Hi to everyone,
I'm experiencing a strange behaviour on some STCK1A8LFC device.
At our software house we developed a Qt5 based digital signage solution running under Ubuntu Linux.
This software works perfectly under Ubuntu 14.04.3 on another intel based machine, powered by Intel HD Graphics driver.
Under STCK1A8LFC we experience random freezes of HDMI output (note, just the output is frozen because the device stills respond to network requests and if I capture a screenshot using "xwd" command the screenshot is correctly showing the image that should be on the screen). The screen looks frozen, but the device still work correctly. The problem happens randomly after a few days of uninterrupted working; I can alto cause it (but not every time) if I disconnect and reconnect the HDMI cable, or if the monitor itself goes into power-saving mode. The defect happens more frequently if we use VAAPI acceleration.
We tried everything to avoid this strange behaviour; the only one solution we found is a reboot every day, but the application should run h24 without interruption. On other platforms (eg: Intel NUC) everything works correctly.
We are using kernel 3.16.0-57-generic from Ubuntu.
Under Xorg we enabled SNA acceleration and TearFree mode, but the problem still exists with TearFree disabled.
Xorg logs about the driver shows this:
[ 12.836] (II) Module intel: vendor="X.Org Foundation"
[ 12.836] compiled for 1.16.0, module version = 2.99.914
[ 12.836] Module class: X.Org Video Driver
[ 12.836] ABI class: X.Org Video Driver, version 18.0
[ 12.836] (II) intel: Driver for Intel(R) Integrated Graphics Chipsets:
i810, i810-dc100, i810e, i815, i830M, 845G, 854, 852GM/855GM, 865G,
915G, E7221 (i915), 915GM, 945G, 945GM, 945GME, Pineview GM,
Pineview G, 965G, G35, 965Q, 946GZ, 965GM, 965GME/GLE, G33, Q35, Q33,
GM45, 4 Series, G45/G43, Q45/Q43, G41, B43
[ 12.837] (II) intel: Driver for Intel(R) HD Graphics: 2000-6000
[ 12.837] (II) intel: Driver for Intel(R) Iris(TM) Graphics: 5100, 6100
[ 12.837] (II) intel: Driver for Intel(R) Iris(TM) Pro Graphics: 5200, 6200, P6300
[ 12.837] (--) using VT number 7
[ 12.855] (II) intel(0): Using Kernel Mode Setting driver: i915, version 1.6.0 20080730
[ 12.855] (II) intel(0): SNA compiled for use with valgrind
[ 12.856] (--) intel(0): Integrated Graphics Chipset: Intel(R) HD Graphics
[ 12.856] (--) intel(0): CPU: x86-64, sse2, sse3, ssse3, sse4.1, sse4.2
Thanks to everyone who can help.
I have discovered another fact.
If, with the monitor frozen, I use xrandr to set it off via the
xrandr --output HDMI1 --off
and then on with
xrandr --output HDMI1 --auto
the system return to work without problem and screen "unfreezes".
Thank you for joining the Intel communities.
So at this point is your system working fine or you are still having the problem?
Thank you for the answer.
System is working fine for about 24-36 hours, but after I need to restore HDMI output by using the 2 xrandr commands I told before.
The same problemi happens if I disconnect the HDMI cable during accelerated video decoding; if I reattach the cable video appears frozen and I need the sequence of xrandr commands to restore it.
This is disappointing for a Digital signage system, specially because I cannot detect this situation (nothing logged in system.log, xorg.0.log, kernel.log) without watching the video...
