On 04/05/2013 04:59 AM, Kirill A. Shutemov wrote: > + if (unlikely(khugepaged_enter(vma))) > + return VM_FAULT_OOM; ... > + ret = vma->vm_ops->huge_fault(vma, &vmf); > + if (unlikely(ret & VM_FAULT_OOM)) > + goto uncharge_out_fallback; > + if (unlikely(ret & (VM_FAULT_ERROR | VM_FAULT_NOPAGE | VM_FAULT_RETRY))) > + goto uncharge_out; > + > + if (unlikely(PageHWPoison(vmf.page))) { > + if (ret & VM_FAULT_LOCKED) > + unlock_page(vmf.page); > + ret = VM_FAULT_HWPOISON; > + goto uncharge_out; > + } One note on all these patches, but especially this one is that I think they're way too liberal with unlikely()s. You really don't need to do this for every single error case. Please reserve them for places where you _know_ there is a benefit, or that the compiler is doing things that you _know_ are blatantly wrong. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx"> email@xxxxxxxxx </a>