Bug 76542 - [BYT] System fail to enter package c6 using 2014Q1 release kernel ver 3.13.6
Summary: [BYT] System fail to enter package c6 using 2014Q1 release kernel ver 3.13.6
Status: CLOSED WONTFIX
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: All Linux (All)
: medium normal
Assignee: Paulo Zanoni
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
: 76543 (view as bug list)
Depends on:
Blocks:
 
Reported: 2014-03-24 08:24 UTC by wendy.wang
Modified: 2015-05-13 07:39 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:

Attachments
dmesg log file (85.56 KB, text/plain)
2014-03-24 08:24 UTC, wendy.wang 		no flags Details
View All

Description wendy.wang 2014-03-24 08:24:04 UTC 
Created attachment 96273 [details]
dmesg log file

Load the 2014Q1 release package kernel and other GFX components drivers:
Kernel_version: 	 3.13.6
Libdrm:        		 (master)libdrm-2.4.52
Mesa:           		(10.1) mesa-10.1
Xserver:                	(server-1.15-branch)xorg-server-1.15.0
Xf86_video_intel:     (master)2.99.910
Cairo:          		(master)1.12.6
Libva:          		1.3.0.pre1
Libva_intel_driver:   1.3.0.pre1
Intel-gpu-tools: 	1.6

1. Set the disk link power management policies to min_power.
For each host${i} directory inside /sys/class/scsi_host/, echo min_power > host${i}/link_power_management_policy
2. Run ./turbostat tool to wait for the PC6 results, but we did not see any PC6 residency always show 0
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.22 0.65 2.00   0   0.48   0.00  99.30   32   0.00   0.00
  0   0   0.30 0.65 2.00   0   0.37   0.00  99.33   32   0.00   0.00
  1   1   0.14 0.67 2.00   0   0.60   0.00  99.27   32
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.21 0.72 2.00   0   4.27   0.00  95.51   32   0.00   0.00
  0   0   0.30 0.74 2.00   0   0.26   0.00  99.44   32   0.00   0.00
  1   1   0.12 0.68 2.00   0   8.28   0.00  91.59   32
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.31 0.85 2.00   0   0.31   0.00  99.38   32   0.00   0.00
  0   0   0.49 0.89 2.00   0   0.36   0.00  99.16   32   0.00   0.00
  1   1   0.13 0.69 2.00   0   0.27   0.00  99.60   32
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.21 0.68 2.00   0   0.35   0.00  99.44   32   0.00   0.00
  0   0   0.30 0.68 2.00   0   0.28   0.00  99.42   32   0.00   0.00
  1   1   0.13 0.67 2.00   0   0.42   0.00  99.46   32
Comment 1 Chris Wilson 2014-03-24 09:57:20 UTC 
*** Bug 76543 has been marked as a duplicate of this bug. ***
Comment 2 wendy.wang 2014-03-31 01:28:45 UTC 
Test the nightly branch kernel: Linux x-byt01 3.14.0-rc7_drm-intel-nightly_842955_20140325+ #1006 SMP Tue Mar 25 13:26:22 CST 2014 x86_64 x86_64 x86_64 GNU/Linux

BYT system also cannot enter PC6 status.
The reproduce steps are the same as above.
Comment 3 wendy.wang 2014-04-10 01:14:29 UTC 
QA team did not find a workable kernel commit which can let BYT enter PC6, both tried recent and old kernel:
Recent Kernel : 3.14.0_drm-intel-nightly_7cd8b8_20140407&#43
Old Kernel : 3.12.0_drm-intel-fixes_1dca22_20131115_&#43

@Pan, Jacob jun <jacob.jun.pan@intel.com>
Suggested by Jesse, you might be the correct developer about this, would you pls take a look at this bug, thanks!
Comment 4 Daniel Vetter 2014-04-11 14:40:13 UTC 
*** Bug 76877 has been marked as a duplicate of this bug. ***
Comment 5 Daniel Vetter 2014-05-19 16:33:47 UTC 
Is this still an issue with latest kernel+igt?
Comment 6 wendy.wang 2014-05-20 02:24:49 UTC 
We do have igt test to cover PC6.
Run the latest version of igt pm_rc6_residency case, the result is PASS.
IGT-Version: 1.6-gd71add5 (x86_64) (Linux: 3.15.0-rc3_drm-intel-nightly_f79ba7_20140519+ x86_64)

Test pc6 using turbostat tool, still did not see Baytrail enter PC6.
Kernel:Linux x-byt06 3.15.0-rc3_drm-intel-nightly_f79ba7_20140519+ #2774 SMP Mon May 19 11:24:27 CST 2014 x86_64 x86_64 x86_64 GNU/Linux

cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.18 1.00 2.00   0   1.07   0.00  98.75   37   0.00   0.00
  0   0   0.32 1.02 2.00   0   2.00   0.00  97.68   37   0.00   0.00
  1   1   0.04 0.88 2.00   0   0.14   0.00  99.82   37
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.15 0.70 2.00   0   0.21   0.00  99.63   37   0.00   0.00
  0   0   0.26 0.69 2.00   0   0.32   0.00  99.42   37   0.00   0.00
  1   1   0.05 0.81 2.00   0   0.11   0.00  99.84   37
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.15 0.74 2.00   0   0.23   0.00  99.62   37   0.00   0.00
  0   0   0.26 0.71 2.00   0   0.36   0.00  99.38   37   0.00   0.00
  1   1   0.05 0.89 2.00   0   0.09   0.00  99.86   37
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.14 0.70 2.00   0   0.20   0.00  99.66   36   0.00   0.00
  0   0   0.24 0.68 2.00   0   0.27   0.00  99.49   36   0.00   0.00
  1   1   0.04 0.84 2.00   0   0.13   0.00  99.83   36
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.15 0.74 2.00   0   2.35   0.00  97.50   37   0.00   0.00
  0   0   0.25 0.69 2.00   0   0.25   0.00  99.49   37   0.00   0.00
  1   1   0.05 0.97 2.00   0   4.45   0.00  95.50   37
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.14 0.77 2.00   0   4.37   0.00  95.48   37   0.00   0.00
  0   0   0.25 0.72 2.00   0   0.30   0.00  99.46   37   0.00   0.00
  1   1   0.04 1.04 2.00   0   8.44   0.00  91.51   37
Comment 7 wendy.wang 2014-05-20 02:28:05 UTC 
Correct comment#6 typo error:
We do NOT have igt test to cover PC6
Comment 8 wendy.wang 2014-06-03 06:34:46 UTC 
After Ping with Jacob, his comment as following:"BYT uses different MSRs for PC6. Therefore,turbostat tool needs to be updated to correctly capture the residency count. BYT uses PC7 counter for PC6 :(, but since the program is so late, it will unlikely to be fixed in punit fw."

