Following code was excuted in xv6(risc-v). I'm little confused. WHy we need to madvise() after malloc()? Is it for page table entry swaping in/out?
PS. vmprint() print the page table.
#include "kernel/types.h"
#include "kernel/stat.h"
#include "user/user.h"
#include "kernel/vm.h"
#define PG_SIZE 4096
#define NR_PG 16
int main(int argc, char *argv[]) {
vmprint();
char *ptr = malloc(NR_PG * PG_SIZE);
vmprint();
madvise((uint64) ptr + 10*PG_SIZE, 2*PG_SIZE , MADV_NORMAL);
vmprint();
madvise((uint64) ptr + 10*PG_SIZE, 2*PG_SIZE , MADV_DONTNEED);
vmprint();
madvise((uint64) ptr + 10*PG_SIZE, 2*PG_SIZE , MADV_WILLNEED);
vmprint();
exit(0);
}
After allocating this memory block by malloc() function, the madvise() function is used to advise the kernel about the usage of specific memory regions.
The first madvise() call advises the kernel that the memory region starting at ptr + 10*PG_SIZE and extending for 2*PG_SIZE bytes should be treated as MADV_NORMAL, indicating normal memory access patterns. This call doesn't cause any immediate change to memory allocation.
The second indicated that the specified memory range can be discarded if it's not in use.
And the third, indicated that the specified memory range will be accessed soon, so the kernel can proactively bring the pages into memory to minimize latency when accessing them.