CCS/EVMK2H: Code section causes DSP to run into endless loop

The picture don't show ,so I put these again.Sorry

Here is what  I see in CCS

The function where dead loop is

The code cause dead loop

DSP will  hold  when execute  this line of code ,I  check assembly code  and find it go into dead loop

It always execute this there assembly code, and dsp will hold.

When I  change the name of code section ,code  won't go into dead loop and dsp keep running.

assembly code change

But "ramisr" and "runfastram" code section are in the same memory ,defferent name.

MEMORY{

...................

dsp_L2_ram:       o = 0x00800C00  l = 0x0006F400

.................

}

runfastram : > dsp_L2_ram

ramisr : > dsp_L2_ram

Question:

1,I don't know why assembly code will create dead loop.I don't create any loop in my C code.

2, Why different name of code section will cause different assembly code?

If anyone know anything  ,please tell me .Thanks very much

Hi,Yordan

I don't change Proceesor SDK.I always use   ti-processor-sdk-rtos-k2hk-evm-04.01.00.06

Here is my linker .cmd

Jacky_Linker.zip

And I didn't see .cfg files in my project. I used to change configure in CCS .Maybe can you show me how to get .cfg file?

Thanks for reply again.

Best Regards,

Jacky

Hi Jacky,

The .cfg file should be part of the project (as in the Processor SDK RTOS examples), see the screenshot bellow:

Best Regards,
Yordan

Hi , Yordan

I search my project again.and other example of EVMK2H. I could not find .cfg file .

I saw your example is running SYSBIOS .Does  only SYSBIOS create the .cfg file ?

My project didn't run OS on DSP. But if you want other configure of CCS, I would share with you as possible as I could.

Best Regards

Jacky

Hi,Yordan

Thanks for reply.I will try place "ramisr" in external_ram memrory.

I just don't know why cause different assembly code.

I am worried about next time I add other code , and it will create other dead loop.

Do you know some ways or tools to see how CCS create assembly executed code?So I could trace why it happen.

Best Regards

Jacky

Hi,Yordan

Thanks for your help. I will wait for your reply.

Best Regards

Jacky

Hi,Rahul

I try my code in v8.1.3 compiler version

And the code section is the only thing I modifed

Thanks for analyze this issue for me .I am confused about that in a long time

If you have any advise or need other about my setting, please let me know.

Best Regards

Jacky

George Mock said:

We say this situation is an endless loop.  Not a death loop, or a dead loop.

George Mock said:

I would like to know when this bad BNOP instruction first appears.

Hi ,George

I got the disassemble file,and capture where endless loop occurs.

Here is the txt file

00828a50             $C$L300:
00828a50       8e41           ADD.L2        B4,-4,B4
00828a52       124d           LDH.D2T2      *B4[0],B4
00828a54   01c52428           MVK.S1        0xffff8a48,A3
00828a58   018040e8           MVKH.S1       0x810000,A3
00828a5c   e2830000           .fphead       n, l, W, H, nobr, nosat, 0010100b
00828a60       2c6e           NOP           2
00828a62       5227           CMPGT.L2      0,B4,B0
00828a64   30148120    [!B0]  BNOP.S1       $C$L301 (PC+40 = 0x00828a88),4
00828a68   0204a35a           MVK.L2        1,B4
00828a6c   000c0264           LDW.D1T1      *+A3[0],A0
00828a70       95d1           ADD.L2X       A3,4,B5
00828a72       4c6e           NOP           3
00828a74       a53a    [!A0]  BNOP.S1       $C$L301 (PC+40 = 0x00828a88),5
00828a76       10dd           LDW.D2T2      *B5[0],B5
00828a78   018c0264 ||        LDW.D1T1      *+A3[0],A3
00828a7c   e6208800           .fphead       n, l, W, BU, br, nosat, 0110001b
00828a80       6c6e           NOP           4
00828a82       bdc9           CMPLTU.L2X    B5,A3,B0
00828a84   3016a120    [!B0]  BNOP.S1       $C$L302 (PC+44 = 0x00828aac),5
00828a88             $C$L301:
00828a88   02815b6e           LDW.D2T2      *+B14[347],B5
00828a8c   0345222a           MVK.S2        0xffff8a44,B6
00828a90   030040ea           MVKH.S2       0x810000,B6
00828a94       136d           LDH.D2T2      *B6[0],B6
00828a96       0c6e           NOP           1
00828a98       32dd           LDH.D2T2      *B5[1],B5
00828a9a       6c6e           NOP           4
00828a9c   ec230000           .fphead       n, l, W, H, nobr, nosat, 1100001b
00828aa0   0014ca7a           CMPEQ.L2      B6,B5,B0
00828aa4   2006a120    [ B0]  BNOP.S1       $C$L302 (PC+12 = 0x00828aac),5
00828aa8   0200a35a           MVK.L2        0,B4
00828aac             $C$L302:
00828aac   0281926e           LDW.D2T2      *+B14[402],B5
00828ab0   02148ffa           OR.L2         B4,B5,B4
00828ab4   0201927e           STW.D2T2      B4,*+B14[402]

00828aa0   0014ca7a           CMPEQ.L2      B6,B5,B0
00828aa4   2006a120    [ B0]  BNOP.S1       $C$L302 (PC+12 = 0x00828aac),5
00828aa8   0200a35a           MVK.L2        0,B4

