• Locked Locked
  • Replies 6 replies
  • Subscribers 53 subscribers
  • Views 1153 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Optimization prevents access to extended addresses

Robert Lane
Expert 1225 points
Other Parts Discussed in Thread: MSP430FR6989, MSP430FR5989

Hi,

MCU: MSP430FR6989 
IDE: Code Composer Studio 6.1.1
Compiler version: v4.4.6

I am working with the MSP430FR6989  and am using a large portion of the onboard FRAM for data storage. This ticket gives some background into how I updated the linker file in order to store data in the FRAM segment starting at address 0x10000:

MSP430FR6989, Unable to use FRAM2 with #pragma PERSISTENT


For the extended address range, I have declared a single uint8_t array (m_stepData2) as PERSISTENT and set its length so that it occupies all of the space from 0x10000 - 0x24000:

#define STEP_DATA_BUF_2_SIZE_BYTES  0x14000U

#pragma PERSISTENT
uint8_t SMemory::m_stepData2[STEP_DATA_BUF_2_SIZE_BYTES] = {0};

I also have a function called WriteStepData that writes to this array. When testing in my debug build with the optimization level set to 0, everything works fine and if I look in the Memory Allocation GUI I can see that m_stepData2 is taking up all of the DATA_FRAM2 segment we declared in the linker:

The problem I have now is if I change the optimization level from 0 to level 4 (Whole Program Optimizations), my calls to WriteStepData result in this warning:

I seem to have narrowed it down to changing the optimization level that generates this warning. The problem is that I need to be able to use a high level of optimization to reduce the code size to a manageable level. 

Do you have any insight into how I can still use the extended address space AND still keep the optimization level 4 enabled?

  • TI__Guru**** 249420 points

    That error message says the tools cannot encode the call instruction from main to the function SMemory::m_stepData2.  Does that make sense?

    Are you building with the compiler option --code_model=large?  If you are, then I have to suspect a bug in the compiler or linker.

    Thanks and regards,

    -George

  • in reply to George Mock
    Expert 1225 points
    Hi George. Yeah, that's how I interpreted it. And so I did make sure that both the large code and large memory compiler options were set. Even with both of them set, I still get the error.

    Here is my compiler option summary:
    -vmspx --abi=eabi --code_model=large --data_model=large --near_data=none -O4 --opt_for_speed=0 --use_hw_mpy=F5 --include_path="C:/ti/ccsv6/ccs_base/msp430/include" --include_path="C:/Foucault/Firmware/src" --include_path="C:/Foucault/Firmware/src/TI_DriverLib" --include_path="C:/ti/ccsv6/tools/compiler/ti-cgt-msp430_4.4.6/include" --advice:power="all" --advice:hw_config=all --define=__MSP430FR5989__ --define=_MPU_ENABLE --display_error_number --diag_wrap=off --diag_warning=225 --small_enum --silicon_errata=CPU21 --silicon_errata=CPU22 --silicon_errata=CPU40 --printf_support=minimal

    Here is my linker option summary:
    -vmspx --abi=eabi --code_model=large --data_model=large --near_data=none -O4 --opt_for_speed=0 --use_hw_mpy=F5 --advice:power="all" --advice:hw_config=all --define=__MSP430FR5989__ --define=_MPU_ENABLE --display_error_number --diag_wrap=off --diag_warning=225 --small_enum --silicon_errata=CPU21 --silicon_errata=CPU22 --silicon_errata=CPU40 --printf_support=minimal -z -m"Foucault.map" --stack_size=250 --heap_size=0 --cinit_hold_wdt=on -i"C:/ti/ccsv6/ccs_base/msp430/include" -i"C:/ti/ccsv6/tools/compiler/ti-cgt-msp430_4.4.6/lib" -i"C:/ti/ccsv6/tools/compiler/ti-cgt-msp430_4.4.6/include" -i"C:/ti/ccsv6/ccs_base/msp430/lib/5xx_6xx_FRxx" -i"C:/ti/ccsv6/ccs_base/msp430/lib/FR59xx" --reread_libs --priority --define=_MPU_ENABLE --diag_wrap=off --warn_sections --display_error_number --xml_link_info="Foucault_linkInfo.xml" --use_hw_mpy=F5 --rom_model

    A side note is that I also tried playing with the --near_data option. But it didn't make any difference whether I set it to "none" or "globals"
  • in reply to Robert Lane
    TI__Guru**** 249420 points

    I think the only way to figure this out is for you to send us your CCS project.  Please create a project zip file as described here, and attach that to your next post.

    Thanks and regards,

    -George

  • in reply to George Mock
    Expert 1225 points

    Hi George,

    I had to cut the project down and remove some sensitive names, but it compiles fine and still throws the same error. In main I make a single call to a function that tries to access the offending array.

    Let me know if you have any questions. Anxiously looking forward to what you can find!

    Thanks!

    Test_Project.zip

  • in reply to Robert Lane
    TI__Guru**** 249420 points

    Thank you for the test case.  I can reproduce the same results.  I filed SDSCM00052683 in the SDOWP system to have this investigated.  You are welcome to follow it with the SDOWP link below in my signature.

    A workaround is to build with --opt_level=3 or lower.  Hopefully, that makes code size small enough.

    Another idea to consider ... See if using Optimizer Assistant can help you find a better setting for the compiler options --opt_level or --opt_for_speed (which also affects size).

    Thanks and regards,

    -George

  • in reply to George Mock
    Expert 1225 points

    Hi George,

    Thanks for submitting that bug report. Not being able to use optimization level 4 is going to be rough. Our code size grows by about 3k when going from level 4 to level 3. 

    Hopefully this can be resolved soon!