Vision kit JOY sample stops working #346
Comments
|
I had the same issue. Ran Kernel oops continues even after upgrade |
|
More logs when i run the face detection demo . The errors happen even with the 20180418 image [ 82.575807] Stopping timer. |
|
I have the same issue but haven’t had the chance to check dmesg. I just bought it from Target today, and it seems to randomly freeze, although it usually works for a half a minute or so. Using a 2.1 amp supply. I hope this gets resolved soon. Stinks it’s not working after I just bought it. |
|
It just now froze in the midst of starting and pulsing the piezo, so the piezo beeped until I pulled the plug. Seems random. |
|
I’m starting to think it’s a hardware issue, because the longer I leave it off before starting it back up, the longer it works without freezing. I’m going to exchange it and see if that fixes it. |
|
I am seeing the same thing. Also leaving the back open after it's cooled
off makes it run longer. I am going to pull mine out of the box and put a
cpu fan on it
…On Sat, May 5, 2018, 9:39 AM Cody Rigney ***@***.***> wrote:
I’m starting to think it’s a hardware issue, because the longer I leave it
off before starting it back up, the longer it works without freezing. I’m
going to exchange it and see if that fixes it.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#346 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AKFuv_75Rw8sJbC8-A5yMGtkpMnELRIzks5tvat7gaJpZM4TqiA2>
.
|
|
Status update. I exchanged it at Target and the new one works flawlessly! It’s as if there are some that have an overheating issue. But the new one is fully enclosed and all, no mods, yet works fine. Shows me for always skipping the first item on the shelf for fear of buying a man handled product :D |
|
I bought 3 of them. 2 were faulty and I had to return it and the third one was fine. So it is definitely some marginal movidius hardware causing us grief. |
|
Hmm, that’s insightful. I wonder who’s in charge of production quality? Anyone from Google have any input? |
|
We are investigating the issue. But in re: production, the manufacturing line producing the units is the same for all kits. The chipsets from Intel Movidius have to pass the QA when they go off their manufacturing line. So I doubt Intel wouldnt catch a problem if there was one. On our production line we have manufacturing tests that cover all the aspects of the hardware. We test the boards before and after the components soldering, and have different set of tests for every part of the device functionality. |
|
Thanks for the insight @PeterMalkin. I'm guessing you guys test the SD cards as well? I know we've had an issue with faulty SD cards at a company I worked at previously. We also had issues where our product worked well in production testing, but then it would fail when the customer got it (it turned out to be a voltage spike randomly frying the board upon plugging it in). Not suggesting anything like that here. All that was to say, I have empathy for the grief this is probably causing you guys. |
|
@cmrigney, since we rely on raspberry pi to provide the power to the bonnet, I doubt it is likely we would have power spike issues. ESD is likely however - if a user accumulates a static charge during unboxing and zaps the board - that could cause damage. |
|
Hey, we have hard time reproducing this on our end reliably. If any of you happen to see this again - would you be willing to mail the kit over? I will mail back a new kit and a $20 gift card to compensate for shipping. That would really help us to get to the bottom of this. Thanks!! |
|
@saket424, @bryantqo what's the exact set up when you're running into those issues? @bryantqo, do you also have Pi connected to 2.1+A usb power supply? |
|
Yes power supply is sufficient. I will attempt to run the command however I
think the last time I ran it I ended up with a kernel panic which flat
lined the system. If I ssh into the board and stop the joy demo before the
crash the os keeps running.
HDMI: No
WiFi: Yes
Power: power micro USB
Disassembled: No
Facing the desk
…On Tue, May 8, 2018, 5:58 PM Leonid Lobachev ***@***.***> wrote:
@saket424 <https://github.com/saket424>, @bryantqo
<https://github.com/bryantqo> what's the exact set up when you're running
into those issues?
In particular:
Is display connected to mini hdmi port on Pi?
Is Pi connected to wifi and/or connected via micro usb data port?
What does your camera face? I.e. your face, monitor, wall?
