Created attachment 115469 [details]
dmesg

System Environment==
--------------------------
Regression: Yes.Good commit:d600654ab94b325f253e267422dcf60302120ea0


==Bug detailed description==
-----------------------------
1.echo mem > /sys/power/state
2.resume the machine
3.repeat above steps for 5 times
From the second time,the machine can't suspend.

dmesg shows:
[   44.768274] [drm] stuck on render ring
[   44.768658] [drm] GPU HANG: ecode 6:0:0x00d3ffe0, reason: Ring hung, action: reset
[   44.768659] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[   44.768659] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[   44.768660] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[   44.768660] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[   44.768661] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[   44.768675] [drm:i915_reset_and_wakeup] resetting chip
[   44.768697] i915 0000:00:02.0: GEM idle failed, resume might fail
[   44.768712] pci_pm_suspend(): i915_pm_suspend+0x0/0x3f [i915] returns -11
[   44.768715] dpm_run_callback(): pci_pm_suspend+0x0/0xf6 returns -11
[   44.768718] PM: Device 0000:00:02.0 failed to suspend async: error -11
[   44.768753] PM: Some devices failed to suspend, or early wake event detected
[   44.768952] rtc_cmos 00:02: System wakeup disabled by ACPI
[   44.769550] serial 00:04: activated
[   44.769995] drm/i915: Resetting chip after gpu hang
[   44.790393] sd 0:0:0:0: [sda] Starting disk
[   44.793317] tpm_tis 00:05: TPM is disabled/deactivated (0x7)
[   44.929257] PM: resume of devices complete after 160.318 msecs
[   44.942465] PM: Finishing wakeup.
[   44.942466] Restarting tasks ... done.

==Reproduce steps==
---------------------------- 
1.echo mem > /sys/power/state
2.resume the machine
3.repeat above steps for 5 times
From the second time,the machine can't suspend.