Re: [PATCH v1] maintainers: add explicit maintainer for some bbdev PMDs
On Wed, Mar 9, 2022 at 12:23 AM Thomas Monjalon wrote: > > 03/03/2022 19:51, Nicolas Chautru: > > These were implicit from DPDK script but adding > > separate reference to make it explicit > > > > Signed-off-by: Nicolas Chautru > > --- > > MAINTAINERS | 4 > > 1 file changed, 4 insertions(+) > > > > diff --git a/MAINTAINERS b/MAINTAINERS > > index 7577834..004ec91 100644 > > --- a/MAINTAINERS > > +++ b/MAINTAINERS > > @@ -421,6 +421,10 @@ T: git://dpdk.org/next/dpdk-next-crypto > > F: lib/bbdev/ > > F: doc/guides/prog_guide/bbdev.rst > > F: drivers/baseband/ > > I think you can remove the above catch-all. > > > +F: drivers/baseband/turbo_sw/ > > +F: drivers/baseband/fpga_lte_fec/ > > +F: drivers/baseband/fpga_5gnr_fec/ > > +F: drivers/baseband/acc100/ > > I would prefer a separate section for Intel baseband drivers. We now have a baseband drivers section with a NXP driver. It is more clear to have Intel drivers in this section like we do for other classes of devices: one section for API, one section for drivers. Btw, the baseband drivers section is missing a global T: entry too. It means that updates on nxp (atm) and any (in the future) bbdev driver ends up in the main repo scope instead of next-crypto. > > > F: doc/guides/bbdevs/ This too is a catch all. For the API section, I'd rather see: -F: drivers/baseband/ -F: doc/guides/bbdevs/ +F: doc/guides/bbdevs/features/default.ini > > F: app/test-bbdev/ > > F: doc/guides/tools/testbbdev.rst -- David Marchand
RE: [PATCH] net/mlx5: set correct CPU socket ID for mlx5_rxq_ctrl
Hi, Thinh
Thank you for the patch, the code looks OK to me, but commit message is not
compliant:
- it should contain "fix" keyword in the title, like this:
"net/mlx5: fix CPU socket ID for Rx queue creation"
- could you, please, make problem description less personal and less wordy?
"The default CPU socket ID was used while creating the Rx queue and this caused
creation failure in case if hardware was not resided on the default socket.
The patch sets the correct CPU socket ID for the mlx5_rxq_ctrl before
calling the mlx5_rxq_create_devx_rq_resources() which eventually calls
mlx5_devx_rq_create() with correct CPU socket ID."
- please add tags:
Cc: sta...@dpdk.org
Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
With best regards,
Slava
> -Original Message-
> From: Raslan Darawsheh
> Sent: Tuesday, March 8, 2022 14:25
> To: Dmitry Kozlyuk
> Cc: Thinh Tran ; dev@dpdk.org;
> d...@linux.vnet.ibm.com
> Subject: RE: [PATCH] net/mlx5: set correct CPU socket ID for mlx5_rxq_ctrl
>
>
> > -Original Message-
> > From: Dmitry Kozlyuk
> > Sent: Tuesday, March 8, 2022 2:23 PM
> > To: Raslan Darawsheh
> > Cc: Thinh Tran ; dev@dpdk.org;
> > d...@linux.vnet.ibm.com
> > Subject: Re: [PATCH] net/mlx5: set correct CPU socket ID for
> > mlx5_rxq_ctrl
> >
> > Hi Raslan,
> >
> > > Missing:
> > > Fixes tag:
> > >
> > > Fixes: 5ceb3a02b000 ("net/mlx5: move Rx queue DevX resource")
> > > Cc: xuemi...@nvidia.com
> >
> > I believe the bug originates from my earlier commit, not Xueming's one:
> >
> > Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
> Yes I think you are correct, my mistake 😊
>
> Kindest regards,
> Raslan Darawsheh
[PATCH v2] doc: fix telemetry example for cryptodev
Blank line added to the final telemetry example for the
cryptography device library as to fix the example
rendering.
Fixes: 1c559ee8469 ("cryptodev: add telemetry endpoint for capabilities")
Cc: sta...@dpdk.org
Signed-off-by: Sean Morrissey
---
doc/guides/prog_guide/cryptodev_lib.rst | 1 +
1 file changed, 1 insertion(+)
diff --git a/doc/guides/prog_guide/cryptodev_lib.rst
b/doc/guides/prog_guide/cryptodev_lib.rst
index ba4c592b84..edd11059bc 100644
--- a/doc/guides/prog_guide/cryptodev_lib.rst
+++ b/doc/guides/prog_guide/cryptodev_lib.rst
@@ -1211,6 +1211,7 @@ are shown below.
"enqueue_err_count": 0, "dequeue_err_count": 0}}
#. Get the capabilities of a particular Crypto device::
+
--> /cryptodev/caps,0
{"/cryptodev/caps": {"crypto_caps": [], "crypto_caps_n": }}
--
2.25.1
[PATCH] net/mlx5: fix push VLAN action validation
Flow domain and direction was validated when OF_PUSH_VLAN action
appears in flow actions. Flow was rejected whenever this action:
- was used in NIC domain, in ingress direction;
- was used in FDB domain, in ingress direction, on ConnectX-5.
This validation logic rejected a valid case when the OF_PUSH_VLAN
action was used when directing traffic to the hairpin queue,
configured in TX implicit mode.
This patch moves code responsible for OF_PUSH_VLAN validation of
domain and direction from flow_dv_validate_push_vlan() to
flow_dv_validate(). Domain and direction are now validated when either
non-hairpin queue is used or hairpin queue is configured in Tx explicit
mode.
