Created attachment 95144 [details] dmesg System Environment: -------------------------- Platform: Broadwell kernel: drm-intel-nightly/0dfe3ce67bbd11c5b5e53250b0552ac2f36ac668 Bug detailed description: --------------------------- Some pm_pc8 subcases fail on Broadwell with -nightly kernel. The latest known good commit: 42aa254c65fa47fbea38a268814f815166adc056 The latest known bad commit: 072d358bf03d0c39dc83bce0b93801d030d2324f run ./pm_pc8 --run-subtest debugfs-read output: IGT-Version: 1.5-g072d358 (x86_64) (Linux: 3.14.0-rc4_drm-intel-nightly_0dfe3c_20140303+ x86_64) Runtime PM support: 0 PC8 residency support: 1 Test assertion failure function debugfs_read_subtest, file pm_pc8.c:981: Last errno: 5, Input/output error Failed assertion: wait_for_suspended() Subtest debugfs-read: FAIL Reproduce steps: ---------------------------- 1. ./pm_pc8 --run-subtest debugfs-read
Can you please attach the output from pm_pc8 from a working kernel? My guess is that the test doesn't properly skip here and 3.14 likely gained pm counter support for bdw or something like this ...
Yes, It doesn't properly skip. output(42aa254c65f): IGT-Version: 1.5-g42aa254 (x86_64) (Linux: 3.14.0-rc4_drm-intel-nightly_0dfe3c_20140303+ x86_64) Runtime PM support: 0 PC8 residency support: 0 Test requirement not met in function setup_environment, file pm_pc8.c:766: Last errno: 5, Input/output error Test requirement: (!(has_runtime_pm || has_pc8)) Subtest debugfs-read: SKIP
That is expected. The "regression" is that I told IGT that Broadwell should have PC8 support. Then, when you ran the test, it detected we did not support runtime PM, so it tried to test everything by checking PC8 residencies, and failed, as expected. Now that we merged runtime PM support for Broadwell, only the subtests with "pc8" on their names should fail. I will remove the word "regression" form the title since the root cause is that we started running tests instead of skipping, and we shouldn't be skipping since we want PC8 support on BDW anyway.
This should be fixed now. Please re-open if you still see a problem.
Created attachment 100073 [details] dmesg Some subcases still fail. # ./pm_rpm --run-subtest pc8-residency IGT-Version: 1.6-ge4ba3b7 (x86_64) (Linux: 3.15.0-rc7_drm-intel-nightly_d13557_20140529+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Test assertion failure function pc8_residency_subtest, file pm_rpm.c:734: Last errno: 2, No such file or directory Failed assertion: pc8_plus_residency_changed(120) Machine is not reaching PC8+ states, please check its configuration. Subtest pc8-residency: FAIL
Hi Please go to the BIOS, Intel Advanced Menu -> Power and Performance -> CPU Power Management Control -> Package C state Limit, then set it to "C10". Then test this tree: http://cgit.freedesktop.org/~pzanoni/linux/log/?h=bdw-pc8-fixes . Don't forget to switch to the bdw-pc8-fixes branch. Please also boot with drm.debug=0xe and send the dmesg output in case there's any problem. Thanks, Paulo
(In reply to comment #6) > Hi > > Please go to the BIOS, Intel Advanced Menu -> Power and Performance -> CPU > Power Management Control -> Package C state Limit, then set it to "C10". > > Then test this tree: > http://cgit.freedesktop.org/~pzanoni/linux/log/?h=bdw-pc8-fixes . Don't > forget to switch to the bdw-pc8-fixes branch. > > Please also boot with drm.debug=0xe and send the dmesg output in case > there's any problem. > > Thanks, > Paulo It still fails. output: IGT-Version: 1.7-g7ef5372 (x86_64) (Linux: 3.16.0-rc2_kcloud_619f97_20140701+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Test assertion failure function pc8_residency_subtest, file pm_rpm.c:734: Last errno: 2, No such file or directory Failed assertion: pc8_plus_residency_changed(120) Machine is not reaching PC8+ states, please check its configuration. Subtest pc8-residency: FAIL
Created attachment 102041 [details] dmesg(bdw-pc8-fixes branch)
