Messages in this thread | | | Date | Fri, 20 Apr 2012 20:37:18 +0200 | From | Oleg Nesterov <> | Subject | Re: [RFC 0/6] uprobes: kill uprobes_srcu/uprobe_srcu_id |
| |
On 04/20, Peter Zijlstra wrote: > > On Mon, 2012-04-16 at 23:47 +0200, Oleg Nesterov wrote: > > (And perhaps we can stop right here? I mean how often this can > > slow down the debugger which installs int3 in the same mm?) > > > > Now we need to clear MMF_HAS_UPROBE somehowe, when the last > > uprobe goes away. Lets ignore uprobe_map/unmap for simplicity. > > > > - We add another flag, MMF_UPROBE_RECALC, it is set by > > remove_breakpoint(). > > > > - We change handle_swbp(). Ignoring all details it does: > > > > if (find_uprobe(vaddr)) > > process_uprobe(); > > else if (test_bit(MMF_HAS_UPROBE) && test_bit(MMF_UPROBE_RECALC)) > > recalc_mmf_uprobe_flag(); > > > > where recalc_mmf_uprobe_flag() checks all vmas and either > > clears both flags or MMF_UPROBE_RECALC only. > > > > This is the really slow O(n) path, but it can only happen after > > unregister, and only if we hit another non-uprobe breakpoint > > in the same mm. > > > > Something like this. What do you think? > > I think I can live with the simple set MMF_HAS_UPROBE and leave it at > that.
Sure, I agree.
A false positive MMF_HAS_UPROBE can only slow down the non-uprobe int3 in the same ->mm, I think we can tolerate this.
> The better optimization seems to be to not install breakpoints > when ->filter() excludes the task..
Ah, this is another story. And I agree this is more important.
So far I do not understand what we should do. Of course, it would be simple to add the filtering when we install the breakpoint but I don't think it is that simple, even if we ignore the nasty complications with multiple consumers with different filters.
Say, a user wants to probe /sbin/init only. What if init forks? We should remove breakpoints from child->mm somehow.
And then we also need the filtering in uprobe_mmap() at least.
But yes, I agree, it would be very nice to do this.
Oleg.
| |