And test through reading the raw counter,we got:

[root@x-byt01 performance_pnp]# ./turbostat -c 0x3fa -i 2 -v
turbostat v3.7 Feb 6, 2014 - Len Brown <lenb@kernel.org>
CPUID(0): GenuineIntel 11 CPUID levels; family:model:stepping 0x6:37:3 (6:55:3)
CPUID(6): APERF, DTS, EPB
SLM BCLK: 133.3 Mhz
RAPL: 4581 sec. Joule Counter Range, at 30 Watts
cpu0: MSR_NHM_PLATFORM_INFO: 0x40008001000
4 * 133 = 533 MHz max efficiency
16 * 133 = 2133 MHz TSC frequency
cpu0: MSR_IA32_POWER_CTL: 0x00000000 (C1E auto-promotion: DISabled)
cpu0: MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x0010000f (UNlocked: pkg-cstate-limit=7: pc7)
cpu0: MSR_NHM_TURBO_RATIO_LIMIT: 0x00000000
cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
cpu0: MSR_RAPL_POWER_UNIT: 0x00000505 (0.031250 Watts, 0.000032 Joules, 0.000977 sec.)
cpu0: MSR_PKG_POWER_LIMIT: 0x003880fa (UNlocked)
cpu0: PKG Limit #1: ENabled (7.812500 Watts, 262144.000000 sec, clamp DISabled)
cpu0: PKG Limit #2: DISabled (0.000000 Watts, 0.000977* sec, clamp DISabled)
cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x00690000 (105 C)
cpu0: MSR_IA32_THERM_STATUS: 0x884b0000 (30 C +/- 1)
cpu1: MSR_IA32_THERM_STATUS: 0x884a0000 (31 C +/- 1)
    Core     CPU Avg_MHz   %Busy Bzy_MHz TSC_MHz     SMI  count 0x3FA  CPU%c1  CPU%c6 CoreTmp PkgWatt CorWatt
       -       -      76    3.60    2102    2133       0   3685522176    0.53   95.87      30    0.43    0.27
       0       0      55    2.66    2060    2133       0   3685522176    0.86   96.49      30    0.43    0.27
       1       1      96    4.53    2127    2133       0   3685522176    0.21   95.25      30
    Core     CPU Avg_MHz   %Busy Bzy_MHz TSC_MHz     SMI  count 0x3FA  CPU%c1  CPU%c6 CoreTmp PkgWatt CorWatt
       -       -     152    7.20    2111    2133       0   2996674304    0.83   91.98      32    0.67    0.51
       0       0      65    3.17    2045    2133       0   2996674304    1.23   95.60      30    0.67    0.51
       1       1     239   11.22    2130    2133       0   2996674304    0.43   88.35      32
    Core     CPU Avg_MHz   %Busy Bzy_MHz TSC_MHz     SMI  count 0x3FA  CPU%c1  CPU%c6 CoreTmp PkgWatt CorWatt
       -       -     154    7.28    2119    2133       0   3677171712    0.35   92.38      30    0.49    0.33
       0       0      45    2.21    2054    2133       0   3677171712    0.65   97.13      30    0.49    0.33
       1       1     263   12.34    2130    2133       0   3677171712    0.04   87.62      30
    Core     CPU Avg_MHz   %Busy Bzy_MHz TSC_MHz     SMI  count 0x3FA  CPU%c1  CPU%c6 CoreTmp PkgWatt CorWatt
       -       -     217   10.24    2118    2133       0   3005754112    0.68   89.08      32    0.54    0.38
       0       0     109    5.23    2078    2133       0   3005754112    1.17   93.60      30    0.54    0.38
       1       1     325   15.26    2132    2133       0   3005754112    0.19   84.56      32
    Core     CPU Avg_MHz   %Busy Bzy_MHz TSC_MHz     SMI  count 0x3FA  CPU%c1  CPU%c6 CoreTmp PkgWatt CorWatt
       -       -       1    0.11    1281    2133       0   3061620992    8.21   91.67      30    0.56    0.40
       0       0       2    0.17    1221    2133       0   3061620992    0.40   99.43      30    0.56    0.40
       1       1       1    0.05    1469    2133       0   3061620992   16.03   83.92      30

Then calculate PC6 residency thought this formula: pc6_3fa_count/(TSC_Mhz*1000000*sample_peroid_sec), I can see BYT entries PC6 state, so will close this bug as will not fix.

Used kernel is: 
[root@x-byt01 performance_pnp]# uname -a
Linux x-byt01 3.15.0-rc7_drm-intel-nightly_085391_20140602+ #3182 SMP Mon Jun 2 11:24:51 CST 2014 x86_64 x86_64 x86_64 GNU/Linux

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.