Bug 13420

Summary: When moving from linux-yocto 5.0.7 to linux-yocto 5.0.13, kernel boot hangs after "Starting kernel ..."
Product: [Yocto Project Subprojects] Kernel Reporter: Matthias Schoepfer <matthias.schoepfer>
Component: linux-yoctoAssignee: Matthias Schoepfer <matthias.schoepfer>
Status: RESOLVED INVALID QA Contact:
Severity: major    
Priority: Medium CC: kevin.tian, randy.macleod
Version: 2.7.2   
Target Milestone: 2.8   
Hardware: Other   
OS: arm64   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: No (bug/feature does not impact docs)

Description Matthias Schoepfer 2019-06-26 13:22:19 UTC
We are using yocto with the dragonboard 410c. Linux-yocto works with version 5.0.7, but when updating poky to commit 0cc214e2d25eb59ef36fa39145097f7dcc30830a and beyond, the resulting Kernel will hang right after u-boot passed the control to the kernel (last message is "Starting kernel ..."

I am well aware, that this might be an upstream bug in the kernel?!
Comment 1 Matthias Schoepfer 2019-06-27 14:22:16 UTC
Okay, so, if I devtool modify linux-yocto, the kernel will run. If I do not do it, kernel will hang. Must have to do with some of the magic with yocto-kernel-cache, I guess?
Comment 2 Bruce Ashfield 2019-07-17 14:20:30 UTC
I honestly don't have any ideas about this. I'm not seeing any boot issues on any of my boards.

devtool modify + rebuild shouldn't have had an impact. If you are still seeing this behaviour, can you provide a diff of the configs that are produced ?
Comment 3 Matthias Schoepfer 2019-07-23 11:50:30 UTC
Actually, I am quite puzzled. I have spend a week with this. Kernel size from devtool or regular build match, difference is only few bytes, that are for example build time or a path. Same with the fitImage. But there must be something broken right when the jump into the kernel happens, as not a single line of kernel messages is printed. 

Basically, I sticked to 5.0.7 for now, need to get some other work done before. 

Its okay with me if you close this issue. If I ever can come up with more insight, I will add a comment here...
Comment 4 Randy MacLeod 2019-07-25 14:50:46 UTC
Closing as per Mattias's suggestion. Reopen if the situation become more clear.