Hi Przemek,
Thank you for your response, It's really provided us breakthrough.
After setting up DPDK on compute node for stable/kilo, Trying to set up
Openstack stable/liberty all-in-one setup, At present not able to get the IP
allocation for the vhost type instances through DHCP. Also tried
Interrupt based request of PF reset from PF is supported by
enabling the adminq event process in VF driver.
Users can register a callback for this interrupt event to get
informed, when a PF reset request detected like:
rte_eth_dev_callback_register(portid,
Currently, i40evf PMD uses a global static buffer to send virtchnl
command to host driver. It is shared by multi VFs.
This patch changed to allocate virtchnl cmd buffer for each VF.
Signed-off-by: Jingjing Wu
---
drivers/net/i40e/i40e_ethdev.h| 2 +
drivers/net/i40e/i40e_ethdev_vf.c | 181
v2 changes:
remove the change on vf reset status checking
add pf event report support in release note
If DPDK is used on VF while the host is using Linux Kernel
driver as PF driver on FVL NIC, some setting on PF will trigger
VF reset. DPDK VF need to know the event.
This patch set makes the
Currently, i40evf PMD uses a global static buffer to send virtchnl
command to host driver. It is shared by multi VFs.
This patch changed to allocate virtchnl cmd buffer for each VF.
Signed-off-by: Jingjing Wu
---
drivers/net/i40e/i40e_ethdev.h| 2 +
drivers/net/i40e/i40e_ethdev_vf.c | 181
Interrupt based request of PF reset from PF is supported by
enabling the adminq event process in VF driver.
Users can register a callback for this interrupt event to get
informed, when a PF reset request detected like:
rte_eth_dev_callback_register(portid,
v2 changes:
remove the change on vf reset status checking
add pf event report support in release note
If DPDK is used on VF while the host is using Linux Kernel
driver as PF driver on FVL NIC, some setting on PF will trigger
VF reset. DPDK VF need to know the event.
This patch set makes the
> -Original Message-
> From: Richardson, Bruce
> Sent: Wednesday, January 27, 2016 4:17 AM
> To: Wang, Xiao W
> Cc: Chen, Jing D ; dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH 15/16] fm10k: use default mailbox message
> handler for pf
>
> On Mon, Jan 25, 2016 at 02:31:05AM +,
Ping?
On Jan 19, 2016 5:16 PM, "Santosh Shukla" wrote:
> - virtio_recv_pkts_vec and other virtio vector friend apis are written for
> sse/avx instructions. For arm64 in particular, virtio vector
> implementation
> does not exist(todo).
>
> So virtio pmd driver wont build for targets like
Ping.
On Jan 19, 2016 5:16 PM, "Santosh Shukla" wrote:
> vfio_pci_mmap() try to map all pci bars. ioport region are not mapped in
> vfio/kernel so ignore mmaping for ioport.
>
> Signed-off-by: Santosh Shukla
> ---
> lib/librte_eal/linuxapp/eal/eal_pci_vfio.c | 20
> 1
Ping
On Jan 19, 2016 5:16 PM, "Santosh Shukla" wrote:
> make sure sys/io.h used only for x86 archs. This fixes build error
> arm64/arm case.
>
> Signed-off-by: Santosh Shukla
> ---
> drivers/net/virtio/virtio_pci.h |2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git
On Wed, Jan 27, 2016 at 07:53:21AM +0530, Santosh Shukla wrote:
> Ping?
I was on vacation late last week. And I was quite busy till now after
the vacation. So, sorry that I still don't have time to do more detailed
reviews in 1 or 2 days. Hopefully I can make it by this Friday.
BTW, I had a very
On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> + if (vq->vhost_hlen == sizeof(struct virtio_net_hdr_mrg_rxbuf)) {
> + *(struct virtio_net_hdr_mrg_rxbuf *)(uintptr_t)desc_addr = hdr;
> + } else {
> + *(struct virtio_net_hdr *)(uintptr_t)desc_addr = hdr.hdr;
> + }
On 1/11/2016 2:43 AM, Tan, Jianfeng wrote:
> Add a new virtual device named eth_cvio, it can be used just like
> eth_ring, eth_null, etc.
>
> Configured parameters include:
> - rx (optional, 1 by default): number of rx, only allowed to be
> 1 for now.
