Mailing List Archive: SIGSEGV in pfring zc balancer worker thread bursts

SIGSEGV in pfring zc balancer worker thread bursts

Feb 19, 2019, 2:56 AM

Post #1 of 10 (1648 views)

Hi, We are currently using the following hardware and software combinations:- 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710

2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710

3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710

4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599 We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4. We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”. Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts(). Thoughts?

Regards,A.
--
Andrew Howard
ah@andrew-howard.me.uk

Re: SIGSEGV in pfring zc balancer worker thread bursts [ In reply to ]

ah at andrew-howard

Feb 19, 2019, 2:58 AM

Post #2 of 10 (1648 views)

Permalink

Apologies for the poor formatting - was written on a phone.
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Tue, Feb 19, 2019 10:56

Hi, We are currently using the following hardware and software combinations:- 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710

2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710

3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710

4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599 We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4. We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”. Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts(). Thoughts?

Regards,A.
--
Andrew Howard
ah@andrew-howard.me.uk

Re: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts [ In reply to ]

cardigliano at ntop

Feb 19, 2019, 6:00 AM

Post #3 of 10 (1646 views)

Permalink

Hi Andrew
let’s focus on systems 1 and 2 which are running latest stable, which is what we can support,
platform 1 is working fine, platform 2 is causing segfault, correct? Are you able to provide the
configuration (e.g. using zbalance_ipc) and if needed also a pcap file to be able to reproduce
the same issue?

Thank you
Alfredo

> On 19 Feb 2019, at 11:58, Andrew Howard <ah@andrew-howard.me.uk> wrote:
>
>
> Apologies for the poor formatting - was written on a phone.
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Tue, Feb 19, 2019 10:56
>
>
> Hi,
>
> We are currently using the following hardware and software combinations:-
>
> 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710
>
> 2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710
>
> 3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710
>
> 4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599
>
>
> We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4.
>
> We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”.
>
> Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts().
>
> Thoughts?
>
> Regards,
> A.
> --
> Andrew Howard
> ah@andrew-howard.me.uk
> _______________________________________________
> Ntop mailing list
> Ntop@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop

Re: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts [ In reply to ]

cardigliano at ntop

Feb 19, 2019, 6:00 AM

Post #4 of 10 (1646 views)

Permalink

Re: SIGSEGV in pfring zc balancer worker thread bursts [ In reply to ]

ah at andrew-howard

Feb 20, 2019, 7:52 AM

Post #5 of 10 (1646 views)

Permalink

Hi Alfredo,

Should the pcap file be emailed to your ntop.org address?

I'm unclear on what configuration you require from zbalancer_ipc. Could you elaborate?

Thanks and regards,
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Tue, Feb 19, 2019 10:58

Apologies for the poor formatting - was written on a phone.
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Tue, Feb 19, 2019 10:56

Hi, We are currently using the following hardware and software combinations:- 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710

2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710

3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710

4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599 We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4. We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”. Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts(). Thoughts?

Regards,A.
--
Andrew Howard
ah@andrew-howard.me.uk

Re: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts [ In reply to ]

cardigliano at ntop

Feb 20, 2019, 7:53 AM

Post #6 of 10 (1646 views)

Permalink

Hi Andrew

> On 20 Feb 2019, at 16:52, Andrew Howard <ah@andrew-howard.me.uk> wrote:
>
>
> Hi Alfredo,
>
> Should the pcap file be emailed to your ntop.org address?

Yes, feel free to use your preferred transfer method.

> I'm unclear on what configuration you require from zbalancer_ipc. Could you elaborate?

Please provide the zbalancer_ipc command line you use to reproduce this.