Hi Thanks for testing. I just updated the tree, including a patch that was missing. Can you please try again? Also please disconnect all the USB devices before running the test. I see you have an USB mouse that keeps waking up the Kernel. And if you have wireless on the machine, please also disconnect it: you might want to physically remove the wireless device, if possible. The BDW environment for PC8 testing should be the same as the HSW environment. Thanks, Paulo
Created attachment 102182 [details] dmesg(commit 8eb06d) IGT-Version: 1.7-g67e29a3 (x86_64) (Linux: 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Test assertion failure function pc8_residency_subtest, file pm_rpm.c:734: Last errno: 2, No such file or directory Failed assertion: pc8_plus_residency_changed(120) Machine is not reaching PC8+ states, please check its configuration. Subtest pc8-residency: FAIL
(In reply to comment #10) > Created attachment 102182 [details] > dmesg(commit 8eb06d) > > IGT-Version: 1.7-g67e29a3 (x86_64) (Linux: > 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) > Runtime PM support: 1 > PC8 residency support: 1 > Test assertion failure function pc8_residency_subtest, file pm_rpm.c:734: > Last errno: 2, No such file or directory > Failed assertion: pc8_plus_residency_changed(120) > Machine is not reaching PC8+ states, please check its configuration. > Subtest pc8-residency: FAIL It seems you didn't disconnect all your USB devices before running the test. I still see them in dmesg. Can you please redo the test as described in comment #9?
Created attachment 102227 [details] dmesg IGT-Version: 1.7-g67e29a3 (x86_64) (Linux: 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Test assertion failure function pc8_residency_subtest, file pm_rpm.c:734: Last errno: 2, No such file or directory Failed assertion: pc8_plus_residency_changed(120) Machine is not reaching PC8+ states, please check its configuration. Subtest pc8-residency: FAIL
QA team checked another BDW-U system, and build kernel from Paulo's branch http://cgit.freedesktop.org/~pzanoni/linux/log/?h=bdw-pc8-fixes, we can successfully see system entry into PC8, passed IGT PC8 subtests: [root@x-bdw02 tests]# ./pm_rpm --run-subtest pc8-residency IGT-Version: 1.7-g8d60b82 (x86_64) (Linux: 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Subtest pc8-residency: SUCCESS [root@x-bdw02 tests]# ./pm_rpm --run-subtest modeset-pc8-residency-stress IGT-Version: 1.7-g8d60b82 (x86_64) (Linux: 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Subtest modeset-pc8-residency-stress: SUCCESS [root@x-bdw02 tests]# ./pm_rpm --run-subtest gem-execbuf-stress-pc8 IGT-Version: 1.7-g8d60b82 (x86_64) (Linux: 3.16.0-rc2_kcloud_8eb06d_20140703+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Subtest gem-execbuf-stress-pc8: SUCCESS
Since the test passes on some of QA's machines and we believe our code is complete, I will close this as fixed. BDW machines that can't hit PC8 are probably either misconfigured, or have a too-old BIOS, or have a too-old processor. If you think the bug is not fixed yet, please reopen.
Re-verified on another failed BDW unit: x-bdw01 Upgrade the BIOS to V86.R00 and change the internal display into eDP panel. used Kernel is : Linux x-bdw01 3.16.0-rc6_drm-intel-nightly_e967a5_20140728+ #868 SMP Mon Jul 28 11:24:20 CST 2014 x86_64 x86_64 x86_64 GNU/Linux pc8-residency and debugfs-read subtests all verified as PASS, log message show in below. [root@x-bdw01 tests]# ./pm_rpm --run-subtest pc8-residency IGT-Version: 1.7-gfcbc502 (x86_64) (Linux: 3.16.0-rc6_drm-intel-nightly_e967a5_20140728+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Subtest pc8-residency: SUCCESS You have new mail in /var/spool/mail/root [root@x-bdw01 tests]# ./pm_rpm --run-subtest debugfs-read IGT-Version: 1.7-gfcbc502 (x86_64) (Linux: 3.16.0-rc6_drm-intel-nightly_e967a5_20140728+ x86_64) Runtime PM support: 1 PC8 residency support: 1 Subtest debugfs-read: SUCCESS
Closing old verified.
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.