#define uint32_t unsigned long int

#define CCMBase 0x01C20000 //clock module
#define Def_APB0_Gating CCMBase + 0x68

//#define Def_APB0_Gating 0x01C20068

#define GPIOBase 0x01C20800
#define GPIOIncrement 0x24

#define ALLWINNER_GPIO_A GPIOBase
#define ALLWINNER_GPIO_B GPIOBase + (1*GPIOIncrement)
#define ALLWINNER_GPIO_C GPIOBase + (2*GPIOIncrement)
#define ALLWINNER_GPIO_D GPIOBase + (3*GPIOIncrement)
#define ALLWINNER_GPIO_E GPIOBase + (4*GPIOIncrement)
#define ALLWINNER_GPIO_F GPIOBase + (5*GPIOIncrement)
#define ALLWINNER_GPIO_G GPIOBase + (6*GPIOIncrement)
#define ALLWINNER_GPIO_H GPIOBase + (7*GPIOIncrement)
#define ALLWINNER_GPIO_I GPIOBase + (8*GPIOIncrement)

#define Port_CFG0 0x00 //pin direction and function 0 to 7
#define Port_CFG1 0x04 //pin direction and function 8 to 15
#define Port_CFG2 0x08 //pin direction and function 16 to 24
#define Port_CFG3 0x0C //pin direction and function 25 to 32
#define Port_DAT 0x10
#define Port_DRV0 0x14
#define Port_DRV1 0x18
#define Port_PUL0 0x1C


void main(void)
{
int i;
int j;
int foo;

uint32_t *portGConfig;
uint32_t *portGData;
uint32_t *APB0Gating;

//setup pointers for registers
portGConfig = (uint32_t *)(ALLWINNER_GPIO_G + Port_CFG1);
portGData = (uint32_t *)(ALLWINNER_GPIO_G + Port_DAT);
APB0Gating = (uint32_t *)(Def_APB0_Gating);

//enable clocking for GPIO
*APB0Gating |= 0x0020;
//configure port G pin 9 to output
*portGConfig |= 0x10;
//set output to on
*portGData |= 0x0200;

Loop:

foo = 0;
for (j = 0; j < 100; j++)
{
//toggle output
*portGData ^= 0x0200;

foo=j;
for (i = 0; i < 30000000; i++)
{
if(i>10)
{
foo++;
}
}
}

goto Loop;

}

MEMORY
{
RAM (XRW) : ORIGIN = 0x00000000, LENGTH = 0x00004000 /* 16 KB */
}

SECTIONS
{
.start    : { *(.start)    } > RAM
.text     : { *(.text)     } > RAM
.bss      : { *(.bss)      } > RAM
.fill     :
{
  FILL(0x00);
  . = ORIGIN(RAM) + LENGTH(RAM) - 1;
  BYTE(0x00);
} > RAM
/DISCARD/ : { *(.dynstr*)  }
/DISCARD/ : { *(.dynamic*) }
/DISCARD/ : { *(.plt*)     }
/DISCARD/ : { *(.interp*)  }
/DISCARD/ : { *(.gnu*)     }
/DISCARD/ : { *(.note*)    }
}

test.elf:     file format elf32-littlearm


Disassembly of section .text.startup:

00000000 <main>:
   0:    e59f3030     ldr    r3, [pc, #48]; 38 <main+0x38>
   4:    e5932068     ldr    r2, [r3, #104]; 0x68
   8:    e3822020     orr    r2, r2, #32
   c:    e5832068     str    r2, [r3, #104]; 0x68
  10:    e59328dc     ldr    r2, [r3, #2268]; 0x8dc
  14:    e3822010     orr    r2, r2, #16
  18:    e58328dc     str    r2, [r3, #2268]; 0x8dc
  1c:    e59328e8     ldr    r2, [r3, #2280]; 0x8e8
  20:    e3822c02     orr    r2, r2, #512; 0x200
  24:    e58328e8     str    r2, [r3, #2280]; 0x8e8
  28:    e59f300c     ldr    r3, [pc, #12]; 3c <main+0x3c>
  2c:    e2533001     subs    r3, r3, #1
  30:    1afffffd     bne    2c <main+0x2c>
  34:    eafffffb     b    28 <main+0x28>