Fixes: 96f85ec489db ("net/mlx5: check VLAN push/pop support")
Cc: dongz...@nvidia.com
Cc: sta...@dpdk.org
Signed-off-by: Dariusz Sosnowski
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5_flow_dv.c | 46 -
1 file changed, 22 insertions(+), 24 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 29751e7eda..1746ef37bd 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -2873,8 +2873,6 @@ flow_dv_validate_action_push_vlan(struct rte_eth_dev *dev,
{
const struct rte_flow_action_of_push_vlan *push_vlan = action->conf;
const struct mlx5_priv *priv = dev->data->dev_private;
- struct mlx5_dev_ctx_shared *sh = priv->sh;
- bool direction_error = false;
if (push_vlan->ethertype != RTE_BE16(RTE_ETHER_TYPE_VLAN) &&
push_vlan->ethertype != RTE_BE16(RTE_ETHER_TYPE_QINQ))
@@ -2886,22 +2884,6 @@ flow_dv_validate_action_push_vlan(struct rte_eth_dev
*dev,
RTE_FLOW_ERROR_TYPE_ACTION, action,
"wrong action order, port_id should "
"be after push VLAN");
- /* Push VLAN is not supported in ingress except for CX6 FDB mode. */
- if (attr->transfer) {
- bool fdb_tx = priv->representor_id != UINT16_MAX;
- bool is_cx5 = sh->steering_format_version ==
- MLX5_STEERING_LOGIC_FORMAT_CONNECTX_5;
-
- if (!fdb_tx && is_cx5)
- direction_error = true;
- } else if (attr->ingress) {
- direction_error = true;
- }
- if (direction_error)
- return rte_flow_error_set(error, ENOTSUP,
- RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
- NULL,
- "push vlan action not supported for
ingress");
if (!attr->transfer && priv->representor)
return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_UNSPECIFIED, NULL,
@@ -7996,6 +7978,28 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL, "encap and decap "
"combination aren't
supported");
+ /* Push VLAN is not supported in ingress except for NICs newer
than CX5. */
+ if (action_flags & MLX5_FLOW_ACTION_OF_PUSH_VLAN) {
+ struct mlx5_dev_ctx_shared *sh = priv->sh;
+ bool direction_error = false;
+
+ if (attr->transfer) {
+ bool fdb_tx = priv->representor_id !=
UINT16_MAX;
+ bool is_cx5 = sh->steering_format_version ==
+ MLX5_STEERING_LOGIC_FORMAT_CONNECTX_5;
+
+ if (!fdb_tx && is_cx5)
+ direction_error = true;
+ } else if (attr->ingress) {
+ direction_error = true;
+ }
+ if (direction_error)
+ return rte_flow_error_set(error, ENOTSUP,
+
RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+ NULL,
+ "push VLAN action not
supported "
+ "for ingress");
+ }
if (!attr->transfer && attr->ingress) {
if (action_flags & MLX5_FLOW_ACTION_ENCAP)
return rte_flow_error_set
@@ -8003,12 +8007,6 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
RTE_FLOW_ERROR_TYPE_ACTION,
NULL, "encap is not supported"
" for ingre
RE: [PATCH 2/6] net/hns3: fix inconsistent enabled RSS behavior
Hi Lihusiong,
> -Original Message-
> From: lihuisong (C)
> Sent: Wednesday, March 9, 2022 10:03 AM
> Subject: Re: [PATCH 2/6] net/hns3: fix inconsistent enabled RSS behavior
>
>
> 在 2022/3/3 10:47, lihuisong (C) 写道:
> > Hi, Ori,
> >
> > 在 2022/3/2 22:07, Ori Kam 写道:
> >> Hi Lihuisong,
> >>
> >>> -Original Message-
> >>> From: lihuisong (C)
> >>> Sent: Wednesday, March 2, 2022 4:10 AM
> >>> Subject: Re: [PATCH 2/6] net/hns3: fix inconsistent enabled RSS
> >>> behavior
> >>>
> >>>
> >>> 在 2022/3/1 0:42, Ferruh Yigit 写道:
> On 2/28/2022 3:21 AM, Min Hu (Connor) wrote:
> > From: Huisong Li
> >
> > RSS will not be enabled if the RTE_ETH_MQ_RX_RSS_FLAG isn't be set in
> > dev_configure phase. However, if this flag isn't set, RSS can be
> > enabled
> > through the ethdev ops and rte_flow API. This behavior is contrary to
> > each
> > other.
> >
> > Fixes: c37ca66f2b27 ("net/hns3: support RSS")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Huisong Li
>
> Hi Huisong, Connor,
>
> Let's get a little more feedback for this patch, cc'ed more people.
>
>
> To enable RSS, multi queue mode should be set to
> 'RTE_ETH_MQ_RX_RSS_FLAG'.
>
> But I wonder if it is required to configure RSS via flow API,
> >>> I do not know the original purpose of adding the RSS configuration in
> >>> flow API.
> >>>
> >> The purpose is simple, this allow to create RSS per rule and not a
> >> global one.
> >> For example create RSS that sends TCP to some queues while othe RSS
> >> will send
> >> UDP traffic to different queues.
> > I'm a little confused now. The "per rule" also seems to be a global
> > configuration.
> > Example:
> > - start PMD with 0,1,2,3
> > - create TCP packets to 2,3 queues. At this moment, only 2,3 queues
> > can be received for other types of packets.
> > Because this rule is implemented by modifying the entry of the
> > redirection table which is global for this device.
> Hi, Ori and Stephen.
> Can you help me clear up the confusion above? If some NICs behave like
> this, what should we do about it?
I'm not sure I understand the issue, maybe it is releated to some HW/PMD
limitation.
In your example non TCP traffic will be routed to one of the 4 queues (0,1,2,3),
While TCP traffic will only be routed to queues 2,3.
Now I can add new rule that matches on UDP packet and RSS to queue 0 and 3 in
this case:
TCP packets will be routed to queues 0,3.
UDP packets will be routed to queues 2,3.
All the rest of the traffic will be routed to queues 0,1,2,3
And just to be clear if now I add a rule to match all packets in higher
priority,
with RSS to queues 1,2. Then all traffic will be routed to queues 1,2.
At least this is what is expected, from API point of view.
Best,
Ori
> >>
> >>> However, as far as I know, the hash algorithm can be configured via
> >>> this
> >>> API,
> >>>
> >>> but not via ethdev ops API.
> >>>
> and if other PMDs check this configuration for flow API?
> >>> Some PMDs already have similar check in RSS releated ops or rte_flow
> >>> API.
> >>>
> >>> For example, hinic, axbge, bnxt, cnxk, otx2, and ena.
> >>>
> .
> >> From my view point those are two different settings.
> >> The RTE_ETH_MQ_RX_RSS_FLAG is global per port while
> >> rte_flow is per rule.
> >>
> >> I think, that if a PMD needs this flag, in order to enable it also
> >> for rte_flow then
> >> it should be documented in the release note of the PMD.
> >> It is a valid use case that the default traffic will not have RSS and
> >> only rules created by
> >> rte_flow will have the RSS, for matching traffc.
> >>
> >> Best,
> >> Ori
RE: [PATCH v2] app/testpmd : fix testpmd quit error
> -Original Message-
> From: Yigit, Ferruh
> Sent: 2022年3月5日 0:19
> To: Wu, WenxuanX ; Yang, Qiming
> ; Zhang, Qi Z ; Li, Xiaoyun
> ; Singh, Aman Deep ;
> Zhang, Yuying
> Cc: dev@dpdk.org
> Subject: Re: [PATCH v2] app/testpmd : fix testpmd quit error
>
> On 3/4/2022 2:37 AM, wenxuanx...@intel.com wrote:
> > From: wenxuan wu
> >
> > When testpmd use func get_eth_dev_info() while related resource had
> > been released.
> >
>
> Is 'eth_dev_info_get_print_err()' fails at this stage?
> What resource is released, the 'slave_port' itself?
Yeah, 1PF ,2VF_repr.
Close port pf ,ok.
Close port vf , error.
In port_close() func,
Is_bonding_slave() call eth_dev_info_get_print_err() to confirm whether it
is a slave or not , but when port is a vf port ,and pf had been released, this
eth_dev_info_get_print_err(vf_id) would read a freed memory ,result in this bug.
>
> And there may be another logic wrong, it shouldn't try to detect if a released
> port is bonding port or not.
>
> > Change the logic of func port_is_bonding_slave, this func
> > eth_dev_info_get_print_err while pf is released would result in this
> > error. Use ports instead to avoid this bug.
> >
>
> This relies to application level stored value to decide about port, not sure
> if
> this is reliable.
