Lock or unlock a capability

A capability whose type matches the source-type subfield of a given lock, and has at least the same access bits on as in the access field check bits subfield of the lock, may be converted to a capability with:

• those access-bits on which are on in the given lock,
• type as given in the target-type subfield of the given lock,
• datum part unchanged.

This is referred to as locking the capability with the given lock. Similarly, a capability may be unlocked if its type matches the target-type subfield of a given key. We also require that the access bit field of the capability have at least the same bits on as the access field of the key. The resulting capability will have:

• access-bits as in the access field check bits subfield of the key
• type given as source-type in the key
• datum part unchanged

Thus, the condition:

• any two capabilities, of type either lock or key, with the same target type, have identical source types and access field check bits;

will assure that any capability resulting from an unlock operation will have previously existed, possibly with more access bits on. (We explain below how to assure this condition.) If the MOT-index unique name representation for segment capabilities is used, then a locked segment capability will contain only the MOT-index and unique name. Unlocking such a capability will cause the base-bound pair to be re-computed. Figure 7 contains an example of locking and unlocking a capability.

 

Figure 7: Example of locking and unlocking.

14#14