm_intel snd_intel_sst_acpi aesni_intel snd_intel_sst_core snd_soc_sst_atom_hifi2_platform
Feb 21 18:42:33 fluorine kernel: [28025.197003] snd_soc_rt5645 snd_hwdep snd_hda_core aes_x86_64 snd_soc_rl6231 crypto_simd snd_soc_acpi snd_soc_acpi_intel_match glue_helper cryptd snd_seq_midi snd_soc_core snd_seq_midi_event kvm snd_compress snd_rawmidi snd_pcm_dmaengine snd_pcm_oss snd_mixer_oss snd_seq snd_pcm lpc_ich irqbypass snd_seq_device mfd_core snd_timer processor_thermal_device shpchp intel_soc_dts_iosf snd int3403_thermal int340x_thermal_zone int3400_thermal acpi_thermal_rel soundcore chromeos_pstore atmel_mxt_ts 8250_dw mac_hid autofs4 btrfs xor zstd_decompress zstd_compress xxhash raid6_pq dm_mirror dm_region_hash dm_log i915 mmc_block video i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm sdhci_acpi sdhci [last unloaded: cfg80211]
Feb 21 18:42:33 fluorine kernel: [28025.197061] CPU: 0 PID: 26987 Comm: kworker/u4:0 Tainted: G W 4.14.14-galliumos #1
Feb 21 18:42:33 fluorine kernel: [28025.197062] Hardware name: GOOGLE Celes, BIOS 01/19/2017
Feb 21 18:42:33 fluorine kernel: [28025.197101] Workqueue: events_unbound intel_atomic_commit_work [i915]
Feb 21 18:42:33 fluorine kernel: [28025.197103] task: ffff8edc0e57ea00 task.stack: ffffa71a03c40000
Feb 21 18:42:33 fluorine kernel: [28025.197141] RIP: 0010:intel_atomic_commit_tail+0xe5a/0xe80 [i915]
Feb 21 18:42:33 fluorine kernel: [28025.197143] RSP: 0018:ffffa71a03c43dd0 EFLAGS: 00010286
Feb 21 18:42:33 fluorine kernel: [28025.197146] RAX: 000000000000001c RBX: ffff8edc74a48000 RCX: 0000000000000001
Feb 21 18:42:33 fluorine kernel: [28025.197147] RDX: 0000000080000001 RSI: ffffffff90eb9282 RDI: 00000000ffffffff
Feb 21 18:42:33 fluorine kernel: [28025.197149] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000cb7
Feb 21 18:42:33 fluorine kernel: [28025.197150] R10: ffffa71a03c43dd0 R11: 0000000000000cb7 R12: 0000000000000000
Feb 21 18:42:33 fluorine kernel: [28025.197152] R13: 0000000000000000 R14: ffff8edc748f7000 R15: 0000000000000001
Feb 21 18:42:33 fluorine kernel: [28025.197154] FS: 0000000000000000(0000) GS:ffff8edc7fc00000(0000) knlGS:0000000000000000
Feb 21 18:42:33 fluorine kernel: [28025.197156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Feb 21 18:42:33 fluorine kernel: [28025.197157] CR2: 00007f538a7cb000 CR3: 0000000153532000 CR4: 00000000001006f0
Feb 21 18:42:33 fluorine kernel: [28025.197159] Call Trace:
Feb 21 18:42:33 fluorine kernel: [28025.197175] ? finish_task_switch+0x80/0x220
Feb 21 18:42:33 fluorine kernel: [28025.197180] ? wait_woken+0x80/0x80
Feb 21 18:42:33 fluorine kernel: [28025.197184] process_one_work+0x148/0x420
Feb 21 18:42:33 fluorine kernel: [28025.197188] worker_thread+0x47/0x420
Feb 21 18:42:33 fluorine kernel: [28025.197192] kthread+0xfc/0x130
Feb 21 18:42:33 fluorine kernel: [28025.197195] ? process_one_work+0x420/0x420
Feb 21 18:42:33 fluorine kernel: [28025.197197] ? kthread_create_on_node+0x40/0x40
Feb 21 18:42:33 fluorine kernel: [28025.197201] ret_from_fork+0x32/0x40
Feb 21 18:42:33 fluorine kernel: [28025.197205] Code: 24 50 4c 89 04 24 48 83 c7 08 e8 32 7b e1 cf 4c 8b 04 24 4d 85 c0 0f 85 08 fe ff ff 8d 75 41 48 c7 c7 40 fa 31 c0 e8 21 06 e3 cf <0f> ff e9 f2 fd ff ff 8d 70 41 48 c7 c7 10 fa 31 c0 e8 0b 06 e3
Feb 21 18:42:33 fluorine kernel: [28025.197257] ---[ end trace 826d9a59e8414bce ]---
Looping over an over. This eventually leads to a different error looping:
60111 Feb 21 09:52:47 fluorine kernel: [29027.550802] atmel_mxt_ts i2c-ATML0000:01: __mxt_read_reg: i2c transfer failed (-110)
60112 Feb 21 09:52:47 fluorine kernel: [29027.550811] atmel_mxt_ts i2c-ATML0000:01: Failed to read T44 and T5 (-110)
Interesting. My syslog is quiet when this happens (at least until I start messing with it by restarting the driver).