I can add also that If I try to "unfreeze" the system via the "xrandr --output HDMI1 --off" command, in syslog this error appears_
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838557] WARNING: CPU: 2 PID: 1229 at /var/lib/dkms/oem-audio-i915-baytrail/0.20150605/build/i915/intel_display.c:3324 intel_crtc_wait_for_pending_flips+0x171/0x180 [i915]()
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838568] Modules linked in: ctr ccm hci_uart(OE) rfcomm arc4 bnep bluetooth 6lowpan_iphc nls_iso8859_1 gpio_keys ath9k_htc joydev ath9k_common intel_rapl ath9k_hw intel_powerclamp hid_multitouch ath coretemp kvm_intel kvm crct10dif_pclmul mac80211 crc32_pclmul cfg80211 aesni_intel aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd mei_txe lpc_ich hdmi_audio(OE) mei snd_seq_midi snd_seq_midi_event snd_rawmidi iosf_mbi snd_soc_rt5640 snd_soc_rl6231 snd_soc_core snd_compress snd_pcm_dmaengine i2c_hid snd_pcm snd_seq snd_seq_device snd_timer dw_dmac dw_dmac_core soc_button_array snd soundcore parport_pc ppdev i2c_designware_platform acpi_pad i2c_designware_core snd_soc_sst_acpi mac_hid spi_pxa2xx_platform 8250_dw lp parport pwm_lpss btrfs xor raid6_pq hid_generic usbhid hid mmc_block i915(OE) i2c_algo_bit drm_kms_helper drm video sdhci_acpi sdhci lz4 lz4_compress
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838846] CPU: 2 PID: 1229 Comm: Xorg Tainted: G W OE 3.16.0-57-generic # 77~14.04.1-Ubuntu
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838858] Hardware name: Intel Corporation STCK1A8LFC/STCK1A8LFC, BIOS FCBYT10H.86A.0031.2015.1027.1417 10/27/2015
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838868] 0000000000000009 ffff8800390bfa80 ffffffff8176891e 0000000000000000
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838888] ffff8800390bfab8 ffffffff8106ee9d 0000000000000000 ffff880031383000
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838906] ffff880035948230 ffff8800358aa000 ffff8800358aa000 ffff8800390bfac8
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838925] Call Trace:
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838953]  dump_stack+0x45/0x56
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838977]  warn_slowpath_common+0x7d/0xa0
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.838997]  warn_slowpath_null+0x1a/0x20
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839110]  intel_crtc_wait_for_pending_flips+0x171/0x180 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839132]  ? prepare_to_wait_event+0x100/0x100
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839245]  intel_crtc_disable_planes+0x33/0x1c0 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839264]  ? delayacct_end+0x80/0x90
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839376]  i9xx_crtc_disable+0x51/0x3f0 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839488]  __intel_set_mode+0x2a7/0x1590 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839600]  ? intel_crtc_set_config+0x112/0xdb0 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839712]  intel_set_mode+0x16/0x30 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839824]  intel_crtc_set_config+0x8ed/0xdb0 [i915]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839901]  drm_mode_set_config_internal+0x61/0xe0 [drm]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.839977]  drm_mode_setcrtc+0xe1/0x570 [drm]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840038]  drm_ioctl+0x1ec/0x660 [drm]
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840067]  do_vfs_ioctl+0x2e0/0x4c0
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840088]  ? restore_altstack+0x19/0x30
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840106]  SyS_ioctl+0x81/0xa0
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840127]  system_call_fastpath+0x1a/0x1f
Jan 22 10:00:58 sipario-STCK1A8LFC kernel: [171801.840139] ---[ end trace 670bf2c09190389d ]---
I can repeat the process every time the display is frozen and the error is always the same.
Maybe you can try updating the bios.
Please download the latest BIOS version for this unit at:
https://downloadcenter.intel.com/download/25506/BIOS-Update-FCBYT10H-86A- Download BIOS Update [FCBYT10H.86A]
In this case I recommend doing a BIOS recovery and you can follow the steps at:
http://www.intel.com/content/www/us/en/support/boards-and-kits/intel-compute-stick/000005902.html Intel® Compute Stick BIOS Recovery Instructions
Please let me know if that can help you.
Please try with the recovery method:
If the problem persists, submit a web ticket for warranty replacement at the following link:
Thank for your answer; I've already tried to restore device to factory defaults; the problem still exists.
I think I could ask Intel for a replacement unit, but the problem still exists on more than 10 devices... so I'm almost sure it's a software (video drivers ?) problem..
Is there something else I could try to solve it ?
I apologize for this inconvenience and I hope you can have a better experience with the Intel® Compute Stick from now on.
At this point I suggest requesting a warranty replacement for the units; you can submit a web ticket at the following link:
I've opened a case with Intel, but after the first questions they stopped responding telling they are working on the problem...
In the meanwhile I've done some tests by myself (since out business is seriously impacted by this problem); I've found that the ICS, when used with a MicroSD card and a simple USB device (eg: an ethernet card) is under-powered. So HDMI freeze problems start to appear.
I've found a mitigating factor by replacing the power-supply shipped with the ICS with a more powerful one (5V-3A). This reduces (but doesn't eliminate) the problem.
Note that this behaviour is common to many different ICS we bought; so it seems an hardware design fault.
It's very annoying and force us to think again about the hardware platform for our digital-signage project, avoiding Intel hardware for HDMI output since it's unrealiable.
Thank for the support,
I'm sorry how do you have connect your hardware to the Intel® Compute Stick, could you please provide us with more detailed information about that?
thank you for your answer.
