Re: [PATCH 0/3] uprobes: teach uprobes to clear MMF_HAS_UPROBES

[Srikar Dronamraju <srikar@linux.vnet.ibm.com>]

* Oleg Nesterov <oleg@redhat.com> [2012-08-19 18:40:08]:

> Hello.
> 
> On top of "kill uprobes_state->count, add MMF_HAS_UPROBES"
> series.
> 
> Once again, I am not sure we really need this (except 1/3).
> Perhaps we could remove uprobe_munmap() instead and forget
> about the false-positive MMF_HAS_UPROBES's.
> 
> OTOH, imho it looks a bit more clean this way. Please comment.


I am agree with the current approach.

The other approach could be to forget about MMF_RECALC_UPROBES and
uprobe_munmap() (essentially patch 2). But in patch 3 we will end up
calling mmf_recalc_uprobes() everytime a process with MM_HAS_UPROBES
hits a non-uprobe breakpoint.


> 
> diffstat reports the code bloat, but note that the previous
> series removed more code.
> 
> Oleg.
> 
>  include/linux/sched.h   |    3 +-
>  kernel/events/uprobes.c |   82 +++++++++++++++++++++++++++++++++++++----------
>  2 files changed, 67 insertions(+), 18 deletions(-)
>