XO-4/Kernel/Testing

From OLPC
< XO-4
Revision as of 21:37, 28 January 2013 by Greenfeld (talk | contribs) (929a97d)
Jump to: navigation, search

Final testing push default configuration:

  • fs-update os26
  • change-tag TS DEVL \ enable aggressive mode with repeated runs

Debugging configuration A (with wireless):

  • log the output of the serial console,
  • olpc-dev-kernel
  • no_console_suspend

Debugging configuration B (without wireless):

  • log the output of the serial console,
  • olpc-dev-kernel
  • no_console_suspend
  • touch /runin/no-wireless
  • echo blacklist mwifiex_sdio >> /etc/modprobe.d/olpc.conf

See also:

Testing is being tracked below by kernel git hash.

Kernel developers, put your new kernel in a new section here. Latest at top. Testers place links to reports indented under the kernel. Test reports should include Open Firmware and EC versions.


0aff862

  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, default serial configuration, default memtouse configuration, completed by James:
    • SKU296, 3833 cycles,
    • SKU295, 3863 cycles,
    • SKU292 with 8787, 3623 cycles,
    • SKU293 with 8686, 3638 cycles.

734fcfa

  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1 plus the memtouse workaround from <trac>12285</trac>, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, default serial configuration, by James:
    • SKU296, 3792 cycles, completed,
    • SKU295, 3946 cycles, completed,
    • SKU292 with 8787, 3618 cycles, completed,
    • SKU293 with 8686, <trac>12285</trac> (out of memory kills memtester), fail at 3050 cycles, 73424 seconds boot time, despite the memtouse workaround.

ae2933a

  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1 plus the memtouse workaround from <trac>12285</trac>, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, default serial configuration, completed by James:
    • SKU296, 3830 cycles,
    • SKU295, 3826 cycles,
    • SKU292 with 8787, 3609 cycles,
    • SKU293 with 8686, 3690 cycles.
  • (while trying to recover runin logs, ae2933a was unreliable compared to 929a97d with respect to longevity of 8787 connection in the absence of suspend and resume. after several minutes needed to re-establish connection by using disconnect/reconnect in sugar network view. the symptom is lack of packet flow, not sure which direction, but can't ping into the host or out of it.)

929a97d

  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1 plus the memtouse workaround from <trac>12285</trac>, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, default serial configuration, by James:
    • SKU296, 3790 cycles, completed,
    • SKU295, 3820 cycles, completed,
    • SKU292 with 8787, 3616 cycles, completed,
    • SKU293 with 8686, 3609 cycles, completed.
  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, extreme, watchdog, default serial configuration, by James:
    • SKU296, 14030 cycles, completed,
    • SKU295, 13731 cycles, completed,
    • SKU292 with 8787, 12773 cycles, completed,
    • SKU293 with 8686, <trac>12285</trac> (out of memory kills memtester), fail at 5773 cycles, 38131 seconds boot time.
  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, default serial configuration, by James:
    • SKU296, 3777 cycles, completed, <trac>12397</trac> (out of memory in audio) observed,
    • SKU295, 3861 cycles, completed, <trac>12397</trac> observed,
    • SKU292 with 8787, 3581 cycles, completed, <trac>12397</trac> observed,
    • SKU293 with 8686, <trac>12285</trac> (out of memory kills memtester), fail at 3219 cycles, 76344 seconds boot time.
    • SKU 295 with 8787, aggressive suspend but otherwise normal settings, completed 8 consecutive 24-hour runs, #12397 seen, by Samuel.
    • SKU 296 with 8787, aggressive suspend but otherwise normal settings, completed 3 consecutive 24-hour runs, #12397 seen, by Samuel. First two cases of #12456 and this XO also seen with this build.
    • SKU 292 with 8686, aggressive suspend but otherwise normal settings, hung after 17 cycles, SDIO/MMC kernel panic on resume, by Samuel. Same XO went on to do 5 consecutive 24-hour runs, #12397 also seen.

c87dda3

  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, watchdog, in progress by James:
    • SKU296, 3801 cycles, completed,
    • SKU295, 3914 cycles, completed,
    • SKU292 with 8787, 3622 cycles, completed,
    • SKU293 with 8686, 3718 cycles, completed.