> -
On Wed, Jan 27, 2016 at 02:46:39AM +, Xie, Huawei wrote:
> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> > + if (vq->vhost_hlen == sizeof(struct virtio_net_hdr_mrg_rxbuf)) {
> > + *(struct virtio_net_hdr_mrg_rxbuf *)(uintptr_t)desc_addr = hdr;
> > + } else {
> > + *(struct
On Tue, Jan 26, 2016 at 10:30:12AM +, Xie, Huawei wrote:
> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> > Signed-off-by: Yuanhan Liu
> > ---
> > lib/librte_vhost/vhost_rxtx.c | 287
> > +-
> > 1 file changed, 113 insertions(+), 174 deletions(-)
>
>
On Thu, Jan 21, 2016 at 02:50:01PM +0100, J?r?me Jutteau wrote:
> Hi Yuanhan,
>
> 2015-12-14 2:47 GMT+01:00 Yuanhan Liu :
> > Right, I should move it in the beginning of this function.
>
> Any news about this refactoring ?
Hi J?r?me,
Thanks for showing interests in this patch set; I was
On Thu, Jan 21, 2016 at 12:49:10PM +0100, Thomas Monjalon wrote:
> 2016-01-19 16:12, Yuanhan Liu:
> > int
> > vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
> > {
> > - hw->vtpci_ops = _ops;
> > + hw->dev = dev;
> > +
> > + /*
> > +* Try if we can succeed
On Thu, Jan 21, 2016 at 12:37:42PM +0100, Thomas Monjalon wrote:
> 2016-01-19 16:12, Yuanhan Liu:
> > +#define IO_READ_DEF(nr_bits, type) \
> > +static inline type \
> > +io_read##nr_bits(type *addr) \
> > +{
v2:
* Put the two extra fix patches ahead of the base code patches.
Wang Xiao W (16):
fm10k: use default mailbox message handler for pf
fm10k/base: add macro definitions that are needed
fm10k/base: cleanup namespace pollution and correct typecast
fm10k/base: use bitshift for itr_scale
The new share code makes fm10k_msg_update_pvid_pf function static, so we can
not refer to it now in fm10k_ethdev.c. The registered pf handler is almost the
same as the default pf handler, removing it has no impact on mailbox.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/fm10k_ethdev.c | 17
Some macros such as FM10K_RXINT_TIMER_SHIFT are removed in the share
code drop, but they are needed in dpdk/fm10k. This patch put all these
necessary macros into fm10k_osdep.h
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_osdep.h | 30 ++
1 file
Correct typecast in fm10k_update_xc_addr_pf.
Make functions that are only referenced locally static.
And fix the function header comment for fm10k_tlv_attr_nest_stop() while
we're at it.
Wrap fm10k_msg_data fm10k_iov_msg_data_pf[] in the new ifndef
NO_DEFAULT_SRIOV_MSG_HANDLERS so that drivers
Upstream community wishes us to use bitshift instead of a divisor,
because this is faster, and prevents any need for a '0' check. In our
case, this even works out because default Gen3 will be 0.
Because of this, we are also able to remove the check for non-zero value
in the vf code path since
VF drivers must detect how many queues are available. Previously, the
driver assumed that each VF has at minimum 1 queue. This assumption is
incorrect, since it is possible that the PF has not yet assigned the
queues to the VF by the time the VF checks. To resolve this, we added a
check first to
Add comments which properly explain the undocumented use of bits in
TDLEN register prior to VF initializing it to the correct value. Note
that the mechanism is entirely software-defined and explain its purpose
to help reduce confusion in the future.
Signed-off-by: Wang Xiao W
---
Cleanup lines over 80 characters.
Cleanup useless else, checkpatch warns that else is not generally
useful after a break or return.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_mbx.c | 2 +-
drivers/net/fm10k/base/fm10k_pf.c | 19 ++-
2 files changed, 11
The upstream Linux kernel community prefers avoiding CamelCase in
variables, function names, etc.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_type.h | 14 +++---
drivers/net/fm10k/fm10k_ethdev.c| 24
2 files changed, 19 insertions(+), 19
Use memcpy instead of copying MAC address byte-by-byte.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_pf.c | 7 ++-
1 file changed, 2 insertions(+), 5 deletions(-)
diff --git a/drivers/net/fm10k/base/fm10k_pf.c
b/drivers/net/fm10k/base/fm10k_pf.c
index 7d48210..a1469aa
The Linux Kernel provides the OS a call "pcie_get_minimum_link" which
can crawl the PCIe tree and determine the actual minimum link speed of a
device which is a more general check than provided by
is_slot_appropriate. Thus, the upstream driver does not use or want the
is_slot_appropriate function
The vid variable name is shorthand for VLAN ID, so we should use this in
comments explaining what is happening.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_pf.c | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/net/fm10k/base/fm10k_pf.c
The comment here was changed during review of upstream patch, and the
new wording is slightly more clear. Re-write the comment in SHARED code
based on this new wording.