>
> > Fixes: 0a0821bcf312 ("app/testpmd: remove most uses of internal ethdev
> > array")
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: wenxuan wu
> > ---
> > app/test-pmd/testpmd.c | 12 +---
> > 1 file changed, 1 insertion(+), 11 deletions(-)
> >
> > diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c index
> > e1da961311..37038c9183 100644
> > --- a/app/test-pmd/testpmd.c
> > +++ b/app/test-pmd/testpmd.c
> > @@ -3824,19 +3824,9 @@ void clear_port_slave_flag(portid_t slave_pid)
> > uint8_t port_is_bonding_slave(portid_t slave_pid)
> > {
> > struct rte_port *port;
> > - struct rte_eth_dev_info dev_info;
> > - int ret;
> >
> > port = &ports[slave_pid];
> > - ret = eth_dev_info_get_print_err(slave_pid, &dev_info);
> > - if (ret != 0) {
> > - TESTPMD_LOG(ERR,
> > - "Failed to get device info for port id %d,"
> > - "cannot determine if the port is a bonded slave",
> > - slave_pid);
> > - return 0;
> > - }
> > - if ((*dev_info.dev_flags & RTE_ETH_DEV_BONDED_SLAVE) || (port-
> >slave_flag == 1))
> > + if ((*port->dev_info.dev_flags & RTE_ETH_DEV_BONDED_SLAVE) ||
> > +(port->slave_flag == 1))
> > return 1;
> > return 0;
> > }
[PATCH] net/mlx5: fix the sample flow failure on the trusted device
The flow with sample action will be split into two sub flows,
and a tag action was added implicitly in the sample prefix sub flow,
the reserved metadata regC index was used for this tag action.
The reserved metadata regC was shared with metering action,
for Connect-5 trusted device(VF/SF), the reserved metadata regC was
invalid since PF only supported the legacy metering.
This patch adds the checking for the tag index and back to use the
application tag if a failure happened.
Fixes: a9b6ea45bed6 ("net/mlx5: fix tag ID conflict with sample action")
Cc: sta...@dpdk.org
Signed-off-by: Jiawei Wang
Acked-by: Viacheslav Ovsiienko
---
doc/guides/nics/mlx5.rst| 2 ++
drivers/net/mlx5/mlx5_flow.c| 6 ++
drivers/net/mlx5/mlx5_flow_dv.c | 10 ++
3 files changed, 18 insertions(+)
diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
index a734d10d3f..ca7ebbe9cd 100644
--- a/doc/guides/nics/mlx5.rst
+++ b/doc/guides/nics/mlx5.rst
@@ -400,6 +400,8 @@ Limitations
sample actions list.
- For E-Switch mirroring flow, supports ``RAW ENCAP``, ``Port ID``,
``VXLAN ENCAP``, ``NVGRE ENCAP`` in the sample actions list.
+ - For ConnectX-5 trusted device, the application metadata with SET_TAG index 0
+is not supported before ``RTE_FLOW_ACTION_TYPE_SAMPLE`` action.
- Modify Field flow:
diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
index ffcaef0baa..e2bc6ce8ad 100644
--- a/drivers/net/mlx5/mlx5_flow.c
+++ b/drivers/net/mlx5/mlx5_flow.c
@@ -5881,6 +5881,12 @@ flow_sample_split_prep(struct rte_eth_dev *dev,
append_index++;
set_tag = (void *)(actions_pre + actions_n + append_index);
ret = mlx5_flow_get_reg_id(dev, MLX5_SAMPLE_ID, 0, error);
+ /* Trust VF/SF on CX5 not supported meter so that the reserved
+* metadata regC is REG_NON, back to use application tag
+* index 0.
+*/
+ if (unlikely(ret == REG_NON))
+ ret = mlx5_flow_get_reg_id(dev, MLX5_APP_TAG, 0, error);
if (ret < 0)
return ret;
mlx5_ipool_malloc(priv->sh->ipool
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 29751e7eda..ac21676416 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -6931,6 +6931,7 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
bool def_policy = false;
bool shared_count = false;
uint16_t udp_dport = 0;
+ uint32_t tag_id = 0;
if (items == NULL)
return -1;
@@ -7407,6 +7408,8 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
++actions_n;
if (action_flags & MLX5_FLOW_ACTION_SAMPLE)
modify_after_mirror = 1;
+ tag_id = ((const struct rte_flow_action_set_tag *)
+ actions->conf)->index;
action_flags |= MLX5_FLOW_ACTION_SET_TAG;
rw_act_num += MLX5_ACT_NUM_SET_TAG;
break;
@@ -7841,6 +7844,13 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
error);
if (ret < 0)
return ret;
+ if ((action_flags & MLX5_FLOW_ACTION_SET_TAG) &&
+ tag_id == 0 && priv->mtr_color_reg == REG_NON)
+ return rte_flow_error_set
+ (error, EINVAL,
+ RTE_FLOW_ERROR_TYPE_ACTION,
+ NULL,
+ "Sample after Tag action cause metadata
tag index 0 corruption");
action_flags |= MLX5_FLOW_ACTION_SAMPLE;
++actions_n;
break;
--
2.18.1
RE: [PATCH v3 1/2] app/testpmd: do not poll stopped queues
> Calling Rx/Tx functions on a stopped queue is not supported.
> Do not run packet forwarding for streams that use stopped queues.
>
> Each stream has a read-only "disabled" field,
> so that lcore function can skip such streams.
> Forwarding engines can set this field
> using a new "stream_init" callback function
> by checking relevant queue states,
> which are stored along with queue configurations
> (not all PMDs implement rte_eth_rx/tx_queue_info_get()
> to query the state from there).
>
> Fixes: 5f4ec54f1d16 ("testpmd: queue start and stop")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Dmitry Kozlyuk
> Acked-by: Matan Azrad
> ---
Pinging testpmd maintainers for this bugfix.
The sibling ethdev patch will have to wait until the next release cycle.
[PATCH] crypto/ipsec_mb: fix GMAC parameters setting
AES-GMAC requires plaintext length to be 0 when using AES-GCM,
so only AAD data is used.
Fixes: a501609ea646 ("crypto/ipsec_mb: fix length and offset settings")
Cc: pablo.de.lara.gua...@intel.com
Cc: sta...@dpdk.org
Signed-off-by: Pablo de Lara
---
drivers/crypto/ipsec_mb/pmd_aesni_mb.c | 16 +---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
index c974886032..afa0b6e3a4 100644
--- a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
+++ b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
@@ -1296,9 +1296,9 @@ set_mb_job_params(IMB_JOB *job, struct ipsec_mb_qp *qp,
op->sym->aead.data.offset;
job->msg_len_to_hash_in_bytes =
op->sym->aead.data.length;
- } else {
- job->msg_len_to_cipher_in_bytes = 0;
+ } else { /* AES-GMAC only, only AAD used */
job->msg_len_to_hash_in_bytes = 0;
+ job->hash_start_src_offset_in_bytes = 0;
}
job->iv = rte_crypto_op_ctod_offset(op, uint8_t *,
@@ -1378,8 +1378,18 @@ set_mb_job_params(IMB_JOB *job, struct ipsec_mb_qp *qp,
job->msg_len_to_cipher_in_bits =
op->sym->cipher.data.length;
break;
- case IMB_CIPHER_CCM:
case IMB_CIPHER_GCM:
+ if (session->cipher.mode == IMB_CIPHER_NULL) {
+ /* AES-GMAC only (only AAD used) */
+ job->msg_len_to_cipher_in_bytes = 0;
+ job->cipher_start_src_offset_in_bytes = 0;
+ } else {
+ job->cipher_start_src_offset_in_bytes =
+ op->sym->aead.data.offset;
+ job->msg_len_to_cipher_in_bytes =
op->sym->aead.data.length;
+ }
+ break;
+ case IMB_CIPHER_CCM:
case IMB_CIPHER_CHACHA20_POLY1305:
job->cipher_start_src_offset_in_bytes =
op->sym->aead.data.offset;
--
2.25.1
Re: [PATCH v2 2/2] build: hide local symbols in shared libraries
Hi Thomas,
On 08/03/2022 14:24, Thomas Monjalon wrote:
The symbols which are not listed in the version script
are exported by default.
