Bug 108986 - system unable to suspend due to "i915 0000:00:02.0: GEM idle failed, suspend/resume might fail"
Summary: system unable to suspend due to "i915 0000:00:02.0: GEM idle failed, suspend/...
Status: CLOSED WORKSFORME
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: DRI git
Hardware: x86-64 (AMD64) Linux (All)
: medium normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard: Triaged
Keywords:
Depends on:
Blocks:
 
Reported: 2018-12-09 10:09 UTC by nanericwang
Modified: 2019-05-02 12:43 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg log when unable to suspend. (772.45 KB, text/x-log)
2018-12-16 13:56 UTC, nanericwang
no flags Details

Description nanericwang 2018-12-09 10:09:07 UTC
dmesg:

[  +0.000118] i915 0000:00:02.0: GEM idle failed, suspend/resume might fail
[  +0.000004] device_prepare(): 0xffffffff91261160 returns -512
[  +0.000002] PM: Device 0000:00:02.0 not prepared for power transition: code -512
[  +0.000001] PM: Some devices failed to suspend, or early wake event detected


kernel parameters:
initrd=\intel-ucode.img initrd=\initramfs-linux-clear.img root=LABEL=ROOT acpi_osi=Linux acpi_enforce_resources=lax elevator=bfq add_efi_memmap

i915 module parameters:
    alpha_support       = "Y"
    disable_display     = "N"
    disable_power_well  = "1"
    dmc_firmware_path   = "(null)"
    edp_vswing          = "0"
    enable_dc           = "-1"
    enable_dp_mst       = "Y"
    enable_dpcd_backlight= "N"
    enable_fbc          = "1"
    enable_guc          = "3"
    enable_gvt          = "N"
    enable_hangcheck    = "Y"
    enable_ips          = "1"
    enable_ppgtt        = "3"
    enable_psr          = "1"
    fastboot            = "Y"
    force_reset_modeset_test= "N"
    guc_firmware_path   = "(null)"
    guc_log_level       = "1"
    huc_firmware_path   = "(null)"
    invert_brightness   = "0"
    load_detect_test    = "N"
    lvds_channel_mode   = "0"
    mmio_debug          = "0"
    modeset             = "1"
    nuclear_pageflip    = "N"
    panel_use_ssc       = "-1"
    prefault_disable    = "N"
    reset               = "2"
    vbt_firmware        = "(null)"
    vbt_sdvo_panel_type = "-1"
    verbose_state_checks= "Y"
Comment 1 Chris Wilson 2018-12-09 10:15:50 UTC
-512 == you sent a signal to stop suspend.
Comment 2 nanericwang 2018-12-09 14:28:44 UTC
I see. It should be because that i killed systemd-sleep. The real problem is that systemd-sleep hangs so i kill it. What flag can i use for more verbose of i915 logs during suspend?
Comment 3 Jani Saarinen 2018-12-12 07:56:32 UTC
Reporter, what kernel version and system this was and have you seen issue with latest drm-tip? If not please try to reproduce the error using drm-tip (https://cgit.freedesktop.org/drm-tip) and kernel parameters drm.debug=0x1e log_buf_len=4M, and if the problem persists attach the full dmesg from boot.
Comment 4 nanericwang 2018-12-16 13:56:06 UTC
Created attachment 142821 [details]
dmesg log when unable to suspend.

It is really hard to repo. Hope this log helps.
Comment 5 Lakshmi 2018-12-18 05:07:34 UTC
(In reply to nanericwang from comment #0)
> dmesg:
> 
> [  +0.000118] i915 0000:00:02.0: GEM idle failed, suspend/resume might fail
> [  +0.000004] device_prepare(): 0xffffffff91261160 returns -512
> [  +0.000002] PM: Device 0000:00:02.0 not prepared for power transition:
> code -512
> [  +0.000001] PM: Some devices failed to suspend, or early wake event
> detected
> 
You get this message as you killed the systemd-sleep. Logs when systemd-sleep hangs are needed for investigation. Also, try to verify with Kernel 4.20.
Comment 6 nanericwang 2018-12-18 09:54:21 UTC
Lakshmi, Logs when systemd-sleep hangs have been attached per request already.
Comment 7 Jani Nikula 2018-12-28 12:17:22 UTC
Side note, the dmesg is a PITA to read when you have CONFIG_KALLSYMS=n and use dmesg --no-time option.
Comment 8 Jani Nikula 2018-12-28 12:21:36 UTC
Why do you have 

acpi_osi=Linux acpi_enforce_resources=lax elevator=bfq add_efi_memmap

in kernel parameters? Can you try without?
Comment 9 nanericwang 2018-12-28 12:48:42 UTC
I tried with no these parameters at beginning, it did not resolve the issue. I don't think they are the root cause.
Comment 10 Jani Nikula 2018-12-31 08:47:24 UTC
(In reply to nanericwang from comment #9)
> I tried with no these parameters at beginning, it did not resolve the issue.
> I don't think they are the root cause.

That doesn't really answer the question why you have them to begin with! ;)
Comment 11 Lakshmi 2019-02-22 08:13:45 UTC
Reporter, any specific reason why you have acpi_osi=Linux acpi_enforce_resources=lax elevator=bfq add_efi_memmap in kernel parameters?
Comment 12 nanericwang 2019-02-22 10:03:53 UTC
For 'elevator=bfq', I need BFQ low-latency to reduce application boot time. Forget about the rest of the kernel parameters, i copied them from somewhere and i don't know them, and they did not help either.
Comment 13 Lakshmi 2019-03-25 11:24:51 UTC
Reporter, can you verify this issue with drmtip https://cgit.freedesktop.org/drm-tip) and kernel parameters drm.debug=0x1e log_buf_len=4M, and if the problem persists attach the full dmesg from boot.
Comment 14 Lakshmi 2019-04-12 09:45:15 UTC
Any results from drmtip?
If it appears on dmrtip, what is the impact of this issue? Is it just the log messages or any other other impact you see?
Comment 15 Lakshmi 2019-05-02 12:43:16 UTC
(In reply to Lakshmi from comment #14)
> Any results from drmtip?
> If it appears on dmrtip, what is the impact of this issue? Is it just the
> log messages or any other other impact you see?

No feedback with drmtip for more than a month. Closing this bug as WORKSFORME. If the issue occurs again with drmtip, please provide all the above mentioned information.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.