How can I use Cortex-M3 bit-banding from C code?

The Cortex-M3 processor has a feature known as "bit-banding". This allows an individual bit in a memory-mapped mailbox or peripheral register to be set/cleared by a single store/load instruction to an bit-band aliased memory address, rather than using a conventional read/modify/write instruction sequence.

There are no special compiler extensions to support this, but it is easy to write C macros to handle bit-banding in a straightforward way.

#define BITBAND_SRAM_REF   0x20000000
#define BITBAND_SRAM_BASE  0x22000000
#define BITBAND_SRAM(a,b) ((BITBAND_SRAM_BASE +
 (a-BITBAND_SRAM_REF)*32 + (b*4)))  // Convert SRAM address
#define BITBAND_PERI_REF   0x40000000
#define BITBAND_PERI_BASE  0x42000000
#define BITBAND_PERI(a,b) ((BITBAND_PERI_BASE +
 (a-BITBAND_PERI_REF)*32 + (b*4)))  // Convert PERI address

#define MAILBOX   0x20004000
#define TIMER     0x40004000
#define MBX_B0    *((volatile unsigned int *)
 (BITBAND_SRAM(MAILBOX,0))) // Bit 0
#define MBX_B7    *((volatile unsigned int *)
 (BITBAND_SRAM(MAILBOX,7))) // Bit 7
#define TIMER_B0  *((volatile unsigned char *)
 (BITBAND_PERI(TIMER,0)))   // Bit 0
#define TIMER_B7  *((volatile unsigned char *)
 (BITBAND_PERI(TIMER,7)))   // Bit 7

int main(void)
{
    unsigned int temp = 0;

    MBX_B0 = 1;      // Word write
    temp = MBX_B7;   // Word read
    TIMER_B0 = temp; // Byte write
    return TIMER_B7; // Byte read
}

  
The code generated for Cortex-M3 is:

main    LDR     r0,|L1.16|       // Pointer to MAILBOX
        MOVS    r1,#1
        STR     r1,[r0,#0]
        LDR     r0,[r0,#0x1c]
        LDR     r1,|L1.20|       // Pointer to TIMER
        STRB    r0,[r1,#0]
        LDRB    r0,[r1,#0x1c]
        BX      lr
|L1.16| DCD     0x22080000       // Bitband addr of MAILBOX
|L1.20| DCD     0x42080000       // BitBand addr of TIMER