ST20–SWC

14/21

4

Execution analysis tools

Three profiling tools are supplied for analyzing the behavior of application programs; the execution

profiler, the utilization monitor, and the test coverage and block profiling tool. The monitoring data is

stored in the target processor ’s memory, so the profiling tools have little execution overhead on the

application. After the program has completed execution, the monitoring data is extracted from the

processor and is analyzed to provide displays on the program execution.

The tools provided are the execution profiler, the utilization monitor, and the test coverage and block

profiling tool. The execution profiler estimates the time spent in each function and procedure, the

processor idle time and various other statistics. The utilization monitor displays a Gantt chart of the

CPU activity of the processor as time progresses. The test coverage and block profiling tool counts

how many times each block of code is executed.

4.1

Execution profiler

The execution profiler gives an analysis of the total time spent executing each function on each

processor.

It provides the following information on program execution:–

The percentage time spent executing each low priority function.

The percentage time spent executing at high priority.

The percentage idle time of the processor.

The number of low priority calls of each function and where it was called from.

Processor ”Root”

Idle time 35.3% (19516)

High time 0.1% (37)

Wptr Misses 0

Iptr Misses 0

Resolution 4

-----------------------------------------------------------------------------

Process ”ex” (99.9% processor) (35.666s)

Stack 100.0% (35666)

Heap 0.0% (0)

Static 0.0% (0)

Function Name

| Process | Processor |Samples

-----------------------------------------------------------------------------

libc.lib/getc

|

11.4 |

11.4 |4081

cc/pp.c/pp_rdch0

|

10.1 |

10.1 |3605

cc/bind.c/globalize_memo

|

6.9 |

6.9 |2467

cc/pp.c/pp_process

|

4.3 |

4.3 |1525

cc/pp.c/pp_rdch3

|

4.2 |

4.2 |1497

cc/pp.c/pp_rdch2

|

3.9 |

3.9 |1380

cc/pp.c/pp_rdch1

|

3.8 |

3.8 |1354

cc/pp.c/pp_rdch

|

3.5 |

3.5 |1252

cc/pp.c/pp_nextchar

|

3.3 |

3.3 |1189

cc/pp.c/pp_checkid

|

3.2 |

3.2 |1150

cc/lex.c/next_basic_sym

|

2.7 |

2.7 |979

libc.lib/strcmp

|

2.3 |

2.3 |812

libc.lib/DummySemWait

|

2.2 |

2.2 |784

libc.lib/sub_vfprintf

|

1.7 |

1.7 |617

Figure 10

Example output from the execution profiler

The execution of the user program is monitored by a profiling kernel. The presence of the profiler

kernel will slow the execution of the program by less than 5%.

4.2

Utilization monitor

The utilization monitor shows in graphical form the utilization of the processor over the time of the

program execution. This is displayed by an interactive program that draws a chart of processor