On 04/15/14 11:46, Oleg Nesterov wrote:
But. Please do not add copy_to_user_page() into copy_to_page() (as your patch did). This is certainly not what uprobe_write_opcode() wants, we do not want or need "flush" in this case. The same for __create_xol_area().
It looked me like a call to a new __copy_to_user_page(current->mm, ...) in xol_get_insn_slot() would be in line with David Miller's suggestion and would cure the problem on ARM (and hopefuly be more philosophically correct for all architectures):
diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c index 04709b6..b418626 100644 --- a/kernel/events/uprobes.c +++ b/kernel/events/uprobes.c @@ -1287,6 +1287,7 @@ static unsigned long xol_get_insn_slot(struct uprobe *uprobe) { struct xol_area *area; unsigned long xol_vaddr; + void *kaddr;
area = get_xol_area(); if (!area) @@ -1297,13 +1298,11 @@ static unsigned long xol_get_insn_slot(struct uprobe *uprobe) return 0;
/* Initialize the slot */ - copy_to_page(area->page, xol_vaddr, - &uprobe->arch.ixol, sizeof(uprobe->arch.ixol)); - /* - * We probably need flush_icache_user_range() but it needs vma. - * This should work on supported architectures too. - */ - flush_dcache_page(area->page); + kaddr = kmap_atomic(area->page); + __copy_to_user_page(current->mm, area->page, xol_vaddr, + kaddr + (xol_vaddr & ~PAGE_MASK), + &uprobe->arch.ixol, sizeof(uprobe->arch.ixol), true); + kunmap_atomic(kaddr);
return xol_vaddr; }
Opinions? It's possible this approach isn't good enough. Cache operations and VM are not my strong suit.
-dl