Fix a number of mailbox comment issues with function header comments,
lower-case acronyms (i.e. FIFO, TLV), incorrect function
Per comments from an upstream patch, and looking at how TLV LE_STRUCT
code works, we actually want these structures to be 4byte aligned, not
1byte aligned. In practice, 1byte alignment has worked so far because
all our structures end up being a multiple of 4. But if a future TLV
structure were
The upstream Linux kernel community prefers constants are to the right of
binary operators.
Signed-off-by: Wang Xiao W
---
drivers/net/fm10k/base/fm10k_pf.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/net/fm10k/base/fm10k_pf.c
On 1/27/2016 11:22 AM, Yuanhan Liu wrote:
> On Wed, Jan 27, 2016 at 02:46:39AM +, Xie, Huawei wrote:
>> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
>>> + if (vq->vhost_hlen == sizeof(struct virtio_net_hdr_mrg_rxbuf)) {
>>> + *(struct virtio_net_hdr_mrg_rxbuf *)(uintptr_t)desc_addr =
On Wed, Jan 27, 2016 at 05:56:56AM +, Xie, Huawei wrote:
> On 1/27/2016 11:22 AM, Yuanhan Liu wrote:
> > On Wed, Jan 27, 2016 at 02:46:39AM +, Xie, Huawei wrote:
> >> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> >>> + if (vq->vhost_hlen == sizeof(struct virtio_net_hdr_mrg_rxbuf)) {
> >>> +
On 1/27/2016 11:26 AM, Yuanhan Liu wrote:
> On Tue, Jan 26, 2016 at 10:30:12AM +, Xie, Huawei wrote:
>> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
>>> Signed-off-by: Yuanhan Liu
>>> ---
>>> lib/librte_vhost/vhost_rxtx.c | 287
>>> +-
>>> 1 file changed,
On Wed, Jan 27, 2016 at 06:12:22AM +, Xie, Huawei wrote:
> On 1/27/2016 11:26 AM, Yuanhan Liu wrote:
> > On Tue, Jan 26, 2016 at 10:30:12AM +, Xie, Huawei wrote:
> >> On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> >>> Signed-off-by: Yuanhan Liu
> >>> ---
> >>> lib/librte_vhost/vhost_rxtx.c |
On 1/27/2016 2:02 PM, Yuanhan Liu wrote:
> On Wed, Jan 27, 2016 at 05:56:56AM +, Xie, Huawei wrote:
>> On 1/27/2016 11:22 AM, Yuanhan Liu wrote:
>>> On Wed, Jan 27, 2016 at 02:46:39AM +, Xie, Huawei wrote:
On 12/3/2015 2:03 PM, Yuanhan Liu wrote:
> + if (vq->vhost_hlen ==
On Wed, Jan 27, 2016 at 06:16:37AM +, Xie, Huawei wrote:
> On 1/27/2016 2:02 PM, Yuanhan Liu wrote:
> > On Wed, Jan 27, 2016 at 05:56:56AM +, Xie, Huawei wrote:
> >> On 1/27/2016 11:22 AM, Yuanhan Liu wrote:
> >>> On Wed, Jan 27, 2016 at 02:46:39AM +, Xie, Huawei wrote:
> On
Laurent, have you resolved this problem?
I'm using the same NIC as yours (i.e. Intel 82598EB 10G NIC) and faced the
same problem as you.
Here is parts of my log and it says that PMD cannot enable RX queue for my
NIC.
I'm using DPDK 2.2.0 and used 'null' for the 4th parameter in calling
On Wed, 2016-01-27 at 15:50 +0900, Moon-Sang Lee wrote:
>
>
> Laurent, have you resolved this problem?
> I'm using the same NIC as yours (i.e. Intel 82598EB 10G NIC) and faced
> the same problem as you.
> Here is parts of my log and it says that PMD cannot enable RX queue
> for my NIC.
> I'm
Moon-Sang
Were you using pktgen or else application?
Could you help to share with me the detailed steps of your reproducing that
issue?