Adding a local section with a wildcard make non-listed functions
and variables as hidden, as it should be in all version.map files.
These are the changes done in the shared libraries:
- DF .text Base auxiliary_add_device
- DF .text Base auxiliary_dev_exists
- DF .text Base auxiliary_dev_iterate
- DF .text Base auxiliary_insert_device
- DF .text Base auxiliary_is_ignored_device
- DF .text Base auxiliary_match
- DF .text Base auxiliary_on_scan
- DF .text Base auxiliary_scan
- DO .bss Base auxiliary_bus_logtype
- DO .data Base auxiliary_bus
- DO .bss Base gpu_logtype
There is no impact on regexdev library.
Because these local symbols were exported as non-internal
in DPDK 21.11, any change in these functions would break the ABI.
Exception rules are added for these experimental libraries,
so the ABI check will skip them until the next ABI version.
A check is added to avoid such miss in future.
Fixes: 1afce3086cf4 ("bus/auxiliary: introduce auxiliary bus")
Fixes: 8b8036a66e3d ("gpudev: introduce GPU device class library")
Cc: sta...@dpdk.org
If I take this 2/2 for 21.11.1, then I also need to backport [0] so I
won't have errors for common_mlx5.
Any problem with taking both?
[0]
commit c2e3059a10f2389b791d5d485fe71e666984c193
Author: Michael Baum
Date: Fri Feb 25 01:25:06 2022 +0200
common/mlx5: consider local functions as internal
Signed-off-by: Thomas Monjalon
---
devtools/check-symbol-maps.sh | 7 +++
devtools/libabigail.abignore | 8
drivers/bus/auxiliary/version.map | 2 ++
lib/gpudev/version.map| 2 ++
lib/regexdev/version.map | 2 ++
5 files changed, 21 insertions(+)
diff --git a/devtools/check-symbol-maps.sh b/devtools/check-symbol-maps.sh
index 5bd290ac97..8266fdf9ea 100755
--- a/devtools/check-symbol-maps.sh
+++ b/devtools/check-symbol-maps.sh
@@ -53,4 +53,11 @@ if [ -n "$duplicate_symbols" ] ; then
ret=1
fi
+local_miss_maps=$(grep -L 'local: \*;' $@)
+if [ -n "$local_miss_maps" ] ; then
+echo "Found maps without local catch-all:"
+echo "$local_miss_maps"
+ret=1
+fi
+
exit $ret
diff --git a/devtools/libabigail.abignore b/devtools/libabigail.abignore
index 18c11c80c6..c618f20032 100644
--- a/devtools/libabigail.abignore
+++ b/devtools/libabigail.abignore
@@ -32,3 +32,11 @@
; Ignore changes in common mlx5 driver, should be all internal
[suppress_file]
soname_regexp = ^librte_common_mlx5\.
+
+; Ignore visibility fix of local functions in experimental auxiliary driver
+[suppress_file]
+soname_regexp = ^librte_bus_auxiliary\.
+
+; Ignore visibility fix of local functions in experimental gpudev library
+[suppress_file]
+soname_regexp = ^librte_gpudev\.
diff --git a/drivers/bus/auxiliary/version.map
b/drivers/bus/auxiliary/version.map
index a52260657c..dc993e84ff 100644
--- a/drivers/bus/auxiliary/version.map
+++ b/drivers/bus/auxiliary/version.map
@@ -4,4 +4,6 @@ EXPERIMENTAL {
# added in 21.08
rte_auxiliary_register;
rte_auxiliary_unregister;
+
+ local: *;
};
diff --git a/lib/gpudev/version.map b/lib/gpudev/version.map
index b23e3fd6eb..a2c8ce5759 100644
--- a/lib/gpudev/version.map
+++ b/lib/gpudev/version.map
@@ -39,4 +39,6 @@ INTERNAL {
rte_gpu_get_by_name;
rte_gpu_notify;
rte_gpu_release;
+
+ local: *;
};
diff --git a/lib/regexdev/version.map b/lib/regexdev/version.map
index 988b909638..3c6e9fffa1 100644
--- a/lib/regexdev/version.map
+++ b/lib/regexdev/version.map
@@ -26,6 +26,8 @@ EXPERIMENTAL {
rte_regexdev_xstats_get;
rte_regexdev_xstats_names_get;
rte_regexdev_xstats_reset;
+
+ local: *;
};
INTERNAL {
RE: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
> -Original Message-
> From: De Lara Guarch, Pablo
> Sent: Wednesday, March 9, 2022 10:45 AM
> To: Nicolau, Radu ; Buckley, Daniel M
> ; Zhang, Roy Fan
> Cc: dev@dpdk.org; De Lara Guarch, Pablo ;
> sta...@dpdk.org
> Subject: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
>
> AES-GMAC requires plaintext length to be 0 when using AES-GCM,
> so only AAD data is used.
>
> Fixes: a501609ea646 ("crypto/ipsec_mb: fix length and offset settings")
> Cc: pablo.de.lara.gua...@intel.com
> Cc: sta...@dpdk.org
>
> Signed-off-by: Pablo de Lara
> ---
> drivers/crypto/ipsec_mb/pmd_aesni_mb.c | 16 +---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> index c974886032..afa0b6e3a4 100644
> --- a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> +++ b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> @@ -1296,9 +1296,9 @@ set_mb_job_params(IMB_JOB *job, struct
> ipsec_mb_qp *qp,
> op->sym->aead.data.offset;
> job->msg_len_to_hash_in_bytes =
> op->sym->aead.data.length;
> - } else {
> - job->msg_len_to_cipher_in_bytes = 0;
> + } else { /* AES-GMAC only, only AAD used */
> job->msg_len_to_hash_in_bytes = 0;
> + job->hash_start_src_offset_in_bytes = 0;
> }
>
> job->iv = rte_crypto_op_ctod_offset(op, uint8_t *,
> @@ -1378,8 +1378,18 @@ set_mb_job_params(IMB_JOB *job, struct
> ipsec_mb_qp *qp,
> job->msg_len_to_cipher_in_bits =
> op->sym->cipher.data.length;
> break;
> - case IMB_CIPHER_CCM:
> case IMB_CIPHER_GCM:
> + if (session->cipher.mode == IMB_CIPHER_NULL) {
> + /* AES-GMAC only (only AAD used) */
> + job->msg_len_to_cipher_in_bytes = 0;
> + job->cipher_start_src_offset_in_bytes = 0;
> + } else {
> + job->cipher_start_src_offset_in_bytes =
> + op->sym->aead.data.offset;
> + job->msg_len_to_cipher_in_bytes = op->sym-
> >aead.data.length;
> + }
> + break;
> + case IMB_CIPHER_CCM:
> case IMB_CIPHER_CHACHA20_POLY1305:
> job->cipher_start_src_offset_in_bytes =
> op->sym->aead.data.offset;
> --
> 2.25.1
Acked-by: Fan Zhang
Thanks Pablo!
