30 SCM common problems and solutions!

One 、 Problem recurrence

Stable recurrence of the problem can correctly locate the problem 、 Solve and verify . Generally speaking , The easier it is to reproduce the problem, the easier it is to solve .

1.1 Simulated recurrence conditions

Some problems exist under specific conditions , You only need to simulate the problem conditions to reproduce . For conditions that rely on external input , If the conditions are complex and difficult to simulate, you can consider entering the corresponding state directly by default in the program .

1.2 Increase the frequency of relevant tasks

For example, if an exception occurs only after a task runs for a long time, the execution frequency of the task can be increased .

1.3 Increase the test sample size

An exception occurred after the program ran for a long time , It's hard to repeat the problem , The test environment can be built, and multiple sets of equipment can be tested at the same time .

Two 、 Problem location

Narrow down the scope of investigation , Confirm the task of introducing the problem 、 function 、 sentence .

2.1 Print LOG

According to the phenomenon of the problem , Add... To the code in question LOG Output , To track the program execution process and the value of key variables , Observe whether it is consistent with expectations .

2.2 Online debugging

Online debugging can play a role in and printing LOG Similar effects , In addition, this method is particularly suitable for troubleshooting program crash classes BUG, When the program falls into an abnormal interrupt (HardFault, Watchdog interrupt, etc ) You can directly STOP see call stack And the value of the kernel register , Quickly locate problem points .

2.3 Version rollback

When using the version management tool, you can locate the version that introduces the problem for the first time by constantly backing back the version and testing and verification , After that, you can check the code added and modified in this version .

2.4 Dichotomous notes

The second note is Comment out part of the code in a way similar to binary search , To determine whether the problem is caused by the commented out part of the code .

The specific method is to comment out half of the code irrelevant to the problem , See if the problem is solved , If it is not solved, note the other half , If the problem is solved, continue to narrow the scope of the comment in half , And so on, gradually narrow the scope of the problem .

2.5 Save kernel register snapshot

Cortex M When the kernel gets into an abnormal interrupt, it will push the values of several kernel registers onto the stack , Here's the picture ：

Uploading … Re upload cancel

We can write the value of the kernel register on the stack to... When we fall into an abnormal interrupt RAM In the area where the default value is retained after a period of reset , Perform the reset operation and then start from RAM Read out and analyze this information , adopt PC、LR Confirm the function executed at that time , adopt R0-R3 Analyze whether the variables processed at that time are abnormal , adopt SP Analyze whether there may be stack overflow, etc .

3、 ... and 、 Problem analysis and treatment

Analyze the cause of the problem by combining the problem phenomenon and the location of the problem code .

3.1 The program continues to run

3.1.1 The value is abnormal

3.1.1.1 Software problems

1、 An array

Subscript exceeds array length when writing array , The corresponding address content is modified . as follows ：

Such problems usually need to be combined with map Document analysis , adopt map The file observes the array near the address of the tampered variable , Check whether there is unsafe code as shown in the figure above when writing to the array , Change it to safe code .

2、 Stack overflow

0x20001ff8	g_val
0x20002000	At the bottom of the stack
…………	Stack space
0x20002200	To the top of the stack

Pictured above , Such problems also need to be combined with map Document analysis . Suppose the stack grows from high address to low address , If stack overflow occurs , be g_val The value of is overwritten by the value on the stack .

In case of stack overflow, analyze the maximum usage of the stack , Too many function call layers , Interrupt the function call in the service function , Large temporary variables declared inside the function may lead to stack overflow .

There are the following ways to solve such problems ：

• Memory resources should be allocated reasonably in the design stage , Set the appropriate size for the stack ;

• Add... To the larger temporary variable in the function ”static” Keywords are converted to static variables , Or use malloc() Dynamic allocation , Put it on the pile ;

• Change the method of function call , Reduce the number of call layers .

3、 Judge whether the sentence condition is written incorrectly

Judging the condition of a statement is easy to put the equality operator “==” Write as assignment operator “=” Causes the value of the variable to be judged to be changed , This kind of error will not be reported at compile time and always returns true .

It is recommended to write the variable to be judged to the right of the operator , In this way, an error will be reported during compilation when it is written as an assignment operator . You can also use some static code checking tools to find such problems .

4、 Synchronization problem

For example, when operating a queue , An interruption occurred during the execution of the out of queue operation ( Task switching ), And interrupt ( Task after switching ) The queue structure may be damaged if the queue entry operation is performed in , For this kind of situation, you should turn off the interrupt during operation （ Use mutex to synchronize ）.

