Our sed v2 (sed2; code: ch5/sed2/) mini-project is very similar to our sed1 project. The key difference is that this time, we'll carry out the "work" via a kernel thread created by the driver for just this purpose. The key differences between this version and the previous one are as follows:
- There's just one global shared memory buffer for holding the metadata, along with the payload; that is, the message to encrypt/decrypt. This is the struct sed_ds->shmem member within our driver context structure, struct stMyCtx.
- The work of encryption/decryption is now performed within a kernel thread (that this driver spawns); we keep the kernel thread asleep. Only when work arises does the driver wake up the kthread and have it consume (execute) the work.
- We now run the kernel timer within the kthread's context and show if it expires prematurely (indicating that the deadline wasn't met).
- A quick test reveals that eliminating...