[PATCH] maintainers: update for KNI, UIO and pcap
Won't able to allocate time on these components, resigning from maintaining them. Signed-off-by: Ferruh Yigit --- Cc: David Marchand --- MAINTAINERS | 4 1 file changed, 4 deletions(-) diff --git a/MAINTAINERS b/MAINTAINERS index 4ba4c8d67c43..fdea1c623f5f 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -318,7 +318,6 @@ F: lib/eal/linux/ F: doc/guides/linux_gsg/ Linux UIO -M: Ferruh Yigit F: drivers/bus/pci/linux/*uio* Linux VFIO @@ -587,7 +586,6 @@ F: app/test/test_link_bonding* F: examples/bond/ Linux KNI -M: Ferruh Yigit F: kernel/linux/kni/ F: lib/kni/ F: doc/guides/prog_guide/kernel_nic_interface.rst @@ -964,7 +962,6 @@ F: doc/guides/nics/avp.rst F: doc/guides/nics/features/avp.ini PCAP PMD -M: Ferruh Yigit F: drivers/net/pcap/ F: doc/guides/nics/pcap_ring.rst F: doc/guides/nics/features/pcap.ini @@ -975,7 +972,6 @@ F: doc/guides/nics/tap.rst F: doc/guides/nics/features/tap.ini KNI PMD -M: Ferruh Yigit F: drivers/net/kni/ F: doc/guides/nics/kni.rst -- 2.35.1
Re: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
On 3/9/2022 10:45 AM, Pablo de Lara wrote:
AES-GMAC requires plaintext length to be 0 when using AES-GCM,
so only AAD data is used.
Fixes: a501609ea646 ("crypto/ipsec_mb: fix length and offset settings")
Cc: pablo.de.lara.gua...@intel.com
Cc: sta...@dpdk.org
Signed-off-by: Pablo de Lara
---
Tested-by: Radu Nicolau
Acked-by: Radu Nicolau
RE: [PATCH] app/testpmd: register metadata dynfield on modify field
Hi Ferruh,
> -Original Message-
> From: Ferruh Yigit
> Sent: Thursday, March 3, 2022 13:21
> To: Dariusz Sosnowski ; Ori Kam
> ; Xiaoyun Li ; Aman Singh
> ; Yuying Zhang
> Cc: dev@dpdk.org; Slava Ovsiienko
> Subject: Re: [PATCH] app/testpmd: register metadata dynfield on modify
> field
>
> External email: Use caution opening links or attachments
>
>
> On 3/1/2022 11:51 AM, Dariusz Sosnowski wrote:
> > This patch adds implicit registration of metadata dynamic field and
> > flag
>
> Hi Dariusz,
>
> metaday dynamic field is explicitly registered when testpmd command used
> to enable tx metadata, or rte flow rule created with "set_meta" action.
>
> Can you please document more when this implicit enablement is required?
> And why that case doesn't cover above explicit enable cases?
Before this patch, when a user inserted a flow rule with MODIFY_FIELD action,
which modified packet metadata, the metadata dynamic field was not registered,
as opposed to
what happened with SET_META action. Goal of this patch is to make the behavior
consistent
between these two actions.
Maybe using "implicit" in the commit message was misleading here.
What do you think about rewording the commit message to something like the one
below?
"This patch adds registration of metadata dynamic field and flag
whenever a MODIFY_FIELD action with META as source and/or destination
field is used. It makes the behavior consistent with SET_META action, where
metadata dynamic field and flag is registered on flow rule creation."
> > whenever a modify_field action with META as source and/or destination
> > field is used.
> >
>
> According below code it is only registered in the DST_TYPE block, not is
> 'else'
> (which seems src) leg, is this OK?
>
> > Signed-off-by: Dariusz Sosnowski
> > Reviewed-by: Viacheslav Ovsiienko
> > ---
> > app/test-pmd/cmdline_flow.c | 11 +--
> > 1 file changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/app/test-pmd/cmdline_flow.c b/app/test-pmd/cmdline_flow.c
> > index 4f7a9f17f9..dd38a635b0 100644
> > --- a/app/test-pmd/cmdline_flow.c
> > +++ b/app/test-pmd/cmdline_flow.c
> > @@ -8347,6 +8347,7 @@ parse_vc_modify_field_id(struct context *ctx,
> const struct token *token,
> > {
> > struct rte_flow_action_modify_field *action_modify_field;
> > unsigned int i;
> > + int ret;
> >
> > (void)token;
> > (void)buf;
> > @@ -8362,9 +8363,15 @@ parse_vc_modify_field_id(struct context *ctx,
> const struct token *token,
> > if (!ctx->object)
> > return len;
> > action_modify_field = ctx->object;
> > - if (ctx->curr == ACTION_MODIFY_FIELD_DST_TYPE_VALUE)
> > + if (ctx->curr == ACTION_MODIFY_FIELD_DST_TYPE_VALUE) {
> > action_modify_field->dst.field = (enum rte_flow_field_id)i;
> > - else
> > + if (action_modify_field->dst.field == RTE_FLOW_FIELD_META) {
> > + ret = rte_flow_dynf_metadata_register();
> > + if (ret < 0)
> > + return -1;
> > + }
> > +
> > + } else
> > action_modify_field->src.field = (enum rte_flow_field_id)i;
> > return len;
> > }
No, I should add registering for source field as well.
Best regards,
Dariusz Sosnowski
[PATCH v1] net/mlx5: disallow multiple ASO actions in a single flow
For now, only one ASO action is supported in a single flow.
Flow with more than one ASO action should be rejected in the
validation stage.
Flow with action non-shared AGE and COUNT together should be
treated as non-ASO because AGE will fall back to use HW counter,
not ASO hit object.
Group 0 will use HW counter for AGE action even if no COUNT action.
This commit will reject patterns (no matter which group if transfer)
like:
1. group 1 pattern... / end actions age / meter / end
2. group 1 pattern... / end actions conntrack / meter / end
3. group 1 pattern... / end actions age / conntrack... / end
If AGE comes together with COUNT in the above patterns, it's allowed.
