From: Robert Sanford
Signed-off-by: Robert Sanford
---
lib/librte_timer/rte_timer.c | 45 +++--
1 files changed, 29 insertions(+), 16 deletions(-)
diff --git a/lib/librte_timer/rte_timer.c b/lib/librte_timer/rte_timer.c
index 8e9243a..51e6038 100644
--- a/
From: Robert Sanford
Signed-off-by: Robert Sanford
---
app/test/Makefile |1 +
app/test/test_timer_racecond.c | 209
2 files changed, 210 insertions(+), 0 deletions(-)
create mode 100644 app/test/test_timer_racecond.c
diff --git a/app
From: Robert Sanford
Signed-off-by: Robert Sanford
---
app/test/test_timer.c | 149 ++---
1 files changed, 116 insertions(+), 33 deletions(-)
diff --git a/app/test/test_timer.c b/app/test/test_timer.c
index 73da5b6..dd63421 100644
--- a/app/test/tes
From: Robert Sanford
This patchset fixes a bug in timer stress test 2, adds a new stress test
to expose a race condition bug in API rte_timer_manage(), and then fixes
the rte_timer_manage() bug.
--
Patch 1, app/test timer stress test 2: Sometimes this test fails and
seg-faults because the slave
2015-07-22 19:28, Konstantin Ananyev:
> + if (!rte_eth_dev_is_valid_port(port_id)) {
> + PMD_DEBUG_TRACE("Invalid port_id=%d\n", port_id);
> + return -EINVAL;
> + }
Please use VALID_PORTID_OR_ERR_RET.
> + * Ethernet device RX queue information strcuture.
Typo here
Anyone?
On 17.07.2015 16:48, ciprian.barbu wrote:
> Hi,
>
> I'm seeing a strange behavior when calling rte_eth_tx_burst with len ==
> 0. I'll explain the reason for this situation further bellow. But what
> I'm seeing is that after doing this call my application keeps returning
> from eth_igb_xmit
2015-07-23 10:52, Ananyev, Konstantin:
> From: Stephen Hemminger [mailto:stephen at networkplumber.org]
> > Konstantin Ananyev wrote:
> > > Add the ability for the upper layer to query RX/TX queue information.
[...]
> > Since all this data should be rxconf already, Is it possible
> > to do a gener
As noticed when reviewing the changes for hotplugging ring PMD,
http://dpdk.org/ml/archives/dev/2015-July/021872.html
a driver is considered in ethdev and EAL as a PCI driver:
* Each Ethernet driver acts as a PCI driver and is represented by a generic
* *eth_driver* structure that holds:
Hi, Pablo
Is there any performance data for this change?
Thanks,
Michael
-Original Message-
From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Pablo de Lara
Sent: Thursday, July 23, 2015 9:12 AM
To: dev at dpdk.org
Subject: [dpdk-dev] [PATCH] examples/l3fwd: increase lookup burst size
With the new hash implementation, the minimum lookup burst size
to get good performance is 8, since its internal pipeline
consists of 4 stages of 2 entries each, so to avoid
duplication, burst size should be 8 or more entries.
Signed-off-by: Pablo de Lara
---
examples/l3fwd/main.c | 234
The vector/SSE pmd used a different element type for the tx queue sw_ring
entries. This led to lots of typecasts in the code which required specific
use of bracketing, leading to subtle errors.
For example, in the original code:
txe = (struct ixgbe_tx_entry_v *)&txq->sw_ring[i];
instead nee
The function inside the vector/SSE poll-mode driver for releasing
the mbufs on the TX queues had the same name as another function
inside the regular PMD. To keep consistency and avoid confusion,
rename the vector PMD version to have a "_vec" suffix.
Signed-off-by: Bruce Richardson
---
drivers/n
The calculations of what mbufs were valid in the RX and TX queues were
incorrect when freeing the mbufs for the vector PMD. This led to crashes
due to invalid reference counts when mbuf debugging was turned on, and
possibly other more subtle problems (such as mbufs being freed when in use)
in other
The two fields for vector RX rearming in the rx queue structure were
incorrectly labelled. Switching the comments on each around makes things
clearer.
Signed-off-by: Bruce Richardson
---
drivers/net/ixgbe/ixgbe_rxtx.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/driver
This reverts commit b35d0d80f0a809939549ddde99c1a76b7e38bff3.
The bug fix was incorrect as it did not take account of the fact that
the mbufs that were previously freed may have since be re-allocated.
Signed-off-by: Bruce Richardson
---
drivers/net/ixgbe/ixgbe_rxtx.c | 3 +--
drivers/net/ixg
Konstantin has correctly pointed out that the previously applied fix:
b35d0d80f0a8 ("ixgbe: check mbuf refcnt when clearing a ring")
is not a proper fix for the reported issue at all.
Ref: http://permalink.gmane.org/gmane.comp.networking.dpdk.devel/21932
This patch set reverts the original fix, a
The RX of VMXNET3 PMD will have deadlock when a lot of traffic coming in.
The root cause is due to mbuf allocation fail in vmxnet3_post_rx_bufs() and
there is no error handling when it is called
from vmxnet3_recv_pkts(). The RXD will not have "free" mbuf for it but the
counter still increment.
Fi
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Thomas Monjalon
> Sent: Thursday, July 23, 2015 4:43 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] management of non-PCI devices
>
> As noticed when reviewing the changes for hotplugging ring PMD,
> htt
RTE_LIBRTE_IXGBE_RX_ALLOW_BULK_ALLOC config option is not really
necessary, as bulk alloc rx function can be used anyway, as long as the
necessary conditions are satisfied, which are checked already
in the library.
Signed-off-by: Pablo de Lara
---
config/common_bsdapp | 1 -
config/co
> > > Fixes: f6bf669b990004dc69231476d9fcf0b80574e568. Bug fix to remove
> > > (rxnfgpc - hw_stats->gprc) which does not account for phy errors.
> > > hw_stats->gprc is the Number of good (non-erred) Rx packets (from the
> > > network) that pass L2 filtering and has a legal length as defined by
> >
Never mind, I've digged a bit more in the sink's driver stats, it turned
out there are a lot of RX misses when it receives through the test box.
Increasing the RX ring from 512 to 2048 solved the issue. I guess the
change in the burst pattern caused this whole issue.
Regards,
Zoltan
On 23/07/
2015-07-22 19:28, Konstantin Ananyev:
> Add the ability for the upper layer to query RX/TX queue information.
> Right now supported for:
> ixgbe, i40e, e1000 PMDs.
Thank you for working on it.
The feature is now in the roadmap for version 2.2.
Please postpone any new version to the 2.2 release cyc
Hi Thomas and all,
This patch set postponed from v2.0, and widely reviewed during this release
cycle.
The packet I/O interrupt framework is the prerequisite of all PMDs to support
packet I/O intr.
There's no significant change since the last three version(v13~v15), the
changes are about ABI and
2015-07-23 10:29, Gonzalez Monroy, Sergio:
> On 23/07/2015 09:12, Thomas Monjalon wrote:
> > 2015-07-23 08:34, Gonzalez Monroy, Sergio:
> >> On 22/07/2015 11:40, Thomas Monjalon wrote:
> >>> Sergio,
> >>>
> >>> As the maintainer of memory allocation, would you consider using
> >>> libhugetlbfs in D
This is just a reminder that on August 17 the DPDK open source community will
gather in San Francisco for the 2nd annual DPDK Summit. It's a one day event
where the community will share some of their developments, applications, and
usage of DPDK. The event will include talks from HP, IBM Asper
Hi,
I've seen an odd behaviour in my test setup, which affected my test
results, so I set up a much simpler scenario.
I'm using netmap pktgen as a packet source, it creates a steady 14.2
Mpps of 64 byte UDP packets over one port of a 82599ES dual port card.
This traffic then goes to an another
Hi Sergio
>Then there is the issue with multi-process; because they return a file
> descriptor while
>unlinking the file, we would need some sort of Inter-Process
That's kind of what VFIO does for multiprocess, so you may want to look at that
in case you decide to get down that route :)
v2 changes:
-changed item on LPM table to add LPM IPv6
-removed item for ACL table and replaced with item on table ops
-added item for hash tables
Signed-off-by: Cristian Dumitrescu
---
doc/guides/rel_notes/deprecation.rst | 10 ++
1 files changed, 10 insertions(+), 0 deletions(-)
di
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Cristian Dumitrescu
> Sent: Thursday, July 23, 2015 1:00 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v2] announce ABI change for librte_table
>
> v2 changes:
> -changed item on LPM table to add LPM IPv
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Cristian Dumitrescu
> Sent: Thursday, July 23, 2015 1:00 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v2] announce ABI change for librte_table
>
> v2 changes:
> -changed item on LPM table to add LPM IPv
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Cristian Dumitrescu
> Sent: Thursday, July 23, 2015 12:00 PM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v2] announce ABI change for librte_table
>
> v2 changes:
> -changed item on LPM table to add LPM
The example includes an ethtool library and two applications:
one application is a non- DPDK process (nic-control)
and the other is a DPDK l2fwd applicaiton (l2fwd-app).
The nic-control process sends ethtool alike device management
requests to l2fwd-app through a named pipe IPC. This example
is des
From: "Andrew G. Harvey"
Signed-off-by: Andrew G. Harvey
---
mk/rte.extlib.mk | 2 ++
mk/rte.lib.mk| 6 ++
2 files changed, 8 insertions(+)
diff --git a/mk/rte.extlib.mk b/mk/rte.extlib.mk
index ba066bc..d2a9b6d 100644
--- a/mk/rte.extlib.mk
+++ b/mk/rte.extlib.mk
@@ -31,6 +31,8 @@
M
This implementation is designed to provide an example illlustrating how to
create a
user-space ethtool library from existing ethdev APIs. In contrast to kernel
version
of same API (such as ops defined in KNI), the user-space APIs enable a fast-path
(no kernel API calls) for device query and data
> -Original Message-
> From: Stephen Hemminger [mailto:stephen at networkplumber.org]
> Sent: Wednesday, July 22, 2015 8:48 PM
> To: Ananyev, Konstantin
> Cc: dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCHv4 1/5] ethdev: add new API to retrieve RX/TX
> queue information
>
> On Wed, 22
On Thu, Jul 16, 2015 at 01:50:12AM +0200, Thomas Monjalon wrote:
> A packet is tunnelled if the tunnel type is identified or if it has
> an inner part.
>
> Fix also a typo in RTE_PTYPE_INNER_L3_MASK.
>
> Fixes: f295a00a2b44 ("mbuf: add definitions of unified packet types")
>
> Signed-off-by: Tho
Hi,
On 07/22/2015 03:57 PM, Yuichi Nakai wrote:
> __mempool_get_trailer() calculated header's address.
> The address of trailer should set after element area.
> This patch fixes this calculating.
>
> This issue was mixed with the following commit:
> Fixes: 97e7e685bfcd ("mempool: add structure f
On 7/23/15 8:29 AM, O'Driscoll, Tim wrote:
> This is just a reminder that on August 17 the DPDK open source community will
> gather in San Francisco for the 2nd annual DPDK Summit. It's a one day event
> where the community will share some of their developments, applications, and
> usage of DP
On 23/07/2015 09:12, Thomas Monjalon wrote:
> 2015-07-23 08:34, Gonzalez Monroy, Sergio:
>> On 22/07/2015 11:40, Thomas Monjalon wrote:
>>> Sergio,
>>>
>>> As the maintainer of memory allocation, would you consider using
>>> libhugetlbfs in DPDK for Linux?
>>> It may simplify a part of our memory a
Acked-by: Cristian Dumitrescu
2015-07-23 08:34, Gonzalez Monroy, Sergio:
> On 22/07/2015 11:40, Thomas Monjalon wrote:
> > Sergio,
> >
> > As the maintainer of memory allocation, would you consider using
> > libhugetlbfs in DPDK for Linux?
> > It may simplify a part of our memory allocator and avoid some potential
> > bugs whic
On Thu, 23 Jul 2015 17:42:58 +0200
Thomas Monjalon wrote:
> As noticed when reviewing the changes for hotplugging ring PMD,
> http://dpdk.org/ml/archives/dev/2015-July/021872.html
> a driver is considered in ethdev and EAL as a PCI driver:
>
> * Each Ethernet driver acts as a PCI driver a
On Thu, 23 Jul 2015 16:51:32 +0800
Marco Lee wrote:
> The RX of VMXNET3 PMD will have deadlock when a lot of traffic coming in.
> The root cause is due to mbuf allocation fail in vmxnet3_post_rx_bufs() and
> there is no error handling when it is called
> from vmxnet3_recv_pkts(). The RXD will no
From: marco
---
drivers/net/vmxnet3/vmxnet3_rxtx.c | 54 +++-
1 file changed, 53 insertions(+), 1 deletion(-)
mode change 100644 => 100755 drivers/net/vmxnet3/vmxnet3_rxtx.c
diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c
b/drivers/net/vmxnet3/vmxnet3_rxtx.c
On Thu, 23 Jul 2015 07:31:01 +
Vithal S Mohare wrote:
> +/* VMXNET3
> + * In the above loop, vmxnet3_post_rx_bufs would fai if all the mbufs
> currently allocated.
> + * In such scenarios where hw device hasn't left with any of 'rx'
> descriptors, packets from
> + * network
Signed-off-by: David Marchand
---
app/test-pmd/cmdline.c| 1 -
lib/librte_eal/bsdapp/eal/eal_pci.c | 1 -
lib/librte_eal/linuxapp/eal/eal_pci.c | 1 -
lib/librte_eal/linuxapp/eal/eal_pci_uio.c | 1 -
4 files changed, 4 deletions(-)
diff --git a/app/test-pmd/cmdline.
There is no reason why eal should depend on drivers.
Remove remaining (unused) references.
Signed-off-by: David Marchand
---
lib/librte_eal/bsdapp/eal/Makefile | 3 ---
lib/librte_eal/bsdapp/eal/eal.c | 1 -
lib/librte_eal/linuxapp/eal/Makefile | 5 -
lib/librte_eal/linuxapp/eal/eal.c
Here is some trivial cleanup for eal.
Nothing really astounding but this will ease an unfinished patchset I have
been working on for some time (getting rid of any driver specific stuff in
eal).
--
David Marchand
David Marchand (2):
eal: remove dependency on drivers
eal: remove useless head
On 23/07/2015 08:09, Rapelly, Varun wrote:
> Hi,
>
> I'm facing problems with "iommu=pt intel_iommu=on" in ProLiant DL380p Gen8
> server[RHEL 7.1]. But on ProLiant DL380 G7[RHEL 7.1] server, not facing this
> issue.
>
> When I add the above options to the kernel boot line and configure OVS bridge
On 22/07/2015 11:40, Thomas Monjalon wrote:
> Sergio,
>
> As the maintainer of memory allocation, would you consider using
> libhugetlbfs in DPDK for Linux?
> It may simplify a part of our memory allocator and avoid some potential
> bugs which would be already fixed in the dedicated lib.
I did have
How about the below changes? I have been using below changes and helping to
resolve the issue.
===
= dpdk/lib/librte_pmd_vmxnet3/vmxnet3_ring.h#3 edit (text) =
@@ -155,10 +155,11 @@ typedef struct vmxnet3_tx_q
Hi,
I'm facing problems with "iommu=pt intel_iommu=on" in ProLiant DL380p Gen8
server[RHEL 7.1]. But on ProLiant DL380 G7[RHEL 7.1] server, not facing this
issue.
When I add the above options to the kernel boot line and configure OVS bridge
[with -dpdk option], not able to send the packets out
>
>Changes in this version (v5):
> - rebase to 2.1.0-rc1
> - fix version map for 2.1
> - fix doxygen comments
> - add BCM57410 ids
>
>Stephen Hemminger (4):
> eal: provide functions to access PCI config
> bnx2x: driver core
> bnx2x: driver support routines
> bnx2x: enable PMD build
>
> MAIN
Hi Stephen,? ? I have question about point 1 and want to discuss with you. Why
it will cause deadlock if just stop receiving it in this cases? ?And I also
found VMXNET3 PMD have the same bug in RX. I will rework the patch and submit
later. Thanks!
Best Regards,Marco
Stephen Hemminger ?
> -Original Message-
> From: Jeff Venable, Sr. [mailto:jeff at vectranetworks.com]
> Sent: Wednesday, July 22, 2015 5:47 PM
> To: Zhang, Helin; lhffjzh; 'Thomas Monjalon'
> Cc: dev at dpdk.org
> Subject: RE: [dpdk-dev] Why only rx queue "0" can receive network packet by
> i40e NIC
>
> Is
Hi,
> -Original Message-
> From: Klaus Degner [mailto:kd at allegro-packets.com]
> Sent: Wednesday, July 22, 2015 5:08 PM
> To: dev at dpdk.org
> Cc: Lu, Wenzhuo; Klaus Degner
> Subject: [PATCH] e1000: enable jumbo frame support for Intel 82583V
>
> This patch enables jumbo frame support
Is the I40E incapable of operating RSS with ETH_RSS_IP (i.e. hashing without L4
ports)?
Thanks,
Jeff
-Original Message-
From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Zhang, Helin
Sent: Saturday, February 28, 2015 6:34 AM
To: lhffjzh; 'Thomas Monjalon'
Cc: dev at dpdk.org; maintai
57 matches
Mail list logo