Making the timer emulation more robust

[bernd-ulmann]

The timer emulation has problems when a newly initialized timer is immediately deactivated again as the timer thread has not been fully established and thus the thread cancel fails.

[sy2002]

The goal would be to remove again this "jump over the nasty part" section in test_programs/timer_test.asm, i.e. remove the RBRA START, 1. It works in hardware, but not in the emulator, yet.

        RBRA    START, 1

        MOVE    0xFFFF, @R0
        MOVE    0xEEEE, @R1
        MOVE    0xDDDD, @R2
        MOVE    0xCCCC, @R3
        MOVE    0xBBBB, @R4
        MOVE    0xAAAA, @R5

        CMP     0xDDDD, @R2
        RBRA    A_HALT, !Z
        MOVE    0x0000, @R2     ; stop timer 0
        CMP     0xAAAA, @R5
        RBRA    A_HALT, !Z
        MOVE    0x0000, @R5     ; stop timer 1
        CMP     0xFFFF, @R0
        RBRA    A_HALT, !Z
        CMP     0xEEEE, @R1
        RBRA    A_HALT, !Z
        CMP     0xCCCC, @R3
        RBRA    A_HALT, !Z
        CMP     0xBBBB, @R4
        RBRA    A_HALT, !Z

START