• State Not Answered
  • Locked Locked
  • Replies 3 replies
  • Subscribers 21 subscribers
  • Views 1455 views
  • Users 0 members are here
2025 survey

We are running a survey to help us improve the experience for all of our members. If you see the survey appear, please take the time to tell us about your experience if you can.

Options
Related
How was your experience today?
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

MICROLIB + Initialised stack

David Henretty

I want to use MICROLIB as (from my initial measurements) it appears to offer space savings of around 1.4KB.

However, I am having problems getting it to work with an initialised STACK. I want to initialise the stacks to known values (e.g. "STAC") for debugging and stack sizing purposes and have modified "startup.s" accordingly. This worked fine without MICROLIB, but using MICROLIB causes a crash. What appears to happen is as follows :-

1) At the start of _main, library code is called to perform initialisation (R/W and ZI).
2) LR (and R4) are pushed on the stack.
3) The stack contents are then initialised to the specified value, overwriting the saved LR (and R4).
4) The LR (and R4) are popped of the stack and the code attempts to branch to the value stored in LR (now "STAC").


I have tried adding an uninitialised ("NOINIT") 8 bytes at the end of the initialised stack, and can configure it so that this area is used for the two register pushes, but these 8 bytes are still being zero-initialised - I don't know why.

The stack declaration looks like this :-

                 GBLA     count
                 AREA     STACK, READWRITE, ALIGN=3
Stack_Mem

count            SETA     0
                 WHILE    count < (Stack_Size / 4)
count            SETA     (count + 1)

                 DCD      STACK_INIT_VAL

                 WEND

                 AREA     STACK, NOINIT, READWRITE, ALIGN=3
                 SPACE    8

__initial_sp
Stack_Top        EQU      (Stack_Mem + Stack_Size + 8)

Any ideas how I can either actually prevent the initialisation of a "NOINIT" section, or get around this in another way ?

Also, is there any easier way to initialise a block of data than the above, which seems overly complex compared to other assemblers I have used. I would expect some sort of declaration that allowed me to specify the size, number of items and initial value. I'm sure I'm missing something obvious but can't find it in the online help.

Thanks,

David.

Parents
  • 0 in reply to Robert  

    Thanks Robert,

    Using the Target dialog would force me to define absolute addresses for the uninitialised data. This may not be a bad thing, but is very inflexible during development. It also kind of defeats the original objective which is to have an initialised stack.

    I found a reference in the online manual which states :-

    Only the attributes of the first AREA directive of a particular name are applied.

    This explains why, in my original second STACK AREA declaration, the NOINIT attribute is ignored.

    I think I have managed to achieve what I wanted by using the GROUP attribute to specify two AREA declarations as part of a single STACK GROUP. The manual states :-

    Sections within a group are kept or discovered together.

    I'm not sure exactly what that's supposed to mean, but it seems to do the job.

    I changed my declaration to :-

                    AREA    STACKINIT, READWRITE, ALIGN=2, GROUP=STACK
Stack_Mem

count           SETA    0
                WHILE   count < (Stack_Size / 4)
count           SETA    (count + 1)
                DCD     STACK_INIT_VAL
                WEND

Stack_Top       EQU     (Stack_Mem + Stack_Size)

                AREA    STACKNOINIT, NOINIT, READWRITE, ALIGN=2, GROUP=STACK

                SPACE   8
__initial_sp

    This appears to work but still needs more testing and tuning.

    Any comments on the other question in my original post regarding better ways to define blocks of data containing a single initialisation value ?

    Thanks again,
    David.

Reply
  • 0 in reply to Robert  

    Thanks Robert,

    Using the Target dialog would force me to define absolute addresses for the uninitialised data. This may not be a bad thing, but is very inflexible during development. It also kind of defeats the original objective which is to have an initialised stack.

    I found a reference in the online manual which states :-

    Only the attributes of the first AREA directive of a particular name are applied.

    This explains why, in my original second STACK AREA declaration, the NOINIT attribute is ignored.

    I think I have managed to achieve what I wanted by using the GROUP attribute to specify two AREA declarations as part of a single STACK GROUP. The manual states :-

    Sections within a group are kept or discovered together.

    I'm not sure exactly what that's supposed to mean, but it seems to do the job.

    I changed my declaration to :-

                    AREA    STACKINIT, READWRITE, ALIGN=2, GROUP=STACK
Stack_Mem

count           SETA    0
                WHILE   count < (Stack_Size / 4)
count           SETA    (count + 1)
                DCD     STACK_INIT_VAL
                WEND

Stack_Top       EQU     (Stack_Mem + Stack_Size)

                AREA    STACKNOINIT, NOINIT, READWRITE, ALIGN=2, GROUP=STACK

                SPACE   8
__initial_sp

    This appears to work but still needs more testing and tuning.

    Any comments on the other question in my original post regarding better ways to define blocks of data containing a single initialisation value ?

    Thanks again,
    David.

Children