priority inversion

A phenomenon which can arise in a {concurrent}

programming environment where a high priority task (H) is

blocked by a low priority task (L), e.g. because L has locked

some resource needed by H, and L then has to wait for a medium

priority task (M). The net result is that H ends up waiting

for M instead of the other way round - the priorities become

inverted.

This can be a problem if, for example, M takes a long time,

causing H to miss a deadline.

A possible cure is to have tasks inherit the maximum priority

of any task that is waiting for them. In that case L

temporarily becomes high priority until H can procede, thus

preventing M from running in place of H.

(1997-12-13)