wolf fails to resume screen properly after suspend |
||||||||||
Issue descriptionForked from https://crbug.com/859358#c16: "Add a Dell Wolf to the list of affected models; mine started failing to resume July 18th, though it has been on the latest stable v67 since it was released late June. Always powers up correctly on lid open if shut down though. Will submit a feedback report tonight." http://feedback/#/Report/85568355203 is a report of a wolf system at R67-10575.58.0 (67.0.3396.99) freezing on suspend and coming back after Alt+F10+x. Here's the end of powerd.PREVIOUS: ... [0725/214215:INFO:daemon.cc(514)] Lid closed ... [0725/214216:INFO:main.cc(258)] Running "/usr/bin/powerd_setuid_helper --action=suspend --suspend_wakeup_count_valid --suspend_wakeup_count=6554" <garbage indicating unclean shutdown> eventlog: 212 | 2018-07-25 21:42:21 | ACPI Enter | S3 213 | 2018-07-25 21:52:56 | EC Event | Lid Open 214 | 2018-07-25 21:52:56 | ACPI Wake | S3 215 | 2018-07-25 21:52:56 | Wake Source | PCI Express | 0 216 | 2018-07-25 21:53:07 | System boot | 60 syslog: ... 2018-07-25T21:42:16.477177-04:00 NOTICE powerd_suspend[3277]: Going to suspend-to-RAM state: args=--suspend_duration=-1 --nosuspend_to_idle --wakeup_count=6554 2018-07-25T21:42:16.557260-04:00 NOTICE powerd_suspend[3334]: Skipping disable BT HCI mode change event on non-Intel BT 2018-07-25T21:42:16.560073-04:00 NOTICE powerd_suspend[3337]: Available order 3 pages: 42075 2018-07-25T21:42:16.563160-04:00 NOTICE powerd_suspend[3340]: Finalizing suspend <garbage> 2018-07-25T21:53:08.848062-04:00 INFO kernel: [ 0.000000] Initializing cgroup subsys cpu ... The wake from S3 in eventlog is suspicious. Did the system resume but fail to reconfigure displays, resulting in the screen remaining blank and the user forcing a reboot before logs got synced?
,
Jul 27
Additionally need clarification ... user reports after 'Alt+F10+x' but doesn't explicitly mention how many times the key combo was pressed. If it were only once or twice I'd expect to see the chrome restart or sysrq signatures in the logs ... but don't. So I'd assume it was multiple times to induce cold reboot.
,
Jul 28
I just sent another report with cbrug 868097 in the description. It came back to the login screen immediately after one alt vol+ x this time.
,
Jul 28
Thanks for additional report (report:85571712227) and for clarification. I do see the sysrq in this case, 2018-07-27T21:34:28.550223-04:00 INFO kernel: [ 2580.960553] SysRq : Cros dump and crash 2018-07-27T21:34:28.550248-04:00 INFO kernel: [ 2580.960593] sysrq_x_cros_signal_process: signal 6 chrome pid 812 tgid 812 And prior to that a successful resume, 2018-07-27 21:34:00.256 6 kernel : [ 2553.236472] PM: resume of devices complete after 122.941 msecs 2018-07-27 21:34:00.256 7 kernel : [ 2553.236710] PM: Finishing wakeup. ... 2018-07-27 21:34:00.317 5 powerd_suspend[5702]: Resume finished So resume itself took < 1sec as expected but after ~28sec of dark screen user induced SysRq to recover. These drm related log messages look suspicious during the suspend, 2018-07-27 21:34:00.255 3 kernel : [ 2552.922623] [drm:i915_write32] *ERROR* Unknown unclaimed register before writing to 44024 2018-07-27 21:34:00.255 3 kernel : [ 2552.922641] [drm:i915_write32] *ERROR* Unclaimed write to 44024 from (b/35521315#comment3 marcheu@): Haswell and up have a feature where the GPU can maintain a list of mmio registers which are valid/invalid at a given point in time (this is the "unclaimed register" errors). So if a register is invalid (it doesn't exist on this hw) or if the corresponding hardware block is off, then you will get such an error. However the same drm messages occur on all suspends not just the failing s2r so they may be benign. +Gfx folks for any help determining why screen remains black after successful resume.
,
Jul 29
I also did send a report. Stepping to a beta version (now version 68.0.3440.76 (Official build) beta (64-bits)) did not help. What seems to help to make the problem less offten appear is shift+search+L before closing the lid. The big question imho is if it goes wrong by closing the lid (some kind of forced crash)... or by opening the lid (some kind of graphics thing). Hope my report did help.
,
Aug 2
I just sent another report. This time the screen came up after I waited about a minute, then pressed a random key, ctrl in this case, It seems to only fail to turn on the screen the first resume after a shutdown/boot. Then will resume correctly after that ever time if brought back by ctrl volup, x until the next reboot. Will see if it continues to resume correctly by "waiting it out"
,
Aug 2
edit above, alt, volup, x; not ctrl as stated above.
,
Aug 3
It seems consistent for me that it only remains at a black screen the first resume after a reboot or shutdown/restart. If the screen is brought back by alt, volup, x, or by "waiting it out", the CB resumes perfectly every time until the next reboot.
,
Aug 8
Upgraded to new stable release 68.0.3440.87 and same symptoms. Sent a report with Issue 868097 in description.
