This is the mail archive of the
systemtap@sources.redhat.com
mailing list for the systemtap project.
Re: kprobes_stack.patch
amavin wrote:
> I thought the idea was to see how necessary a separate stack would
> be in a situation where we have hundreds of kprobes on a
> system. [...]
No, it was because kprobes can be theoretically triggered at any point
in the kernel. Only an unknown, and possibly tiny amount of kernel
stack space remains free for the kprobes infrastructure as well as the
handlers. So the problem is not that there may be many kprobes being
concurrently handled, nor that many may sit passively inserted. It is
that when any one runs, it could exhaust its stack without trying too
hard.
- FChE