Patchwork net: fix incorrect counting in __scm_destroy()

login
register
mail settings
Submitter Amerigo Wang
Date Nov. 4, 2009, 10:04 a.m.
Message ID <20091104100717.4785.57149.sendpatchset@localhost.localdomain>
Download mbox | patch
Permalink /patch/37548/
State Rejected
Delegated to: David Miller
Headers show

Comments

Amerigo Wang - Nov. 4, 2009, 10:04 a.m.
It seems that in __scm_destroy() we forgot to decrease
the ->count after fput(->fp[i]), this may cause some
problem when we recursively call fput() again.

Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: David S. Miller <davem@davemloft.net>

---
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet - Nov. 4, 2009, 10:29 a.m.
Amerigo Wang a écrit :
> It seems that in __scm_destroy() we forgot to decrease
> the ->count after fput(->fp[i]), this may cause some
> problem when we recursively call fput() again.
> 
> Signed-off-by: WANG Cong <amwang@redhat.com>
> Cc: David S. Miller <davem@davemloft.net>
> 
> ---
> diff --git a/net/core/scm.c b/net/core/scm.c
> index b7ba91b..fa53219 100644
> --- a/net/core/scm.c
> +++ b/net/core/scm.c
> @@ -120,8 +120,10 @@ void __scm_destroy(struct scm_cookie *scm)
>  				fpl = list_first_entry(&work_list, struct scm_fp_list, list);
>  
>  				list_del(&fpl->list);
> -				for (i=fpl->count-1; i>=0; i--)
> +				for (i = fpl->count-1; i >= 0; i--) {
>  					fput(fpl->fp[i]);
> +					fpl->count--;
> +				}
>  				kfree(fpl);
>  			}
>  

Hmm, your patch seems suspicious.

Are you fixing a real crash/bug, or is it something you discovered in a code review ?

Given we kfree(fpl) at the end of loop, we cannot recursively call __scm_destroy()
on same fpl, it would be a bug anyway ?

So you probably need something better, like testing fpl->list being not re-included
in current->scm_work_list before kfree() it 

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - Nov. 4, 2009, 12:41 p.m.
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 04 Nov 2009 11:29:05 +0100

> Given we kfree(fpl) at the end of loop, we cannot recursively call
> __scm_destroy() on same fpl, it would be a bug anyway ?
> 
> So you probably need something better, like testing fpl->list being
> not re-included in current->scm_work_list before kfree() it

I can't even see what the problem is.

The code is designed such that the ->count only matters for
the top level.

If we recursively fput() and get back here, we'll see that
there is someone higher in the call chain already running
the fput() loop and we'll just list_add_tail().

The inner while() loop will make sure we process such
entries once we get back to the top level and exit the
for() loop.

Amerigo, please show us the problematic code path where the counts go
wrong and this causes problems.

Thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Amerigo Wang - Nov. 10, 2009, 6:12 a.m.
David Miller wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Wed, 04 Nov 2009 11:29:05 +0100
> 
>> Given we kfree(fpl) at the end of loop, we cannot recursively call
>> __scm_destroy() on same fpl, it would be a bug anyway ?
>>
>> So you probably need something better, like testing fpl->list being
>> not re-included in current->scm_work_list before kfree() it
> 
> I can't even see what the problem is.
> 
> The code is designed such that the ->count only matters for
> the top level.
> 
> If we recursively fput() and get back here, we'll see that
> there is someone higher in the call chain already running
> the fput() loop and we'll just list_add_tail().
> 
> The inner while() loop will make sure we process such
> entries once we get back to the top level and exit the
> for() loop.
> 
> Amerigo, please show us the problematic code path where the counts go
> wrong and this causes problems.

Hi, all.

Thanks for your replies.

I met a soft lockup around this code on ia64, something like:

  [<a0000001006394e0>] unix_gc+0x240/0x760
                                 sp=e0000260f002fd70 bsp=e0000260f0029560
  [<a000000100634500>] unix_release_sock+0x440/0x460
                                 sp=e0000260f002fdb0 bsp=e0000260f0029508
  [<a000000100634560>] unix_release+0x40/0x60
                                 sp=e0000260f002fdb0 bsp=e0000260f00294e8
  [<a00000010051fba0>] sock_release+0x80/0x1c0
                                 sp=e0000260f002fdb0 bsp=e0000260f00294c0
  [<a00000010051fd60>] sock_close+0x80/0xa0
                                 sp=e0000260f002fdc0 bsp=e0000260f0029498
  [<a000000100172280>] __fput+0x1a0/0x420
                                 sp=e0000260f002fdc0 bsp=e0000260f0029458
  [<a000000100172540>] fput+0x40/0x60
                                 sp=e0000260f002fdc0 bsp=e0000260f0029438
  [<a000000100534a30>] __scm_destroy+0x130/0x1e0
                                 sp=e0000260f002fdc0 bsp=e0000260f0029410
  [<a000000100636370>] unix_destruct_fds+0x70/0xa0
                                 sp=e0000260f002fdd0 bsp=e0000260f00293e8
  [<a00000010052da30>] __kfree_skb+0x1f0/0x320
                                 sp=e0000260f002fe00 bsp=e0000260f00293c0
  [<a00000010052dbf0>] kfree_skb+0x90/0xc0
                                 sp=e0000260f002fe00 bsp=e0000260f00293a0
  [<a000000100634420>] unix_release_sock+0x360/0x460
                                 sp=e0000260f002fe00 bsp=e0000260f0029348
  [<a000000100634560>] unix_release+0x40/0x60
                                 sp=e0000260f002fe00 bsp=e0000260f0029328
  [<a00000010051fba0>] sock_release+0x80/0x1c0
                                 sp=e0000260f002fe00 bsp=e0000260f0029300
  [<a00000010051fd60>] sock_close+0x80/0xa0
                                 sp=e0000260f002fe10 bsp=e0000260f00292d8
  [<a000000100172280>] __fput+0x1a0/0x420
                                 sp=e0000260f002fe10 bsp=e0000260f0029298
  [<a000000100172540>] fput+0x40/0x60
                                 sp=e0000260f002fe10 bsp=e0000260f0029278


Yes, this even happens after commit f8d570a47.

But after doing a bisect, we found another hrtimer patch fixes this
problem, so it's not a bug of __scm_destroy().

Sorry for the noise.

Thanks.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet - Nov. 10, 2009, 6:33 a.m.
Cong Wang a écrit :
> 
> Yes, this even happens after commit f8d570a47.
> 
> But after doing a bisect, we found another hrtimer patch fixes this
> problem, so it's not a bug of __scm_destroy().
> 
> Sorry for the noise.
> 

Thanks for the explanation !
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/core/scm.c b/net/core/scm.c
index b7ba91b..fa53219 100644
--- a/net/core/scm.c
+++ b/net/core/scm.c
@@ -120,8 +120,10 @@  void __scm_destroy(struct scm_cookie *scm)
 				fpl = list_first_entry(&work_list, struct scm_fp_list, list);
 
 				list_del(&fpl->list);
-				for (i=fpl->count-1; i>=0; i--)
+				for (i = fpl->count-1; i >= 0; i--) {
 					fput(fpl->fp[i]);
+					fpl->count--;
+				}
 				kfree(fpl);
 			}