in ccs ti emulator for c6474, ldw insn load value to register immediately

zheng che

Other Parts Discussed in Thread: SYSBIOS

i'm debugging a hello world program in ccs under target c6747, and in somwhere i think it's in middle of print, there's instruction LDW.D1T1 *+A11[0], A4, and when i step over to next instruction, A4 got it's new value immediately. can someone help me out, thanks

over 8 years ago

0 Titusrathinaraj Stalin over 8 years ago

TI__Guru** 116100 points

Hi,
Did you set "No boot" or "emulation" boot mode for debugging the code ?
Please make sure that you have used gel file to initialize.
Is it normal C bare metal code or SYSBIOS ?
Can you provide the complete project to investigate ?

Did you use linker command file in your project for C6474 ?

0 zheng che over 8 years ago in reply to Titusrathinaraj Stalin

Intellectual 280 points

test_c6474.gzHi, thanks for reply.

I think it's normal c bare metal with libc supports. here's my whole project

0 Raja over 8 years ago in reply to zheng che

TI__Guru* 81335 points

Hi Zheng,

Please refer below wiki to debug your application usign CCS. The "Disassembly and mixed source and assembly mode" section will help to debug the issue.

Thank you.

0 zheng che over 8 years ago in reply to Raja

Intellectual 280 points

Hi Raja,
I keep tracing the instruction flow from the very beginning, it steps down all the way with the right register values util this one.
I upload my whole project in floor two, can you please check it for me ? thanks

0 Titusrathinaraj Stalin over 8 years ago in reply to zheng che

TI__Guru** 116100 points

Hi Zheng,
I'm not able to extract your attached file.
Can you zip your complete project and attach again ?
And one more thing, able to get "hello world" print ?

0 zheng che over 8 years ago in reply to Titusrathinaraj Stalin

Intellectual 280 points

Hi Stalin,

here's the zipped file, and the "hello, world\n" can print out. thanks again, and hopefully you can help out.

test_c6474.zip

0 Titusrathinaraj Stalin over 8 years ago in reply to zheng che

TI__Guru** 116100 points

Hi Zheng,

Can you try to run the attached *.out and let me know the results of both.

BTW, what is your CCS version ?

While loading the *.out , able to stop at "main()" (auto break point) ?

Please note that I don't have C6474 board with me but I've tried your code and its running successfully.

out_files.zip

0 zheng che over 8 years ago in reply to Titusrathinaraj Stalin

Intellectual 280 points

Hi Stalin,
I think i can run both of outs you give me, and they do stop at main.
My CCS version is 5.5.0.
My problem is when i set breakpoint in 0x10800b60, like the png i posted at floor 0, and after i step over to 0x10800b68
all the sudden, the A4 register is freshed with new value, just like the ldw instruction executed without delays.

0 Titusrathinaraj Stalin over 8 years ago in reply to zheng che

TI__Guru** 116100 points

Hi,
Did you set "No boot" or "emulation" boot mode for debugging the code ?
Please make sure that you have used gel file to initialize.

Could you please tell me the complete procedure that you followed.

0 Andy Polyakov over 8 years ago in reply to Titusrathinaraj Stalin

Expert 1340 points

Yet another example of TI failing to even grasp the question...

Zheng, your analysis is incorrect. Value is not loaded immediately, but with proper delay of 4 cycles. Note that LDW is paired with BNOP, which has delay argument of 4. This is the delay I presume you expect to see. The fact that debugger steps to next address as if there is no delay is question of representation, not actual cycle timing.

0 zheng che over 8 years ago in reply to Andy Polyakov

Intellectual 280 points

Hi Polyakov,

I don't think BNOP actually executed, because it has false condition [!A0], and i'm at that point A[0] = 1

0 zheng che over 8 years ago in reply to Titusrathinaraj Stalin

Intellectual 280 points

Hi,
I'm working on a open source project of simulating c6474, and use the hello world program as a testing demo.
I record all the registers step by step from the very beginning. And Everything matched before that point, so I just
hanging there, and don't know what to do.
You can set a break there, step over, and watch the A[4]'s changes.

0 Andy Polyakov over 8 years ago in reply to zheng che

Expert 1340 points

zheng che said:

I don't think BNOP actually executed, because it has false condition [!A0], and i'm at that point A[0] = 1

Quoting processor reference guide:

"BNOP instructions may be predicated. The predication condition
controls whether or not the branch is taken, but does not affect the
insertion of NOPs. BNOP always inserts the number of NOPs
specified by N, regardless of the predication condition."

0 zheng che over 8 years ago in reply to Andy Polyakov

Intellectual 280 points

Thanks so so much

0 Titusrathinaraj Stalin over 8 years ago in reply to zheng che

TI__Guru** 116100 points

Thanks Andy for supporting the community members.
I really appreciate that.

Processors

Processors forum

in ccs ti emulator for c6474, ldw insn load value to register immediately