We will find time on that soon later. Thanks!
Regards,
Helin
-Original Message-
From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Laurent GUERBY
>
>I have to practically identical systems, same hypervisor on each (Centos
>7.x). In one, I have a 57800 card which works fine with DPDK with
>SRIOV. In the other, I have a 57810 card which doesn't work with SRIOV.
>
>For the 57810 I have tracked this down to the status block in the VF
>failing
2016-01-27 11:46, Yuanhan Liu:
> On Thu, Jan 21, 2016 at 12:49:10PM +0100, Thomas Monjalon wrote:
> > 2016-01-19 16:12, Yuanhan Liu:
> > > int
> > > vtpci_init(struct rte_pci_device *dev, struct virtio_hw *hw)
> > > {
> > > - hw->vtpci_ops = _ops;
> > > + hw->dev = dev;
> > > +
> >
Hello Jingjing,
On Wed, Jan 27, 2016 at 2:49 AM, Jingjing Wu wrote:
> Interrupt based request of PF reset from PF is supported by
> enabling the adminq event process in VF driver.
> Users can register a callback for this interrupt event to get
> informed, when a PF reset request detected like:
>
Fixed issue of byte order in ethdev library that the structure
for setting fdir's mask and flow entry is inconsist and made
inputs of mask be in big endian.
fixes: 76c6f89e80d4 ("ixgbe: support new flow director masks")
2d4c1a9ea2ac ("ethdev: add new flow director masks")
Reported-by:
v6 changes:
reflect the changes in release notes and library version map file
revise our duff's code style a bit to make it more readable
v5 changes:
add comment about duff's device and our variant implementation
v4 changes:
fix a silly typo in error handling when rte_pktmbuf_alloc fails
v3
v6 changes:
reflect the changes in release notes and library version map file
revise our duff's code style a bit to make it more readable
v5 changes:
add comment about duff's device and our variant implementation
v3 changes:
move while after case 0
add context about duff's device and why we
v4 changes:
fix a silly typo in error handling when rte_pktmbuf_alloc fails
reported by haifeng
pre-allocate a bulk of mbufs instead of allocating one mbuf a time on demand
Signed-off-by: Gerald Rogers
Signed-off-by: Huawei Xie
Acked-by: Konstantin Ananyev
Acked-by: Yuanhan Liu
Tested-by:
Hi Anatoly,
Few small comments.
The comments "function pointer typedef" or "structure to hold" don't
bring new information. Please keep it short.
2016-01-13 12:36, Anatoly Burakov:
> +/* function pointer typedef for DMA mapping functions */
-> DMA mapping function type
It would be
2016-01-27 16:37, Jingjing Wu:
> Fixed issue of byte order in ethdev library that the structure
> for setting fdir's mask and flow entry is inconsist and made
> inputs of mask be in big endian.
Please be more precise. Which one is big endian?
Wasn't it tested before?
> fixes: 76c6f89e80d4
v2 changes:
add missed commit message in v1
fix the error reported by checkpatch:
"ERROR: return is not a function, parentheses are not required"
also removed other extra parentheses like:
"return val == 0"
"return (rte_mempool_lookup(...))"
Signed-off-by: Huawei Xie
---
2015-12-16 16:38, Stephen Hemminger:
> rte_get_log_type and rte_get_log_level functions has been avaliable
> for many versions. But they are missing from the shared library map
> and therefore do not get exported correctly.
>
> Signed-off-by: Stephen Hemminger
> ---
>
On 1/26/2016 10:58 AM, Tetsuya Mukawa wrote:
> On 2016/01/25 19:15, Xie, Huawei wrote:
>> On 1/22/2016 6:38 PM, Tetsuya Mukawa wrote:
>>> On 2016/01/22 17:14, Xie, Huawei wrote:
On 1/21/2016 7:09 PM, Tetsuya Mukawa wrote:
> virtio: Extend virtio-net PMD to support container environment
On Tue, Jan 26, 2016 at 4:50 PM, Jan Viktorin
wrote:
> What about the RTE_VERIFY? I think, it's more appropriate here.
Well, here, I am removing useless checks in static functions.
But for the rest of ethdev api, I agree we could add some RTE_VERIFY.