Have you tried a large swap file?
On Thu, Feb 22, 2018 at 2:52 PM, Russell Parker [email protected] wrote:
Not sure if this is related but I'm running RW_LEGACY with testing enabled and have recently started running into issues with the trackpad pointer freezing.
At boot, get atmel_mxt_ts i2c-ATML0000:01: Direct firmware load for maxtouch.cfg failed with error -2 too.
But I'm also seeing a lot of errors in /var/log/syslog:
Feb 21 18:42:33 fluorine kernel: [28025.196806] pipe A vblank wait timed out Feb 21 18:42:33 fluorine kernel: [28025.196882] ------------[ cut here ]------------ Feb 21 18:42:33 fluorine kernel: [28025.196941] WARNING: CPU: 0 PID: 26987 at drivers/gpu/drm/i915/intel_display.c:12195 intel_atomic_commit_tail+0xe5a/0xe80 [i915] Feb 21 18:42:33 fluorine kernel: [28025.196943] Modules linked in: iwlmvm mac80211 iwlwifi cfg80211 xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack libcrc32c ipt_REJECT nf_reject_ipv4 rfcomm ccm xt_tcpudp bridge stp llc ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter ip_tables x_tables msr bnep uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 btusb videobuf2_core btrtl btbcm btintel videodev lz4 bluetooth lz4_compress media ecdh_generic zram binfmt_misc arc4 snd_soc_sst_cht_bsw_rt5645 intel_rapl intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel joydev pcbc snd_hda_codec_hdmi snd_hda_intel snd_hda_codec kvm_intel snd_intel_sst_acpi aesni_intel snd_intel_sst_core snd_soc_sst_atom_hifi2_platform Feb 21 18:42:33 fluorine kernel: [28025.197003] snd_soc_rt5645 snd_hwdep snd_hda_core aes_x86_64 snd_soc_rl6231 crypto_simd snd_soc_acpi snd_soc_acpi_intel_match glue_helper cryptd snd_seq_midi snd_soc_core snd_seq_midi_event kvm snd_compress snd_rawmidi snd_pcm_dmaengine snd_pcm_oss snd_mixer_oss snd_seq snd_pcm lpc_ich irqbypass snd_seq_device mfd_core snd_timer processor_thermal_device shpchp intel_soc_dts_iosf snd int3403_thermal int340x_thermal_zone int3400_thermal acpi_thermal_rel soundcore chromeos_pstore atmel_mxt_ts 8250_dw mac_hid autofs4 btrfs xor zstd_decompress zstd_compress xxhash raid6_pq dm_mirror dm_region_hash dm_log i915 mmc_block video i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm sdhci_acpi sdhci [last unloaded: cfg80211] Feb 21 18:42:33 fluorine kernel: [28025.197061] CPU: 0 PID: 26987 Comm: kworker/u4:0 Tainted: G W 4.14.14-galliumos #1 Feb 21 18:42:33 fluorine kernel: [28025.197062] Hardware name: GOOGLE Celes, BIOS 01/19/2017 Feb 21 18:42:33 fluorine kernel: [28025.197101] Workqueue: events_unbound intel_atomic_commit_work [i915] Feb 21 18:42:33 fluorine kernel: [28025.197103] task: ffff8edc0e57ea00 task.stack: ffffa71a03c40000 Feb 21 18:42:33 fluorine kernel: [28025.197141] RIP: 0010:intel_atomic_commit_tail+0xe5a/0xe80 [i915] Feb 21 18:42:33 fluorine kernel: [28025.197143] RSP: 0018:ffffa71a03c43dd0 EFLAGS: 00010286 Feb 21 18:42:33 fluorine kernel: [28025.197146] RAX: 000000000000001c RBX: ffff8edc74a48000 RCX: 0000000000000001 Feb 21 18:42:33 fluorine kernel: [28025.197147] RDX: 0000000080000001 RSI: ffffffff90eb9282 RDI: 00000000ffffffff Feb 21 18:42:33 fluorine kernel: [28025.197149] RBP: 0000000000000000 R08: 0000000000000001 R09: 0000000000000cb7 Feb 21 18:42:33 fluorine kernel: [28025.197150] R10: ffffa71a03c43dd0 R11: 0000000000000cb7 R12: 0000000000000000 Feb 21 18:42:33 fluorine kernel: [28025.197152] R13: 0000000000000000 R14: ffff8edc748f7000 R15: 0000000000000001 Feb 21 18:42:33 fluorine kernel: [28025.197154] FS: 0000000000000000(0000) GS:ffff8edc7fc00000(0000) knlGS:0000000000000000 Feb 21 18:42:33 fluorine kernel: [28025.197156] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 Feb 21 18:42:33 fluorine kernel: [28025.197157] CR2: 00007f538a7cb000 CR3: 0000000153532000 CR4: 00000000001006f0 Feb 21 18:42:33 fluorine kernel: [28025.197159] Call Trace: Feb 21 18:42:33 fluorine kernel: [28025.197175] ? finish_task_switch+0x80/0x220 Feb 21 18:42:33 fluorine kernel: [28025.197180] ? wait_woken+0x80/0x80 Feb 21 18:42:33 fluorine kernel: [28025.197184] process_one_work+0x148/0x420 Feb 21 18:42:33 fluorine kernel: [28025.197188] worker_thread+0x47/0x420 Feb 21 18:42:33 fluorine kernel: [28025.197192] kthread+0xfc/0x130 Feb 21 18:42:33 fluorine kernel: [28025.197195] ? process_one_work+0x420/0x420 Feb 21 18:42:33 fluorine kernel: [28025.197197] ? kthread_create_on_node+0x40/0x40 Feb 21 18:42:33 fluorine kernel: [28025.197201] ret_from_fork+0x32/0x40 Feb 21 18:42:33 fluorine kernel: [28025.197205] Code: 24 50 4c 89 04 24 48 83 c7 08 e8 32 7b e1 cf 4c 8b 04 24 4d 85 c0 0f 85 08 fe ff ff 8d 75 41 48 c7 c7 40 fa 31 c0 e8 21 06 e3 cf <0f> ff e9 f2 fd ff ff 8d 70 41 48 c7 c7 10 fa 31 c0 e8 0b 06 e3 Feb 21 18:42:33 fluorine kernel: [28025.197257] ---[ end trace 826d9a59e8414bce ]---
Looping over an over. This eventually leads to a different error looping:
60111 Feb 21 09:52:47 fluorine kernel: [29027.550802] atmel_mxt_ts i2c-ATML0000:01: __mxt_read_reg: i2c transfer failed (-110) 60112 Feb 21 09:52:47 fluorine kernel: [29027.550811] atmel_mxt_ts i2c-ATML0000:01: Failed to read T44 and T5 (-110)
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-367800541, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9feaxZvARjj_saAPZd0B9cgGDjuM3ks5tXcV-gaJpZM4Rvb5Q .
--
THOMAS BOUTELL, CHIEF SOFTWARE ARCHITECT P'UNK AVENUE | (215) 755-1330 | punkave.com
Hey there @boutell, have you tested a large swap? got any success? My trackpad has not given more trouble since then. :)
I will as soon as I get home.
