search

PhotonVision / photonvision

PhotonVision is the free, fast, and easy-to-use computer vision solution for the FIRST Robotics Competition.
https://photonvision.org
GNU General Public License v3.0
263 stars 184 forks source link

PhotonVision OjectDetection camera sometimes missing on boot #1273

Closed gartaud closed 6 months ago

gartaud commented 6 months ago

Describe the bug This is a follow up from this tread on Chief Delphi: https://www.chiefdelphi.com/t/photonvision-ojectdetection-camera-sometimes-missing-on-boot/458083 (note: I am not the OP but we observe similar symptoms, so copied original text herehunder)

We have PhotonVision running on an Orange Pi 5 with one AprilTag and one ObjectDetection camera. We have an intermittent issue where the ObjectDetection pipeline doesn’t work on boot.

To Reproduce The code reports the camera isn’t present and there is no data in the network table. When we check the web UI, it shows the yellow spinning image for that pipeline and the camera never comes up. The camera also does not appear on the Camera Settings page.

Screenshots / Videos See CD thread

Platform:

Additional context In our case we think that things were working fine for days before everything fell apart. We did not knowingly change anything that could explain the change.

mcm001 commented 6 months ago

Duplicate of #1182.

gartaud commented 6 months ago

Hi. I'm not sure why this issue was closed and am appealing the decision.

https://github.com/PhotonVision/photonvision/issues/1182 talks about UI crashing "after a few minutes of using object detection".

While a few posts under the CD thread do reference symptoms that fully align with the ones described in 1182, the main symptoms described in the top CD post in that thread clearly mention that the "object detection camera [is] sometimes missing on boot".

"on boot" does not highly correlate with "a few minutes".

Please confirm that you think 1273 and 1182 are duplicate.

mcm001 commented 6 months ago

All the dmesg logs we've seen in Discord have shown the same symptoms at failure (something similar to this):

[ 2087.275215] Unable to handle kernel paging request at virtual address ffff000000000000
[ 2087.276237] Mem abort info:
[ 2087.276770]   ESR = 0x96000146
[ 2087.277365]   EC = 0x25: DABT (current EL), IL = 32 bits
[ 2087.278138]   SET = 0, FnV = 0
[ 2087.278718]   EA = 0, S1PTW = 0
[ 2087.279310] Data abort info:
[ 2087.279873]   ISV = 0, ISS = 0x00000146
[ 2087.280541]   CM = 1, WnR = 1
[ 2087.281128] swapper pgtable: 4k pages, 48-bit VAs, pgdp=0000000001d91000
[ 2087.282066] [ffff000000000000] pgd=00000000efff8003, p4d=00000000efff8003, pud=00000000efff7003, pmd=0000000000000000
[ 2087.283383] Internal error: Oops: 96000146 [#1] SMP

If this isn't the same symptoms as the issue you're describing, we can reopen for tracking. I'd prefer we keep things in one place for now though.

gartaud commented 6 months ago

Ok, thanks.

I'm not sure if you noticed the suggestion to run https://scan.coverity.com/ for free on your codebase in the linked CD thread. That should not take too long to put in place and might point out to things in your code that you oversaw.