> Otherwise, feel free to add:
>
>
On 1/21/2016 7:09 PM, Tetsuya Mukawa wrote:
> + /* Set BAR region */
> + for (i = 0; i < NB_BAR; i++) {
> + switch (dev->bar[i].type) {
> + case QTEST_PCI_BAR_IO:
> + case QTEST_PCI_BAR_MEMORY_UNDER_1MB:
> + case QTEST_PCI_BAR_MEMORY_32:
> +
The passthrough pipeline implementation is extended with load balancing
function. This function allows uniform distribution of the packets among
its output ports. For packets distribution, any application level logic
can be applied. For instance, in this implementation, hash value
computed over
2016-01-27 10:08, Burakov, Anatoly:
> > Why a new file for these functions?
>
> Well, my thought was to make future extensions easier by way of avoiding
> mixing irrelevant and/or general code with driver-specific code. I can change
> it back if that's not OK.
No strong opinion here.
David?
On Wed, Jan 27, 2016 at 11:12 AM, Thomas Monjalon
wrote:
> 2016-01-27 10:08, Burakov, Anatoly:
>> > Why a new file for these functions?
>>
>> Well, my thought was to make future extensions easier by way of avoiding
>> mixing irrelevant and/or general code with driver-specific code. I can
>>
> >> > Why a new file for these functions?
> >>
> >> Well, my thought was to make future extensions easier by way of
> avoiding mixing irrelevant and/or general code with driver-specific code. I
> can
> change it back if that's not OK.
> >
> > No strong opinion here.
> > David?
>
> Hum, no
This patch adds a new function to the EAL API:
int rte_eal_primary_proc_alive(const char *path);
The function indicates if a primary process is alive right now.
This functionality is implemented by testing for a write-
lock on the config file, and the function tests for a lock.
The use case for
> From: Richardson, Bruce
> > Agreed, however hiding it totally removes the flexibility of waiting for a
> > primary
> > that is starting with --file-prefix (aka: in a non-default location).
> > Imposing
> > a limit on only monitoring primary procs in the default location seems
> > wrong.
>
>
Ping?
On Tue, Jan 19, 2016 at 5:16 PM, Santosh Shukla wrote:
> For non-x86 arch, Compiler will throw build error for in/out apis. Including
> dummy api function so to pass build.
>
> Note that: For virtio to work for non-x86 arch - RTE_EAL_VFIO is the only
> supported method. RTE_EAL_IGB_UIO is
On Tue, Jan 26, 2016 at 9:51 PM, Santosh Shukla wrote:
> On Tue, Jan 26, 2016 at 7:58 PM, Thomas Monjalon
> wrote:
>> 2016-01-26 19:35, Santosh Shukla:
>>> On Tue, Jan 26, 2016 at 6:30 PM, Thomas Monjalon
>>> wrote:
>>> > 2016-01-26 15:56, Santosh Shukla:
>>> >> In my observation, currently
This patch fixes the socket-id error in ip_pipeline sample
application running over uni-processor systems.
Signed-off-by: Jasvinder Singh
Acked-by: Cristian Dumitrescu
---
v2:
- used SOCKET_ID_ANY instead of -1
examples/ip_pipeline/init.c | 14 +++---
1 file changed, 11 insertions(+),
Hi Abhijeet,
It seems you are almost there!
When booting the VM?s do you request hugepage memory for them (by setting
hw:mem_page_size=large in flavor extra_spec)?
If not then please do, if yes then please look into libvirt logfiles for the
VM?s (in /var/log/libvirt/qemu/instance-xxx), I think
Hi Neil,
On 1/26/2016 10:19 PM, Neil Horman wrote:
> On Tue, Jan 26, 2016 at 10:22:18AM +0800, Tan, Jianfeng wrote:
>> Hi Neil,
>>
>> On 1/25/2016 9:46 PM, Neil Horman wrote:
>>> On Mon, Jan 25, 2016 at 02:49:53AM +0800, Jianfeng Tan wrote:
>> ...
--
2.1.4
>>> This doesn't
2016-01-07 16:17, Panu Matilainen:
> On 01/03/2016 07:56 PM, Huawei Xie wrote:
> > v2 changes:
> > change LOG level from ERR to INFO
> >
> > virtio PMD could use IO port to configure the virtio device without
> > using uio driver.
> >
> > There are two issues with previous implementation:
> > 1)
On Fri, Jan 22, 2016 at 4:27 PM, David Marchand
wrote:
> The 4th patch introduces a change in linux eal.
> Before, if a pci device was bound to no kernel driver, eal would set kdrv
> to "unknown". With this change, kdrv is set to "none".
> This might make it possible to avoid the old issue of
Hi Alex,
> On 01/23/2016 04:23 AM, Alex Williamson wrote:
> > Using iommu_present() to determine whether an IOMMU group is real or
> > fake has some problems. First, apparently Power systems don't
> > register an IOMMU on the device bus, so the groups and containers get
> > marked as noiommu and
On 01/26/2016 06:15 PM, Marcin Kerlin wrote:
> This patch adds new function rte_jobstats_abort. It marks *job* as finished
> and time of this work will be add to management time instead of execution
> time.
> This function should be used instead of rte_jobstats_finish if condition
> occure,
>
2016-01-20 21:15, Matthew Hall:
> On 1/20/16 7:27 AM, Thomas Monjalon wrote:
> > Hi Matthew,
> >
> > RTE_SDK_BIN is an internal variable and should not be overriden.
> >
> > Have you installed DPDK somewhere? Example:
> > make install O=mybuild DESTDIR=mylocalinstall
> >
> > Then you should
On Wed, Jan 27, 2016 at 6:51 PM, Burakov, Anatoly
wrote:
> Hi Alex,
>
>> On 01/23/2016 04:23 AM, Alex Williamson wrote:
>> > Using iommu_present() to determine whether an IOMMU group is real or
>> > fake has some problems. First, apparently Power systems don't
>> > register an IOMMU on the
On 01/26/2016 07:03 PM, Huawei Xie wrote:
> v6 changes:
> reflect the changes in release notes and library version map file
> revise our duff's code style a bit to make it more readable
>
> v5 changes:
> add comment about duff's device and our variant implementation
>
> v3 changes:
> move
This commit is adding a generic mechanism to support multiple IOMMU
types. For now, it's only type 1 (x86 IOMMU) and no-IOMMU (a special
VFIO mode that doesn't use IOMMU at all), but it's easily extended
by adding necessary definitions into eal_pci_init.h and a DMA
mapping function to
Apologies, lost the signoff from Santosh Shukla and also the commit message
still mentions the file that is now non-existent, so I'll submit a v4.
Thanks,
Anatoly
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Anatoly Burakov
> Sent: Wednesday, January
This commit is adding a generic mechanism to support multiple IOMMU
types. For now, it's only type 1 (x86 IOMMU) and no-IOMMU (a special
VFIO mode that doesn't use IOMMU at all), but it's easily extended
by adding necessary definitions into eal_pci_init.h and a DMA
mapping function to
2016-01-21 12:23, David Marchand:
> free() already handles NULL pointer.
>
> Signed-off-by: David Marchand
Applied, thanks
2016-01-22 15:06, David Marchand:
> It was first a preparation step for future patchsets, but I am not sure what
> will become of them, so sending this anyway since it does not hurt to clean
> this now.
>
> Changes since v1:
> - rebased on HEAD (previous patchset was based on another patch I sent
2016-01-17 22:05, Zhihong Wang:
> This patch set optimizes DPDK memcpy for AVX512 platforms, to make full
> utilization of hardware resources and deliver high performance.
On a related note, your expertise would be very valuable to review
these patches please:
(memcpy)
> Zhihong Wang (5):
> lib/librte_eal: Identify AVX512 CPU flag
> mk: Predefine AVX512 macro for compiler
> lib/librte_eal: Optimize memcpy for AVX512 platforms
> app/test: Adjust alignment unit for memcpy perf test
> lib/librte_eal: Tune memcpy for prior platforms
>
>
On Wed, Jan 27, 2016 at 4:11 PM, Santosh Shukla wrote:
> On Tue, Jan 26, 2016 at 9:51 PM, Santosh Shukla wrote:
>> On Tue, Jan 26, 2016 at 7:58 PM, Thomas Monjalon
>> wrote:
>>> 2016-01-26 19:35, Santosh Shukla:
On Tue, Jan 26, 2016 at 6:30 PM, Thomas Monjalon
wrote:
>
2016-01-27 21:02, Santosh Shukla:
> 1. virtio currently works for vfio+noiommu and likely will work for
> vfio+iommu in near future.
> 2. So remove __noiommu suffix and always use default.
> 3. Introduce vfio resource parsing global function, That function
> suppose to do parsing for default vfio
On Wed, Jan 27, 2016 at 9:09 PM, Thomas Monjalon
wrote:
> 2016-01-27 21:02, Santosh Shukla:
>> 1. virtio currently works for vfio+noiommu and likely will work for
>> vfio+iommu in near future.
