Accessing sysenter page by kernel syscalls
Hi,
is there any reason why copy_from_user does not punch hole in
kernel space to get it possible to access sysenter page? I do not think that it would add too big overhead to copy_*_user (which is already long), but I cannot think about any other reason why sysenter page is not accessible through write().
Today I wanted to confirm whether my kernel uses sysenter or int 0x80,
and I found that simple program below does not work and I had to copy page first to some temporary location, and then write it from this temporary location to stdout.
void main(void) {
write(1, (void*)0xFFFFE000, 4096);
}
Thanks,
Petr Vandrovec