Fixes: daed4b6e ("net/mlx5: use aging by counter when counter exists")
Cc: sta...@dpdk.org
Signed-off-by: Rongwei Liu
Acked-by: Xiaoyu Min
Acked-by: Matan Azrad
---
drivers/net/mlx5/mlx5_flow_dv.c | 20 +++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 29751e7eda..fccc4a9301 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -6862,7 +6862,7 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
bool external, int hairpin, struct rte_flow_error *error)
{
int ret;
- uint64_t action_flags = 0;
+ uint64_t aso_mask, action_flags = 0;
uint64_t item_flags = 0;
uint64_t last_item = 0;
uint8_t next_protocol = 0xff;
@@ -6931,6 +6931,8 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
bool def_policy = false;
bool shared_count = false;
uint16_t udp_dport = 0;
+ const struct rte_flow_action_age *non_shared_age = NULL;
+ const struct rte_flow_action_count *count = NULL;
if (items == NULL)
return -1;
@@ -7468,6 +7470,7 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
attr, error);
if (ret < 0)
return ret;
+ count = actions->conf;
action_flags |= MLX5_FLOW_ACTION_COUNT;
++actions_n;
break;
@@ -7773,6 +7776,7 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
++actions_n;
break;
case RTE_FLOW_ACTION_TYPE_AGE:
+ non_shared_age = actions->conf;
ret = flow_dv_validate_action_age(action_flags,
actions, dev,
error);
@@ -8048,6 +8052,20 @@ flow_dv_validate(struct rte_eth_dev *dev, const struct
rte_flow_attr *attr,
"cannot be done before meter action");
}
}
+ /*
+* Only support one ASO action in a single flow.
+* non-shared AGE + counter will fallback to use HW counter, no ASO hit
object.
+* Group 0 uses HW counter for AGE too even if no counter action.
+*/
+ aso_mask = (action_flags & MLX5_FLOW_ACTION_METER &&
priv->sh->meter_aso_en) << 2 |
+ (action_flags & MLX5_FLOW_ACTION_CT && priv->sh->ct_aso_en)
<< 1 |
+ (action_flags & MLX5_FLOW_ACTION_AGE &&
+ !(non_shared_age && count) &&
+ (attr->group || (attr->transfer && priv->fdb_def_rule)) &&
+ priv->sh->flow_hit_aso_en);
+ if (__builtin_popcountl(aso_mask) > 1)
+ return rte_flow_error_set(error, ENOTSUP,
RTE_FLOW_ERROR_TYPE_ACTION,
+ NULL, "unsupported combining AGE,
METER, CT ASO actions in a single flow");
/*
* Hairpin flow will add one more TAG action in TX implicit mode.
* In TX explicit mode, there will be no hairpin flow ID.
--
2.27.0
RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
> > -Original Message-
> > From: Bronowski, PiotrX
> > Sent: Monday, March 7, 2022 3:33 PM
> > To: dev@dpdk.org
> > Cc: Zhang, Roy Fan ; tho...@monjalon.net;
> > gak...@marvell.com; Yigit, Ferruh ; Doherty,
> > Declan ; Bronowski, PiotrX
> > ; sta...@dpdk.org
> > Subject: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
> >
> > This patch removes coverity defect CID 375828:
> > Untrusted value as argument (TAINTED_SCALAR)
> >
> > Coverity issue: CID 375828
> > Fixes: 918fd2f1466b ("crypto/ipsec_mb: move aesni_mb PMD")
> >
> > Signed-off-by: Piotr Bronowski
> >
> > Cc: sta...@dpdk.org
> >
> > ---
> > v2: use a different logic to check digest length
> > ---
> Acked-by: Fan Zhang
Acked-by: Kai Ji
[PATCH] lib/power: power pmd errata with RTM and gcc-9
An errata exists where users may see reduced power savings when using PMD Power Management. This issue occurs when compiling DPDK applications with gcc-9 on platforms with TSX enabled. In rte_power_monitor_multi(), the function may return without successfully starting the RTM transaction (the _xbegin() fails). Signed-off-by: David Hunt --- doc/guides/rel_notes/release_22_03.rst | 11 +++ 1 file changed, 11 insertions(+) diff --git a/doc/guides/rel_notes/release_22_03.rst b/doc/guides/rel_notes/release_22_03.rst index 60e5b4f9aa..34d6f34dba 100644 --- a/doc/guides/rel_notes/release_22_03.rst +++ b/doc/guides/rel_notes/release_22_03.rst @@ -313,6 +313,17 @@ Known Issues Also, make sure to start the actual text at the margin. === +* **Possible reduced power saving with PMD Power Management.** + + Users may see reduced power savings when using PMD Power Management. + This issue occurs when compiling DPDK applications with gcc-9 on + platforms with TSX enabled. In rte_power_monitor_multi(), the + function may return without successfully starting the RTM + transaction (the _xbegin() fails). + + There are three workarounds for this issue. Either build DPDK with gcc-11 + or newer, build with shared libraries, or build DPDK with fewer drivers. + Tested Platforms -- 2.17.1
DTS Improvement Working Group Minutes (March 2)
Attendees - Owen Hilyard - Honnappa Nagarahalli - Lijuan Tu - Lincoln Lavoie Agenda - Additions to the agenda - Review/Carry forward pending action items - Review DTS DPDK Modifications excel sheet - Merging DTS into DPDK - Review additional work items MinutesReview/Carry forward pending action items - Honnappa - Patches to DPDK to address changing the constants and printfs under progress. - Still in progress - Honnappa - The changes to DPDK need to be backported to LTS releases. Discuss with Lijuan and the community. - Patches will be backported - Owen - Owen to review Juraj’s feedback. - Done - Honnappa - Follow up with DPDK governing board to get the approval/legal opinion on including the GPL licensed files from DTS in the DPDK tree. - Governing board has not met - Owen - Look at making DTS build using the common DPDK repo in addition to the existing tarball option. - Looks to be possible, blocked by decisions on the placement of DTS inside of DPDK - Lijuan - The makefile removal patch is ready to get merged ( http://patchwork.dpdk.org/project/dts/list/?series=20610). Merge it in the 22.03 DTS release. - Additional issues discovered - Meson configuration changed improperly - Makefiles will be removed as the meson issues are fixed - Juraj - Postponed: Convert the makefile build for documentation to use meson build after merging the repositories or the directory structure is known. - Not present - Honnappa - Discuss merge permissions for Owen for the DTS tree with the Tech Board - Merge permissions have been granted - Lijuan/Owen - Check if dep/scapy_modules/Dot1BR.py file is required to be in DTS. If it is required, understand options for replacement/removing this file. - Look into rewriting the file so it’s not GPL - Lijuan/Owen - Check if dep/QMP/qmp.py can be installed on DUT as part of the qemu-guest-agent package. - In progress - Honnappa/Owen - Discuss Python bindings for DPDK in March 9th Techboard call. - Meeting has not happened yet Merging DTS into DPDK - Are we going to integrate DTS with the DPDK build system? No resolution yet. - Try for 22.07 - Focus on framework and test cases used in CI - Do a deep dive on Feb 9th - identify what needs to be done and estimate the work. DPDK Modifications Review - Some issues are fixed - Many changes are still present. Action Items - Honnappa - Patches to DPDK to address changing the constants and printfs under progress. - Owen - Write a script to run formatters (Black + Isort). Pin versions in dts requirements file. As part of the patch series, format DTS again. - Jurai - Rewrite dep/scapy_modules/Dot1BR.py from scratch so we can change the license. - Lijuan/Owen - Check if dep/QMP/qmp.py can be installed on DUT as part of the qemu-guest-agent package. - Lijuan/Owen - Remove ABI_stable test suite - Lijuan/Owen - Investigate DTS test suites to check for redundant tests - Unit testing (handled by meson now) - “Does it compile?” tests - etc Any other business - Next Meeting: March 9, 2022
RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
Hi Piotr,
> -Original Message-
> From: Zhang, Roy Fan
> Sent: Monday, March 7, 2022 4:27 PM
> To: Bronowski, PiotrX ; dev@dpdk.org
> Cc: tho...@monjalon.net; gak...@marvell.com; Yigit, Ferruh
> ; Doherty, Declan ;
> sta...@dpdk.org
> Subject: RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
>
> > -Original Message-
> > From: Bronowski, PiotrX
> > Sent: Monday, March 7, 2022 3:33 PM
> > To: dev@dpdk.org
> > Cc: Zhang, Roy Fan ; tho...@monjalon.net;
> > gak...@marvell.com; Yigit, Ferruh ; Doherty,
> Declan
> > ; Bronowski, PiotrX
> > ; sta...@dpdk.org
> > Subject: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
> >
> > This patch removes coverity defect CID 375828:
> > Untrusted value as argument (TAINTED_SCALAR)
> >
> > Coverity issue: CID 375828
> > Fixes: 918fd2f1466b ("crypto/ipsec_mb: move aesni_mb PMD")
> >
> > Signed-off-by: Piotr Bronowski
> >
> > Cc: sta...@dpdk.org
> >
> > ---
> > v2: use a different logic to check digest length
> > ---
> Acked-by: Fan Zhang
Sorry I missed a point in your change and thanks for Ciara pointing this out.