>> 2. So remove __noiommu suffix and always use default.
>> 3. Introduce vfio resource parsing global
> -Original Message-
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Panu Matilainen
> Sent: Wednesday, January 27, 2016 2:38 PM
> To: Kerlin, MarcinX ; dev at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH V1 1/1] jobstats: added function abort for job
>
> On 01/26/2016 06:15 PM,
On Wed, 2016-01-27 at 07:32 +, Harish Patil wrote:
> >
> >I have to practically identical systems, same hypervisor on each
> (Centos
> >7.x).??In one, I have a 57800 card which works fine with DPDK with
> >SRIOV.??In the other, I have a 57810 card which doesn't work with
> SRIOV.
> >
> >For
Hi Thomas,
> > +/* DMA mapping function prototype.
> > + * Takes VFIO container fd as a parameter.
> > + * Returns 0 on success, -1 on error.
> > + * */
> > +typedef int (*vfio_dma_func_t)(int);
> > +
> > +struct vfio_iommu_type {
> > + int type_id;
> > + const char *name;
> > +
Hi Przemek,
Thanks for the quick response. Now able to get the DHCP ip's for 2
vhostuser instances and able to ping each other. Isssue was a bug in
cirros 0.3.0 images which we were using in openstack after using 0.3.1
image as given in the URL(
This work is to make DPDK ports more visible and to enable using common
Linux tools to configure DPDK ports.
Patch is based on KNI but contains only control functionality of it,
also this patch does not include any Linux kernel network driver as
part of it.
Basically with the help of a kernel
This kernel module is based on KNI module, but this one is stripped
version of it and only for control messages, no data transfer
functionality provided.
This Linux kernel module helps userspace application create virtual
interfaces and when a control command issued into that virtual
interface,
This library gets control messages form kernelspace and forwards them to
librte_ether and returns response back to the kernelspace.
Library does:
1) Trigger Linux virtual interface creation
2) Initialize the netlink socket communication
3) Provides process() API to the application that does
Control interface APIs added into the sample application.
To have the support corresponding kernel module (KCP) needs to be inserted.
If kernel module is not there, application will run as it is without
kernel control path support.
When KCP module inserted, running application creates a virtual
Hi Thomas,
> > Is it possible (is it better) to declare these functions with
> > vfio_dma_func_t?
>
> Yeah, sure. Or maybe the other way around - maybe we could do away with
> the typedef. I'll go for the former though.
No, we can't declare the functions with a function pointer. At least I
This is slow data path communication implementation based on existing KNI.
Difference is: librte_kni converted into a PMD, kdp kernel module is almost
same except all control path functionality removed and some simplification done.
Motivation is to simplify slow path data communication.
Now any
This kernel module is based on KNI module, but this one is stripped
version of it and only for data messages, no control functionality
provided.
FIFO implementation of the KNI is kept exact same, but ethtool related
code removed and virtual network management related code simplified.
This module
This patch provides slow data path communication to the Linux kernel.
Patch is based on librte_kni, and heavily re-uses it.
The main difference is librte_kni library converted into a PMD, to
provide ease of use for applications.
Now any application can use slow path communication without any
2015-12-30 16:59, Zhihong Wang:
> Zhihong Wang (3):
> app/test-pmd: Handle SIGINT and SIGTERM in testpmd
> examples/l2fwd: Handle SIGINT and SIGTERM in l2fwd
> examples/l3fwd: Handle SIGINT and SIGTERM in l3fwd
Applied, thanks
On 1/21/2016 7:09 PM, Tetsuya Mukawa wrote:
> +qtest_find_pci_device(struct qtest_session *s, uint16_t bus, uint8_t device)
> +{
> + struct qtest_pci_device *dev;
> + uint32_t val;
> +
> + val = qtest_pci_inl(s, bus, device, 0, 0);
> + TAILQ_FOREACH(dev, >head, next) {
> +
On Wed, Jan 27, 2016 at 9:26 PM, Santosh Shukla wrote:
> On Wed, Jan 27, 2016 at 9:09 PM, Thomas Monjalon
> wrote:
>> 2016-01-27 21:02, Santosh Shukla:
>>> 1. virtio currently works for vfio+noiommu and likely will work for
>>> vfio+iommu in near future.
>>> 2. So remove __noiommu suffix and
1 - 100 of 124 matches
Mail list logo
