Which variable can be used to access start of .text segment and size of .text segment?

[jamit]

Hi,

I am using MetrowerksCodeWarrior v2.7.3 and MSC8122 for my project. I want to use the start address and the size of the .text section
during program execution. Is there any variable that stores the start and the size of .text section.
I couldn't get anything related to this in the manuals.

Regards
Amit

[CrasyCat · Accepted solution]

Hello

Did you try to use the Expressions and Symbols described in SC100 linker reference.pdf, chapter Linker Command File section Linker Command File Syntax > Expressions and Symbols.

You can define a symbol in the .lcf file using this syntax and use that symbol in your C source file.

Be careful symbol must start with an _ in the .lcf file because of internal name mangling.

If your symbol is called _Start_of_text, it needs to be declared as Start_of_text in your C source file.

CrasyCat

[jamit]

Thanks for your help.

I have a further query.

How can I use the size of .text segment during program execution of C code? The Expressions and Symbols defines @segsize(text) as the size of the text segment. But how can I access this variable during program execution.

Regards

Amit

[CrasyCat]

Hello

Just define the following symbols in your linker file:

.provide _TextStart, @secaddr(".text")
.provide _TextSize, @secsize(".text")

And you can refer to then as follows in your .c source file:

/* Defined in linker control file */

extern unsigned long TextStart[];
extern unsigned long TextSize[];

/* Global variables used to hold .text start address and size.*/

unsigned char *StartOfCode;
unsigned int   SizeOfCode;

void foo(void)
{
   StartOfCode = (unsigned char *)TextStart;
   SizeOfCode  = (unsigned int) TextSize;
}

CrasyCat

[jamit]

Thank you so much for your help.

Regards

Amit

[jamit]

Hi,

I am using MSC8122. The memory models is:- M1 memory contains core specific program code (isr etc.). Program code common to all the cores is placed in M2 memory. I use .lcf files for linking and common file is used to generate the .map file for all the 4 cores.

As discussed earlier, I did the following to access start of M2 .text segment and size.

extern unsigned long M2_Text_Start[];

extern unsigned long M2_Text_Size[];

extern unsigned char* g_M2_Start_Of_Code;

extern unsinged int g_M2_Size_Of_Code;

In the C file I do:-

g_M2_Size_Of_Code = (unsinged int)M2_Text_Size;

But I get the following error:-

[LNK,2,6999,-1]: Error: "_M2_Text_Size" is unresolved, referenced from:

 file.eln

[LNK,2,6999,-1]: Error: "_M2_Text_Size" is unresolved, referenced from:

 file.eln

[LNK,2,6999,-1]: Error: "_M2_Text_Size" is unresolved, referenced from:

 file.eln

I assume it is coming from 3 cores but I am not able to resolve it.

Pls let me know how to resolve the issue.

Regards

Amit

[CrasyCat]

Hello

How does the definition look like in the .lcf file?

CrasyCat

[jamit]

Hi,

Following is a snippet of the .lcf file.

.unit c0,""

.space shared_m2, M2_start, M2_end  ".m2_text"

.export "shared_m2"

.org M2_start

.segment .m2_text, ".text",

.set _M2_Text_Start, @segaddr(.m2_text)

.set _M2_Text_Size, @segsize(.m2_text)

.unit c1, "core1"

.import "c0'shared_m2"

.unit c2, "core2"

.import "c0'shared_m2"

.unit c3, "core3"

.import "c0'shared_m2"

Regards

Amit

[CrasyCat]

Hello

According to your lcf snippet,  _M2_Text_Start and  _M2_Text_Size are only defined for core 0.

They are undefined in core 1, 2 and 3.

This is the reason you are getting the message.

CrasyCat

[jamit]

Hi,

I understood your point and checked the .map files for all the 4 cores. I see the symbol _M2_Text_Start and _M2_Text_Size only in the memory map for core0 and not in the other three cores.

But my .lcf files defines M2 for core0 only and then exports the shared_m2. The other cores import the shared_m2. So won't this automatically mean the symbols _M2_Text_Start and _M2_Text_Size will be accessible to all the cores.

I tried to modify my .lcf file as 

.unit c1, "core1"

.import "c0'shared_m2"

.set _M2_Text_Start, @segaddr(.m2_text)

.set _M2_Text_Size, @segsize(.m2_text)

But then i get the following error:-

[LNK,2,6003,37,msc8122.lcf]: Error: segment ".m2_text" does not exist

My assumption was by exporting and importing c0'shared_m2 all the cores have access to .m2_text and the associated symbols. I am not sure how to do this.

Further, is it possible that in my C code I restrict the _M2_Text_Start and _M2_Text_Size to core0 only using some pragmas e.g #pragma pgm_seg_name 

Thanks

Amit

[CrasyCat]

Hello

  This does not seem to work.

  What you can try is define the symbols StartOfCode and SizeOfCode into shared memory. Initialize them on core 0 only

and then access the symbols from any other core.

CrasyCat