This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [RFC] [PATCH 2.6.37-rc5-tip 8/20] 8: uprobes: mmap and fork hooks.
- From: Srikar Dronamraju <srikar at linux dot vnet dot ibm dot com>
- To: Peter Zijlstra <peterz at infradead dot org>
- Cc: Ingo Molnar <mingo at elte dot hu>, Steven Rostedt <rostedt at goodmis dot org>, Linux-mm <linux-mm at kvack dot org>, Arnaldo Carvalho de Melo <acme at infradead dot org>, Linus Torvalds <torvalds at linux-foundation dot org>, Ananth N Mavinakayanahalli <ananth at in dot ibm dot com>, Christoph Hellwig <hch at infradead dot org>, Masami Hiramatsu <masami dot hiramatsu dot pt at hitachi dot com>, Oleg Nesterov <oleg at redhat dot com>, LKML <linux-kernel at vger dot kernel dot org>, SystemTap <systemtap at sources dot redhat dot com>, Jim Keniston <jkenisto at linux dot vnet dot ibm dot com>, Frederic Weisbecker <fweisbec at gmail dot com>, Andi Kleen <andi at firstfloor dot org>, Andrew Morton <akpm at linux-foundation dot org>, "Paul E. McKenney" <paulmck at linux dot vnet dot ibm dot com>
- Date: Wed, 26 Jan 2011 14:33:46 +0530
- Subject: Re: [RFC] [PATCH 2.6.37-rc5-tip 8/20] 8: uprobes: mmap and fork hooks.
- References: <20101216095714.23751.52601.sendpatchset@localhost6.localdomain6> <20101216095848.23751.73144.sendpatchset@localhost6.localdomain6> <1295957739.28776.717.camel@laptop>
- Reply-to: Srikar Dronamraju <srikar at linux dot vnet dot ibm dot com>
> On Thu, 2010-12-16 at 15:28 +0530, Srikar Dronamraju wrote:
> > +void uprobe_mmap(struct vm_area_struct *vma)
> > +{
> > + struct list_head tmp_list;
> > + struct uprobe *uprobe, *u;
> > + struct mm_struct *mm;
> > + struct inode *inode;
> > +
> > + if (!valid_vma(vma))
> > + return;
> > +
> > + INIT_LIST_HEAD(&tmp_list);
> > +
> > + /*
> > + * The vma was just allocated and this routine gets called
> > + * while holding write lock for mmap_sem. Function called
> > + * in context of a thread that has a reference to mm.
> > + * Hence no need to take a reference to mm
> > + */
> > + mm = vma->vm_mm;
> > + up_write(&mm->mmap_sem);
>
> Are you very very sure its a good thing to simply drop the mmap_sem
> here? Also, why?
>
I actually dont like to release the write_lock and then reacquire it.
write_opcode, which is called thro install_uprobe, i.e to insert the
actual breakpoint instruction takes a read lock on the mmap_sem.
Hence uprobe_mmap gets called in context with write lock on mmap_sem
held, I had to release it before calling install_uprobe.
Another solution, I thought of was to pass a context to write_opcode to
say that map-sem is already acquired by us. But I am not sure that
idea is good enuf.
> > + mutex_lock(&uprobes_mutex);
> > +
> > + inode = vma->vm_file->f_mapping->host;
>
> Since you just dropped the mmap_sem, what's keeping that vma from going
> away?
>
How about dropping the mmap_sem after add_to_temp_list and cachng the
vma->vm_start value before calling add_to_temp_list?
Or if you have better ideas, then that would be great.
> > + add_to_temp_list(vma, inode, &tmp_list);
> > +
> > + list_for_each_entry_safe(uprobe, u, &tmp_list, pending_list) {
> > + mm->uprobes_vaddr = vma->vm_start + uprobe->offset;
> > + install_uprobe(mm, uprobe);
> > + list_del(&uprobe->pending_list);
> > + }
> > + mutex_unlock(&uprobes_mutex);
> > + down_write(&mm->mmap_sem);
> > +}
>
>