Quick fix in .96, but breaks at least some versions of strace. Changed
in .98 - strace works again, but I think there may be a race condition
if read() passes the checks, sleeps on memcpy_tofs() due to a swap-in,
and the target process execs a setuid program in this time window.
> If I revamp my /proc-paranoia patch for 1.3.x, will it get accepted,
> or should I not bother ?
Ask Linus :-). He might not like encoding the flag in the sticky bit of
/proc - maybe add a struct proc_sb_info containing mount flags (there
might be more in the future) like other filesystems do.
Marek