This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: kprobe fault handling
hunt wrote:
> [...]
> function copy_more:long(addr:long)
> %{
> THIS->__retvalue = _stp_copy_from_user (buf,
> (char __user *)(long)THIS->addr, 2048);
> %}
> [...]
As mentioned several times, this runtime routine "_stp_copy_from_user"
is not safe, and should be rewritten or removed.
This is a separate matter from the kprobes fault handler doing the
right thing. Once it works again the way it did in dprobes, an access
fault will most likely be made to cause a probe handler abort, not
just a polite local -EFAULT result code.
- FChE