> A strange thing is that __mm_populate() is only called by do_mlock() from mm/mlock.c,
> which makes len PAGE_ALIGN already. That VM_BUG_ON should not be triggered.
Unless I overlooked something, __mm_populate() gets called from:
load_elf_library() -> vm_brk() -> vm_brk_flags():
vm_brk_flags() {
...
populate = ((mm->def_flags & VM_LOCKED) != 0);
...
if (populate && !ret)
mm_populate(addr, len);
}
mm_populate() -> __mm_populate():
__mm_populate() {
...
VM_BUG_ON(len != PAGE_ALIGN(len));
...
}
In load_elf_library(), we have:
len = ELF_PAGESTART(eppnt->p_filesz + eppnt->p_vaddr +
ELF_MIN_ALIGN - 1);
bss = eppnt->p_memsz + eppnt->p_vaddr;
if (bss > len) {
error = vm_brk(len, bss - len);
if (error)
goto out_free_ph;
}
So len gets page aligned, but not bss (eppnt->p_memsz + eppnt->p_vaddr), maybe that's the problem?
--
Oscar Salvador
SUSE L3
