Code: Select all
void* free; //<- initialize this with a block of free size obtained by any
void* end_of_free_mem; // detection technique
void* bottom_up_alloc(int size) {
void *ret=NULL;
if (free+size<end_of_free_mem) {
ret=free;
free+=size;
}
return ret;
}
There's a much better way than this. When linking your kernel together, use a linker script. The linker script allows you to define veriables that are exported into your kernel, so if you want to know where the end of your kernel is, you can export a symbol after the last section. For example, a simple loader script might be:James Buchanan wrote: When you compile your kernel you could define a couple of bytes, or a couple of words even, like a boot signature scheme determines a valid bootsector, and stick it at the very end of your compiled kernel image. Start at address 0x00000000 and scan up, looking for that signature, counting from the start of your kernel.
Code: Select all
SECTIONS
{
__start = .;
.text 0x100000
{
/* kernel code */
*(.text)
}
.data
{
/* kernel data */
*(.data)
}
.bss
{
/* kernel BSS */
*(.bss)
*(COMMON)
}
__end = .;
}
That's cause I was looking up an example... 
I really wish I understood some of this paging info, but I guess age 14 is too young for writing an OS. hum, except for the 'ROM & VRAM' hole between 0xA0000 and 0x100000, are there any other 'holes' ? from what i've learned from PCI bus, the PCI devices get mapped in memory where you tell the to map to ... and ususally, this is from 0xffff0000 ... so ?mystran wrote: Actually, it's not safe to assume that the memory is continuous.
There might be holes, especially when there is a lot of memory in the system.