Did you attempt to disassemble the kit? It's quite easy to damage vision
bonnet when removing white standoffs from it.
If that happens again could you run 'vcgencmd measure_temp' on Pi and
report back the output please?
@bryantqo <https://github.com/bryantqo>, do you also have Pi connected to
2.1+A usb power supply?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#346 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AKFuv71RqS1aZ8bBHdaJTpoaMot7_TkBks5twhUSgaJpZM4TqiA2>
.
|
|
I am experiencing the same thing. @PeterMalkin, I am willing to mail you my vision kit rather than exchange it at Target. (For Science. And for the community.) Overheating seems to be a likely culprit. I was debating on slapping heat sink on the zero to see if that would fix the problem, but will hold off until I hear back from you. In the meantime I'll work with the Voice Kit. Would love to try the Android Things setup. And DialogueFlow and Firebase and sill hafta try gVisor... So much to do, so little time! |
|
@ScottBriening |
|
I'm also having an issue with my Vision Kit where very shortly after starting the joy detection demo my ssh session becomes unresponsive. It seems related to the board getting too hot (which board?): I should note that I was seeing |
|
@dtreskunov What kind of You max temperature (53.0 C) is actually pretty low. In my experiments VisionKit was working fine when Pi was at 90.0 C. Everything below 85.0 C is considered normal I guess. |
|
@dmitriykovalev Here's the |
|
Running a stress test on my Raspberry Pi doesn't result in any sort of badness. I think it's a hardware problem with the Vision Bonnet. I'm planning to return my AIY Vision Kit for a replacement. |
|
@dtreskunov Could you try to add at the beginning of |
|
@dmitriykovalev Thanks so much! You just saved me a trip to the store. The problem seems to be gone after adding At idle, the CPU clock is 700MHz and voltage is 1.2V. While For my own reference, here's some documentation on how to take these measurements using the |
|
@dtreskunov Great, I'm glad this solution helped! @ScottBriening, thank you very much for sending your board to us for experiments, it has the same issue and the same fix removed the problem. We still need time to understand why :) It's interesting that you didn't use |
|
@dmitriykovalev @PeterMalkin YAY!!! I just sent you an email suggesting a similar solution, but it looks like you beat me to the punch. Nice work guys. Now let's make some mAgIc! |
|
@dmitriykovalev This is interesting indeed. So it's just the camera with pi0 combo that is the problem. I'm surprised this hasn't popped up more often / addressed by the folks at Pi. From the raspberrypi.org docs, the default over_voltage setting for the pi zero is 6 (1.35V) and over_voltage_min is 0. By setting over_voltage to 4 (1.3V) it should decrease the voltage supplied by 0.05V. Setting over_voltage_min to 4 should ensure the voltage supplied goes no lower than 1.3 when the system is idle. It's great that these settings allow cameras to work with pi0, but... Why? (As it says in the link you provided "I don't know what to believe".) Would love to tinker. Frying a couple pi0s isn't a big deal, but I certainly don't wanna lose the vision hat! Oh and you're welcome, of course. Happy to help! Got the new kit put together and so far so good. Oh, save the buzzer. Either I don't have enough joy or it's a dud. I'll get around to replacing it sometime. |
|
I had this problem but it's fixed based on the over_voltage fix. Thanks. |
|
Same issue here but not fixed with over_voltage. Returning soon. |
|
@phillipeloher If possible, can you provide any additional details about your crash? e.g. dmesg output |
|
I'm getting the same issue if I allow the joy detection demo to run after bootup. I have HDMI connected to a monitor and mini unpowered usb hub to connect my logitech wireless keyboard/trackpad (1 usb dongle connected). Same issue after applying the voltage fix to /boot/config.txt. dmesg output with kernel error right before it locked up below. I bought this from target about a month ago and just got around to playing with it today. Running apt-get upgrade to see if that helps.