5、 optimization problem

Uploading … Re upload cancel

As shown in the above figure, the program , The original intention is to wait irq No more execution after interruption foo() function , But after being optimized by the compiler , During actual operation flg May be loaded into a register and determine the value in the register each time without re starting from ram Read in flg Value , Cause even irq Interruption occurs foo() Has been running , Here we need to be in flg Add... Before your statement “volatile” keyword , Force every time from ram get flg Value .

3.1.1.2 Hardware problem

1、 chip BUG

The chip itself exists BUG, In some specific cases, it returns an incorrect value to the MCU , The program needs to judge the read back value , Filter outliers .

2、 Communication timing error

For example, power management chip Isl78600, Let's say two pieces are cascading , When the voltage sampling data of two chips are read at the same time , The high-end chip will transmit data to the low-end chip through the daisy chain in a fixed cycle , There is only one buffer on the low-end chip .

If the MCU does not read the data on the low-end chip within the specified time, the new data will overwrite the current data when it comes , Cause data loss . Such problems require careful analysis of the data book of the chip , Strictly meet the timing requirements of chip communication .

3.1.2 Abnormal action

3.1.2.1 Software problems

1、 Design problems

There are errors or omissions in the design , Design documents need to be reviewed again .

2、 The implementation is inconsistent with the design

The implementation of the code is inconsistent with the design document. It is necessary to add unit tests to cover all conditional branches , Cross code review.

3、 State variable exception

For example, the variable recording the current state of the state machine is tampered with , The method of analyzing this kind of problem is the same as the numerical anomaly part above .

3.1.2.2 Hardware problem

1、 Hardware failure

The goal is IC invalid , Do not act after receiving the control command , Need to check the hardware .

2、 Abnormal communication

And target IC Communication error , Unable to execute control command correctly , You need to use an oscilloscope or logic analyzer to observe the communication sequence , Analyze whether the signal sent is wrong or subject to external interference .

3.2 Program crash

3.2.1 Stop running

3.2.1.1 Software problems

1、HardFault

The following conditions can cause HardFault：

• Operate the register of the peripheral when the peripheral clock gate is not enabled ;

• Jump function address is out of bounds , It usually happens when the function pointer is tampered with , The troubleshooting method is the same as that for abnormal values ;

• Alignment problem when dereferencing pointer ：

Take the small end sequence as an example , If we declare a forcibly aligned structure, it is as follows ：

Address	0x00000000	0x00000001	0x00000002	0x00000003
Variable name	Val0	Val1_low	Val1_high	Val2
value	0x12	0x56	0x34	0x78

here a.val1 The address for 0x00000001, If the uint16_t Type to dereference this address will enter... Due to alignment problems HardFault, If you must manipulate the variable in pointer mode, you should use memcpy().

2、 The interrupt flag... Is not cleared in the interrupt service function

The interrupt flag is incorrectly cleared before the interrupt service function exits , When the program execution exits from the interrupt service function, it will immediately enter the interrupt service function , Show procedural “ Feign death ” The phenomenon .

3、NMI interrupt

Encountered during debugging SPI Of MISO Pin reuse NMI function , When passed SPI When the connected peripherals are damaged MISO Be pulled high , Cause the single-chip microcomputer to reset after NMI The pins are configured to SPI Enter directly before the function NMI interrupt , The program hangs in NMI In interruption . This can happen in NMI Disable... In the interrupt service function of NMI Function to exit NMI interrupt .

3.2.1.2 Hardware problem

1、 The crystal oscillator does not start

2、 Insufficient supply voltage

3、 The reset pin is pulled low

3.2 .2 Reset

3.2.2.1 Software problems

1、 Watchdog reset

In addition to the reset caused by dog feeding timeout , Also pay attention to the special requirements of watchdog configuration , With Freescale KEA For example, SCM , The MCU watchdog needs to execute unlocking sequence when configuring （ Write two different values continuously to its register ）, The unlocking sequence must be in 16 Complete within a bus clock , Timeout will cause the watchdog to reset . Such problems can only be familiar with the MCU data manual , Pay attention to similar details .

3.2.2.2 Hardware problem

1、 The supply voltage is unstable

2、 Insufficient load capacity of power supply

Four 、 regression testing

After the problem is solved, regression test is needed , On the one hand, confirm whether the problem will not recur , On the other hand, make sure that the modification will not introduce other problems .

5、 ... and 、 Summary of experience

Summarize the causes of this problem and the methods to solve it , Think about how to prevent similar problems in the future , Whether the same platform products are worth learning from , To draw inferences from one case , Learn from failure .