driver/net/pcap fix: pcap fd leak
Checks
Commit Message
pcap fd was opend when vdev probed,
but not closed when vdev removed.
This bug appears in dpdk-pdump
Signed-off-by: ZhangTengfei <zypscode@outlook.com>
---
drivers/net/pcap/rte_eth_pcap.c | 1 +
1 file changed, 1 insertion(+)
Comments
On 2/26/2021 4:20 PM, ZhangTengfei wrote:
> pcap fd was opend when vdev probed,
> but not closed when vdev removed.
> This bug appears in dpdk-pdump
>
> Signed-off-by: ZhangTengfei <zypscode@outlook.com>
> ---
> drivers/net/pcap/rte_eth_pcap.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/pcap/rte_eth_pcap.c b/drivers/net/pcap/rte_eth_pcap.c
> index 90f5d75ea..fb01ea924 100644
> --- a/drivers/net/pcap/rte_eth_pcap.c
> +++ b/drivers/net/pcap/rte_eth_pcap.c
> @@ -1597,6 +1597,7 @@ pmd_pcap_remove(struct rte_vdev_device *dev)
> if (eth_dev == NULL)
> return 0; /* port already released */
>
> + eth_dev_stop(eth_dev);
> eth_dev_close(eth_dev);
> rte_eth_dev_release_port(eth_dev);
>
>
Thanks for the fix,
the cleanup seems missing in 'eth_dev_close()' too, what do you think moving
'eth_dev_stop(eth_dev);' inside the 'eth_dev_close()'?
So both 'close' and 'remove' will be covered.
Btw, you have same patch with both "ZhangTengfei <zhangtengfei@oppo.com>" sign
and "ZhangTengfei <zypscode@outlook.com>" sign (this one), can you please
clarify which one do you prefer?
I think your idea is fine
What do you think just record file path in "pmd_pcap_probe()",
Perform an open operation only in "eth_dev_start()"?
When the secondary process add pcap vdev,
it send the request to primary process,
the primary process probe pcap vdev too ,
Both the two process open the same file (in function "pmd_pcap_probe()")
It's not necessary
I prefer "ZhangTengfei <zypscode@outlook.com>" sign (this one)
发送自 Windows 10 版邮件<https://go.microsoft.com/fwlink/?LinkId=550986>应用
发件人: Ferruh Yigit<mailto:ferruh.yigit@intel.com>
发送时间: 2021年2月27日 0:46
收件人: ZhangTengfei<mailto:zypscode@outlook.com>
抄送: dev@dpdk.org<mailto:dev@dpdk.org>
主题: Re: [PATCH] driver/net/pcap fix: pcap fd leak
On 2/26/2021 4:20 PM, ZhangTengfei wrote:
> pcap fd was opend when vdev probed,
> but not closed when vdev removed.
> This bug appears in dpdk-pdump
>
> Signed-off-by: ZhangTengfei <zypscode@outlook.com>
> ---
> drivers/net/pcap/rte_eth_pcap.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/net/pcap/rte_eth_pcap.c b/drivers/net/pcap/rte_eth_pcap.c
> index 90f5d75ea..fb01ea924 100644
> --- a/drivers/net/pcap/rte_eth_pcap.c
> +++ b/drivers/net/pcap/rte_eth_pcap.c
> @@ -1597,6 +1597,7 @@ pmd_pcap_remove(struct rte_vdev_device *dev)
> if (eth_dev == NULL)
> return 0; /* port already released */
>
> + eth_dev_stop(eth_dev);
> eth_dev_close(eth_dev);
> rte_eth_dev_release_port(eth_dev);
>
>
Thanks for the fix,
the cleanup seems missing in 'eth_dev_close()' too, what do you think moving
'eth_dev_stop(eth_dev);' inside the 'eth_dev_close()'?
So both 'close' and 'remove' will be covered.
Btw, you have same patch with both "ZhangTengfei <zhangtengfei@oppo.com>" sign
and "ZhangTengfei <zypscode@outlook.com>" sign (this one), can you please
clarify which one do you prefer?
Please do not top post, message moved down.