It seems that it is ok. But I use my XDS pro to trace again. The endless loop still happen

Here is the part I get from the Disassembly window

2160                  stop_condition |= ((SectorRwControl.stop_sector < 0) && (SectorRwControl.stop_count != 0)
          $C$L300:
00828a50:   8E41                ADD.L2        B4,-4,B4
00828a52:   124D                LDH.D2T2      *B4[0],B4
00828a54:   01C52428            MVK.S1        0xffff8a48,A3
00828a58:   018040E8            MVKH.S1       0x810000,A3
00828a5c:   E2830000            .fphead       n, l, W, H, nobr, nosat, 0010100b
00828a60:   2C6E                NOP           2
00828a62:   5227                CMPGT.L2      0,B4,B0
00828a64:   30148120     [!B0]  BNOP.S1       $C$L301 (PC+40 = 0x00828a88),4
00828a68:   0204A35A            MVK.L2        1,B4
00828a6c:   000C0264            LDW.D1T1      *+A3[0],A0
00828a70:   95D1                ADD.L2X       A3,4,B5
00828a72:   4C6E                NOP           3
00828a74:   A53A         [!A0]  BNOP.S1       $C$L301 (PC+40 = 0x00828a88),5
00828a76:   10DD                LDW.D2T2      *B5[0],B5
00828a78:   018C0264 ||         LDW.D1T1      *+A3[0],A3
00828a7c:   E6208800            .fphead       n, l, W, BU, br, nosat, 0110001b
00828a80:   6C6E                NOP           4
00828a82:   BDC9                CMPLTU.L2X    B5,A3,B0
00828a84:   3016A120     [!B0]  BNOP.S1       $C$L302 (PC+44 = 0x00828aac),5
          $C$L301:
00828a88:   02815B6E            LDW.D2T2      *+B14[347],B5
00828a8c:   0345222A            MVK.S2        0xffff8a44,B6
00828a90:   030040EA            MVKH.S2       0x810000,B6
00828a94:   136D                LDH.D2T2      *B6[0],B6
00828a96:   0C6E                NOP           1
00828a98:   32DD                LDH.D2T2      *B5[1],B5
00828a9a:   6C6E                NOP           4
00828a9c:   EC230000            .fphead       n, l, W, H, nobr, nosat, 1100001b
00828aa0:   0014CA7A            CMPEQ.L2      B6,B5,B0
00828aa4:   0001A120            BNOP.S1       0x828aa2,5
00828aa8:   0200A35A            MVK.L2        0,B4
          $C$L302:
00828aac:   0281926E            LDW.D2T2      *+B14[402],B5
00828ab0:   02148FFA            OR.L2         B4,B5,B4
00828ab4:   0201927E            STW.D2T2      B4,*+B14[402]

Besides,I did compile my code again but not change any code . I don't know why

00828aa2:    CMPGT.L2    0,B4,B0

Change to

00828aa0:    CMPEQ.L2    B6,B5,B0

But it still run into endless loop

It doesn't matter that you point out my mistake in my English. It is good to use common way to describe

If you have any concern, please tell me freely

Thanks for reply and Best Regards

Jacky 

Hi, Chester

         Thanks for reply, It is a good advise .But I have something odd in my disassembly window

I try your way to debug my program.

I set my breakpoint, check 00828aa4 in the disassembly window after downloading the program .It seems to be right.

But when I run my code, my program stop at the  00828aa4 again.And I check my disassembly view .The content of 00828aa4  is not changed.

It is so odd. My program still stop at the 00828aa4.

Do you see this odd issue before?

I feel confused and I try several time.I still have some issue

Maybe something cause my dissembly view doesn't display message on real time?

Best Regards

Jacky

JackyXiaodong said:

My program still stop at the 00828aa4.

Since the problem might be caused by the instruction at that address being overwritten by the running program need to set a Hardware Watchpoint for writes to that address. To do that, after downloading the program right click in the Breakpoints view and select Breakpoint (Code Composer Studio) -> Hardware Watchpoint. In the Hardware Watchpoint dialogue:

a. Set the location to 0x00828aa4

b. Set the memory to Write.

Hi, Chester

I try as you say. But I have two situation  when I set hardware watchpoint and locate the memory of 0x00828aa4

First, the content of 0x00828aa4 is correct. But it seems to run into the same endless loop

And I try some assembly step into. CCS print error message and stop

Second,when I run my code ,my program can't be started. And program stop the area I don't know.

So I think  maybe some content cross over the border of memory and cause damage and modification.

Could the CCS have some tool to catch overflow? Or do you have any idea about that?

Best Regards

Jacky 

Hi,George

Do you have methods to get all assembly code in Dissembly window?

I only can get a little part.

So I can compare that with dissemble code which generated from .out file. 

To figure out that  are there  others content of address also being changed  when running my program.

Thanks and Best Regards

Jacky

Hi, Chester

Thanks for your advise .You are correct that address 00828aa4 is being overwritten by something in my program.

I couldn't find what change contents of  the address 00828aa4 when using hardware watchpoint.

But I monitor the address 00828aa4 on Memory Brower and run program step by step .

I found the pointer of a array did change the content of  address 00828aa4

Thanks

Best Regards

Jacky

Hi, George

Thanks for your help .

And I found the pointer of a array did overwrite the content of  address 00828aa4.

Thanks

Best Regards

Jacky