| |||
Home > Functional Description > Functional operation > Using locked transaction sequences |
If a master performs locked transaction sequences, a transaction might stall, or an AXI protocol violation might occur when:
If a locked transaction sequence crosses a 4KB boundary and the regions have different region permissions, the TZASC might prevent access to the second region and therefore the slave would not receive the latter part of the locked transaction sequence.
The AXI protocol recommends that locked transaction sequences do not cross a 4KB address boundary.
During a locked transaction sequence, if a master changes the state of arprots[1] or awprots[1] and the region has different region permissions for Secure state and Non-secure state, the TZASC might deny a transaction and therefore the slave would not receive the latter part of the locked transaction sequence.
During a locked transaction sequence, if a master performs reads and writes to a region, depending on the region permissions, the TZASC might deny a transaction and therefore the slave would not receive the latter part of the locked transaction sequence.