Thank you
Alfredo

>
> Thanks and regards,
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Tue, Feb 19, 2019 10:58
>
>
> Apologies for the poor formatting - was written on a phone.
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Tue, Feb 19, 2019 10:56
>
>
> Hi,
>
> We are currently using the following hardware and software combinations:-
>
> 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710
>
> 2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710
>
> 3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710
>
> 4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599
>
>
> We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4.
>
> We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”.
>
> Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts().
>
> Thoughts?
>
> Regards,
> A.
> --
> Andrew Howard
> ah@andrew-howard.me.uk
> _______________________________________________
> Ntop mailing list
> Ntop@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop

Re: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts [ In reply to ]

cardigliano at ntop

Feb 20, 2019, 7:53 AM

Post #7 of 10 (1646 views)

Permalink

Re: SIGSEGV in pfring zc balancer worker thread bursts [ In reply to ]

ah at andrew-howard

Feb 21, 2019, 1:10 AM

Post #8 of 10 (1646 views)

Permalink

Hi Alfredo,

Forget this - I've realisrf what the problem is. The distribution function signature has changed from an int32 return to an int64 return. All is now well.

Thanks & regards,
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Wed, Feb 20, 2019 15:52

Hi Alfredo,

Should the pcap file be emailed to your ntop.org address?

I'm unclear on what configuration you require from zbalancer_ipc. Could you elaborate?

Thanks and regards,
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Tue, Feb 19, 2019 10:58

Apologies for the poor formatting - was written on a phone.
A.

--
Andrew Howard
ah@andrew-howard.me.uk

----- Reply message -----
From: "Andrew Howard" <ah@andrew-howard.me.uk>
To: <ntop@listgateway.unipi.it>
Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
Date: Tue, Feb 19, 2019 10:56

Hi, We are currently using the following hardware and software combinations:- 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710

2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710

3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710

4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599 We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4. We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”. Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts(). Thoughts?

Regards,A.
--
Andrew Howard
ah@andrew-howard.me.uk

Re: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts [ In reply to ]

cardigliano at ntop

Feb 21, 2019, 3:56 AM

Post #9 of 10 (1646 views)

Permalink

Hi Andrew
great, thank you for the update.

Regards
Alfredo

> On 21 Feb 2019, at 10:10, Andrew Howard <ah@andrew-howard.me.uk> wrote:
>
>
> Hi Alfredo,
>
> Forget this - I've realisrf what the problem is. The distribution function signature has changed from an int32 return to an int64 return. All is now well.
>
> Thanks & regards,
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Wed, Feb 20, 2019 15:52
>
>
> Hi Alfredo,
>
> Should the pcap file be emailed to your ntop.org address?
>
> I'm unclear on what configuration you require from zbalancer_ipc. Could you elaborate?
>
> Thanks and regards,
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Tue, Feb 19, 2019 10:58
>
>
> Apologies for the poor formatting - was written on a phone.
> A.
>
> --
> Andrew Howard
> ah@andrew-howard.me.uk
>
> ----- Reply message -----
> From: "Andrew Howard" <ah@andrew-howard.me.uk>
> To: <ntop@listgateway.unipi.it>
> Subject: SIGSEGV in __pfring_zc_balancer_worker_thread_bursts
> Date: Tue, Feb 19, 2019 10:56
>
>
> Hi,
>
> We are currently using the following hardware and software combinations:-
>
> 1. PF_RING 7.4.0 on Centos7 (Linux 3.10.0), i40e driver on Intel X710
>
> 2. PF_RING 7.4.0 on RHEL 6.10 (Linux 2.6.32), i40e driver on Intel X710
>
> 3. Our own snapshot of PF_RING 7.0.0 branch taken 14-Feb-2018 on RHEL 6.10, i40e on X710
>
> 4. PF_RING v6.4.1 on RHEL 6.10, ixgbe driver on Intel 82599
>
>
> We are successfully receiving traffic via the ZC API (i.e. calling pfring_zc_open_device() with a device name of the format “zc:eth0”) on platforms 1, 3 and 4.
>
> We can successfully write traffic via the ZC API on platform 1 and 4, but an attempt to do this on platform 3 produces “Socket type not supported”.
>
> Prior to trying writes on platform 2, we attempted reads, which produces signal 11 in __pfring_zc_balancer_worker_thread_bursts().
>
> Thoughts?
>
> Regards,
> A.
> --
> Andrew Howard
> ah@andrew-howard.me.uk
> _______________________________________________
> Ntop mailing list
> Ntop@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop