[dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[John Daley]

rte_flow structures were not being freed when destroyed or flushed.

Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled")
Cc: stable@dpdk.org

Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
---
 drivers/net/enic/enic_flow.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/enic/enic_flow.c b/drivers/net/enic/enic_flow.c
index 0cf04aefd..9b612f1d5 100644
--- a/drivers/net/enic/enic_flow.c
+++ b/drivers/net/enic/enic_flow.c
@@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev, struct rte_flow *flow,
 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
 	LIST_REMOVE(flow, next);
 	rte_spinlock_unlock(&enic->flows_lock);
+	rte_free(flow);
 	return 0;
 }
 
@@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct rte_flow_error *error)
 		flow = LIST_FIRST(&enic->flows);
 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
 		LIST_REMOVE(flow, next);
+		rte_free(flow);
 	}
 	rte_spinlock_unlock(&enic->flows_lock);
 	return 0;
-- 
2.16.2

[dpdk-stable] [PATCH v2 1/3] net/enic: fix flow API memory leak

[John Daley]

rte_flow structures were not being freed when destroyed or flushed.

Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled")
Cc: stable@dpdk.org

Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Signed-off-by: John Daley <johndale@cisco.com>
---
v2: fix signoff

 drivers/net/enic/enic_flow.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/enic/enic_flow.c b/drivers/net/enic/enic_flow.c
index 0cf04aefd..9b612f1d5 100644
--- a/drivers/net/enic/enic_flow.c
+++ b/drivers/net/enic/enic_flow.c
@@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev, struct rte_flow *flow,
 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
 	LIST_REMOVE(flow, next);
 	rte_spinlock_unlock(&enic->flows_lock);
+	rte_free(flow);
 	return 0;
 }
 
@@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct rte_flow_error *error)
 		flow = LIST_FIRST(&enic->flows);
 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
 		LIST_REMOVE(flow, next);
+		rte_free(flow);
 	}
 	rte_spinlock_unlock(&enic->flows_lock);
 	return 0;
-- 
2.16.2

Re: [dpdk-stable] [dpdk-dev] [PATCH v2 1/3] net/enic: fix flow API memory leak

[Ferruh Yigit]

On 9/28/2018 4:08 AM, John Daley wrote:
> rte_flow structures were not being freed when destroyed or flushed.
> 
> Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> Signed-off-by: John Daley <johndale@cisco.com>

Series applied to dpdk-next-net/master, thanks.

(Patch 3/3 merged into 2/3 while merging, to keep relevant documentation and
implementation is same patch.)

Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[John Daley (johndale)]

Got it, thanks Yongseok for the clarification. I may have sent it twice by accident. 
-john