|
|
Check out my comment to #418 in the forum. I was able to resolve the same type of problem at least for now by explicitly setting the CPU speed in config.txt to less than 1 GHz. Note: I had previously tried the over_voltage=4 "fix" that was suggested with no success. The joy app is working and I'm not getting any lockups of kernel oops failures. Hopefully this will help while Google tries to figure out if it is a bad batch of pi w boards or the bonnet that doesn't like to run at 1 GHz. |
|
@burtbick Thank you for the detailed comment! VisionBonnet has it's own SoC and clock source, so most likely that's a low level Pi issue (root cause is still unclear). Usually everything works just fine but some boards have issues. By the way, did you try |
|
Thanks Dmitriy, Yes, I tried the over_voltage=4 setting first. That didn't make ANY difference at all. I also tried adding over_voltage_min=4 as was suggested in another thread and that also didn't make any difference. The only thing that seems to help is setting the CPU speed to less than 1 GHz.. Note that the Pi W seems to keep running just fine at 1 GHz with other applications. It is only when the vision related applications are running that it starts choking. If I kill the joy demo then I can run the Pi W at full speed and it seems to work with no kernel oops lock ups. By dropping the Pi speed then the vision demos appear to work just fine, but I haven't run for more than an hour. With everything back in the closed box and running a video app the temp appears to stabilize out at 62.7 degrees C. When I get some more time I'll try swapping in a different Pi W and see if that makes any difference. And I plan on also trying a different SD card to see if that makes any difference. |
|
My kit was freezing too, but right after boot if I opened terminal or anything else. Here's what I've done:
sudo nano /boot/config.txt add the following lines to the end of the file: Ctrl+O (save); Ctrl+W (exit)
sudo apt-get dist-upgrade
Now it's flawless! |
|
@lufiaraujo Right, that solution works and already described in https://github.com/google/aiyprojects-raspbian/blob/aiyprojects/docs/vision.md. @burtbick Can you try to decrease CPU frequency (check #418 for details)? Looks like that also works sometimes. |
|
As per the image titled "aiyprojects-2018-11-16.img.xz" I can confirm that the original problem of the OS crashing is solved by adding the following to /boot/config.txt over_voltage=4 Do not run the upgrade afterwards. Don't run the upgrade before, either. If you upgrade raspbian you will lose the AIY software functionality with this AIY disk image. |
I have a new Vision Kit from Target that I assembled. After working initially for a few minutes, the vision kit JOY sample stops working with the following output in the dmesg log. I have the unit powered with a 5V 2.5 A powersupply. Any suggestions ?
[ 252.549751] Unregistered device pwm22
[ 317.266061] ------------[ cut here ]------------
[ 317.266115] kernel BUG at mm/slub.c:3873!
[ 317.266126] Internal error: Oops - BUG: 0 [#1] ARM
