On Fri, Jan 21, 2005 at 07:01:55PM +0300, Oleg Nesterov wrote: > Andi Kleen wrote: > > > > I suppose this simpler patch has the same effect (also untested). > > > > if (flags & ~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR)) > > return -EINVAL; > >@@ -502,6 +502,10 @@ > > pol = vma->vm_ops->get_policy(vma, addr); > > else > > pol = vma->vm_policy; > >+ pol2 = mpol_copy(pol); > >+ up_read(&mm->mmap_sem); > >+ if (IS_ERR(pol2)) > >+ return PTR_ERR(pol2); > > > > I don't think so. With MPOL_F_ADDR|MPOL_F_NODE sys_get_mempolicy > calls lookup_node()->get_user_pages() few lines below, so we can't > up_read(&mm->mmap_sem) here.
True. > > > It's hard to figure out what your patch actually does because > > of all the gratious white space changes. > > For your convenience here is the code with the patch applied. Looks reasonable. -Andi P.S.: Again if you really care about these class of deadlocks take a look at tasklist_lock. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
