Introducing the DEV_TX_OFFLOAD_MT_LOCKFREE TX capability flag.
if a PMD advertises DEV_TX_OFFLOAD_MT_LOCKFREE capable, multiple threads
can invoke rte_eth_tx_burst() concurrently on the same tx queue without
SW lock. This PMD feature will be useful in the following use cases and
found in the OCTEON
On Wednesday 05 July 2017 05:25 AM, Gaetan Rivet wrote:
> From: Thomas Monjalon
>
> The bus name was stored with embedded double quotes.
> Indeed the bus name is given with a string in a macro,
> which is not used elsewhere.
> These macros are useless because the buses are drivers,
> so they must
Yes sure, I will send it
> -Original Message-
> From: Ferruh Yigit [mailto:ferruh.yi...@intel.com]
> Sent: Wednesday, July 05, 2017 18:32
> To: Vasily Philipov ; dev@dpdk.org
> Cc: Adrien Mazarguil ; Nélio Laranjeiro
>
> Subject: Re: [dpdk-dev] [PATCH v8 1/4] app/testpmd: add isolated mod
-Original Message-
> Date: Wed, 05 Jul 2017 19:46:54 +0200
> From: Thomas Monjalon
> To: Jerin Jacob
> Cc: dev@dpdk.org, bruce.richard...@intel.com, harry.van.haa...@intel.com,
> hemant.agra...@nxp.com, gage.e...@intel.com, nipun.gu...@nxp.com,
> santosh.shu...@caviumnetworks.com
> Subj
Currently, setting MTU will fail if MTU requires the support
of scattered packets before scattered_rx is enabled.
This patch allows setting this special MTU when device is
stopped, cause scattered_rx will be re-configured during
starting port.
Signed-off-by: Beilei Xing
---
v3 changes:
-Check de
-Original Message-
> Date: Wed, 5 Jul 2017 12:15:51 +0100
> From: "Hunt, David"
> To: Jerin Jacob
> CC: dev@dpdk.org, harry.van.haa...@intel.com, Gage Eads
> , Bruce Richardson
> Subject: Re: [PATCH v5 1/3] examples/eventdev_pipeline: added sample app
> User-Agent: Mozilla/5.0 (Windows
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Ferruh Yigit
> Sent: Wednesday, July 5, 2017 10:00 PM
> To: Xing, Beilei ; Lu, Wenzhuo
>
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] net/ixgbe: remove MTU setting
> limitation when port's stopped
>
>
On Wed, Jul 05, 2017 at 11:48:20PM +0200, Thomas Monjalon wrote:
> Hi,
>
> Following community requests, the Linux Foundation tried to get
> the ownership of the DPDK account on GitHub which was owned
> by someone.
> Good news, it worked:
> https://github.com/dpdk
> Thanks Trishan for the he
> -Original Message-
> From: Rybalchenko, Kirill
> Sent: Thursday, July 6, 2017 12:15 AM
> To: Zhang, Roy Fan ; Doherty, Declan
>
> Cc: dev@dpdk.org; Rybalchenko, Kirill
> Subject: [PATCH v3] crypto/scheduler: add multicore scheduling mode
>
> Multi-core scheduling mode is a mode where s
20/06/2017 10:34, De Lara Guarch, Pablo:
> > > + if (stats_period != 0) {
> > > + uint64_t prev_tsc = 0, cur_tsc, timer_tsc = 0;
> > > + uint64_t timer_period;
> > > +
> > > + /* Convert to number of cycles */
> > > + timer_per
> > > Add parameter to start forwarding sending first a burst of packets,
> > > which is useful when testing a loopback connection.
> > >
> > > This was already implemented as an internal command, but adding it as
> > > a parameter is interesting, as it allows the user to test a loopback
> > > conn
Any comment, please?
12/06/2017 11:38, Shijith Thotton:
> Set UIO info device file operations open and release. Call pci reset
> function inside open and release to clear device state at start and end.
> Copied this behaviour from vfio_pci kernel module code. With this patch,
> it is not mandatory
05/07/2017 19:58, Stephen Hemminger:
> In order to support VMBUS, the ethernet device name needs to be increased.
> What would be good timing; vmbus support will be in 17.11.
> Either break ABI now, and vmbus won't have to do it later.
> Or wait and make it part of of VMBUS patches?
Please send a
05/07/2017 18:55, Stephen Hemminger:
> Changes to PCI and EAL to handle full width of PCI domain
> value. It turns out that PCI domain is only 16 bits in ACPI,
> but some environments use the extra bits in PCI standard to allow
> encoding of PCI buses that are outside of ACPI.
>
> This version fix
Radu, Fan,
You have change the API as announced in the deprecation notice.
Please, could you update the release notes as well in the
"API Changes" section?
Thank you
---
crypto/scheduler: remove deprecated functions
This patc
14/06/2017 12:37, Andrew Rybchenko:
> On 05/25/2017 08:40 PM, Stephen Hemminger wrote:
> > On Thu, 25 May 2017 16:57:53 +0100
> > Andrew Rybchenko wrote:
> >
> >> From: Roman Zhukov
> >>
> >> Check that numbers of Rx and Tx descriptors satisfy descriptors limits
> >> from the Ethernet device info
> On Jul 5, 2017, at 11:12 AM, Yongseok Koh wrote:
>
> To make vectorized burst routines enabled, it is required to run on x86_64
> architecture which can support at least SSE4.1. If all the conditions are
> met, the vectorized burst functions are enabled automatically. The decision
> is made in
04/05/2017 16:10, Ferruh Yigit:
> -u accepts "dpdk" argument to unbind all devices bound to a DPDK driver.
>
> Usage:
> usertools/dpdk-devbind.py -u dpdk
>
> Example:
> $ usertools/dpdk-devbind.py -s
>
> Network devices using DPDK-compatible driver
>
09/06/2017 20:36, Ferruh Yigit:
> Device name resides in two different locations, in rte_device->name and
> in ethernet device private data.
>
> For now, the copy in the ethernet device private data is required for
> multi process support, the name is the how secondary process finds about
> primar
Hi,
Following community requests, the Linux Foundation tried to get
the ownership of the DPDK account on GitHub which was owned
by someone.
Good news, it worked:
https://github.com/dpdk
Thanks Trishan for the help.
First of all, it is not planned to use GitHub and its pull request model,
On Tue, 4 Jul 2017 17:13:28 +0100
Ferruh Yigit wrote:
> +
> + if (nl_msg->err == 0 && recv_len != expected_len)
> + pr_info("Expected and received len not match "
> + "%zu - %zu\n", recv_len, expected_len);
> +
On Tue, 4 Jul 2017 17:13:29 +0100
Ferruh Yigit wrote:
> +static void unci_net_stats64(struct net_device *dev,
> + struct rtnl_link_stats64 *stats)
> +{
> + int err;
> +
> + err = unci_nl_exec(UNCI_REQ_GET_STATS, dev, NULL, 0,
> + stats, sizeof(struct rtnl_
On Tue, 4 Jul 2017 17:13:29 +0100
Ferruh Yigit wrote:
> +#if (KERNEL_VERSION(3, 9, 0) <= LINUX_VERSION_CODE)
> +static int unci_net_change_carrier(struct net_device *dev, bool new_carrier)
> +{
> + if (new_carrier)
> + netif_carrier_on(dev);
> + else
> + netif_car
On Tue, 4 Jul 2017 17:13:30 +0100
Ferruh Yigit wrote:
> +static int unci_check_if_running(struct net_device *dev)
> +{
> + return 0;
> +}
Don't define stub for this. If you leave ethtool_get_link as NULL
it will do the right thing.
On Tue, 4 Jul 2017 17:13:28 +0100
Ferruh Yigit wrote:
> +int unci_nl_exec(u32 cmd, struct net_device *dev, void *in_data,
> + size_t in_data_len, void *out_data, size_t out_data_len)
> +{
> + struct unci_dev *unci = netdev_priv(dev);
> + int err = -EINVAL;
> + int ret;
>
On Tue, 4 Jul 2017 17:13:30 +0100
Ferruh Yigit wrote:
> +static void unci_get_drvinfo(struct net_device *dev,
> + struct ethtool_drvinfo *info)
> +{
> + int ret;
> +
> + ret = unci_nl_exec(info->cmd, dev, NULL, 0,
> + info, sizeof(struct ethtool_drvinfo));
From: Tomasz Kulasek
This patch removes restrictions in bonded device library which prevent a
bonded device to be added to another bonded device with the limitation
that 802.3ad mode is not supported if one or more slaves is also a
bonded device,
Signed-off-by: Declan Doherty
---
v4:
- rebase
From: Tomasz Kulasek
This patch fixes the management of link properties in the bonded device.
In all mode except mode 4 a bonded device link will default to reporting
the link as full duplex and auto-neg. The link speed for a bond port is
calculated on it's active slaves and the particular mode
On 7/5/2017 3:21 AM, Qi Zhang wrote:
> The patchset is based on
> http://dpdk.org/dev/patchwork/patch/25292/
>
> v4:
> - fix missing rename from roughly to fuzzy
>
> v3:
> - will check fuzzy match's last and mask value.
>
> v2:
> - add comment to explain macro replacement
> - fix wrong flow
To make vectorized burst routines enabled, it is required to run on x86_64
architecture which can support at least SSE4.1. If all the conditions are
met, the vectorized burst functions are enabled automatically. The decision
is made individually on RX and TX. There's no PMD option to make a
selecti
The callbacks are global to a device but the seletion is made every queue
configuration, which is redundant.
Signed-off-by: Yongseok Koh
---
drivers/net/mlx5/mlx5_ethdev.c | 23 +++
drivers/net/mlx5/mlx5_rxq.c | 2 --
drivers/net/mlx5/mlx5_trigger.c | 3 +++
drivers/ne
This is to introduce more efficient Rx/Tx burst functions using SIMD
instructions. Currently it is only supported by 64bit x86 having SSE4.1.
>From functional perspective, Rx burst function is equivalent to the
existing mlx5_rx_burst() except for scatter support, which will be added
soon. Tx burst
When searching LKEY, if search key is mempool pointer, the 2nd cacheline
has to be accessed and it even requires to check whether a buffer is
indirect per every search. Instead, using address for search key can reduce
cycles taken. And caching the last hit entry is beneficial as well.
Signed-off-b
When processing Tx completion, it is more efficient to free buffers in bulk
using rte_mempool_put_bulk() if buffers are from a same mempool.
Signed-off-by: Yongseok Koh
Acked-by: Nelio Laranjeiro
---
drivers/net/mlx5/mlx5_rxtx.c | 33 -
1 file changed, 24 inserti
For Tx SW ring (txq->elts[]), indexes are kept and used in
txq->elts_head/tail. Because of this, one entry must always be left unused
and it also makes code complex. Changed to store counters instead of
indexes in order to make the code simpler and to reduce a few calculations.
Signed-off-by: Yong
On Wed, 5 Jul 2017 01:55:23 +0200
Gaetan Rivet wrote:
> case RTE_DEVTYPE_WHITELISTED_PCI:
> case RTE_DEVTYPE_BLACKLISTED_PCI:
> /* try to parse pci identifier */
> - if (eal_parse_pci_BDF(buf, &devargs->pci.addr) != 0 &&
> - eal_parse_pci_Dom
In order to support VMBUS, the ethernet device name needs to be increased.
What would be good timing; vmbus support will be in 17.11.
Either break ABI now, and vmbus won't have to do it later.
Or wait and make it part of of VMBUS patches?
13/06/2017 05:07, Qi Zhang:
> Add new meta pattern item RTE_FLOW_TYPE_ITEM_FUZZY in flow API.
>
> This is for device that support fuzzy match option.
> Usually a fuzzy match is fast but the cost is accuracy.
> i.e. Signature Match only match pattern's hash value, but it is
> possible that two diff
27/04/2017 13:00, Jerin Jacob:
> -Original Message-
> > Date: Mon, 24 Apr 2017 14:34:59 +0200
> > From: Thomas Monjalon
> > To: Jerin Jacob
> > Cc: dev@dpdk.org, bruce.richard...@intel.com, harry.van.haa...@intel.com,
> > hemant.agra...@nxp.com, gage.e...@intel.com, nipun.gu...@nxp.com,
On Wed, Jul 05, 2017 at 10:21:26AM +0200, Nélio Laranjeiro wrote:
> On Tue, Jul 04, 2017 at 05:38:44PM -0700, Yongseok Koh wrote:
> > On Tue, Jul 04, 2017 at 10:58:52AM +0200, Nélio Laranjeiro wrote:
> > > Yongseok, some comments in this huge and great work,
> > >
> > > On Fri, Jun 30, 2017 at 12:
On 7/5/2017 5:41 PM, Declan Doherty wrote:
> On 05/07/2017 4:27 PM, Daniel Mrzyglod wrote:
>> ring network driver is limited to 32 characters
>> the name of device was extened to net_ring_unit_test_mode4_slave_0 which is
>> 33
>>
>> Fixes: 5e41ab250dfa ("app/test: unit tests for bonding mode 4")
>
07/06/2017 12:34, Pavan Nikhilesh:
> From: Pavan Nikhilesh Bhagavatula
>
> If ethdev has already started it should return -EBUSY instead of 0 when
> rte_eth_dev_start is called.
It is an API change and must be announced in a deprecation notice.
> lib/librte_ether/rte_ethdev.c | 2 +-
> 1 file
On 7/5/2017 6:09 PM, Declan Doherty wrote:
> From: Tomasz Kulasek
>
> This patch fixes the management of link properties in the bonded device.
>
> In all mode except mode 4 a bonded device link will default to reporting
> the link as full duplex and auto-neg. The link speed for a bond port is
>
From: Tomasz Kulasek
This patch fixes the management of link properties in the bonded device.
In all mode except mode 4 a bonded device link will default to reporting
the link as full duplex and auto-neg. The link speed for a bond port is
calculated on it's active slaves and the particular mode
In some environments, the PCI domain can be larger than 16 bits.
For example, a PCI device passed through in Azure gets a synthetic domain
id which is internally generated based on GUID. The PCI standard does
not restrict domain to be 16 bits.
This change breaks ABI for API's that expose PCI addr
The PCI domain in Azure maybe 32 bits. When device is passed through
the domain is synthesize from the internal GUID. The code that
parses for PCI address in Mellanox drivers needs to use wider
scanf format.
Signed-off-by: Stephen Hemminger
---
v3 - add MLX4
drivers/net/mlx4/mlx4.c| 2 +
Changes to PCI and EAL to handle full width of PCI domain
value. It turns out that PCI domain is only 16 bits in ACPI,
but some environments use the extra bits in PCI standard to allow
encoding of PCI buses that are outside of ACPI.
This version fixes both Mellanox drivers.
Stephen Hemminger (3):
The function strtoul returns unsigned long and can be directly
assigned to a smaller type. Removing the casts allows easier
expansion of PCI domain.
Signed-off-by: Stephen Hemminger
---
lib/librte_eal/linuxapp/eal/eal_pci.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --gi
On 05/07/2017 4:27 PM, Daniel Mrzyglod wrote:
ring network driver is limited to 32 characters
the name of device was extened to net_ring_unit_test_mode4_slave_0 which is 33
Fixes: 5e41ab250dfa ("app/test: unit tests for bonding mode 4")
Signed-off-by: Daniel Mrzyglod
---
test/test/test_link_b
From: Tomasz Kulasek
This patch removes restrictions in bonded device library which prevent a
bonded device to be added to another bonded device with the limitation
that 802.3ad mode is not supported if one or more slaves is also a
bonded device,
Signed-off-by: Declan Doherty
---
v4:
- rebase
From: Tomasz Kulasek
This patch fixes the management of link properties in the bonded device.
In all mode except mode 4 a bonded device link will default to reporting
the link as full duplex and auto-neg. The link speed for a bond port is
calculated on it's active slaves and the particular mode
On 7/5/2017 10:56 AM, Pablo de Lara wrote:
Currently, the cryptodev library requires the session mempools to be created
inside each device.
This can be a waste of memory, as several devices with the same driver could be
using the same session.
This patchset makes changes in the library to allo
while registering driver to dpaa2, hard coded string is used.
It is now updated as per the latest changes in string name.
Signed-off-by: Akhil Goyal
---
drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/crypto/dpaa2_sec/dpaa2_
Multi-core scheduling mode is a mode where scheduler distributes
crypto operations in a round-robin base, between several core
assigned as workers.
Signed-off-by: Kirill Rybalchenko
---
Changes in v3:
- Rebased against dpdk-next-crypto
- Fixed op status check in the reordering ring dequeuing
- Mo
On 6/13/2017 4:07 AM, Qi Zhang wrote:
> Add new meta pattern item RTE_FLOW_TYPE_ITEM_FUZZY in flow API.
>
> This is for device that support fuzzy match option.
> Usually a fuzzy match is fast but the cost is accuracy.
> i.e. Signature Match only match pattern's hash value, but it is
> possible tha
> -Original Message-
> From: Yigit, Ferruh
> Sent: Wednesday, July 5, 2017 2:20 PM
> To: Mcnamara, John
> Cc: dev@dpdk.org; Yigit, Ferruh ; Olivier Matz
>
> Subject: [PATCH v3] doc: document NIC features
>
> Document NIC features, add more information about them and add more
> implemen
On 7/5/2017 9:14 AM, Vasily Philipov wrote:
> Providing this parameter requests flow API isolated mode on all ports at
> initialization time. It ensures all traffic is received through the
> configured flow rules only (see flow command).
>
> Ports that do not support this mode are automatically di
-Original Message-
> Date: Wed, 5 Jul 2017 11:14:01 +0200
> From: Maxime Coquelin
> To: Santosh Shukla ,
> tho...@monjalon.net, bruce.richard...@intel.com, dev@dpdk.org
> CC: jerin.ja...@caviumnetworks.com, hemant.agra...@nxp.com,
> shreyansh.j...@nxp.com, gaetan.ri...@6wind.com
> Subjec
ring network driver is limited to 32 characters
the name of device was extened to net_ring_unit_test_mode4_slave_0 which is 33
Fixes: 5e41ab250dfa ("app/test: unit tests for bonding mode 4")
Signed-off-by: Daniel Mrzyglod
---
test/test/test_link_bonding_mode4.c | 8
1 file changed, 4 i
On 7/5/2017 4:18 PM, Ferruh Yigit wrote:
> On 7/5/2017 9:14 AM, Vasily Philipov wrote:
>> Providing this parameter requests flow API isolated mode on all ports at
>> initialization time. It ensures all traffic is received through the
>> configured flow rules only (see flow command).
>>
>> Ports tha
On 05/07/2017 6:26 AM, Pablo de Lara wrote:
Currently, the cryptodev library requires the session mempools to be created
inside each device.
This can be a waste of memory, as several devices with the same driver could be
using the same session.
This patchset makes changes in the library to all
On 7/5/2017 9:14 AM, Vasily Philipov wrote:
> Providing this parameter requests flow API isolated mode on all ports at
> initialization time. It ensures all traffic is received through the
> configured flow rules only (see flow command).
>
> Ports that do not support this mode are automatically di
From: Tomasz Kulasek
This patch removes restrictions in bonded device library which prevent a
bonded device to be added to another bonded device with the limitation
that 802.3ad mode is not supported if one or more slaves is also a
bonded device,
---
v3: reworked link management changes and spli
From: Tomasz Kulasek
This patch fixes the management of link properties in the bonded device.
In all mode except mode 4 a bonded device link will default to reporting
the link as full duplex and auto-neg. The link speed for a bond port is
calculated on it's active slaves and the particular mode
On 7/5/2017 3:55 PM, Ferruh Yigit wrote:
> On 6/29/2017 8:35 PM, Charles (Chas) Williams wrote:
>> Refactor vmxnet3_post_rx_bufs() to call vmxnet3_renew_desc()
>> to update the newly allocated mbufs. While here, relocate the
>> relevant comments to vmxnet3_renew_desc().
>>
>> Signed-off-by: Chas W
On 6/29/2017 8:35 PM, Charles (Chas) Williams wrote:
> Refactor vmxnet3_post_rx_bufs() to call vmxnet3_renew_desc()
> to update the newly allocated mbufs. While here, relocate the
> relevant comments to vmxnet3_renew_desc().
>
> Signed-off-by: Chas Williams
Reviewed-by: Ferruh Yigit
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of mstolarchuk
> Sent: Friday, May 26, 2017 1:31 PM
> To: bruce.richard...@intel.co
> Cc: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH] __rte_hash_add_key_with_hash not releasing
> multiwriter_lock in failure paths
>
>
On Wed, Jul 05, 2017 at 11:14:11AM +0300, Vasily Philipov wrote:
> This commit adds support for the flow API RSS action with the following
> limitations:
>
> - Only supported when isolated mode is enabled.
> - The number of queues specified by the action (rte_flow_action_rss.num)
>must be a
On Wed, Jul 05, 2017 at 11:14:09AM +0300, Vasily Philipov wrote:
> The user must request isolated mode before device configuration.
>
> Signed-off-by: Vasily Philipov
Acked-by: Adrien Mazarguil
--
Adrien Mazarguil
6WIND
On Wed, Jul 05, 2017 at 11:14:10AM +0300, Vasily Philipov wrote:
> A special "parent" queue must be allocated in addition to a group of
> standard Rx queues for RSS to work. This is done automatically outside of
> isolated mode by the PMD when applications request several Rx queues.
>
> Since each
On Wed, Jul 05, 2017 at 11:14:08AM +0300, Vasily Philipov wrote:
> Providing this parameter requests flow API isolated mode on all ports at
> initialization time. It ensures all traffic is received through the
> configured flow rules only (see flow command).
>
> Ports that do not support this mode
On 7/5/2017 3:35 PM, Adrien Mazarguil wrote:
> On Wed, Jul 05, 2017 at 02:49:22PM +0300, Vasily Philipov wrote:
>> In debug mode, all mbuf ol_flags are temporarily enabled while sitting in
>> the Rx queue to detect otherwise silent data corruption, however some of
>> them are special (indirect and
On 7/5/2017 7:48 AM, Qi Zhang wrote:
> Internal Rx packet bytes is over counted, CRC size is not excluded,
> that cause incorrect Rx bytes or even negative value.
>
> Fixes: 0bcdc44510ef ("net/i40e: exclude internal packet's byte count")
>
> Signed-off-by: Qi Zhang
Applied to dpdk-next-net/mas
On Wed, Jul 05, 2017 at 02:49:22PM +0300, Vasily Philipov wrote:
> In debug mode, all mbuf ol_flags are temporarily enabled while sitting in
> the Rx queue to detect otherwise silent data corruption, however some of
> them are special (indirect and control) and must be cleared before
> returning mb
On Wednesday 05 July 2017 07:15 PM, Gaëtan Rivet wrote:
Hi Shreyansh,
On Wed, Jul 05, 2017 at 07:05:40PM +0530, Shreyansh Jain wrote:
Hello Gaetan,
Some comments inline:
On Wednesday 05 July 2017 05:25 AM, Gaetan Rivet wrote:
Find which bus should be able to parse this device name into an in
>From documentation it is very unclear how VMDq configuration can be
tweaked, and online search offer very poor results.
This patch will ultimately spawn an online documentation page
for the rte_eth_vmdq_rx_conf struct which will eventually add a bit of
documentation about the rx_mode tag and how
On 7/5/2017 3:33 AM, Beilei Xing wrote:
> Currently, setting MTU will fail if MTU requires the support
> of scattered packets before scatter is enabled.
> This patch allows setting this special MTU when device is
> stopped, cause scatter will be re-configured during starting
> port.
>
> Signed-off
Internal Rx packet bytes is over counted, CRC size is not excluded,
that cause incorrect Rx bytes or even negative value.
Fixes: 0bcdc44510ef ("net/i40e: exclude internal packet's byte count")
Cc: sta...@dpdk.org
Signed-off-by: Qi Zhang
---
v2:
- fix some comments.
drivers/net/i40e/i40e_ethdev
Hi Hemant,
On Wednesday 05 July 2017 06:47 PM, Hemant Agrawal wrote:
> On 6/8/2017 4:35 PM, Santosh Shukla wrote:
>> - Moving late bus scanning to up..just after eal_parsing.
>> - Detect iova mapping mode based on user provided eal option
>> (rte_eal_iova_mode) and result of rte_bus_scan_iommu_
Hi Shreyansh,
On Wed, Jul 05, 2017 at 07:05:40PM +0530, Shreyansh Jain wrote:
> Hello Gaetan,
>
> Some comments inline:
>
> On Wednesday 05 July 2017 05:25 AM, Gaetan Rivet wrote:
> >Find which bus should be able to parse this device name into an internal
> >device representation.
> >
> >Signed-
Modified cryptodev library section in Programmer's Guide,
with the recent changes in the crypto sessions.
Signed-off-by: Slawomir Mrozowicz
Signed-off-by: Pablo de Lara
Acked-by: John McNamara
---
doc/guides/prog_guide/cryptodev_lib.rst | 61 ++--
doc/guides/prog_guide/img/cryptodev_
Since now the private session data is initialized after
the session pool is created, there is no need to keep
this PMD function.
Signed-off-by: Pablo de Lara
---
drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c | 8
lib/librte_cryptodev/rte_cryptodev_pmd.h| 16
2 files
The session mempool pointer is needed in each queue pair,
if session-less operations are being handled.
Therefore, the API is changed to accept this parameter,
as the session mempool is created outside the
device configuration function, similar to what ethdev
does with the rx queues.
Signed-off-by
From: Slawomir Mrozowicz
Change crypto device's session management to make it
device independent and simplify architecture when session
is intended to be used on more than one device.
Sessions private data is agnostic to underlying device
by adding an indirection in the sessions private data
usi
From: Slawomir Mrozowicz
Since crypto session will not be attached to a specific
device or driver, the field driver_id is not required
anymore (only used to check that a session was being
handled by the right device).
Signed-off-by: Slawomir Mrozowicz
Signed-off-by: Pablo de Lara
---
doc/guid
From: Slawomir Mrozowicz
Mempool pointer can be obtained from the object itself,
which means that it is not required to actually store the pointer
in the session.
Signed-off-by: Slawomir Mrozowicz
Signed-off-by: Pablo de Lara
---
doc/guides/rel_notes/release_17_08.rst | 1 +
lib/librte_crypto
From: Slawomir Mrozowicz
Device id is necessary in the crypto session,
as it was only used for the functions that attach/detach
a session to a queue pair.
Since the session is not going to be attached to a device
anymore, this is field is no longer necessary.
Signed-off-by: Slawomir Mrozowicz
From: Slawomir Mrozowicz
Device id is going to be removed from session,
as the session will be device independent.
Therefore, the functions that attach/dettach a session
to a queue pair need to be updated, to accept the device id
as a parameter, apart from the queue pair id and the session.
Sign
From: Slawomir Mrozowicz
Provide a function to get the private session size
of any crypto device (specifically, to its crypto driver).
This will be useful once the session mempool is created
outside the library.
Signed-off-by: Slawomir Mrozowicz
Signed-off-by: Pablo de Lara
---
lib/librte_cr
From: Slawomir Mrozowicz
Instead of creating the session mempool while configuring
the crypto device, apps will create the mempool themselves.
This way, it gives flexibility to the user to have a single
mempool for all devices (as long as the objects are big
enough to contain the biggest private
Prior to removing the session pool creation from cryptodev
configure function, session init function needs to be
separated from the pool creation.
Signed-off-by: Pablo de Lara
---
lib/librte_cryptodev/rte_cryptodev.c | 13 +
1 file changed, 5 insertions(+), 8 deletions(-)
diff --git
Cryptodev session structure was a duplication of the
cryptodev symmetric structure.
It was used by some PMDs that should use the symmetric
structure instead.
Since this structure was internal, there is no deprecation
notice required.
Signed-off-by: Pablo de Lara
---
drivers/crypto/kasumi/rte_ka
Currently, the cryptodev library requires the session mempools to be created
inside each device.
This can be a waste of memory, as several devices with the same driver could be
using the same session.
This patchset makes changes in the library to allow session to contain multiple
driver private
Hello Gaetan,
Some comments inline:
On Wednesday 05 July 2017 05:25 AM, Gaetan Rivet wrote:
Find which bus should be able to parse this device name into an internal
device representation.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_bus.c | 21 +
lib
Document NIC features, add more information about them and add more
implementation related support.
Signed-off-by: Ferruh Yigit
Signed-off-by: John McNamara
---
Cc: Olivier Matz
v3:
* received updates from John, Thanks!
v2:
* Add more details, mbuf and API fields
* Formatting added
TODO:
- N
On 6/8/2017 4:35 PM, Santosh Shukla wrote:
- Moving late bus scanning to up..just after eal_parsing.
- Detect iova mapping mode based on user provided eal option
(rte_eal_iova_mode) and result of rte_bus_scan_iommu_class.
Signed-off-by: Santosh Shukla
Signed-off-by: Jerin Jacob
---
lib/libr
23/06/2017 20:20, Daniel Verkamp:
> Some PCI vendor and device IDs have leading zeros.
>
> Signed-off-by: Daniel Verkamp
> ---
> --- a/lib/librte_eal/common/include/rte_pci.h
> +++ b/lib/librte_eal/common/include/rte_pci.h
> @@ -68,6 +68,9 @@ const char *pci_get_sysfs_path(void);
> /** Short for
On Wednesday 05 July 2017 05:25 AM, Gaetan Rivet wrote:
Signed-off-by: Gaetan Rivet
---
Acked-by: Shreyansh Jain
26/06/2017 08:49, Jianfeng Tan:
> When primary process is booted with --file-prefix option, the API,
> rte_eal_primary_proc_alive(), uses a wrong config file path to
> check if primary process is alive.
>
> Fix it by calling helper function to get config file path.
>
> Fixes: dd3e00138d74 ("eal:
> -Original Message-
> From: Yigit, Ferruh
> Sent: Tuesday, June 27, 2017 12:36 PM
> > For 2, does the "use existing methods" include calling xstats? If so,
> > then we can just drop the requirement for 2 to have any new functions
> > implemented in the driver. Instead, have the informati
1 - 100 of 144 matches
Mail list logo