You are changing the gen_digest_size to 64 which is wrong.
Please send v3.
Also instead of ack - Nack this patch.
RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
Hi Piotr,
>-Original Message-
>From: Zhang, Roy Fan
>Sent: Wednesday 9 March 2022 14:35
>To: Zhang, Roy Fan ; Bronowski, PiotrX
>; dev@dpdk.org
>Cc: tho...@monjalon.net; gak...@marvell.com; Yigit, Ferruh
>; Doherty, Declan ;
>sta...@dpdk.org; Power, Ciara
>Subject: RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
>
>Hi Piotr,
>
>> -Original Message-
>> From: Zhang, Roy Fan
>> Sent: Monday, March 7, 2022 4:27 PM
>> To: Bronowski, PiotrX ; dev@dpdk.org
>> Cc: tho...@monjalon.net; gak...@marvell.com; Yigit, Ferruh
>> ; Doherty, Declan ;
>> sta...@dpdk.org
>> Subject: RE: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
>>
>> > -Original Message-
>> > From: Bronowski, PiotrX
>> > Sent: Monday, March 7, 2022 3:33 PM
>> > To: dev@dpdk.org
>> > Cc: Zhang, Roy Fan ; tho...@monjalon.net;
>> > gak...@marvell.com; Yigit, Ferruh ; Doherty,
>> Declan
>> > ; Bronowski, PiotrX
>> > ; sta...@dpdk.org
>> > Subject: [PATCH v2] crypto/ipsec_mb: fix usage of untrusted value
>> >
>> > This patch removes coverity defect CID 375828:
>> > Untrusted value as argument (TAINTED_SCALAR)
>> >
>> > Coverity issue: CID 375828
>> > Fixes: 918fd2f1466b ("crypto/ipsec_mb: move aesni_mb PMD")
>> >
>> > Signed-off-by: Piotr Bronowski
>> >
>> > Cc: sta...@dpdk.org
>> >
>> > ---
>> > v2: use a different logic to check digest length
>> > ---
>> Acked-by: Fan Zhang
>
>Sorry I missed a point in your change and thanks for Ciara pointing this out.
>You are changing the gen_digest_size to 64 which is wrong.
>Please send v3.
>Also instead of ack - Nack this patch.
[CP]
In the v3 I think Fixes line should also be updated to either:
Fixes: 746825e5c0ea ("crypto/ipsec_mb: move aesni_gcm PMD")
Or
Fixes: ceb863938708 ("crypto/aesni_gcm: support all truncated digest sizes")
Cc: pablo.de.lara.gua...@intel.com
(The second one seems to be where the code was introduced before being moved
into the consolidated ipsec_mb PMD in 21.11)
Thanks,
Ciara
[PATCH v3] crypto/ipsec_mb: fix coverity issue
This patch removes coverity defect CID 375828:
Untrusted value as argument (TAINTED_SCALAR)
Coverity issue: CID 375828
Fixes: 918fd2f1466b ("crypto/ipsec_mb: move aesni_mb PMD")
Signed-off-by: Piotr Bronowski
Cc: sta...@dpdk.org
---
v3: use a different logic to check digest length
---
drivers/crypto/ipsec_mb/pmd_aesni_gcm.c | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
b/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
index e5ad629fe5..2c033c6f28 100644
--- a/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
+++ b/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
@@ -96,7 +96,9 @@ aesni_gcm_session_configure(IMB_MGR *mb_mgr, void *session,
sess->iv.length = auth_xform->auth.iv.length;
key_length = auth_xform->auth.key.length;
key = auth_xform->auth.key.data;
- sess->req_digest_length = auth_xform->auth.digest_length;
+ sess->req_digest_length =
+ RTE_MIN(auth_xform->auth.digest_length,
+ DIGEST_LENGTH_MAX);
break;
case IPSEC_MB_OP_AEAD_AUTHENTICATED_ENCRYPT:
case IPSEC_MB_OP_AEAD_AUTHENTICATED_DECRYPT:
@@ -116,7 +118,9 @@ aesni_gcm_session_configure(IMB_MGR *mb_mgr, void *session,
key_length = aead_xform->aead.key.length;
key = aead_xform->aead.key.data;
sess->aad_length = aead_xform->aead.aad_length;
- sess->req_digest_length = aead_xform->aead.digest_length;
+ sess->req_digest_length =
+ RTE_MIN(aead_xform->aead.digest_length,
+ DIGEST_LENGTH_MAX);
break;
default:
IPSEC_MB_LOG(
--
2.30.2
--
Intel Research and Development Ireland Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263
This e-mail and any attachments may contain confidential material for the sole
use of the intended recipient(s). Any review or distribution by others is
strictly prohibited. If you are not the intended recipient, please contact the
sender and delete all copies.
RE: [PATCH v2] doc: fix telemetry example for cryptodev
Hi Sean,
>-Original Message-
>From: Sean Morrissey
>Sent: Wednesday 9 March 2022 09:33
>Cc: dev@dpdk.org; Morrissey, Sean ;
>sta...@dpdk.org
>Subject: [PATCH v2] doc: fix telemetry example for cryptodev
>
>Blank line added to the final telemetry example for the cryptography device
>library as to fix the example rendering.
>
>Fixes: 1c559ee8469 ("cryptodev: add telemetry endpoint for capabilities")
>Cc: sta...@dpdk.org
>
>Signed-off-by: Sean Morrissey
Thanks for this.
