Hi Andreas, Am 18.02.2023 um 20:59 schrieb Andreas Schwab:
On Feb 18 2023, Michael Schmitz wrote:Not sure it's wise to allow init to abort though.When PID 1 exits, the kernel panics.
I might have been a mite subtle there ...Anyway - Finn's intention was to log signals caused by the stack protector mechanism. Reading the description of the abort() function, it appears that allowing init to take the signal through a handler that returns, abort() restores the default handler and retries. That might just manage to abort init after all (I haven't found where PID 1 is treated special in the kernel signal code to rule that out).
Logging ABRT in the kill() syscall path seems much the safer option. I'll see what I can come up with.
Cheers, Michael