On 2/26/2021 5:47 PM, 张 杨 wrote:
>> I think your idea is fine
>>
>> What do you think just record file path in "pmd_pcap_probe()",
>>
>> Perform an open operation only in "eth_dev_start()"?
>>
>> When the secondary process add pcap vdev,
>>
>> it send the request to primary process,
>>
>> the primary process probe pcap vdev too ,
>>
>> Both the two process open the same file (in function "pmd_pcap_probe()")
>>
>> It's not necessary
>>
>> I prefer "ZhangTengfei <zypscode@outlook.com>" sign (this one)
>>
>> 发送自Windows 10 版邮件 <https://go.microsoft.com/fwlink/?LinkId=550986>应用
>>
>> *发件人: *Ferruh Yigit <mailto:ferruh.yigit@intel.com>
>> *发送时间: *2021年2月27日0:46
>> *收件人: *ZhangTengfei <mailto:zypscode@outlook.com>
>> *抄送: *dev@dpdk.org <mailto:dev@dpdk.org>
>> *主题: *Re: [PATCH] driver/net/pcap fix: pcap fd leak
>>
>> On 2/26/2021 4:20 PM, ZhangTengfei wrote:
>>> pcap fd was opend when vdev probed,
>>> but not closed when vdev removed.
>>> This bug appears in dpdk-pdump
>>>
>>> Signed-off-by: ZhangTengfei <zypscode@outlook.com>
>>> ---
>>> drivers/net/pcap/rte_eth_pcap.c | 1 +
>>> 1 file changed, 1 insertion(+)
>>>
>>> diff --git a/drivers/net/pcap/rte_eth_pcap.c b/drivers/net/pcap/rte_eth_pcap.c
>>> index 90f5d75ea..fb01ea924 100644
>>> --- a/drivers/net/pcap/rte_eth_pcap.c
>>> +++ b/drivers/net/pcap/rte_eth_pcap.c
>>> @@ -1597,6 +1597,7 @@ pmd_pcap_remove(struct rte_vdev_device *dev)
>>> if (eth_dev == NULL)
>>> return 0; /* port already released */
>>>
>>> + eth_dev_stop(eth_dev);
>>> eth_dev_close(eth_dev);
>>> rte_eth_dev_release_port(eth_dev);
>>>
>>>
>>
>> Thanks for the fix,
>> the cleanup seems missing in 'eth_dev_close()' too, what do you think moving
>> 'eth_dev_stop(eth_dev);' inside the 'eth_dev_close()'?
>> So both 'close' and 'remove' will be covered.
>>
>>
>> Btw, you have same patch with both "ZhangTengfei <zhangtengfei@oppo.com>" sign
>> and "ZhangTengfei <zypscode@outlook.com>" sign (this one), can you please
>> clarify which one do you prefer?
>>
>
> I think your idea is fine
>
> What do you think just record file path in "pmd_pcap_probe()",
> Perform an open operation only in "eth_dev_start()"?
>
> When the secondary process add pcap vdev,
> it send the request to primary process,
> the primary process probe pcap vdev too ,>
> Both the two process open the same file (in function "pmd_pcap_probe()")
> It's not necessary
>
Opening pcap helps us fail early in probe() if something is wrong, otherwise the
driver probed and the problem detected in the start() when it is too late.
start() also opens pcap if it is not already opened.
In your usecase, if the pcap added by the secondary with the intention to use
only by the secondary, yes primary process also opens the pcap unnecessarily,
but that shouldn't be really a concern, this is one time cost in probe().
> I prefer "ZhangTengfei <zypscode@outlook.com>" sign (this one)
OK, also can you please use "Name Surname <email@address.com>" format, to be
consistent, I guess for you it means as following:
Tengfei Zhang <zypscode@outlook.com>
On 2021/3/1 下午7:40, Ferruh Yigit wrote:
> Please do not top post, message moved down.
>
> On 2/26/2021 5:47 PM, 张 杨 wrote:
>>> I think your idea is fine
>>>
>>> What do you think just record file path in "pmd_pcap_probe()",
>>>
>>> Perform an open operation only in "eth_dev_start()"?
>>>
>>> When the secondary process add pcap vdev,
>>>
>>> it send the request to primary process,
>>>
>>> the primary process probe pcap vdev too ,
>>>
>>> Both the two process open the same file (in function
>>> "pmd_pcap_probe()")
>>>
>>> It's not necessary
>>>
>>> I prefer "ZhangTengfei <zypscode@outlook.com>" sign (this one)
>>>
>>> 发送自Windows 10 版邮件
>>> <https://go.microsoft.com/fwlink/?LinkId=550986>应用
>>>
>>> *发件人: *Ferruh Yigit <mailto:ferruh.yigit@intel.com>
>>> *发送时间: *2021年2月27日0:46
>>> *收件人: *ZhangTengfei <mailto:zypscode@outlook.com>
>>> *抄送: *dev@dpdk.org <mailto:dev@dpdk.org>
>>> *主题: *Re: [PATCH] driver/net/pcap fix: pcap fd leak
>>>
>>> On 2/26/2021 4:20 PM, ZhangTengfei wrote:
>>>> pcap fd was opend when vdev probed,
>>>> but not closed when vdev removed.
>>>> This bug appears in dpdk-pdump
>>>>
>>>> Signed-off-by: ZhangTengfei <zypscode@outlook.com>
>>>> ---
>>>> drivers/net/pcap/rte_eth_pcap.c | 1 +
>>>> 1 file changed, 1 insertion(+)
>>>>
>>>> diff --git a/drivers/net/pcap/rte_eth_pcap.c
>>>> b/drivers/net/pcap/rte_eth_pcap.c
>>>> index 90f5d75ea..fb01ea924 100644
>>>> --- a/drivers/net/pcap/rte_eth_pcap.c
>>>> +++ b/drivers/net/pcap/rte_eth_pcap.c
>>>> @@ -1597,6 +1597,7 @@ pmd_pcap_remove(struct rte_vdev_device *dev)
>>>> if (eth_dev == NULL)
>>>> return 0; /* port already released */
>>>> + eth_dev_stop(eth_dev);
>>>> eth_dev_close(eth_dev);
>>>> rte_eth_dev_release_port(eth_dev);
>>>>
>>>
>>> Thanks for the fix,
>>> the cleanup seems missing in 'eth_dev_close()' too, what do you
>>> think moving
>>> 'eth_dev_stop(eth_dev);' inside the 'eth_dev_close()'?
>>> So both 'close' and 'remove' will be covered.
>>>
>>>
>>> Btw, you have same patch with both "ZhangTengfei
>>> <zhangtengfei@oppo.com>" sign
>>> and "ZhangTengfei <zypscode@outlook.com>" sign (this one), can you
>>> please
>>> clarify which one do you prefer?
>>>
> >
> > I think your idea is fine
> >
> > What do you think just record file path in "pmd_pcap_probe()",
> > Perform an open operation only in "eth_dev_start()"?
> >
> > When the secondary process add pcap vdev,
> > it send the request to primary process,
> > the primary process probe pcap vdev too ,>
> > Both the two process open the same file (in function
> "pmd_pcap_probe()")
> > It's not necessary
> >
>
> Opening pcap helps us fail early in probe() if something is wrong,
> otherwise the driver probed and the problem detected in the start()
> when it is too late.
> start() also opens pcap if it is not already opened.
>
> In your usecase, if the pcap added by the secondary with the intention
> to use only by the secondary, yes primary process also opens the pcap
> unnecessarily, but that shouldn't be really a concern, this is one
> time cost in probe().
>
> > I prefer "ZhangTengfei <zypscode@outlook.com>" sign (this one)
>
> OK, also can you please use "Name Surname <email@address.com>" format,
> to be consistent, I guess for you it means as following:
> Tengfei Zhang <zypscode@outlook.com>
Rarely submit code to the open source community. Thank you for your advice.
I quite agree with your view "fail early".
I will update the patch
@@ -1597,6 +1597,7 @@ pmd_pcap_remove(struct rte_vdev_device *dev)
if (eth_dev == NULL)
return 0; /* port already released */
+ eth_dev_stop(eth_dev);
eth_dev_close(eth_dev);
rte_eth_dev_release_port(eth_dev);