cocalls - second shm calls causing kernel panic

[simonarnell]

Originally raised on Slack, logging here for completeness.

[Currently] experimenting on morello with cheribsd off of the cocalls branch. We’ve built an image that includes [@paulmetzger]'s recent [(3ffdfca0c44fbaaf789ddfd8495b13024f4751ef)] fix for coprocess switcher. However we’re hitting a kernel panic when running our code, we’ve created a reproducer (tarball) to give the simplest case we could think of that demonstrates this. Whilst the code runs fine on the first attempt, on the second attempt we receive a kernel panic. Any thoughts?

Kernel Panic backtrace:

In Slack @kwitaszczyk commented:

7d8cad5e2b54967c3d95ace1e1a8d29ab958ac0e from [@bsdjhb] introduces this check in the cocalls branch (originally coexecve) Based on that commit message, I wonder what's the state of shm in the cocalls branch. cc @brooksdavis @trasz.

To which @trasz responded:

I’m not sure I ever tested this tbh.