netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
@ 2023-06-05  7:01 Hangyu Hua
  2023-06-05 12:31 ` Simon Horman
                   ` (2 more replies)
  0 siblings, 3 replies; 7+ messages in thread
From: Hangyu Hua @ 2023-06-05  7:01 UTC (permalink / raw)
  To: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni
  Cc: netdev, linux-kernel, Hangyu Hua

try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
implement the corresponding function we should call module_put to drop
the refcount.

Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
---
 net/sched/cls_api.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 2621550bfddc..92bfb892e638 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
 		return PTR_ERR(ops);
 	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
 		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
+		module_put(ops->owner);
 		return -EOPNOTSUPP;
 	}
 
-- 
2.34.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-05  7:01 [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add() Hangyu Hua
@ 2023-06-05 12:31 ` Simon Horman
  2023-06-06  2:13   ` Hangyu Hua
  2023-06-06  8:53 ` Larysa Zaremba
  2023-06-06 11:21 ` Jiri Pirko
  2 siblings, 1 reply; 7+ messages in thread
From: Simon Horman @ 2023-06-05 12:31 UTC (permalink / raw)
  To: Hangyu Hua
  Cc: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

On Mon, Jun 05, 2023 at 03:01:58PM +0800, Hangyu Hua wrote:
> try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
> implement the corresponding function we should call module_put to drop
> the refcount.

Hi Hangyu Hua,

Is this correct even if try_module_get() is
not called via tcf_proto_lookup_ops() ?

> Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
> ---
>  net/sched/cls_api.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
> index 2621550bfddc..92bfb892e638 100644
> --- a/net/sched/cls_api.c
> +++ b/net/sched/cls_api.c
> @@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
>  		return PTR_ERR(ops);
>  	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
>  		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
> +		module_put(ops->owner);
>  		return -EOPNOTSUPP;
>  	}
>  
> -- 
> 2.34.1
> 
> 

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-05 12:31 ` Simon Horman
@ 2023-06-06  2:13   ` Hangyu Hua
  2023-06-06  9:33     ` Simon Horman
  0 siblings, 1 reply; 7+ messages in thread
From: Hangyu Hua @ 2023-06-06  2:13 UTC (permalink / raw)
  To: Simon Horman
  Cc: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

On 5/6/2023 20:31, Simon Horman wrote:
> On Mon, Jun 05, 2023 at 03:01:58PM +0800, Hangyu Hua wrote:
>> try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
>> implement the corresponding function we should call module_put to drop
>> the refcount.
> 
> Hi Hangyu Hua,
> 
> Is this correct even if try_module_get() is
> not called via tcf_proto_lookup_ops() ?
> 

tcf_proto_lookup_ops will return error if try_module_get() is not called 
in tcf_proto_lookup_ops(). I am not sure what you mean?

Thanks,
Hangyu

>> Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
>> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
>> ---
>>   net/sched/cls_api.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
>> index 2621550bfddc..92bfb892e638 100644
>> --- a/net/sched/cls_api.c
>> +++ b/net/sched/cls_api.c
>> @@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
>>   		return PTR_ERR(ops);
>>   	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
>>   		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
>> +		module_put(ops->owner);
>>   		return -EOPNOTSUPP;
>>   	}
>>   
>> -- 
>> 2.34.1
>>
>>

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-05  7:01 [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add() Hangyu Hua
  2023-06-05 12:31 ` Simon Horman
@ 2023-06-06  8:53 ` Larysa Zaremba
  2023-06-07  2:20   ` Hangyu Hua
  2023-06-06 11:21 ` Jiri Pirko
  2 siblings, 1 reply; 7+ messages in thread
From: Larysa Zaremba @ 2023-06-06  8:53 UTC (permalink / raw)
  To: Hangyu Hua
  Cc: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

On Mon, Jun 05, 2023 at 03:01:58PM +0800, Hangyu Hua wrote:
> try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
> implement the corresponding function we should call module_put to drop
> the refcount.
> 

Code seems reasonable. But commit message is pretty hard to understand.
Please, replace "corresponding" with "required".
Also change the first sentence, do not use "can". From what I see, successful
execution of tcf_proto_lookup_ops always means we now hold reference to module.

CC me in v2, I'll give you Reviewed-by.

> Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
> ---
>  net/sched/cls_api.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
> index 2621550bfddc..92bfb892e638 100644
> --- a/net/sched/cls_api.c
> +++ b/net/sched/cls_api.c
> @@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
>  		return PTR_ERR(ops);
>  	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
>  		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
> +		module_put(ops->owner);
>  		return -EOPNOTSUPP;
>  	}
>  
> -- 
> 2.34.1
> 
> 

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-06  2:13   ` Hangyu Hua
@ 2023-06-06  9:33     ` Simon Horman
  0 siblings, 0 replies; 7+ messages in thread
From: Simon Horman @ 2023-06-06  9:33 UTC (permalink / raw)
  To: Hangyu Hua
  Cc: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

On Tue, Jun 06, 2023 at 10:13:44AM +0800, Hangyu Hua wrote:
> On 5/6/2023 20:31, Simon Horman wrote:
> > On Mon, Jun 05, 2023 at 03:01:58PM +0800, Hangyu Hua wrote:
> > > try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
> > > implement the corresponding function we should call module_put to drop
> > > the refcount.
> > 
> > Hi Hangyu Hua,
> > 
> > Is this correct even if try_module_get() is
> > not called via tcf_proto_lookup_ops() ?
> > 
> 
> tcf_proto_lookup_ops will return error if try_module_get() is not called in
> tcf_proto_lookup_ops(). I am not sure what you mean?

Thanks,

I think that answers my question.

My concern was if there is a situation where module_put() is
now called, but there was not in fact a reference that should
be released.

...

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-05  7:01 [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add() Hangyu Hua
  2023-06-05 12:31 ` Simon Horman
  2023-06-06  8:53 ` Larysa Zaremba
@ 2023-06-06 11:21 ` Jiri Pirko
  2 siblings, 0 replies; 7+ messages in thread
From: Jiri Pirko @ 2023-06-06 11:21 UTC (permalink / raw)
  To: Hangyu Hua
  Cc: jhs, xiyou.wangcong, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

Mon, Jun 05, 2023 at 09:01:58AM CEST, hbh25y@gmail.com wrote:
>try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
>implement the corresponding function we should call module_put to drop
>the refcount.

Who's "we"? Use imperative mood. Tell the codebase what to do, what to
change, etc.

Code-wise, this is fine. Please fix the patch description.


>
>Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
>Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
>---
> net/sched/cls_api.c | 1 +
> 1 file changed, 1 insertion(+)
>
>diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
>index 2621550bfddc..92bfb892e638 100644
>--- a/net/sched/cls_api.c
>+++ b/net/sched/cls_api.c
>@@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
> 		return PTR_ERR(ops);
> 	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
> 		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
>+		module_put(ops->owner);
> 		return -EOPNOTSUPP;
> 	}
> 
>-- 
>2.34.1
>

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add()
  2023-06-06  8:53 ` Larysa Zaremba
@ 2023-06-07  2:20   ` Hangyu Hua
  0 siblings, 0 replies; 7+ messages in thread
From: Hangyu Hua @ 2023-06-07  2:20 UTC (permalink / raw)
  To: Larysa Zaremba
  Cc: jhs, xiyou.wangcong, jiri, davem, edumazet, kuba, pabeni, netdev,
	linux-kernel

On 6/6/2023 16:53, Larysa Zaremba wrote:
> On Mon, Jun 05, 2023 at 03:01:58PM +0800, Hangyu Hua wrote:
>> try_module_get can be called in tcf_proto_lookup_ops. So if ops don't
>> implement the corresponding function we should call module_put to drop
>> the refcount.
>>
> 
> Code seems reasonable. But commit message is pretty hard to understand.
> Please, replace "corresponding" with "required".
> Also change the first sentence, do not use "can". From what I see, successful
> execution of tcf_proto_lookup_ops always means we now hold reference to module.
> 
> CC me in v2, I'll give you Reviewed-by.
> 

I apologize for my incorrect English expression. I will send a v2 later.

Thanks,
Hangyu

>> Fixes: 9f407f1768d3 ("net: sched: introduce chain templates")
>> Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
>> ---
>>   net/sched/cls_api.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
>> index 2621550bfddc..92bfb892e638 100644
>> --- a/net/sched/cls_api.c
>> +++ b/net/sched/cls_api.c
>> @@ -2952,6 +2952,7 @@ static int tc_chain_tmplt_add(struct tcf_chain *chain, struct net *net,
>>   		return PTR_ERR(ops);
>>   	if (!ops->tmplt_create || !ops->tmplt_destroy || !ops->tmplt_dump) {
>>   		NL_SET_ERR_MSG(extack, "Chain templates are not supported with specified classifier");
>> +		module_put(ops->owner);
>>   		return -EOPNOTSUPP;
>>   	}
>>   
>> -- 
>> 2.34.1
>>
>>

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2023-06-07  2:20 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-06-05  7:01 [PATCH net] net: sched: fix possible refcount leak in tc_chain_tmplt_add() Hangyu Hua
2023-06-05 12:31 ` Simon Horman
2023-06-06  2:13   ` Hangyu Hua
2023-06-06  9:33     ` Simon Horman
2023-06-06  8:53 ` Larysa Zaremba
2023-06-07  2:20   ` Hangyu Hua
2023-06-06 11:21 ` Jiri Pirko

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).