linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] uprobes: Refuse to insert a probe into MAP_SHARED vma
@ 2014-04-28 19:16 Oleg Nesterov
  2014-04-28 19:30 ` Oleg Nesterov
  0 siblings, 1 reply; 2+ messages in thread
From: Oleg Nesterov @ 2014-04-28 19:16 UTC (permalink / raw)
  To: Ingo Molnar
  Cc: Ananth N Mavinakayanahalli, David Long, Peter Zijlstra,
	Srikar Dronamraju, linux-kernel

valid_vma() rejects the VM_SHARED vmas, but this still allows to insert
a probe into the MAP_SHARED but not VM_MAYWRITE vma.

Currently this is fine, such a mapping doesn't really differ from the
private read-only mmap except mprotect(PROT_WRITE) won't work. However,
get_user_pages(FOLL_WRITE | FOLL_FORCE) doesn't allow to COW in this
case, and it would be safer to follow the same conventions as mm even
if currently this happens to work.

After the recent cda540ace6a1 "mm: get_user_pages(write,force) refuse
to COW in shared areas" only uprobes can insert an anon page into the
shared file-backed area, lets stop this and change valid_vma() to check
VM_MAYSHARE instead.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 kernel/events/uprobes.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index d1edc5e..7716c40 100644
--- a/kernel/events/uprobes.c
+++ b/kernel/events/uprobes.c
@@ -127,7 +127,7 @@ struct xol_area {
  */
 static bool valid_vma(struct vm_area_struct *vma, bool is_register)
 {
-	vm_flags_t flags = VM_HUGETLB | VM_MAYEXEC | VM_SHARED;
+	vm_flags_t flags = VM_HUGETLB | VM_MAYEXEC | VM_MAYSHARE;
 
 	if (is_register)
 		flags |= VM_WRITE;
-- 
1.5.5.1



^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] uprobes: Refuse to insert a probe into MAP_SHARED vma
  2014-04-28 19:16 [PATCH] uprobes: Refuse to insert a probe into MAP_SHARED vma Oleg Nesterov
@ 2014-04-28 19:30 ` Oleg Nesterov
  0 siblings, 0 replies; 2+ messages in thread
From: Oleg Nesterov @ 2014-04-28 19:30 UTC (permalink / raw)
  To: Ingo Molnar, Hugh Dickins
  Cc: Ananth N Mavinakayanahalli, David Long, Peter Zijlstra,
	Srikar Dronamraju, linux-kernel

And of course, I forgot to add Hugh ;) Fix that.

On 04/28, Oleg Nesterov wrote:
>
> valid_vma() rejects the VM_SHARED vmas, but this still allows to insert
> a probe into the MAP_SHARED but not VM_MAYWRITE vma.
> 
> Currently this is fine, such a mapping doesn't really differ from the
> private read-only mmap except mprotect(PROT_WRITE) won't work. However,
> get_user_pages(FOLL_WRITE | FOLL_FORCE) doesn't allow to COW in this
> case, and it would be safer to follow the same conventions as mm even
> if currently this happens to work.
> 
> After the recent cda540ace6a1 "mm: get_user_pages(write,force) refuse
> to COW in shared areas" only uprobes can insert an anon page into the
> shared file-backed area, lets stop this and change valid_vma() to check
> VM_MAYSHARE instead.
> 
> Signed-off-by: Oleg Nesterov <oleg@redhat.com>
> ---
>  kernel/events/uprobes.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
> index d1edc5e..7716c40 100644
> --- a/kernel/events/uprobes.c
> +++ b/kernel/events/uprobes.c
> @@ -127,7 +127,7 @@ struct xol_area {
>   */
>  static bool valid_vma(struct vm_area_struct *vma, bool is_register)
>  {
> -	vm_flags_t flags = VM_HUGETLB | VM_MAYEXEC | VM_SHARED;
> +	vm_flags_t flags = VM_HUGETLB | VM_MAYEXEC | VM_MAYSHARE;
>  
>  	if (is_register)
>  		flags |= VM_WRITE;
> -- 
> 1.5.5.1
> 


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2014-04-28 20:21 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-04-28 19:16 [PATCH] uprobes: Refuse to insert a probe into MAP_SHARED vma Oleg Nesterov
2014-04-28 19:30 ` Oleg Nesterov

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).