From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934424Ab1JELLQ (ORCPT ); Wed, 5 Oct 2011 07:11:16 -0400 Received: from e9.ny.us.ibm.com ([32.97.182.139]:47849 "EHLO e9.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S934403Ab1JELLP (ORCPT ); Wed, 5 Oct 2011 07:11:15 -0400 Date: Wed, 5 Oct 2011 16:22:43 +0530 From: Srikar Dronamraju To: Oleg Nesterov Cc: Peter Zijlstra , Ingo Molnar , Steven Rostedt , Linux-mm , Arnaldo Carvalho de Melo , Linus Torvalds , Masami Hiramatsu , Hugh Dickins , Christoph Hellwig , Ananth N Mavinakayanahalli , Thomas Gleixner , Jonathan Corbet , LKML , Jim Keniston , Roland McGrath , Andi Kleen , Andrew Morton Subject: Re: [PATCH v5 3.1.0-rc4-tip 5/26] Uprobes: copy of the original instruction. Message-ID: <20111005105243.GA806@linux.vnet.ibm.com> Reply-To: Srikar Dronamraju References: <20110920115938.25326.93059.sendpatchset@srdronam.in.ibm.com> <20110920120057.25326.63780.sendpatchset@srdronam.in.ibm.com> <20111003162905.GA3752@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <20111003162905.GA3752@redhat.com> User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Oleg Nesterov [2011-10-03 18:29:05]: > On 09/20, Srikar Dronamraju wrote: > > > > +static int __copy_insn(struct address_space *mapping, > > + struct vm_area_struct *vma, char *insn, > > + unsigned long nbytes, unsigned long offset) > > +{ > > + struct file *filp = vma->vm_file; > > + struct page *page; > > + void *vaddr; > > + unsigned long off1; > > + unsigned long idx; > > + > > + if (!filp) > > + return -EINVAL; > > + > > + idx = (unsigned long) (offset >> PAGE_CACHE_SHIFT); > > + off1 = offset &= ~PAGE_MASK; > > + > > + /* > > + * Ensure that the page that has the original instruction is > > + * populated and in page-cache. > > + */ > > Hmm. But how we can ensure? > > > + page_cache_sync_readahead(mapping, &filp->f_ra, filp, idx, 1); > > This schedules the i/o, > > > + page = grab_cache_page(mapping, idx); > > This finds/locks the page in the page-cache, > > > + if (!page) > > + return -ENOMEM; > > + > > + vaddr = kmap_atomic(page); > > + memcpy(insn, vaddr + off1, nbytes); > > What if this page is not PageUptodate() ? Since we do a synchronous read ahead, I thought the page would be populated and upto date. would these two lines after grab_cache_page help? if (!PageUptodate(page)) mapping->a_ops->readpage(filp, page); > > Somehow this assumes that the i/o was already completed, I don't > understand this. > > But I am starting to think I simply do not understand this change. > To the point, I do not underestand why do we need copy_insn() at all. > We are going to replace this page, can't we save/analyze ->insn later > when we copy the content of the old page? Most probably I missed > something simple... > > > > +static struct task_struct *get_mm_owner(struct mm_struct *mm) > > +{ > > + struct task_struct *tsk; > > + > > + rcu_read_lock(); > > + tsk = rcu_dereference(mm->owner); > > + if (tsk) > > + get_task_struct(tsk); > > + rcu_read_unlock(); > > + return tsk; > > +} > > Hmm. Do we really need task_struct? > > > -static int install_breakpoint(struct mm_struct *mm, struct uprobe *uprobe) > > +static int install_breakpoint(struct mm_struct *mm, struct uprobe *uprobe, > > + struct vm_area_struct *vma, loff_t vaddr) > > { > > - /* Placeholder: Yet to be implemented */ > > + struct task_struct *tsk; > > + unsigned long addr; > > + int ret = -EINVAL; > > + > > if (!uprobe->consumers) > > return 0; > > > > - atomic_inc(&mm->mm_uprobes_count); > > - return 0; > > + tsk = get_mm_owner(mm); > > + if (!tsk) /* task is probably exiting; bail-out */ > > + return -ESRCH; > > + > > + if (vaddr > TASK_SIZE_OF(tsk)) > > + goto put_return; > > But this should not be possible, no? How it can map this vaddr above > TASK_SIZE ? > > get_user_pages(tsk => NULL) is fine. Why else do we need mm->owner ? > > Probably used by the next patches... Say, is_32bit_app(tsk). This > can use mm->context.ia32_compat (hopefully will be replaced with > MMF_COMPAT). > We used the tsk struct for checking if the application was 32 bit and for calling get_user_pages. Since we can pass NULL to get_user_pages and since we can use mm->context.ia32_compat or MMF_COMPAT, I will remove get_mm_owner, that way we dont need to be dependent on CONFIG_MM_OWNER. -- Thanks and Regards Srikar