This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: [RFC] [PATCH 2.6.37-rc5-tip 5/20] 5: Uprobes: register/unregister probes.
- From: Peter Zijlstra <peterz at infradead dot org>
- To: Srikar Dronamraju <srikar at linux dot vnet dot ibm dot com>
- 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>, Andrew Morton <akpm at linux-foundation 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>, LKML <linux-kernel at vger dot kernel dot org>, "Paul E. McKenney" <paulmck at linux dot vnet dot ibm dot com>
- Date: Thu, 27 Jan 2011 11:29:36 +0100
- Subject: Re: [RFC] [PATCH 2.6.37-rc5-tip 5/20] 5: Uprobes: register/unregister probes.
- References: <20101216095714.23751.52601.sendpatchset@localhost6.localdomain6> <20101216095817.23751.76989.sendpatchset@localhost6.localdomain6> <1295957744.28776.722.camel@laptop> <20110126075558.GB19725@linux.vnet.ibm.com> <1296036708.28776.1138.camel@laptop> <20110126153036.GN19725@linux.vnet.ibm.com> <1296056756.28776.1247.camel@laptop> <20110126165645.GP19725@linux.vnet.ibm.com> <1296061949.28776.1343.camel@laptop> <20110127100157.GS19725@linux.vnet.ibm.com>
On Thu, 2011-01-27 at 15:31 +0530, Srikar Dronamraju wrote:
>
> > You can, if only to wreck your thing, you can call mmap() as often as
> > you like (until your virtual memory space runs out) and get many many
> > mapping of the same file.
> >
> > It doesn't need to make sense to the linker, all it needs to do is
> > confuse your code ;-)
>
> Currently if there are multiple mappings of the same executable
> code, only one mapped area would have the breakpoint inserted.
Right, so you could use it to make debugging harder..
> If the code were to execute from some other mapping, then it would
> work as if there are no probes. However if the code from the
> mapping that had the breakpoint executes then we would see the
> probes.
>
> If we want to insert breakpoints in each of the maps then we
> would have to extend mm->uprobes_vaddr.
>
> Do you have any other ideas to tackle this?
Supposing I can get my preemptible mmu patches anywhere.. you could
simply call install_uprobe() while holding the i_mmap_mutex ;-)
> Infact do you think we should be handling this case?
I'm really not sure how often this would happen, but dealing with it
sure makes me feel better..