we now access the user buffer in randomread() outside of the lock,
only copying and advancing the chacha state under the lock. this
means we can use randomread() within the fault handling path now
without fearing deadlock. this also allows multiple readers to
generate random numbers in parallel.
we might wake up on a different cpu after the sleep so
delta from machX->ticks - machY->ticks can become negative
giving spurious timeouts. to avoid this always use the
same mach 0 tick counter for the delta.
the manpage states that capabilities time out after a minute,
so we add ticks field into the Caphash struct and record the
time when the capability was inserted. freeing old capabilities
is handled in trimcaps(), which makes room for one extra cap
and frees timed out ones.
we also limit the capuse write size to less than 1024 bytes to
prevent denial of service as we have to copy the user buffer.
(memory exhaustion).
we have to check the from user *before* attempting to remove
the capability! the wrong user shouldnt be able to change any
state. this fixes the memory leak of the caphash.
do the hash comparsion with tsmemcmp(), avoiding timing
side channels.
allocate the capabilities in secret memory pool to prevent
debugger access.
previously, right shift >>0 resulted in >>32 being emited. this
is especially problematic when the shift count comes from a macro
expansion.
we now handle constant shift >>0 as <<0 (no shift) and allow
shift by 32 be specified.
this applies to logical right shift (>>) arithmetic right shift (->)
and right rotate (@>).
the prior implementation was unneccesarily complicated for
no good reason due to me misunderstanding how libc's nrand()
works. in contrast to libc, we already generate the *closest*
power-of-2 random number with mprand() in the sampling loop.