The Compute Sticks are directly connected (in most cases without any cable or using the supplied HDMI cable) to monitors.
The typical setup is equipped as this:
2) 8GB microSD HC card
3) (optional) USB ethernet card (D-Link DUB e-100)
I've found that:
1) On a system with just ICS and MICROSD from Trascend and Samsung (EVO PRO series) the HDMI output freeze very often
1b) If I replace the supplied power-adaptor with a more powerful one (eg: 5V-3A) the defect seems to disappear (the device worked for about 5 days without a freeze, while in the "Normal" configuration the freeze appeared randomly in the first 2 days of working)
1c) If I replace the MicroSD with a "slower" one from Samsung (the same used in test 3) the problem seems to disappear (the device worked for about 3 days without a freeze).
2) On a system with ICS, MICROSD from Samsung, ethernet USB card and "powerful" adapter the problem still appear almost daily
3) On a system with ICS, MICROSD from Samsung, but low performance series (25 Mbyte/sec), ethernet USB card and standard power-supply the problem appeared just 1 time in 4 days
The only solution I've adopted in our software is to periodically (sigh!) reset the video card with an
echo 1 > /sys/kernel/debug/dri/0/i915_wedged
This partially solves the problem, but leave the device in a frozen state for the time between the "freeze" and the manual reset of card.
Our application launches different X processes (to keep them timing-independent) which play 1 video, show an animated web page and bitblt some pictures.
I just need a way to detect freeze and reset the video card, but every inspection in sysfs files shows that everything is working correctly during freeze :-(
If, for example, I capture a screenshot with "xwd" during the freeze, it shows the correct image that the screen should show... But the screen is frozen... After a "wedge" reset it starts working again without problems.
IMHO there's a power design problem, but I hope Intel support will solve it.
Is the problem occurring with balanced or performance mode or both? I'm facing the same problem but only with certain Samsung professional grade screens. On consumer grade TV-s the problem has not occurred.
He could try that to see how it goes, LucaBoss you can also access the BIOS and set the option for power mode to performance or balance to see what will be the behavior.
Press F2 to access the BIOS and you will find power mode under configuration.
Hi to everyone.
I've finally found the culprit; it is the Ubuntu unattended upgrade service.
It spots randomly to download and apply automatically updates from Ubuntu servers, freezing the video card (I don't know yet if it's due to high cpu load or problems with notifications GUI). I've tried to manually stress CPU with the "stress" command: output temporarily freezes, but after a while it starts again so I don't think it's related to high cpu usage. I haven't the time now to investigate the "notification" process (which popup the window telling there are updates to install) since the problem seems solved for me (monitors are operating from more than 100 hours without reboot).
If I disable the service and remove che ubuntu-release-upgrade checker scripts, the players works flawlessly.
Note that I'm using a very simple GUI configuration: lightdm is disabled, my script just starts (in order) X, compositing window manager and my player application (which uses the whole screen size).
Thank you for your help,
I talked too early; today, with a Philips Signage Grade monitor we experienced problem again.
The classic "i915_wedged" periodic fix partially mitigated the problem, but causes random kernel panic (which we manage via the watchdog which hard-reset the device).
Note that the problem (which was happening almost daily on some kind of TV set) has been reduced in frequency by removing the auto-update procedure. so I can bet it's high-CPU usage related.
Have you tried setting the kernel parameter intel_idle.max_cstate=1? This should fix some of the stability issues on baytrail computers. I haven't tested it on the field yet but with the parameter set the stick worked for 3 days without any crashes. I'm using gstreamer for video (GPU acceleration enabled) and webkitgtk for web.
Thank you for your answer;
in the past I tried with the parameter, but after a while (tipically the 2nd day) the problem appeared. It seems related to HDMI signaling because the video-card isn't in hang state (frame sequence numbers increases) and the device responds to network interaction; it's just the HDMI part which freezes.
I've tried to kill the compositor during the freeze (I was thinking it was the culprit) without luck; note that with Sharp Professional grade monitor we don't experience ANY problem; the problem is just with TV set and Philips Signage Grade monitors. With Philips monitor we have a specific installation (2 of them) where the system freezes almost daily. With a periodic (15 minutes) i915_wedged write, however, everything works but I have to manage sporadic system hangs with the hardware watchdog. It's a bit noisy to see the screen "frozen" and, after a while, a reboot of the device..
Maybe Intel changed something in recent bioses to handle power-up with HDMI disconnected; I don't know if this could be the cause.