Hi Olivier and Ananyev,
I like the new packet types and how they work the same for VLAN and QINQ. Just
so I understand your suggestion, X710 (as it seems to work today) would not set
RTE_PTYPE_L2_ETHER_VLAN in dev_supported_ptypes_get() because it does not know
how to determine that packet typ
On 4/23/2016 7:29 PM, Xing, Beilei wrote:
> Add 3 vf ops: rx_queue_count, rxq_info_get and
> txq_info_get. They can reuse corresponding pv APIs.
>
> Signed-off-by: Beilei Xing
> ---
> drivers/net/i40e/i40e_ethdev_vf.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/net/i40
On 4/23/2016 7:29 PM, Xing, Beilei wrote:
> Add 3 vf ops: rx_queue_count, rxq_info_get and
> txq_info_get. They can reuse corresponding pv APIs.
a typo here? pv -> pf ?
>
> Signed-off-by: Beilei Xing
> ---
> drivers/net/i40e/i40e_ethdev_vf.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff
On 4/23/2016 7:26 PM, Xing, Beilei wrote:
> This patch enables configuring MTU for i40e.
> Since changing MTU needs to reconfigure queue, stop port first
> before configuring MTU.
>
> Signed-off-by: Beilei Xing
> ---
> drivers/net/i40e/i40e_ethdev.c | 49
>
Currently, vhost PMD doesn't have linkage for librte_vhost, even though
it depends on librte_vhost APIs. This causes a linkage error if below
conditions are fulfilled.
- DPDK libraries are compiled as shared libraries.
- DPDK application doesn't link librte_vhost.
- Above application tries to l
Issue: virtio's drv_flags are decided by devices types (modern vs legacy),
and which kernel driver is used, and the negotiated features (especially
VIRTIO_NET_STATUS) with backend, which makes it possible to multiple
virtio devices have different versions of drv_flags, but this variable
is currentl
Hi Yuanhan,
> -Original Message-
> From: Yuanhan Liu [mailto:yuanhan.liu at linux.intel.com]
> Sent: Tuesday, April 26, 2016 11:43 AM
> To: Tan, Jianfeng
> Cc: dev at dpdk.org; Xie, Huawei
> Subject: Re: [PATCH v2] virtio: fix segfault when transmit pkts
>
> On Mon, Apr 25, 2016 at 02:37:
Hi Stephen,
On 4/26/2016 12:48 PM, Stephen Hemminger wrote:
> On Thu, 21 Apr 2016 12:36:10 +
> Jianfeng Tan wrote:
>
>> Issue: when using virtio nic to transmit pkts, it causes segment fault.
>>
>> How to reproduce:
>> a. start testpmd with vhost.
>> $testpmd -c 0x3 -n 4 --socket-mem 1024,0 -
On 2016/04/26 12:47, Yuanhan Liu wrote:
> On Mon, Apr 25, 2016 at 12:28:37PM +0300, Panu Matilainen wrote:
> >
>>> Another way is applying a below patch.
>>> --- a/drivers/net/vhost/Makefile
>>> +++ b/drivers/net/vhost/Makefile
>>> @@ -38,6 +38,7 @@ LIB = librte_pmd_vhost.a
>>>
>>> CFLAGS += -O3
Currently, vhost PMD doesn't have linkage for librte_vhost, even though
it depends on librte_vhost APIs. This causes a linkage error if below
conditions are fulfilled.
- DPDK libraries are compiled as shared libraries.
- DPDK application doesn't link librte_vhost.
- Above application tries to l
On 2016/04/26 14:48, Yuanhan Liu wrote:
> On Tue, Apr 26, 2016 at 02:37:37PM +0900, Tetsuya Mukawa wrote:
>> On 2016/04/26 12:47, Yuanhan Liu wrote:
>>> On Mon, Apr 25, 2016 at 12:28:37PM +0300, Panu Matilainen wrote:
>>> >
> Another way is applying a below patch.
> --- a/drivers/net/vhost
Have you tested with it?
I think we need to test it in a longer time, e.g. 1 hour
My commetns inlined.
Thanks,
Helin
> -Original Message-
> From: Yigit, Ferruh
> Sent: Tuesday, April 26, 2016 12:11 AM
> To: dev at dpdk.org
> Cc: Zhang, Helin; Yigit, Ferruh
> Subject: [PATCH] kni: add chai
Thanks,
great that you added more on CC for a wider discussion - I think that is
the only right way to go.
Just to "defend" a bit - solution a) was created under the special
circumstance that I wanted a workaround that would work today.
But that is/was special to what I package with DPDK 2.2 + OVS
Fix issue reported by Coverity.
Coverity ID 13282: Out-of-bounds write
overrun-local: Overrunning array mcfg->memseg of 256 44-byte elements
at element index 257 using index j.
Fixes: af75078fece3 ("first public release")
Signed-off-by: Slawomir Mrozowicz
---
lib/librte_eal/linuxapp/eal/eal_me
On 4/26/2016 7:49 AM, Zhang, Helin wrote:
> Have you tested with it?
Yes, has been tested.
> I think we need to test it in a longer time, e.g. 1 hour
I will make a longevity test before sending next patch.
> My commetns inlined.
>
> Thanks,
> Helin
>
>> -Original Message-
>> From: Yigit
On 26 April 2016 at 00:35, Bruce Richardson
wrote:
> On Wed, Apr 20, 2016 at 09:44:59PM +0800, Jianbo Liu wrote:
>> move SSE-dependent code to new file "ixgbe_rxtx_vec_sse.h"
>>
>> Signed-off-by: Jianbo Liu
>> ---
>> drivers/net/ixgbe/ixgbe_rxtx_vec.c | 369 +
>>
Talking about wording,
2016-04-25 20:43, Yuanhan Liu:
> ---
> Subject: virtio: fix segfault on Tx desc flags setup
I think the english word "crash" is better than "segfault".
2016-04-25 21:16, Yuanhan Liu:
> On Mon, Apr 25, 2016 at 11:18:16AM +0200, Christian Ehrhardt wrote:
> > The API doesn't hold a way to specify a owner/permission set for vhost_user
> > created sockets.
>
> Yes, it's kind of like a known issue. So, thanks for bringing it, with
> a solution, for dic
On Tue, Apr 26, 2016 at 09:44:47AM +0200, Slawomir Mrozowicz wrote:
> Fix issue reported by Coverity.
>
> Coverity ID 13282: Out-of-bounds write
> overrun-local: Overrunning array mcfg->memseg of 256 44-byte elements
> at element index 257 using index j.
>
> Fixes: af75078fece3 ("first public rel
Hi
Looks like there is a bug in API "ixgbevf_remove_mac_addr()". For
deleting a given MAC it deletes all MAC (including permanent MAC) and
while adding (it does after few statements), it skips to add permanent
MAC.
ixgbevf_remove_mac_addr(struct rte_eth_dev *dev, uint32_t index)
{
...
Hi Santosh,
Things are a little bit more complex.
When the permanent MAC address of a VF is assigned by the PF driver, it
cannot be changed nor deleted by the VF through commands issued to the
PF driver. In this case, the function xgbevf_remove_mac_addr() must not
add the permanent MAC address
On 04/26/2016 08:39 AM, Tetsuya Mukawa wrote:
> Currently, vhost PMD doesn't have linkage for librte_vhost, even though
> it depends on librte_vhost APIs. This causes a linkage error if below
> conditions are fulfilled.
>
> - DPDK libraries are compiled as shared libraries.
> - DPDK application d
On 26/04/2016 09:53, Bruce Richardson wrote:
> On Tue, Apr 26, 2016 at 09:44:47AM +0200, Slawomir Mrozowicz wrote:
>> Fix issue reported by Coverity.
>>
>> Coverity ID 13282: Out-of-bounds write
>> overrun-local: Overrunning array mcfg->memseg of 256 44-byte elements
>> at element index 257 using i
On 14/04/16 18:23, Eric Kinzie wrote:
>Instead of a hard-coded maximum receive length, allow the bond interface
>to inherit this limit from the first slave added. This allows
>an application that uses jumbo frames to pass realistic values to
>rte_eth_dev_configure without causing a
If no SSE nor NEON are available the l3fwd should complain loudly to quickly
find out the reason.
Signed-off-by: Jan Viktorin
---
It has happened to me once when I've accidently built a GCC without the NEON
support. It was confusing as at first I thought it is a bug... It is not,
there is just mi
- Removing dependency on nfp_uio kernel module. The igb_uio
kernel modules can be used instead.
Fixes: 80bc1752f16e ("nfp: add guide")
Signed-off-by: Alejandro Lucero
---
doc/guides/nics/nfp.rst | 47 ---
1 file changed, 16 insertions(+), 31 delet
>-Original Message-
>From: Gonzalez Monroy, Sergio
>Sent: Tuesday, April 26, 2016 11:44 AM
>To: Richardson, Bruce ; Mrozowicz, SlawomirX
>
>Cc: david.marchand at 6wind.com; dev at dpdk.org
>Subject: Re: [dpdk-dev] [PATCH] eal: out-of-bounds write
>
>On 26/04/2016 09:53, Bruce Richardson w
On Tue, Apr 26, 2016 at 4:24 AM, Jianfeng Tan wrote:
> Issue: virtio's drv_flags are decided by devices types (modern vs legacy),
> and which kernel driver is used, and the negotiated features (especially
> VIRTIO_NET_STATUS) with backend, which makes it possible to multiple
> virtio devices have
Some apps calling some functions from different threads at the
same time could lead to reconfig problems. Reconfig mechanism is
based on a hardware queue where incrementing a counter signals the
firmware to do the reconfig. If there are two increments before the
first one has been processed the fir
librte_malloc was long since merged into librte_eal, mop up the
leftovers from rarer drivers.
Fixes: 2f9d47013e4d ("mem: move librte_malloc to eal/common")
Signed-off-by: Panu Matilainen
---
drivers/net/cxgbe/Makefile| 2 +-
drivers/net/ena/Makefile | 2 +-
drivers/net/mpipe/Makefile
Issue: When virtio was proposed in DPDK, there is no API to free memzones.
But this has changed since rte_memzone_free() has been implemented by
commit ff909fe21f.
This patch is to make sure memzones in struct virtqueue, like mz and
virtio_net_hdr_mz, are freed when queue is released or setup fail
rx_q fifo may have chained mbufs, merge them into single skb before
handing to the network stack.
Signed-off-by: Ferruh Yigit
---
.../linuxapp/eal/include/exec-env/rte_kni_common.h | 4 +-
lib/librte_eal/linuxapp/kni/kni_net.c | 83 --
2 files changed, 64 insert
Hi,
Since some guys are asking about the status of this patch, I'd like to
ping if anyone still has concerns.
Current conclusion is: with option --avail-cores.
Thanks,
Jianfeng
On 3/4/2016 6:05 PM, Jianfeng Tan wrote:
> This patch adds option, --avail-cores, to use lcores which are available
>
On Mon, Apr 25, 2016 at 10:13:00PM -0700, Rasesh Mody wrote:
> The base driver is the backend module for the QLogic FastLinQ QL4
> 25G/40G CNA family of adapters as well as their virtual functions (VF)
> in SR-IOV context.
>
> The purpose of the base module is to:
> - provide all the common c
Some apps calling some functions from different threads at the
same time could lead to reconfig problems. Reconfig mechanism is
based on a hardware queue where incrementing a counter signals the
firmware to do the reconfig. If there are two increments before the
first one has been processed the fi
On Mon, Apr 25, 2016 at 10:12:59PM -0700, Rasesh Mody wrote:
> Signed-off-by: Harish Patil
> Signed-off-by: Rasesh Mody
> Signed-off-by: Sony Chacko
> ---
> MAINTAINERS |7 +
> doc/guides/nics/index.rst |1 +
> doc/guides/nics/overview.rst | 86 +
mbufs where not properly released when they are chained.
Fixes: b812daadad0d ("nfp: add Rx and Tx")
Signed-off-by: Alejandro Lucero
---
drivers/net/nfp/nfp_net.c | 21 ++---
1 file changed, 10 insertions(+), 11 deletions(-)
diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/n
RTE_PCI_DRV_DETACHABLE is required for detaching a device
during execution.
Signed-off-by: Alejandro Lucero
---
drivers/net/nfp/nfp_net.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index 1259d2c..ea5a2a3 100644
---
On Mon, Apr 25, 2016 at 10:13:00PM -0700, Rasesh Mody wrote:
> The base driver is the backend module for the QLogic FastLinQ QL4
> 25G/40G CNA family of adapters as well as their virtual functions (VF)
> in SR-IOV context.
>
> The purpose of the base module is to:
> - provide all the common c
On Mon, Apr 25, 2016 at 10:13:04PM -0700, Rasesh Mody wrote:
> Physical link is handled by the management Firmware.
> This patch lays the infrastructure for attention handling in the driver,
> as link change notifications arrive via async attentions, as well as the
> handling of such notifications.
Thomas Monjalon writes:
> 2016-04-25 21:16, Yuanhan Liu:
>> On Mon, Apr 25, 2016 at 11:18:16AM +0200, Christian Ehrhardt wrote:
>> > The API doesn't hold a way to specify a owner/permission set for vhost_user
>> > created sockets.
>>
>> Yes, it's kind of like a known issue. So, thanks for bringi
move common code to new file "ixgbe_rxtx_vec_common.h",
and vPMD for x86 is implemented in ixgbe_rxtx_vec.c
Signed-off-by: Jianbo Liu
Suggested-by: Bruce Richardson
---
drivers/net/ixgbe/ixgbe_rxtx_vec.c| 256 +--
drivers/net/ixgbe/ixgbe_rxtx_vec_common.h | 325 +
use ARM NEON intrinsic to implement ixgbe vPMD
Signed-off-by: Jianbo Liu
---
drivers/net/ixgbe/Makefile | 4 +
drivers/net/ixgbe/ixgbe_rxtx_vec_neon.c | 556
2 files changed, 560 insertions(+)
create mode 100644 drivers/net/ixgbe/ixgbe_rxtx_vec_ne
Signed-off-by: Jianbo Liu
---
config/defconfig_arm64-armv8a-linuxapp-gcc | 1 -
1 file changed, 1 deletion(-)
diff --git a/config/defconfig_arm64-armv8a-linuxapp-gcc
b/config/defconfig_arm64-armv8a-linuxapp-gcc
index 9abeca4..98cc054 100644
--- a/config/defconfig_arm64-armv8a-linuxapp-gcc
+++ b
Signed-off-by: Jianbo Liu
---
MAINTAINERS | 1 +
1 file changed, 1 insertion(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 1953ea2..20158e3 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -142,6 +142,7 @@ F: lib/librte_eal/common/include/arch/arm/*_64.h
F: lib/librte_acl/acl_run_neon.*
F: lib/
> > After having removed the deprecated stuff, we can start pushing
> > new fixes and features in the version 16.07.
> >
> > Signed-off-by: Thomas Monjalon
> Acked-by: Bruce Richardson
Applied
On Mon, Apr 25, 2016 at 10:13:06PM -0700, Rasesh Mody wrote:
> This patch enables the QEDE PMD build.
>
> Signed-off-by: Harish Patil
> Signed-off-by: Rasesh Mody
> Signed-off-by: Sony Chacko
> ---
> config/common_base | 12
> drivers/net/Makefile |1 +
> mk/rte.app.mk
On Mon, Apr 25, 2016 at 10:12:59PM -0700, Rasesh Mody wrote:
> Signed-off-by: Harish Patil
> Signed-off-by: Rasesh Mody
> Signed-off-by: Sony Chacko
> ---
> MAINTAINERS |7 +
> doc/guides/nics/index.rst |1 +
> doc/guides/nics/overview.rst | 86 +
> -Original Message-
> From: Richardson, Bruce
> Sent: Tuesday, April 26, 2016 4:04 PM
> To: Rasesh Mody
> Cc: thomas.monjalon at 6wind.com; dev at dpdk.org; ameen.rahman at qlogic.com;
> Harish Patil ; Sony Chacko
> ; Mcnamara, John
> Subject: Re: [PATCH v6 1/8] qede: Add maintainers,
2016-04-26 15:19, Mcnamara, John:
> > From: Richardson, Bruce
> > Ideally, the feature matrix would be updated as each patch adds new
> > features, but since this is a new driver, I'm ok with just having this as
> > part of that core driver patch too - with a note in the commit stating
> > that the
It seems that with gcc >5.x and -O2/-O3 optimization breaks packet grouping
algorithm in l3fwd-thread application causing segfault.
When last packet pointer "lp" and "pnum->u64" buffer points the same
memory buffer, high optimization can cause unpredictable results. It seems
that assignment of pre
Hi,
Thanks for the documentation.
In general you should generate and view the Html output to make sure
everything is okay:
make doc-guides-html
firefox build/doc/html/guides/nics/qede.html &
Other comments below.
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.o
2016-04-11 23:27, John McNamara:
> Changed symbol on NIC overview table from X to Y to help
> clarify the indicated features are supported. The X caused
> confusion for some readers.
>
> Also, added * character to indicate partially supported
> features. This can be used in the future to direct th
On Mon, Apr 25, 2016 at 10:21:32PM +0800, Huawei Xie wrote:
> check merge-able header as it is supported.
> previously we don't support merge-able feature, so non merge-able
> header is checked.
Signed-off-by is missing here. Otherwise, this patch looks good to me
--yliu
On Tue, Apr 26, 2016 at 10:43:35AM +0200, Thomas Monjalon wrote:
> Talking about wording,
>
> 2016-04-25 20:43, Yuanhan Liu:
> > ---
> > Subject: virtio: fix segfault on Tx desc flags setup
>
> I think the english word "crash" is better than "segfault".
Yes.
Acked-by: Yuanhan Liu
And,
Hey-
So a few days ago we were reviewing Davids patch series to introduce the
abiilty to dump hardware support from pmd DSO's in a human readable format.
That effort encountered some problems, most notably the fact that stripping a
DSO removed the required information that the proposed tool
modify PMD_REGISTER_DRIVER so that, when building as a DSO, PMD's emit an
additional set of symbols named this_pmd_driver, where is an incrementing
counter. This gives well known symbol names that external apps can search for
when looking up PMD information. These new symbols are aliased to the
Because the DPDK DSO's are opt-in for symbol export, we need to add the symbols
that the modified PMD_REGISTER_DRIVER macro creates so that external
applications can see them
Signed-off-by: Neil Horman
CC: David Marchand
CC: Stephen Hemminger
CC: "Richardson, Bruce"
CC: Panu Matilainen
CC: Th
For the PMD's which support pci devices, add the appropriate pci table pointer
to the rte_driver structure so external applications can find it.
Also note, some modifications to the em/igb and i40e drivers. These are done to
support an artifact of the use of the __COUNTER__ macro in PMD_REGISTER_
This tool uses the prior infrastructure to provide human readable information to
a user about the devices which a given pmd DSO supports.
Usage:
pmdinfo /path/to/driver/pmd
pmdinfo dlopens the specified file, then iteratively looks up the
this_pmd_driver symbol. For each found symbol in the DSO,
> From: Bruce Richardson [mailto:bruce.richardson at intel.com]
> Sent: Tuesday, April 26, 2016 6:03 AM
>
> On Mon, Apr 25, 2016 at 10:12:59PM -0700, Rasesh Mody wrote:
> > Signed-off-by: Harish Patil
> > Signed-off-by: Rasesh Mody
> > Signed-off-by: Sony Chacko
> > ---
> > MAINTAINERS
2016-04-26 18:27, Rasesh Mody:
> > From: Bruce Richardson [mailto:bruce.richardson at intel.com]
> > Sent: Tuesday, April 26, 2016 6:03 AM
> >
> > On Mon, Apr 25, 2016 at 10:12:59PM -0700, Rasesh Mody wrote:
> > > Signed-off-by: Harish Patil
> > > Signed-off-by: Rasesh Mody
> > > Signed-off-by:
> From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com]
> Sent: Tuesday, April 26, 2016 12:34 PM
>
> 2016-04-26 18:27, Rasesh Mody:
> > > From: Bruce Richardson [mailto:bruce.richardson at intel.com]
> > > Sent: Tuesday, April 26, 2016 6:03 AM
> > >
> > > On Mon, Apr 25, 2016 at 10:12:59PM -
Fixes: 7182d3e7d177 ("enic: expose Rx missed packets counter")
Signed-off-by: John Daley
---
drivers/net/enic/enic_main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/enic/enic_main.c b/drivers/net/enic/enic_main.c
index 60fe765..be4e9e5 100644
--- a/drivers
Data DMA used m->data_off of uninitialized mbufs instead of
RTE_PKTMBUF_HEADROOM, potentially causing Rx data to be
placed at the wrong alignment in the mbuf.
Fixes: 947d860c821f ("enic: improve Rx performance")
Signed-off-by: John Daley
---
drivers/net/enic/enic_main.c | 5 +++--
drivers/net/en
Optimizations and cleanup:
- flatten packet send path
- flatten mbuf free path
- disable CQ entry writing and use CQ messages instead
- use rte_mempool_put_bulk() to bulk return freed mbufs
- remove unnecessary fields vnic_bufs struct, use contiguous array of cache
aligned divisible elements. No
66 matches
Mail list logo