> -----Original Message-----
> From: Yongseok Koh <yskoh@mellanox.com>
> Sent: Thursday, October 04, 2018 4:40 PM
> To: John Daley (johndale) <johndale@cisco.com>
> Cc: johnda888@gmail.com; dpdk stable <stable@dpdk.org>; Hyong Youb Kim
> (hyonkim) <hyonkim@cisco.com>
> Subject: Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak
> 
> 
> > On Oct 4, 2018, at 3:03 PM, John Daley (johndale) <johndale@cisco.com>
> wrote:
> >
> > Hi Yongseok,
> >
> > It should go into 17.11 and hopefully 18.02, 18.05, 18.08 eventually.
> >
> > Not sure what  you mean when you say it was only for stable releases.
> > It was submitted to dev@ with the Cc: stable@ in the commit. The patch
> > is here:
> >
> https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpat
> >
> ches.dpdk.org%2Fpatch%2F45545%2F&amp;data=02%7C01%7Cyskoh%40me
> llanox.c
> >
> om%7Cb5541c8c6a2f4da3a1b908d62a454896%7Ca652971c7d2e4d9ba6a4d1
> 49256f46
> >
> 1b%7C0%7C0%7C636742874397325810&amp;sdata=wBWZ13CGeDZB3dnm%
> 2Bv4fXivc%2
> > BRQGRCpotawCQ3oE6PU%3D&amp;reserved=0
> >
> > When cc'ing stable in dev@ submissions, should we generally follow up
> with an email to stable@ to indicate which stable branches to target? If so,
> should we update the instructions in the contributors guide (section 6.4), or
> did I miss where that process is documented?
> 
> If you send out a patch to dev@dpdk.org with the tag (Cc: stable@dpdk.org),
> the patch will automatically be queued for stable branches. But the reason
> why I asked the question was this specific one has been sent to
> stable@dpdk.org without dev@dpdk.org. It looks you've sent the same patch
> twice?
> 
> Summary.
> 
> 1) If you want to fix a bug in the current upstream (dpdk/master), send the
> fix to dev@dpdk.org with the stable cc tag, then it will be automatically
> queued to be backported to a stable branch if the bug is in the stable branch.
> If you don't think backporting is necessary, you don't need to add the stable
> cc tag.
> Then, nothing will happen.
> 
> 2) If you want to fix a bug and the bug exists only in a stable branch, not in
> the master branch, then you should send it only to stable@dpdk.org with a
> prefix. Please refer to "5.11.1. Backporting patches for Stable Releases" in the
> guide.  E.g.,
> 	git send-email --subject-prefix='PATCH 16.11' ...
> If the bug exists in multiple stable branches, you should send it multiple
> times.
> 
> 
> Thanks,
> Yongseok
> 
> > Thanks,
> > John
> >
> >> -----Original Message-----
> >> From: Yongseok Koh <yskoh@mellanox.com>
> >> Sent: Thursday, October 04, 2018 2:36 PM
> >> To: John Daley (johndale) <johndale@cisco.com>
> >> Cc: johnda888@gmail.com; dpdk stable <stable@dpdk.org>; Hyong Youb
> >> Kim
> >> (hyonkim) <hyonkim@cisco.com>
> >> Subject: Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory
> >> leak
> >>
> >>
> >>> On Sep 27, 2018, at 6:51 PM, John Daley <johndale@cisco.com> wrote:
> >>>
> >>> rte_flow structures were not being freed when destroyed or flushed.
> >>>
> >>> Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced
> >>> filters enabled")
> >>> Cc: stable@dpdk.org
> >>>
> >>> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> >>> Reviewed-by: John Daley <johndale@cisco.com>
> >>> ---
> >>
> >> Hi,
> >>
> >> Is this patch only for stable releases? If so, which one? Or was dev@
> >> absent by mistake?
> >>
> >>
> >> Thanks,
> >> Yongseok
> >>
> >>> drivers/net/enic/enic_flow.c | 2 ++
> >>> 1 file changed, 2 insertions(+)
> >>>
> >>> diff --git a/drivers/net/enic/enic_flow.c
> >>> b/drivers/net/enic/enic_flow.c index 0cf04aefd..9b612f1d5 100644
> >>> --- a/drivers/net/enic/enic_flow.c
> >>> +++ b/drivers/net/enic/enic_flow.c
> >>> @@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev,
> >> struct rte_flow *flow,
> >>> 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
> >>> 	LIST_REMOVE(flow, next);
> >>> 	rte_spinlock_unlock(&enic->flows_lock);
> >>> +	rte_free(flow);
> >>> 	return 0;
> >>> }
> >>>
> >>> @@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev,
> >>> struct
> >> rte_flow_error *error)
> >>> 		flow = LIST_FIRST(&enic->flows);
> >>> 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
> >>> 		LIST_REMOVE(flow, next);
> >>> +		rte_free(flow);
> >>> 	}
> >>> 	rte_spinlock_unlock(&enic->flows_lock);
> >>> 	return 0;
> >>> --
> >>> 2.16.2
> >>>
> >

Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[Yongseok Koh]

> On Oct 4, 2018, at 3:03 PM, John Daley (johndale) <johndale@cisco.com> wrote:
> 
> Hi Yongseok,
> 
> It should go into 17.11 and hopefully 18.02, 18.05, 18.08 eventually.
> 
> Not sure what  you mean when you say it was only for stable releases. It was submitted to dev@ with the Cc: stable@ in the commit. The patch is here: https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpatches.dpdk.org%2Fpatch%2F45545%2F&amp;data=02%7C01%7Cyskoh%40mellanox.com%7Cb5541c8c6a2f4da3a1b908d62a454896%7Ca652971c7d2e4d9ba6a4d149256f461b%7C0%7C0%7C636742874397325810&amp;sdata=wBWZ13CGeDZB3dnm%2Bv4fXivc%2BRQGRCpotawCQ3oE6PU%3D&amp;reserved=0
> 
> When cc'ing stable in dev@ submissions, should we generally follow up with an email to stable@ to indicate which stable branches to target? If so, should we update the instructions in the contributors guide (section 6.4), or did I miss where that process is documented?

If you send out a patch to dev@dpdk.org with the tag (Cc: stable@dpdk.org), the
patch will automatically be queued for stable branches. But the reason why I
asked the question was this specific one has been sent to stable@dpdk.org
without dev@dpdk.org. It looks you've sent the same patch twice?

Summary.

1) If you want to fix a bug in the current upstream (dpdk/master), send the fix
to dev@dpdk.org with the stable cc tag, then it will be automatically queued to
be backported to a stable branch if the bug is in the stable branch. If you
don't think backporting is necessary, you don't need to add the stable cc tag.
Then, nothing will happen.

2) If you want to fix a bug and the bug exists only in a stable branch, not in
the master branch, then you should send it only to stable@dpdk.org with a
prefix. Please refer to "5.11.1. Backporting patches for Stable Releases" in
the guide.  E.g.,
	git send-email --subject-prefix='PATCH 16.11' ...
If the bug exists in multiple stable branches, you should send it multiple times.


Thanks,
Yongseok

> Thanks,
> John
> 
>> -----Original Message-----
>> From: Yongseok Koh <yskoh@mellanox.com>
>> Sent: Thursday, October 04, 2018 2:36 PM
>> To: John Daley (johndale) <johndale@cisco.com>
>> Cc: johnda888@gmail.com; dpdk stable <stable@dpdk.org>; Hyong Youb Kim
>> (hyonkim) <hyonkim@cisco.com>
>> Subject: Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak
>> 
>> 
>>> On Sep 27, 2018, at 6:51 PM, John Daley <johndale@cisco.com> wrote:
>>> 
>>> rte_flow structures were not being freed when destroyed or flushed.
>>> 
>>> Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced
>>> filters enabled")
>>> Cc: stable@dpdk.org
>>> 
>>> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
>>> Reviewed-by: John Daley <johndale@cisco.com>
>>> ---
>> 
>> Hi,
>> 
>> Is this patch only for stable releases? If so, which one? Or was dev@ absent
>> by mistake?
>> 
>> 
>> Thanks,
>> Yongseok
>> 
>>> drivers/net/enic/enic_flow.c | 2 ++
>>> 1 file changed, 2 insertions(+)
>>> 
>>> diff --git a/drivers/net/enic/enic_flow.c
>>> b/drivers/net/enic/enic_flow.c index 0cf04aefd..9b612f1d5 100644
>>> --- a/drivers/net/enic/enic_flow.c
>>> +++ b/drivers/net/enic/enic_flow.c
>>> @@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev,
>> struct rte_flow *flow,
>>> 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
>>> 	LIST_REMOVE(flow, next);
>>> 	rte_spinlock_unlock(&enic->flows_lock);
>>> +	rte_free(flow);
>>> 	return 0;
>>> }
>>> 
>>> @@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct
>> rte_flow_error *error)
>>> 		flow = LIST_FIRST(&enic->flows);
>>> 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
>>> 		LIST_REMOVE(flow, next);
>>> +		rte_free(flow);
>>> 	}
>>> 	rte_spinlock_unlock(&enic->flows_lock);
>>> 	return 0;
>>> --
>>> 2.16.2
>>> 
> 

Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[John Daley (johndale)]

Hi Yongseok,

It should go into 17.11 and hopefully 18.02, 18.05, 18.08 eventually.

Not sure what  you mean when you say it was only for stable releases. It was submitted to dev@ with the Cc: stable@ in the commit. The patch is here: https://patches.dpdk.org/patch/45545/

When cc'ing stable in dev@ submissions, should we generally follow up with an email to stable@ to indicate which stable branches to target? If so, should we update the instructions in the contributors guide (section 6.4), or did I miss where that process is documented?

Thanks,
John

> -----Original Message-----
> From: Yongseok Koh <yskoh@mellanox.com>
> Sent: Thursday, October 04, 2018 2:36 PM
> To: John Daley (johndale) <johndale@cisco.com>
> Cc: johnda888@gmail.com; dpdk stable <stable@dpdk.org>; Hyong Youb Kim
> (hyonkim) <hyonkim@cisco.com>
> Subject: Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak
> 
> 
> > On Sep 27, 2018, at 6:51 PM, John Daley <johndale@cisco.com> wrote:
> >
> > rte_flow structures were not being freed when destroyed or flushed.
> >
> > Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced
> > filters enabled")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> > Reviewed-by: John Daley <johndale@cisco.com>
> > ---
> 
> Hi,
> 
> Is this patch only for stable releases? If so, which one? Or was dev@ absent
> by mistake?
> 
> 
> Thanks,
> Yongseok
> 
> > drivers/net/enic/enic_flow.c | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/net/enic/enic_flow.c
> > b/drivers/net/enic/enic_flow.c index 0cf04aefd..9b612f1d5 100644
> > --- a/drivers/net/enic/enic_flow.c
> > +++ b/drivers/net/enic/enic_flow.c
> > @@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev,
> struct rte_flow *flow,
> > 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
> > 	LIST_REMOVE(flow, next);
> > 	rte_spinlock_unlock(&enic->flows_lock);
> > +	rte_free(flow);
> > 	return 0;
> > }
> >
> > @@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct
> rte_flow_error *error)
> > 		flow = LIST_FIRST(&enic->flows);
> > 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
> > 		LIST_REMOVE(flow, next);
> > +		rte_free(flow);
> > 	}
> > 	rte_spinlock_unlock(&enic->flows_lock);
> > 	return 0;
> > --
> > 2.16.2
> >

Re: [dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[Yongseok Koh]

> On Sep 27, 2018, at 6:51 PM, John Daley <johndale@cisco.com> wrote:
> 
> rte_flow structures were not being freed when destroyed or flushed.
> 
> Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> Reviewed-by: John Daley <johndale@cisco.com>
> ---

Hi,

Is this patch only for stable releases? If so, which one? Or was dev@
absent by mistake?


Thanks,
Yongseok

> drivers/net/enic/enic_flow.c | 2 ++
> 1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/net/enic/enic_flow.c b/drivers/net/enic/enic_flow.c
> index 0cf04aefd..9b612f1d5 100644
> --- a/drivers/net/enic/enic_flow.c
> +++ b/drivers/net/enic/enic_flow.c
> @@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev, struct rte_flow *flow,
> 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
> 	LIST_REMOVE(flow, next);
> 	rte_spinlock_unlock(&enic->flows_lock);
> +	rte_free(flow);
> 	return 0;
> }
> 
> @@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct rte_flow_error *error)
> 		flow = LIST_FIRST(&enic->flows);
> 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
> 		LIST_REMOVE(flow, next);
> +		rte_free(flow);
> 	}
> 	rte_spinlock_unlock(&enic->flows_lock);
> 	return 0;
> -- 
> 2.16.2
> 

[dpdk-stable] [PATCH 1/3] net/enic: fix flow API memory leak

[John Daley]

rte_flow structures were not being freed when destroyed or flushed.

Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled")
Cc: stable@dpdk.org

Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
---
 drivers/net/enic/enic_flow.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/enic/enic_flow.c b/drivers/net/enic/enic_flow.c
index 0cf04aefd..9b612f1d5 100644
--- a/drivers/net/enic/enic_flow.c
+++ b/drivers/net/enic/enic_flow.c
@@ -1532,6 +1532,7 @@ enic_flow_destroy(struct rte_eth_dev *dev, struct rte_flow *flow,
 	enic_flow_del_filter(enic, flow->enic_filter_id, error);
 	LIST_REMOVE(flow, next);
 	rte_spinlock_unlock(&enic->flows_lock);
+	rte_free(flow);
 	return 0;
 }
 
@@ -1555,6 +1556,7 @@ enic_flow_flush(struct rte_eth_dev *dev, struct rte_flow_error *error)
 		flow = LIST_FIRST(&enic->flows);
 		enic_flow_del_filter(enic, flow->enic_filter_id, error);
 		LIST_REMOVE(flow, next);
+		rte_free(flow);
 	}
 	rte_spinlock_unlock(&enic->flows_lock);
 	return 0;
-- 
2.16.2