Created attachment 279283 [details]
dmesg of Linux 4.19.0 with enabled pm_debug_messages
As with HP Elite x2 1013 G3 bug 201579 I trying to follow https://01.org/blogs/qwang59/2018/how-achieve-s0ix-states-linux guide to find, why S0ix don't work.
According to both of powertop and /sys/devices/system/cpu/cpuidle/low_power_idle_cpu_residency_us PC10 works and residency level is high (around 80-90% in short test runs, couple of minutes long) so no problem here. low_power_idle_cpu_residency_us counter increase not only in suspend freeze, but also with display in DPMS mode, so opportunistic sleep works too.
And like in HP Elite x2 1013 G3 case, /sys/kernel/debug/pmc_core/slp_s0_residency_usec is always zero, so here is debug data:
~# cat /sys/class/drm/card0/power/rc6_residency_ms
~# cat /sys/kernel/debug/dri/0/i915_dmc_info
fw loaded: yes
DC3 -> DC5 count: 1427
DC5 -> DC6 count: 1407
program base: 0x09004040
ssp base: 0x00002fc0
~# dmesg | grep i915
[ 6.622332] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[ 6.637093] [drm] Finished loading DMC firmware i915/skl_dmc_ver1_27.bin (v1.27)
[ 7.248588] [drm] Initialized i915 1.6.0 20180719 for 0000:00:02.0 on minor 0
[ 7.252941] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
[ 8.397699] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
[ 9.118709] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
[ 750.511004] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
[ 1258.991673] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
[ 1757.005162] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
Error message most likely related to this issue: https://bugs.freedesktop.org/show_bug.cgi?id=103229
Unfortunately, enabling pm_debug_messages doesn't not produce any output to dmesg, as you can see in attached log. So I not sure why S0ix would not work on this device. Any ideas?
Just in case: suspend freeze was tested with powertop --auto-tune.
Issue is still reproducible on Linux 5.0rc4 with this patches https://patchwork.kernel.org/patch/10714257/
Issue is still reproducible on Linux 5.0rc4 with this patches https://patchwork.kernel.org/project/platform-driver-x86/list/?series=74547
Have you tried Linux v5.4-rcx kernel recently?
Both V5.3 and v5.4 merged improvement patches for S0ix, covering NVMe and EC components.
If issue still can be reproduce with v5.4-rcx, please paste the outputs of
Wendy, this issue is no longer reproducible with Linux 5.5. Is possible, could you please ask developers to look into https://gitlab.freedesktop.org/drm/intel/issues/347 ? It's two and half years old.
The issue you've asked (https://gitlab.freedesktop.org/drm/intel/issues/347 ) is a gfx bug, in linux kernel, gfx bug discussion and debug should work on https://bugs.freedesktop.org/, how about reopen the issue and ask for the comments: https://bugs.freedesktop.org/show_bug.cgi?id=103229 if you see it still can be reproduced.
> how about reopen the issue and ask for the comments:
> https://bugs.freedesktop.org/show_bug.cgi?id=103229 if you see it still can
> be reproduced.
As I understand bugs.freedesktop.org is not longer relevant, as you can see for example here: https://bugs.freedesktop.org/describecomponents.cgi?product=DRI
> Moved to https://gitlab.freedesktop.org/drm/intel
Also as you can see in Comment 59 of https://bugs.freedesktop.org/show_bug.cgi?id=103229 I had to create new bugreport because original reporter don't answer anymore. My new bugreport already has been moved from old bugtracker https://bugs.freedesktop.org/show_bug.cgi?id=111201 to new bugtracker https://gitlab.freedesktop.org/drm/intel/issues/347
Unfortunate result of this back and forth with first bugreport (103229) second bugreport (111201) and third bugreport (347) is that all is happening is I re-test drm-tip and confirm that issue is not fixed for two and half years :(