Hi!

Generally speaking, in sigreturn, we take the saved machine state (as
before signal delivery) from the user stack, and we restore it. Usually
this machine state was saved by the kernel (sendsig), but the user
process may tweak it.

Hence, we usually check that the user process does not try to gain
improper privileges by modifying the signal context.

There is no such check in MIPS version of sigreturn, but the comment
states that such checks should be done. Are such checks useless on MIPS?
Why?

-- 
Emmanuel Dreyfus
manu@netbsd.org