On Fri, Feb 23, 2018 at 12:17 PM, Evandro Begati [email protected] wrote:
Hey there @boutell https://github.com/boutell, have you tested a large swap? got any success? My trackpad has not given more trouble since then. :)
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-368075825, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fcR62qNFYx629LUGEl7lJm2WOoTLks5tXvKWgaJpZM4Rvb5Q .
--
THOMAS BOUTELL, CHIEF SOFTWARE ARCHITECT P'UNK AVENUE | (215) 755-1330 | punkave.com
I have installed the swapfile this morning. I am hopeful...
I noticed that GalliumOS also comes with zram enabled as a swapfile of sorts. zram is interesting - it's a RAM-based compressed block device, so using it as swap is like having somewhat more memory than you really have without the performance impact of touching the disk. But, if it had a bug or some kind of timing issue that triggered a bug in the trackpad...
zram is higher priority than the swapfile. So I would assume the swapfile would never be touched unless zram were exhausted first. That could definitely happen though in the high load situations we've been talking about.
OK, I did some tests this morning with a 2GB swap file in place.
I did my best to mess it up with non-hardware things that have before: Facebooking (to the point where typing slowed down), YouTube videos continuously playing. No problems in about 45 minutes of use.
Then I plugged into a projector via HDMI. No problems while connected. Tried different display settings.
Then I unplugged the HDMI cable. Boom, pointer freeze.
I then checked syslog. In syslog I do see the following, however this started appearing periodically as soon as the HDMI cable was plugged in and didn't wait until later when I unplugged it, so I don't know if it's directly related:
pipe C vblank wait timed out
Followed by a stack trace etc. which I can attach if desired.
Those stopped happening after the HDMI cable was disconnected.
I looked at /proc/swaps a few times and while the use of /dev/zram0 climbed a little, I never noticed use of the swap file on disk.
I will keep testing because a 45 minute test is not really enough, but if I had only this to go on, I would say that the trackpad is a very "high-need" device that gets lost if the OS fails to pay attention to it even once, and the swap file ameliorates the problem by eliminating some scenarios where that happens, but the driver probably really needs to be at a realtime priority.
The HDMI disconnect situation is one where the kernel is especially likely to be thoroughly distracted by the needs of its other children for a moment, and can't pay attention to the trackpad at a critical juncture, and things go off the rails.
I experimented some more. I was not able to freeze the pointer simply by disconnecting HDMI, it hasn't happened again yet.
However, I did notice that I can get a similar vblank error message just by going fullscreen in youtube without a second monitor, at least half of the time:
pipe A vblank wait timed out
I've seen freezes associated with fullscreen in the past, so although it didn't freeze on this pass it does make me suspect I didn't imagine the connection.
I have reproduced the freeze in a "vblank wait timed out" situation (full screen apps, MAME, etc) with no external monitor. So far, I have not reproduced it in the absence of that message since adding swap.
@begati since adding your monster swapfile, have you done much with full screen applications and/or HDMI output? It doesn't happen 100% of the time, but so far those are the only circumstances in which I have experienced a pointer freeze since adding my own swapfile.
Obviously it would be good to fix those situations too...
Of course, with a low-frequency bug like this (since the 4.14.14 kernel), it's hard to be 100% sure I'll never see it again under other conditions.
My latest attempt to resolve this: I disabled zram (Gallium's default swap drive, a RAMdisk... which sounds crazy... except it's a compressed RAMdisk... so it's pretty sane). Pet theory is that there's a condition somewhere in which if the system swaps to zram it can't do something fast enough to suit the keyboard/trackpad hardware and things go south. Obviously swapping to disk should not be better, except that it's been around forever so it might say "hang on just a minit" in a better-debugged fashion.
Hey, it's an idea. I'll let you know.
No pointer freezes so far, but a youtube video fullscreen experience led me to more "vblank timed out" errors in syslog, after which I started to get some key duplication and video/audio got very choppy.
Others suggest the vblank messages can be associated with memory pressure:
https://bugs.freedesktop.org/show_bug.cgi?id=102667
Oooh! This looks really similar despite coming at it from a very different angle:
https://forums.puri.sm/t/is-anyone-else-experiencing-freezing-issues-with-librem-15-v3/1233/28
And I'm getting promising early results from this:
"I’m now running with intel_pstate=disable on kernel args and haven’t had the problem since."