[ 317.266138] Modules linked in: fuse aiy_adc(O) industrialio gpio_aiy_io(O) pwm_aiy_io(O) cmac rfcomm bnep aiy_io_i2c(O) leds_ktd202x(O) hci_uart btbcm bluetooth aiy_vision(O) spidev usb_f_rndis u_ether usb_f_acm u_serial brcmfmac brcmutil snd_soc_bcm2835_i2s(O) regmap_mmio cfg80211 snd_soc_core rfkill snd_compress snd_pcm_dmaengine snd_pcm snd_timer i2c_bcm2835 snd spi_bcm2835 bcm2835_gpiomem uio_pdrv_genirq uio fixed pwm_soft(O) i2c_dev libcomposite dwc2 udc_core ip_tables x_tables ipv6
[ 317.266304] CPU: 0 PID: 259 Comm: python3 Tainted: G O 4.9.59+ #1047
[ 317.266315] Hardware name: BCM2835
[ 317.266328] task: d6806d00 task.stack: cccfa000
[ 317.266368] PC is at kfree+0x144/0x18c
[ 317.266406] LR is at kvfree+0x54/0x5c
[ 317.266419] pc : [] lr : [] psr: 40000013
sp : cccfbe18 ip : cccfbe40 fp : cccfbe3c
[ 317.266432] r10: 0000084b r9 : d8826000 r8 : c00eddd0
[ 317.266442] r7 : c00ee078 r6 : c01071b4 r5 : cccfbe50 r4 : cce13f9d
[ 317.266452] r3 : d7e58c4c r2 : 00000100 r1 : 00000100 r0 : cccfbe50
[ 317.266468] Flags: nZcv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user
[ 317.266479] Control: 00c5387d Table: 0cde4008 DAC: 00000055
[ 317.266490] Process python3 (pid: 259, stack limit = 0xcccfa188)
[ 317.266500] Stack: (0xcccfbe18 to 0xcccfc000)
[ 317.266513] be00: 00000001 c0914e10
[ 317.266529] be20: cce13f80 cce13f9d cccfbe34 cccfbe38 cccfbe4c cccfbe40 c01071b4 c0134e58
[ 317.266545] be40: cccfbe6c cccfbe50 c012514c c010716c cce13f80 00000000 000005d3 ae214018
[ 317.266561] be60: cccfbe7c cccfbe70 c0125240 c01250c8 cccfbe94 cccfbe80 bf413c20 c01251f8
[ 317.266576] be80: d509f400 00000000 cccfbefc cccfbe98 bf4154cc bf413bb0 00000000 d509f470
[ 317.266591] bea0: d509f480 ae214008 00000001 00000018 00003a98 00c00010 0000007e 00000000
[ 317.266608] bec0: d6806d00 c004ff58 cccfbec8 cccfbec8 00000000 ae214008 d6857190 d2ee4960
[ 317.266623] bee0: 00000005 00000005 cccfa000 00000000 cccfbf7c cccfbf00 c01575f0 bf414fdc
[ 317.266638] bf00: cccfbf24 c00703cc cccfbf4c cccfbf18 c00703cc c049d27c 80000000 00000009
[ 317.266653] bf20: 00000189 00000000 00000001 00000081 003eb198 c016211c cd3d5480 00c00010
[ 317.266671] bf40: ae214008 be9cdbd4 c0108903 00000005 cccfbf6c d2ee4961 ae214008 d2ee4960
[ 317.266686] bf60: c0108903 00000005 cccfa000 00000000 cccfbfa4 cccfbf80 c0157d6c c0157560
[ 317.266701] bf80: be9cdbd4 00c00010 00378000 be9cdbd4 00000036 c000ffc4 00000000 cccfbfa8
[ 317.266717] bfa0: c000fe40 c0157d34 00c00010 00378000 00000005 c0108903 ae214008 be9cdbd4
[ 317.266734] bfc0: 00c00010 00378000 be9cdbd4 00000036 00000005 c0108903 ae214008 be9cdbd8
[ 317.266749] bfe0: 003782fc be9cdbc4 000362d4 b6d9f8ac 80000010 00000005 17ffa861 17ffac61
[ 317.266789] [] (kfree) from [] (kvfree+0x54/0x5c)
[ 317.266827] [] (kvfree) from [] (__vunmap+0x90/0xe0)
[ 317.266850] [] (__vunmap) from [] (vfree+0x54/0x94)
[ 317.266897] [] (vfree) from [] (transaction_unref+0x7c/0xa4 [aiy_vision])
[ 317.266956] [] (transaction_unref [aiy_vision]) from [] (visionbonnet_ioctl+0x4fc/0x6a8 [aiy_vision])
[ 317.266994] [] (visionbonnet_ioctl [aiy_vision]) from [] (do_vfs_ioctl+0x9c/0x7d4)
[ 317.267014] [] (do_vfs_ioctl) from [] (SyS_ioctl+0x44/0x6c)
[ 317.267051] [] (SyS_ioctl) from [] (ret_fast_syscall+0x0/0x1c)
[ 317.267069] Code: 1a000003 e5932014 e3120001 1a000000 (e7f001f2)
[ 317.267085] ---[ end trace 8b8f08149c229173 ]---
The text was updated successfully, but these errors were encountered: