Hi,
> Subject: [dpdk-dev] [PATCH] doc: announce ivshmem support removal
>
> There was a prior call with an explanation of what needs to be done:
> http://dpdk.org/ml/archives/dev/2016-June/040844.html
> - Qemu patch upstreamed
> - IVSHMEM PCI device managed by a PCI driver
> - No DPDK
Hi,
> Subject: Re: [dpdk-dev] daemon process problem in DPDK
>
> Much appericated, Get it now.
>
> Thanks,
> Xun
>
> -Original Message-
> From: Stephen Hemminger [mailto:stephen at networkplumber.org]
> Sent: Tuesday, January 13, 2015 3:14 AM
> To: Neil Horman
> Cc: Ni, Xun; dev at
> Subject: Re: [dpdk-dev] mk: fix app linking for combined libs
>
> 2014-12-08 09:53, Neil Horman:
> > On Thu, Oct 23, 2014 at 04:36:44PM +0100, Sergio Gonzalez Monroy wrote:
> > > Building combined shared libraries results in applications being linked
> > > against separeted/individual and
signed cache_size, unsigned private_data_size,
> rte_mempool_ctor_t *mp_init, void *mp_init_arg,
> rte_mempool_obj_ctor_t *obj_init, void *obj_init_arg,
> int socket_id, unsigned flags);
>
>
> Brgs,
> Chi xiaobo
>
>
> -
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
The application should be linked to the single combined library in the
condition that both of CONFIG_RTE_BUILD_COMBINE_LIB and
CONFIG_RTE_BUILD_SHARED_LIB are enabled.
The current makefile generates an application that links to each l
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Only CPU_LDFLAGS is used in mk/rte.sharelib.mk.
It should be LDFLAGS to build the library with correct linkage options.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
mk/rte.sharelib.mk | 2 +-
1 file changed, 1 ins
Hi,
> Subject: [dpdk-dev] [PATCH v2] add one option memory-only for secondary
> processes
>
> From: Chi Xiaobo
>
> Problem: There is one normal DPDK processes deployment scenarios: one primary
> process and several (even hundreds) secondary
> processes; all outside packets/messages are
> Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> recv/xmit
>
> On Wed, Oct 01, 2014 at 11:33:23PM +0000, Hiroshi Shimamoto wrote:
> > > Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> > > recv/xmit
> > >
&
> Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> recv/xmit
>
> On Wed, Oct 01, 2014 at 09:12:44AM +0000, Hiroshi Shimamoto wrote:
> > > Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> > > recv/xmit
> > >
&
> Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> recv/xmit
>
> On Tue, Sep 30, 2014 at 11:52:00PM +0000, Hiroshi Shimamoto wrote:
> > Hi,
> >
> > > Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> > >
Hi Thomas,
> Subject: Re: [dpdk-dev] [memnic PATCH v2 0/7] MEMNIC PMD performance
> improvement
>
> > This patchset improves MEMNIC PMD performance.
> >
> > Hiroshi Shimamoto (7):
> > guest: memnic-tester: PMD benchmark in guest
> > pmd: remove ne
e the message when you apply
this patch?
thanks,
Hiroshi
>
> > -Original Message-
> > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Hiroshi Shimamoto
> > Sent: Tuesday, September 30, 2014 7:15 PM
> > To: dev at dpdk.org
> > Cc: Hayato Momma
> &
Hi,
> Subject: Re: [dpdk-dev] [memnic PATCH v2 6/7] pmd: add branch hint in
> recv/xmit
>
> On Tue, Sep 30, 2014 at 11:14:40AM +0000, Hiroshi Shimamoto wrote:
> > From: Hiroshi Shimamoto
> >
> > To reduce instruction cache miss, add branch condition hint
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
In rte_pktmbuf_free(), there might be cache miss/memory stall issue.
In small packet case, it could harm the performance.
>From the result of memnic-tester, in less than 1024 frame size the
performance could be improved.
Using Xeon E
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
To reduce instruction cache miss, add branch condition hints into
recv/xmit functions. This improves a bit performance.
We can see performance improvements with memnic-tester.
Using Xeon E5-2697 v2 @ 2.70GHz, 4 vCPU.
size | before |
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Prefetch the next packet area to reduce memory stall cycles.
Prefetching the next packet area could hide memory stall, because the next
area will be accessed just after processing the current receive operations.
We can see perfo
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
x86 can keep store ordering with standard operations.
Using memory barrier is much expensive in main packet processing loop.
Removing this improves xmit/recv packet performance.
We can see performance improvements with memnic-tester.
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Do not touch pktmbuf directly.
Intead of direct access, use rte_pktmbuf_pkt_len() and rte_pktmbuf_data_len()
to access the property.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 6 +++---
1 file c
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Because these assignment are done in rte_pktmbuf_alloc(), get rid of them.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/pmd/pmd_memnic.c
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Introduce memnic-tester which benchmarks MEMNIC PMD performance in guest.
It starts with two threads, one thread produces and consumes packets,
other thread receives packets and directly transmits the received
packets. This evaluates MEMN
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
This patchset improves MEMNIC PMD performance.
The first patch introduces a new benchmark test run in guest,
and will be used to evaluate the following patch effects.
This patchset improves the throughput results of memnic-tester.
Usin
> Subject: Re: [dpdk-dev] DPDK doesn't work with iommu=pt
>
>
>
> On Mon, Sep 29, 2014 at 2:53 AM, Hiroshi Shimamoto ct.jp.nec.com> wrote:
> > Hi,
> >
> >> Subject: Re: [dpdk-dev] DPDK doesn't work with iommu=pt
> >>
> >> iom
witch to another OS distribution also removed the symptom
> > 3) even different HW we will not see the symptom. In my case, switch from
> > Engineering board to EPSD board.
> >
> > Regards,
> > Choi, Sy Jong
> > Platform Application Engineer
> >
> >
> &g
bled.
What we want to do is that use SR-IOV functionality and DPDK application
concurrently on the same box.
thanks,
Hiroshi
>
> >-Original Message-
> >From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Hiroshi Shimamoto
> >Sent: Friday, September 26, 2014 5:14
I encountered an issue that DPDK doesn't work with "iommu=pt intel_iommu=on"
on HP ProLiant DL380p Gen8 server. I'm using the following environment;
HW: ProLiant DL380p Gen8
CPU: E5-2697 v2
OS: RHEL7
kernel: kernel-3.10.0-123 and the latest kernel 3.17-rc6+
DPDK: v1.7.1-53-gce5abac
Hi Thomas, Keith,
> Subject: Re: [dpdk-dev] [memnic PATCH 7/7] pmd: split calling mbuf free
>
>
> On Sep 24, 2014, at 10:20 AM, Thomas Monjalon
> wrote:
>
> > 2014-09-11 07:52, Hiroshi Shimamoto:
> >> @@ -408,9 +408,9 @@ retry:
> >>
> >>
> Subject: Re: [dpdk-dev] [memnic PATCH 4/7] pmd: use compiler barrier
>
> 2014-09-11 07:48, Hiroshi Shimamoto:
> > x86 can keep store ordering with standard operations.
>
> Are we sure it's always the case (including old 32-bit CPU)?
> I would prefer to have a reference
> Subject: Re: [dpdk-dev] [memnic PATCH 3/7] pmd: use helper macros
>
> 2014-09-11 07:47, Hiroshi Shimamoto:
> > Do not touch pktmbuf directly.
> >
> > Instead of direct access, use rte_pktmbuf_pkt_len() and
> > rte_pktmbuf_data_len()
> > to access
> Subject: Re: [dpdk-dev] [memnic PATCH 2/7] pmd: remove needless assignment
>
> 2014-09-11 07:47, Hiroshi Shimamoto:
> > Because these assignment are done in rte_pktmbuf_alloc(), get rid of them.
>
> Is it increasing the performances?
I hadn't tried to test, because
Hi Bruce,
> Subject: [dpdk-dev] [PATCH 0/3] eal affinitize low priority threads to lcore 0
>
> This patchset sets things up so that we can affinitize the interrupt,
> vfio management, and hpet timer management threads to lcore 0, so that
> they never interfere with data plane threads.
I don't
Hi Mukawa-san,
> Subject: Re: [dpdk-dev] [memnic PATCH 0/7] MEMNIC PMD performance improvement
>
> Hi Shimamoto-san,
>
>
> (2014/09/11 16:45), Hiroshi Shimamoto wrote:
> > From: Hiroshi Shimamoto
> >
> > This patchset improves MEMNIC PMD performance.
>
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
In rte_pktmbuf_free(), there might be cache miss/memory stall issue.
In small packet case, it could harm the performance.
>From the result of memnic-tester, in less than 1024 frame size the
performance could be improved.
Using Xeon E
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
To reduce instruction cache miss, add branch condition hints into
recv/xmit functions. This improves a bit performance.
We can see performance improvements with memnic-tester.
Using Xeon E5-2697 v2 @ 2.70GHz, 4 vCPU.
size | before |
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Prefetch the next packet area to reduce memory stall cycles.
Prefetching the next packet area could hide memory stall, because the next
area will be accessed just after processing the current receive operations.
We can see perfo
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
x86 can keep store ordering with standard operations.
Using memory barrier is much expensive in main packet processing loop.
Removing this improves xmit/recv packet performance.
We can see performance improvements with memnic-tester.
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Do not touch pktmbuf directly.
Instead of direct access, use rte_pktmbuf_pkt_len() and rte_pktmbuf_data_len()
to access the property.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 4 ++--
1 file c
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Because these assignment are done in rte_pktmbuf_alloc(), get rid of them.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/pmd/pmd_memnic.c
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Introduce memnic-tester which benchmarks MEMNIC PMD performance in guest.
It starts with two threads, one thread produces and consumes packets,
other thread receives packets and directly transmits the received
packets. This evaluates MEMN
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
This patchset improves MEMNIC PMD performance.
The first patch introduces a new benchmark test run in guest,
and will be used to evaluate the following patch effects.
This patchset improves the throughput results of memnic-tester.
Usin
Hi,
> Subject: Re: [dpdk-dev] [PATCH] eal/linuxapp: Add parameter to specify master
> lcore id
>
> 2014-07-23 08:53, Hiroshi Shimamoto:
> > 2014-07-23 09:50, Thomas Monjalon:
> > > 2014-07-22 23:40, Hiroshi Shimamoto:
> > > > does anyone have interest i
Hi all,
does anyone have interest in this functionality?
I think this is important and useful.
Since we should care about core assignment to get high performance
and the master lcore thread is special in DPDK, we will want to
assign the master to the target core.
For example, with hyperthreading
ting MEMNIC-1.2 or better understanding .
> 11.Is there any better application to test MEMNIC for VM-VM or VM to host
> data transfer .
I think the current host-sim doesn't have any packet switching capability, we
need to implement such
a functionality to test MEMNIC.
Actually, I started MEM
Hi,
> Subject: [dpdk-dev] MENNIC1.2 host-sim crashed for me
>
> Hi,
> I want to run MEMNIC 1.2 application .
>
> 1. I compiled DPDK1.6
>
> 2. I compiled memnic.12
>
> 3. And while running memnic-hostsim appgot strucked
>
> 4.
>
> 5. [root at localhost host-sim]#
gt;-Original Message-
> >From: Thomas Monjalon [mailto:thomas.monjalon at 6wind.com]
> >Sent: Wednesday, June 25, 2014 6:05 PM
> >To: Cao, Waterman
> >Cc: dev at dpdk.org; Hiroshi Shimamoto; Hayato Momma
> >Subject: Re: [dpdk-dev] [PATCH] kni: compatibility with RHEL 7
> &
Hi,
> Subject: Re: [dpdk-dev] Testing memnic for VM to VM transfer
>
> 2014-06-18 11:42, Hiroshi Shimamoto:
> > 2014-06-18 19:26, GongJinrong:
> > > Do you have any idea that how to write a host application
> > > to put the data to guest memnic PMD?
> &g
.
Set len and fill ether frame data, then set the status to 2 on host.
Guest PMD may receive the packet.
Proceed to the next packet block.
thanks,
Hiroshi
>
> -----Original Message-
> From: Hiroshi Shimamoto [mailto:h-shimamoto at ct.jp.nec.com]
> Sent: Wednesday, June 18
se it on host.
Could you please try DPDK vSwitch and enables MEMNIC vport?
DPDK vSwitch must handle packets between physical NIC port and MEMNIC vport
exposed to guest with dpdk.org memnic driver.
thanks,
Hiroshi
>
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org]
Hi,
> Subject: [dpdk-dev] Testing memnic for VM to VM transfer
>
> Hi everyone:
> We are interested in testing the performance of the memnic driver
> posted at http://dpdk.org/browse/memnic/refs/.
> We want to compare its performance compared to other techniques to transfer
>
t_hash? was here
skb_set_hash(struct sk_buff *skb, __u32 hash, enum pkt_hash_types type)
thanks,
Hiroshi
>
> Thanks
>
> Waterman
>
> -Original Message-
> >From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Hiroshi Shimamoto
> >Sent: Thursday, June
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
The below commit requires stdio FILE structure.
commit 591a9d7985c1230652d9f7ea1f9221e8c66ec188
Author: Stephen Hemminger
Date: Fri May 2 16:42:56 2014 -0700
add FILE argument to debug functions
Application which includes rte_me
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Compilation in RHEL7 is failed. This fixes the build issue.
RHEL7 has skb_set_hash, the kernel version is 3.10 though.
Don't define skb_set_hash for RHEL7.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
lib/libr
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Add the capability to change MTU.
On MTU change, remember the corresponding frame size and request new
frame size to the host on reset, if the host MEMNIC has that feature.
Don't trust framesz of header in general usage, because host
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Add framesz field in memnic data structure, and initialized with the current
frame size.
Replace length check on TX/RX with the above frame size.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
linux/memnic_net
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
If the MEMNIC framework has the feature MEMNIC_FEAT_FRAME_SIZE and
configured frame size, set request bit and frame size to support
larger frame size on reset.
Don't trust framesz of header in general usage, because host might change
the
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Add framesz field in adapter structure, and initialized with the current
frame size.
Replace length check on TX/RX with the above frame size.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 10 +++
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Update MEMNIC data structure in common header file.
Prepare to support extra features for MEMNIC.
Change name reserved to request which will be used to negotiate between
host and guest, and add feature flag and other definitions.
Sign
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
This patchset provides variable frame size functionality with MEMNIC
extra features framework.
First, update the memnic.h to synchronise upstream data structure which
has extra feature framework.
Next, prepare for changing frame size.
F
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Calling netif_receive_skb() from memnic thread may cause deadlock, if
softirq is not disabled.
The netif_receive_skb() should be called in softirq context, but memnic
thread is not softirq context. That may conflict softirq work like a
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Because DPDK has its own compare and set function to optimize to
dedicated processor type, use that rte_atomic32_cmpset() instead of
cmpxchg macro which is specially introduced for MEMNIC.
Signed-off-by: Hiroshi Shimamoto
---
common/me
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
The MEMNIC header file should be under Dual BSD/GPL license.
Put the license text "Dual BSD/GPL" into the file header.
Signed-off-by: Hiroshi Shimamoto
---
common/memnic.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/c
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
There is a race condition, on transmit to host.
Guest PMD Host
Thread-A Thread-B vSwitch
|idx=0 |idx=0 |p[0] st!=2
|cmpxchg ||
|p[0] st->1 ||
Hi,
> Subject: Re: [dpdk-dev] [memnic PATCH 1/5] pmd: fix race condition
>
> Hi Hiroshi-san,
>
> Please see my comments below.
>
> On 03/11/2014 06:37 AM, Hiroshi Shimamoto wrote:
> > From: Hiroshi Shimamoto
> >
> > There is a race condition, on tran
Hi,
> Subject: Re: [dpdk-dev] [memnic PATCH 3/5] pmd: implement stats of MEMNIC
>
> Hi,
>
> 11/03/2014 05:38, Hiroshi Shimamoto:
> > From: Hiroshi Shimamoto
> >
> > Implement missing feature to account statistics.
> > This patch adds just an infrast
andle the compilation on any kernel, we can just
> remove these functions as they are not needed.
>
> Signed-off-by: Olivier Matz
Acked-by: Hiroshi Shimamoto
thanks,
Hiroshi
> ---
> linux/memnic_net.c | 33 -
> 1 file changed, 33 deletions(-)
&g
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
The current MEMNIC PMD cannot handle multiple segments.
Add the functionality to transmit a mbuf which has multiple segments.
Walk every segment in transmitting mbuf and copy the data to MEMNIC
packet buffer.
Signed-off-by: Hiroshi Shi
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Implement packet accounting of MEMNIC on TX/RX.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 37 +++--
1 file changed, 35 insertions(+), 2 deletions(-)
diff --git
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Implement missing feature to account statistics.
This patch adds just an infrastructure.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
---
pmd/pmd_memnic.c | 45 ++---
1 file chang
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
Drop packets which have invalid length.
Normally this must not happen while vSwitch works fine, however
it's better to put a sentinel to prevent memory corruption.
Signed-off-by: Hiroshi Shimamoto
Reviewed-by: Hayato Momma
--
From: Hiroshi Shimamoto <h-shimam...@ct.jp.nec.com>
There is a race condition, on transmit to vSwitch.
Guest PMD Host
Thread-A Thread-B vSwitch
|idx=0 |idx=0 |p[0] st!=2
|cmpxchg ||
|p[0] st->1 ||
I never noticed about that and I haven't check compilation with newer kernel.
But I think you have completed to test it.
Fine to me.
thanks,
Hiroshi
> Subject: [dpdk-dev] [memnic PATCH] linux: fix build with kernel >= 3.3
>
> Signed-off-by: Olivier Matz
> ---
> linux/memnic_net.c | 28
> Subject: [dpdk-dev] [memnic PATCH] pmd: use memory barrier function instead
> of asm volatile
>
> Use the DPDK specific function rte_mb() instead of
> the GCC statement asm volatile ("" ::: "memory").
Yes, that's preferred for DPDK, I think.
Looks okay to me.
By the way, I was also asked to
> Subject: [dpdk-dev] [memnic PATCH] pmd: fix attributes
>
> Add missing "const" and remove useless "rte_unused" attributes.
Good catch. Looks fine to me.
thanks,
Hiroshi
>
> Signed-off-by: Olivier Matz
> ---
> pmd/pmd_memnic.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
Looks fine to me.
thanks,
Hiroshi
> Subject: [memnic PATCH 3/3] common: remove double underscores
>
> The usage of double underscores is reserved.
>
> Signed-off-by: Thomas Monjalon
> ---
> common/memnic.h |6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git
> Subject: [memnic PATCH 2/3] pmd: remove useless includes
>
> Signed-off-by: Thomas Monjalon
> ---
> common/memnic.h |4
> pmd/pmd_memnic.c |4
> 2 files changed, 8 deletions(-)
>
> diff --git a/common/memnic.h b/common/memnic.h
> index 6ff38a0..58dd019 100644
> ---
Hi,
> Subject: [memnic PATCH 1/3] pmd: remove symlink
>
> No need to have a symbolic link to a common file
> when it can be simply included.
Looks fine to me.
When I prepared the file with a bit complex file path layout,
easy to keep consistency.
Because you separated the code from DPDK
75 matches
Mail list logo