Crossing my fingers this continues. If it has to do with powersave stuff kicking in badly that could explain the unpredictability very well.
Well.. at least I ruled some stuff out! I got a pointer freeze with zram and intel_pstate both disabled. Alas.
Same vblank stuff: pipe A vblank wait timed out
Continuing on this theme of turning other stuff off, rather than assuming the specific drivers for the trackpad/keyboard are at fault, I am very optimistic right now that it might be fixed (: Of course too soon to say for sure...
Here's what I did: I went into compton.conf and changed backend="glx"; to backend="xrender".
With this change, there are no vblank errors, video is much more stable, fullscreen and second monitor situations seem stable so far, and mame fills the second monitor properly (this one is not really relevant, but nice!)
In general it just feels like a happy machine that isn't getting clobbered by some deep issue that causes it to pause, chug and act up.
So my working theory is that opengl, or at least opengl with vsync enabled, is borked on CELES. Using xrender takes care of it, and I still seem to get good framerates; we watched Saturday Night Live with no issues and MAME framerates are great.
There could still be an issue. It's possible that the keyboard/trackpad driver just doesn't handle "hang on I can't help you right now" situations well and the vblank bug/problem/whatever with opengl was just making it really obvious. In which case I will eventually see a freeze anyway in some other situation. But... this looks an awful lot better so far.
Here's a summary of what I've done in total:
- intel power management is disabled at the kernel level (fallback clock speed management kicked in)
- zram is disabled
- There is a large swapfile (2GB) on disk
- compton is set to use xrender rather than opengl
- The vsync option is not configured at all for compton (since the old setting was opengl specific, I commented it out).
My working theory is that only the switch to xrender was actually necessary, but of course it's possible that all of these things are contributing to my happier outcome.
... So far. (: I'll let folks know when I've had this config for a week without freezes.
Only creating a large 2gb swap partition was not enough to stop the problem.
I'm gonna try the other things you've said right now.
Cool! I recommend you start with switching compton to xrender, I suspect that might be the only important one, possibly with the addition of the 4.14.14 kernel if you haven't done that one already.
Already running on 4.14.14 kernel, trying first switching compton to xrender.
.... Aaaaand I just experienced a pointer freeze out of nowhere. headdesk
Certainly the vblanks are gone, this has been well worth it.
Maybe this is related:
mmc0: Tuning timeout, falling back to fixed sampling clock
I bet that error is caused by putting swap on an SD card (mmc0). I am going to try disabling that swapfile and see if I do better with no swap at all or, if that's not practical, bringing back zram. It makes sense that an SD card is probably not the best place for realtime stuff.
I enabled xrender as the Compton backend and didn't make any other changes. Haven't seen any vblank timeouts and Youtube playback has much less stuttering but still getting pointer freezes (often during video playback) and haven't uncovered any useful log output.
Dang! Thanks for trying it.
On Mar 4, 2018 1:09 PM, "Russell Parker" [email protected] wrote:
I enabled xrender as the Compton backend and didn't make any other changes. Haven't seen any vblank timeouts and Youtube playback has much less stuttering but still getting pointer freezes (often during video playback) and haven't uncovered any useful log output.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-370250048, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fYsOYzjGxJN9eOnERtHoSWncmoHtks5tbC3JgaJpZM4Rvb5Q .
We should keep trying turning more stuff off but it may be that the issue really is with the trackpad or keyboard driver itself.
On Mar 4, 2018 1:26 PM, "Tom Boutell" [email protected] wrote:
Dang! Thanks for trying it.
On Mar 4, 2018 1:09 PM, "Russell Parker" [email protected] wrote:
I enabled xrender as the Compton backend and didn't make any other changes. Haven't seen any vblank timeouts and Youtube playback has much less stuttering but still getting pointer freezes (often during video playback) and haven't uncovered any useful log output.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GalliumOS/galliumos-distro/issues/415#issuecomment-370250048, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB9fYsOYzjGxJN9eOnERtHoSWncmoHtks5tbC3JgaJpZM4Rvb5Q .