Bug found in Port driver (plugin) from SW32K3_RTD_4.4_2.0.0

取消
显示结果 
显示  仅  | 搜索替代 
您的意思是: 

Bug found in Port driver (plugin) from SW32K3_RTD_4.4_2.0.0

863 次查看
Artur_S
Contributor II

While working with SW32K3_RTD_4.4_2.0.0 intergrated with Vector's AUTOSAR stack for S32K344 chip, we found a bug in Port driver related to Adc driver configuration.

In function Siul2_Port_Ip_WriteDCMConfiguration:

Artur_S_0-1671028604132.png

 




 

 

 

 

 

 

 

 

 

 

the call of PORT_WRITE32 is unconditional, executed always, therefore overriding previous configuration(s).

We had to move this call and therefore modify this function in Siul2_Port_Ip.c as follows:


Artur_S_1-1671028935287.png

With such a hotfix, the ADC driver works correctly. 
Please consider this finding in the next release of SW32K3_RTD.

标记 (3)
0 项奖励
回复
3 回复数

817 次查看
Adam_C
Contributor I

Lets check original code...

Register DCM_DCMRWF4_ADDR32

03_register.jpg

Configuration (generated):

04_configuration.jpg

Original code

enums (0x0xxx causes set proper bit, 0xFxxx causes clear proper bit)

01_enums_b.jpg

defines (DCM_DCMRWF4_ADC_INTERLEAVE_MASK represents all MUX_MODE bits)

02_bitfields.jpg

Code with comments:

05_code_with_comments.jpg

Final result:
- all MUX_MODE are set to '1' regardless of configuration !
- bits 16..31 (used 16..18 e.g. for glitch) are always cleared


Solution1:
As mentioned in Artur's message

dcmrwf4RegValue &= (adcInterleaves | (~DCM_DCMRWF4_ADC_INTERLEAVE_MASK));
Protects bits 16..31

PORT_WRITE32(DCM_DCMRWF4_ADDR32, dcmrwf4RegValue);
Correct write operation. This is called only once, because adcInterleaves size is always 2 and when one is set then second is always '0':

.adcInterleaves = {
MUX_MODE_EN_ADC2_S9_0,
MUX_MODE_NOT_AVAILABLE // 0
},


Solution2:
Original code with change:
typedef enum
{
MUX_MODE_NOT_AVAILABLE = 0x0000UL, // unsigned long
MUX_MODE_EN_ADC1_S14_1 = 0x0008UL,
MUX_MODE_EN_ADC1_S15_1 = 0x0010UL,
MUX_MODE_EN_ADC0_S8_1 = 0x0002UL,
MUX_MODE_EN_ADC2_S8_1 = 0x0200UL,
MUX_MODE_EN_ADC0_S9_1 = 0x0004UL,
MUX_MODE_EN_ADC2_S9_1 = 0x0400UL,
MUX_MODE_EN_ADC1_S22_1 = 0x0020UL,
MUX_MODE_EN_ADC1_S23_1 = 0x0040UL,

MUX_MODE_EN_ADC1_S14_0 = 0xFFFFFFF7UL, // added FFFF to protect bits 16..31
MUX_MODE_EN_ADC1_S15_0 = 0xFFFFFFEFUL,
MUX_MODE_EN_ADC0_S8_0 = 0xFFFFFFFDUL,
MUX_MODE_EN_ADC2_S8_0 = 0xFFFFFDFFUL,
MUX_MODE_EN_ADC0_S9_0 = 0xFFFFFFFBUL,
MUX_MODE_EN_ADC2_S9_0 = 0xFFFFFBFFUL, // this case
MUX_MODE_EN_ADC1_S22_0 = 0xFFFFFFDFUL,
MUX_MODE_EN_ADC1_S23_0 = 0xFFFFFFBFUL,
} Siul2_Port_Ip_AdcInterleaves;

And of course in the same place as in original code correct write operation:
PORT_WRITE32(DCM_DCMRWF4_ADDR32, dcmrwf4RegValue);

 

 

 

0 项奖励
回复

812 次查看
cuongnguyenphu
NXP Employee
NXP Employee

@Adam_C ,
Thank you for your explanation, finally I can get your points in this case.
Yes, this function should be fixed as your suggestions
I will raise this issue to development team to fix

0 项奖励
回复

841 次查看
cuongnguyenphu
NXP Employee
NXP Employee

Hi @Artur_S ,
As I see, it's not a bug here.
dcmrwf4RegValue will be updated if the pin has Mux mode, and the bit of DCMRWF4 will be update based on Connected pad option, unless this dcmrwf4RegValue will not be updated.
this dcmrwf4RegValue will be accumulated then set into DCMRWF4 register.

Your modification also correct, but it takes multiple times write into DCMRWF4 register.

0 项奖励
回复