,
Aug 29
Thanks for feedback ( report:85590496941 ) I see these error in UI_LOG, [[7145:7145:0808/015918.693748:ERROR:vaapi_wrapper.cc(577)] vaQueryConfigEntrypoints failed VA error: the requested VAProfile is not supported [7145:7145:0808/015918.693828:ERROR:vaapi_wrapper.cc(577)] vaQueryConfigEntrypoints failed VA error: the requested VAProfile is not supported [7020:7020:0808/015918.840382:ERROR:input_method_manager_impl.cc(1080)] IMEEngine for "jkghodnilhceideoidjikpgommlajknk" is not registered device-enumerator: scan all dirs device-enumerator: scanning /sys/bus device-enumerator: scanning /sys/class device-enumerator: scan all dirs device-enumerator: scanning /sys/bus device-enumerator: scanning /sys/class [7145:7153:0808/015920.560040:ERROR:hardware_display_plane_manager.cc(445)] CTM is empty. Expected a 3x3 matrix. [7145:7153:0808/015920.560133:ERROR:drm_display.cc(181)] Failed to set color correction for display: crtc_id = 19 [7020:7020:0808/015926.104951:ERROR:input_method_manager_impl.cc(1080)] IMEEngine for "jkghodnilhceideoidjikpgommlajknk" is not registered [7020:7124:0808/015926.244048:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015926.244235:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015926.244358:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015926.244552:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.101412:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.104746:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.105073:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.105372:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service device-enumerator: scan all dirs device-enumerator: scanning /sys/bus device-enumerator: scanning /sys/classpported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.105073:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service [7020:7124:0808/015927.105372:ERROR:service_manager_context.cc(250)] Attempting to run unsupported native service: /opt/google/chrome/chrome_renderer.service device-enumerator: scan all dirs device-enumerator: scanning /sys/bus device-enumerator: scanning /sys/class
,
Aug 29
,
Oct 10
*** coral - Dell Chromebook 11 5190 are also affected by this issue. *** One of our EDU customer reported the same problem as described hereunder: Do we have any status update about this issue to share? [DESCRIPTION - #17154236] = WORKING ENVIRONMENT= Chrome OS version: 69.0.3497.95 Devices make/model: Dell Chromebook 11 5190 Domain managed devices: Yes = ISSUE DESCRIPTION = When user closed the lid and opened it again, the Chromebook will not wake up and display screen stays black = STEPS TO REPRODUCE = //Please follow the format below, and avoid describing everything in just one paragraph. 1. Turn on device and sign in 2. Close lid 3. Open lid = WHAT'S THE EXPECTED BEHAVIOR: Chrome device should wake up after opening the lid = WHAT'S THE ACTUAL RESULT: Display screen stays black = TIMEFRAME WHEN ISSUE STARTED = After 9/4 light black screens and 9/20 very heavy = DOES IT AFFECT ALL DEVICES = Particular portion of the device = ISOLATED TO = Dell Chromebook 11 5190 = TROUBLESHOOTING STEPS TAKEN = Per customer, "Put on Charge. Hold the power switch when charger in or out. Combination of the key and power button. Let’s them set on the shelf for a few days and they can power on eventually if not then replacing the MB. Feels like power management issue. Some our students certified by Dell and Samsung to repair devices on-site. After they removed the battery out and plugged in the charger then they able to turn on the device. They have tried on 4 devices and works even after installed the battery back. = SUPPORTING INFO = - Chrome Policy / <https://drive.google.com/drive/folders/1g2YexYlK87rUDnCsJvAeOTQetYXf5-R_?ogsrc=32> last modified 10/4/18 with a filename of policies.json - Chrome debug logs / <https://drive.google.com/drive/folders/1g2YexYlK87rUDnCsJvAeOTQetYXf5-R_?ogsrc=32> last modified 10/4.18
,
Oct 11
It just happens that I had a wolf lying around. I tested it and it did repro once out of 10 suspends. It isn't graphics related because the whole kernel didn't come back from suspend (I had no network for example).
,
Nov 2
,
Nov 2
Hi, Do we have any status update about this issue to share? Our Enterprise customer is requesting some updates. == ADDITIONAL NOTES FROM CUSTOMER == - tried several Dell 5190 model to capture the crash ID and under chrome://crashes/? stays blank, nothing there at all - They're seeing the issue on Samsung XE500 C12 K102, Dell 3120 and 3180. They all have the power on/ wake up the issue but Dell 5190 is the leader of this issue == AFFECTED MODEL == - candy - kefka - winky - coral
,
Nov 6
,
Nov 7
No luck repro'ing on candy R71-11056.0.0 On wolf device it was running really old FW need to upgrade that before retesting. Device is also EVT2 which might present other challenges so I'll try to get a PVT/MP unit in parallel.
,
Nov 7
Ran 1 hour test on wolf remotely with device logged in as guest using servo to close/open lid every Xsecs (between 0 & 30sec) without any hard hang/reboot. Seems like most recent feedback reports focus on long resume for first suspend after a reboot. Going to focus my efforts there for repro next.
,
Nov 7
,
Nov 8
Upgraded the FW on wolf device I borrowed and was NOT able to repro either a) black screen upon repeated lid close/opens (tried 10) b) long resume time after reboot and initial lid close (tried 10). I also tried to repro of b) on a candy device w/o success I'm afraid without repro there's little more to debug until there's additional feedback reports that might offer a clue on where the failure is.
,
Nov 9
I haven't had this issue on Wolf since v70 stable.
,
Nov 12
Thanks for feedback. Marking this 'wont fix' as pertains to wolf. For other models, please file separate bug if problem is still occurring on latest-stable and attach logs or feedback links. |
||||||||||
►
Sign in to add a comment |
||||||||||
Comment 1 by trumbull@chromium.org
, Jul 26