arm-none-eabi-gcc     -Os -march=armv7-a -std=c99 -marm -fpic -Wall -fno-common -fno-builtin -ffreestanding -nostdinc -nostdlib -fno-strict-aliasing -mno-thumb-interwork -fno-stack-protector -fno-toplevel-reorder -Wstrict-prototypes -Wno-format-nonliteral -Wno-format-security -c test.c -o test.o

arm-none-eabi-ld      -T test.lds -M -o test.elf test.o

arm-none-eabi-objdump -D test.elf > test.asm

arm-none-eabi-objcopy -O binary test.elf test.bin

mksunxiboot.exe test.bin test-sd.bin

void main(void)
{
volatile int i;
volatile int j;
volatile int foo;

volatile uint32_t *portGConfig;
volatile uint32_t *portGData;
volatile uint32_t *APB0Gating;

//setup pointers for registers
portGConfig = (volatile uint32_t *)(ALLWINNER_GPIO_G + Port_CFG1);
portGData = (volatile uint32_t *)(ALLWINNER_GPIO_G + Port_DAT);
APB0Gating = (volatile uint32_t *)(Def_APB0_Gating);

//enable clocking for GPIO
*APB0Gating |= 0x0020;
//configure port G pin 9 to output
*portGConfig |= 0x10;
//set output to on
*portGData |= 0x0200;

Loop:

foo = 0;
for (j = 0; j < 100; j++)
{
//toggle output
*portGData ^= 0x0200;

foo=j;
for (i = 0; i < 30000000; i++)
{
if(i>10)
{
foo++;
}
}
}

goto Loop;


}

00000000 <main>:
0: e59f3094 ldr r3, [pc, #148]; 9c <main+0x9c>
4: e24dd010 sub sp, sp, #16
8: e5932068 ldr r2, [r3, #104]; 0x68
c: e3822020 orr r2, r2, #32
10: e5832068 str r2, [r3, #104]; 0x68
14: e59328dc ldr r2, [r3, #2268]; 0x8dc
18: e3822010 orr r2, r2, #16
1c: e58328dc str r2, [r3, #2268]; 0x8dc
20: e59328e8 ldr r2, [r3, #2280]; 0x8e8
24: e3822c02 orr r2, r2, #512; 0x200
28: e58328e8 str r2, [r3, #2280]; 0x8e8
2c: e3a02000 mov r2, #0
30: e58d200c str r2, [sp, #12]
34: e58d2008 str r2, [sp, #8]
38: e59d2008 ldr r2, [sp, #8]
3c: e3520063 cmp r2, #99; 0x63
40: cafffff9 bgt 2c <main+0x2c>
44: e59328e8 ldr r2, [r3, #2280]; 0x8e8
48: e2222c02 eor r2, r2, #512; 0x200
4c: e58328e8 str r2, [r3, #2280]; 0x8e8
50: e59d2008 ldr r2, [sp, #8]
54: e58d200c str r2, [sp, #12]
58: e3a02000 mov r2, #0
5c: e58d2004 str r2, [sp, #4]
60: e59d1004 ldr r1, [sp, #4]
64: e59f2034 ldr r2, [pc, #52]; a0 <main+0xa0>
68: e1510002 cmp r1, r2
6c: ca000007 bgt 90 <main+0x90>
70: e59d2004 ldr r2, [sp, #4]
74: e352000a cmp r2, #10
78: c59d200c ldrgt r2, [sp, #12]
7c: c2822001 addgt r2, r2, #1
80: c58d200c strgt r2, [sp, #12]
84: e59d2004 ldr r2, [sp, #4]
88: e2822001 add r2, r2, #1
8c: eafffff2 b 5c <main+0x5c>
90: e59d2008 ldr r2, [sp, #8]
94: e2822001 add r2, r2, #1
98: eaffffe5 b 34 <main+0x34>
9c: 01c20000 biceq r0, r2, r0
a0: 01c9c37f biceq ip, r9, pc, ror r3

sudo dd if=bin/release.bin of=/dev/sdb bs=1024 seek=8

GPIO issue
The above code works but the LED only turns on if printf is used. This appears to be caused by compiling optimization.
Not sure how to fix this with the u boot make files however the make file in the following section has optimization turned down to allow the GPIO setting to not be compiled out.