9e42d55

  • stock kernel in build 26, along with Q7B11, EC 0.3.08, and runin 0.21.0,
  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.09, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, by James:
    • SKU296, with watchdog, 3800 cycles, completed,
    • SKU295, 3754 cycles, completed,
    • SKU295, with watchdog, 1572 cycles, discovered off, no serial log, last kern.log activity was cmd 0x66,
    • SKU292 with 8787, 3770 cycles, completed,
    • SKU292 with 8787, with watchdog, 1225 cycles, discovered off, no serial log, last kern.log activity was cmd 0x66,
    • SKU293 with 8686, with watchdog, 2550 cycles, 1177 cycles (why did it go backwards?).
    • SKU293 with 8686, out of memory fail at 2940 cycles, logs-SHC2380007F-130117_025503.tar.gz
  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.08, runin 0.21.1, blacklisted btmrvl_sdio and mwifiex_sdio, aggressive, by James:
    • SKU296, hung at 2649 cycles, olpc-ec-1.75: transmit FIFO not empty,
    • SKU295, 3869 cycles, completed,
    • SKU292, 3633 cycles, completed,
  • build 24 (with kernel rpm), Q7B11jd, EC 0.3.08, runin 0.21.1, blacklisted btmrvl_sdio, aggressive, by James:
    • SKU293 with 8686, hung at 3122 cycles.
  • build 26 Q7B11, EC 0.3.09, runin 0.20.0
    • SKU 295 with 8787, aggressive suspend but otherwise normal settings, completed 5 consecutive 24-hour runs, #12397 seen, by Samuel.
    • SKU 292 with 8686, aggressive suspend but otherwise normal settings, hung after 17 cycles, SDIO/MMC kernel panic on resume, by Samuel.

6ebcabf

In Progress

  • stock kernel in build 25, along w. Q7B11 and EC 0.3.08
  • build 24 (with kernel rpm), runin 0.21.1, blacklisted btmrvl_sdio only, no-wireless, aggressive, SKU296, 4849 cycles, completed, by James.
  • build 25, aggressive timing, a B1 w. 8686 card, 10K S/R, in progress by wad.
  • build 25, aggressive timing, a B1 w. 8686 card, 6000 cycles, in progress by wad.
  • build 25, no-wireless and blacklisted mwifiex_sdio and btmrvl_sdio drivers, aggressive timing, (2) C1 units w. 8787 cards, all 11K+ cycles, in progress by wad.
  • build 25, C2, with 8787 enabled, with no_console_suspend. Started 2013-01-17 00:07:25 UCT. Logging to runin-with-8787-with-n_c_s-0.log
  • build 25 plus EC 3.09, no_console_suspend debug and aggressive, SKU 298, 2200 S/R cycles, in progress by wad
  • build 25 plus EC 3.09, aggressive timing, a B1 and a 1.2 GHz B0B w. 8686 cards, around 2200 S/R cycles each, in progress by wad.

Failures

  • build 25, aggressive timing, (7) B1 and C1 units w. 8787 cards, all failed within 100 S/R cycles, by wad.
  • build 25, aggressive timing, (2) B1 units w. 8686 cards, hung at 1600 and 160 S/R cycles, by wad.
  • build 25, aggressive timing, a B1 and a 1.2 GHz B0B w. 8686 cards, around 8K S/R cycles each, hung in resume, by wad.
  • build 25, no wireless and blacklisted mwifiex/btmrvl drivers, aggressive timing, (2) C1 and a B1, 8K S/R cycles, hung (lost power due to EC bug), by wad.
  • build 25, no wireless and blacklisted mwifiex/btmrvl drivers, aggressive timing, (1) B1 w. 8787, 8K S/R cycles, hung, by wad.
  • build 25, sku 295, drivers blacklisted, hang on wakeup from idle suspend using keyboard --- traced to another EC wakeup issue by pgf
  • build 24 (with kernel rpm), abbreviated scan, interlocked scan, aggressive timing, by James, all failed within 100 S/R cycles,
  • build 24 (with kernel rpm), blacklisted btmrvl_sdio only, no-wireless, aggressive, SKU295, hang after suspend, after 3018 cycles.
  • build 25, C2, with 8787 enabled, without no_console_suspend, hung at the 70th suspend cycle (last message was "Suspending console(s)...". Next try will be with no_console_suspend. Saving log file as runin-with-8787-without-n_c_s.log. by wmb.

Success

  • Huh ? We never declare success, only "in progress" or "failed"!

62abd44

  • stock kernel in build 24, along w. Q7B11 and EC 0.3.08
  • build 24, abbreviated scan, interlocked scan, aggressive timing, test ended prematurely with iw lockup, 50 suspend cycles on SKU295, 71 suspend cycles on SKU296,

e673868

  • stock kernel in build 23, along w. Q7B11 and EC 0.3.08
  • build 23, abbreviated scan, interlocked scan, 4306 suspend cycles, aggressive timing, by James, passed.
  • build 23, no-wireless and blacklisted mwifiex_sdio and btmrvl_sdio drivers, 14,800 suspend cycles, aggressive timing, SKU 297, by wad, passed