On Mon, Jun 19, 2023 at 02:23:47PM +0200, Greg KH wrote:
On Mon, Jun 19, 2023 at 11:11:16AM +0000, Janne Huttunen (Nokia) wrote:
We recently updated an internal test server from kernel v4.19.273 to v4.19.286 and since then it has already multiple times triggered a kernel panic due to a hard lockup. The lockups look e.g. like this:
Does this also happen on newer 5.4 and 6.4-rc7 releases?
If I am correct, no, it will not happen there. I was surprised
I believe the change that causes this issue was:
b4a34aa6d "ipmi: Fix how the lower layers are told to watch for messages"
which was backported into 4.9. But there was a locking change:
e1891cffd4c4 "ipmi: Make the smi watcher be disabled immediately when not needed"
that came in 5.1. That was required for b4a34aa6d to work correctly, and I didn't realize that. If Janne tests that patch (and the other fix for it), then kernels 5.1 and later are good as they are.
Thanks,
-corey
thanks,
greg k-h