AM2634: When PC is at 0x0, Debugger Thinks Exception Table Symbol is in an Unrelated Source Code File

Kier

Part Number: AM2634

Hello,

After I've loaded the program and reset the core(s), the PC is 0x00000000. The symbol located there is the exception table as expected.

The question: Why does the debugger associate the exception table base address (0x0) with a seemingly irrelevant source code file and line? In this case, line 6183 of Can_30_Mcan.c.

Thank you.

over 2 years ago

0 Sunil Kumar M S over 2 years ago

TI__Genius 12380 points

Hi Kier,

Can you share map file of your program for the analysis?

Regards,

Sunil

0 Kier over 2 years ago in reply to Sunil Kumar M S

Mastermind 8855 points

Hi Sunil,

I can't share my project, sorry, but you can see a similar problem in SDK example: ipc_notify_echo_am263x-cc_system_freertos_nortos

Map file attached.

ipc_notify_echo.Debug.zip

For example, the source code file for R5_0 (ti_dpl_config.c, line 239) is not relevant to address 0x00000000 where _vectors is located. The section _vectors is defined in HwiP_armv7r_vectors_freertos_asm.S not ti_dpl_config.c.

If source code is not available then it should say 'Source Not Available' or similar. It should not map to or open an arbitrary source file.

Kier.

0 Aakash Kedia over 2 years ago in reply to Kier

TI__Mastermind 26145 points

Hi Kier,

Let me work on this and come back to you on this.

Best Regards,
Aakash

0 Aakash Kedia over 2 years ago in reply to Aakash Kedia

TI__Mastermind 26145 points

Hi Kier,

I have replicated the problem. However the problem only exists with 0x0 address and not others.

Let me raise an issue internally. I think this is more towards compiler issue and ELF creation instead of CCS/Debugger issue.

Best Regards,
Aakash

0 Kier over 2 years ago in reply to Aakash Kedia

Mastermind 8855 points

Thanks for working on this Aakash.

0 Aakash Kedia over 1 year ago in reply to Kier

TI__Mastermind 26145 points

Hi Kier,

This is confirmed as an compiler issue which is mentioned in the following thread - https://e2e.ti.com/support/tools/code-composer-studio-group/ccs/f/code-composer-studio-forum/1270435/ccs-theia-does-not-show-holding-at-main-on-debug-start

Best Regards,
Aakash

0 Kier over 1 year ago in reply to Aakash Kedia

Mastermind 8855 points

Hi Aakash,

Thank you. I guess the next question is will this be fixed in the next linker update?

Kier.

+1 Alan Phipps over 1 year ago in reply to Kier

TI__Expert 4680 points

Hi Kier,

The issue is being tracked as https://sir.ext.ti.com/jira/browse/EXT_EP-11458 which you are welcome to follow. It will be fixed in a forthcoming patch release (it will not be fixed in the 3.2.0.LTS release which is being released this week). We believe the issue pertains to how the linker resets debug information for functions that are excluded from the final link. Effectively the PC values are reset to 0, but the source line information is not reset. Consequently, this confuses the debugger.

A potential workaround would be to, when debugging, build your application with:

-fno-function-sections

This will prevent the linker from omitting functions from your application. Note that using this is only recommended for debug builds only since it will increase code size.

Another potential workaround is to use the linker option:

--unused_section_elimination=off

which will have the same effect, but it doesn't require compile-time changes. This option should be passed to the linker directly: "tiarmclang -Wl,--unused_section_elimination=off"

0 Kier over 1 year ago in reply to Alan Phipps

Mastermind 8855 points

Thank you for the explanation Alan.

Arm-based microcontrollers

Arm-based microcontrollers forum

AM2634: When PC is at 0x0, Debugger Thinks Exception Table Symbol is in an Unrelated Source Code File