Acked-by: Ciara Power
RE: [EXT] Re: [PATCH] examples/l3fwd: resolve stack buffer overflow issue
Hi Thomas,
> -Original Message-
> From: Thomas Monjalon
> Sent: Tuesday, March 8, 2022 4:51 PM
> To: Rahul Bhansali
> Cc: dev@dpdk.org; david.march...@redhat.com; Conor Walsh
>
> Subject: [EXT] Re: [PATCH] examples/l3fwd: resolve stack buffer overflow issue
>
> External Email
>
> --
> 11/01/2022 13:50, Rahul Bhansali:
> > This patch fixes the stack buffer overflow error reported from
> > AddressSanitizer.
> > Function send_packetsx4() tries to access out of bound data from
> > rte_mbuf and fill it into TX buffer even in the case where no pending
> > packets (len = 0).
> > Performance impact:- No
> >
> > ASAN error report:-
> > ==819==ERROR: AddressSanitizer: stack-buffer-overflow on address
> > 0xe2c0dcf0 at pc 0x005e791c bp 0xe2c0d7e0 sp
> > 0xe2c0d800 READ of size 8 at 0xe2c0dcf0 thread T0
> > #0 0x5e7918 in send_packetsx4 ../examples/l3fwd/l3fwd_common.h:251
> > #1 0x5e7918 in send_packets_multi ../examples/l3fwd/l3fwd_neon.h:226
>
> This code comes from below commit, so these tags are missing:
> Fixes: 96ff445371e0 ("examples/l3fwd: reorganise and optimize LPM code
> path")
> Cc: sta...@dpdk.org
>
> > Signed-off-by: Rahul Bhansali
> > ---
> > examples/l3fwd/l3fwd_common.h | 4
> > 1 file changed, 4 insertions(+)
> >
> > diff --git a/examples/l3fwd/l3fwd_common.h
> > b/examples/l3fwd/l3fwd_common.h index 7d83ff641a..de77711f88 100644
> > --- a/examples/l3fwd/l3fwd_common.h
> > +++ b/examples/l3fwd/l3fwd_common.h
> > @@ -236,6 +236,9 @@ send_packetsx4(struct lcore_conf *qconf, uint16_t
> > port, struct rte_mbuf *m[],
> >
> > /* copy rest of the packets into the TX buffer. */
> > len = num - n;
> > + if (len == 0)
> > + goto exit;
> > +
>
> I don't understand how it can fix something.
> There is already "while (j < len)" with j and len being 0, the loop should
> not be
> effective in this case.
This Switch will execute Case statement first even before considering the while
condition or anything else before case statement. While condition will be
executed only after all switch cases are executed.
Hence in case of len = 0 and n > 28, it is throwing stack buffer overflow error.
Below is sample code to simulate the while loop behavior inside switch. Checked
it for both x86 and arm64.
https://godbolt.org/z/4Kecqbsde
>
> > j = 0;
> > switch (len % FWDSTEP) {
> > while (j < len) {
> > @@ -258,6 +261,7 @@ send_packetsx4(struct lcore_conf *qconf, uint16_t
> port, struct rte_mbuf *m[],
> > }
> > }
> >
> > +exit:
> > qconf->tx_mbufs[port].len = len;
> > }
>
>
[PATCH v1] gpu/cuda: set unused gdrcopy functions arguments
From: Elena Agostini
Without enabling gdrcopy, dpdk build has warnings about
unused gdrcopy functions input parameters.
This patch set them to __rte_unused .
Fixes: 24c7759 ("gpu/cuda: map GPU memory with GDRCopy")
Signed-off-by: Elena Agostini
---
drivers/gpu/cuda/gdrcopy.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/cuda/gdrcopy.c b/drivers/gpu/cuda/gdrcopy.c
index 96e768b97b..3126d246eb 100644
--- a/drivers/gpu/cuda/gdrcopy.c
+++ b/drivers/gpu/cuda/gdrcopy.c
@@ -80,7 +80,7 @@ gdrcopy_open(gdr_t *g)
#endif
int
-gdrcopy_pin(gdr_t *gdrc_h, __rte_unused gdr_mh_t *mh, uint64_t d_addr, size_t
size, void **h_addr)
+gdrcopy_pin(__rte_unused gdr_t *gdrc_h, __rte_unused gdr_mh_t *mh,
__rte_unused uint64_t d_addr, __rte_unused size_t size, __rte_unused void
**h_addr)
{
#ifdef DRIVERS_GPU_CUDA_GDRCOPY_H
if (*gdrc_h == NULL) {
@@ -116,7 +116,7 @@ gdrcopy_pin(gdr_t *gdrc_h, __rte_unused gdr_mh_t *mh,
uint64_t d_addr, size_t si
}
int
-gdrcopy_unpin(gdr_t gdrc_h, __rte_unused gdr_mh_t mh, void *d_addr, size_t
size)
+gdrcopy_unpin(gdr_t gdrc_h, __rte_unused gdr_mh_t mh, __rte_unused void
*d_addr, __rte_unused size_t size)
{
if (gdrc_h == NULL)
return -EINVAL;
--
2.17.1
RE: [PATCH] examples/l3fwd: resolve stack buffer overflow issue
> This patch fixes the stack buffer overflow error reported
> from AddressSanitizer.
> Function send_packetsx4() tries to access out of bound data
> from rte_mbuf and fill it into TX buffer even in the case
> where no pending packets (len = 0).
> Performance impact:- No
>
> ASAN error report:-
> ==819==ERROR: AddressSanitizer: stack-buffer-overflow on address
> 0xe2c0dcf0 at pc 0x005e791c bp 0xe2c0d7e0 sp 0xe2c0d800
> READ of size 8 at 0xe2c0dcf0 thread T0
> #0 0x5e7918 in send_packetsx4 ../examples/l3fwd/l3fwd_common.h:251
> #1 0x5e7918 in send_packets_multi ../examples/l3fwd/l3fwd_neon.h:226
>
> Signed-off-by: Rahul Bhansali
> ---
> examples/l3fwd/l3fwd_common.h | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/examples/l3fwd/l3fwd_common.h b/examples/l3fwd/l3fwd_common.h
> index 7d83ff641a..de77711f88 100644
> --- a/examples/l3fwd/l3fwd_common.h
> +++ b/examples/l3fwd/l3fwd_common.h
> @@ -236,6 +236,9 @@ send_packetsx4(struct lcore_conf *qconf, uint16_t port,
> struct rte_mbuf *m[],
>
> /* copy rest of the packets into the TX buffer. */
> len = num - n;
> + if (len == 0)
> + goto exit;
> +
> j = 0;
> switch (len % FWDSTEP) {
> while (j < len) {
> @@ -258,6 +261,7 @@ send_packetsx4(struct lcore_conf *qconf, uint16_t port,
> struct rte_mbuf *m[],
> }
> }
>
> +exit:
> qconf->tx_mbufs[port].len = len;
> }
>
> --
Acked-by: Konstantin Ananyev
> 2.25.1
FW: [dpdk] Patch notification: 2 patches updated
Hi Christian/Luca, It was discussed in the mailing list to merge this into stable releases, however the 'Cc' to stable was missing. Can you please make sure this is going into stable releases? Kevin is adding it to 21.11 LTS. Thanks, Honnappa -Original Message- From: DPDK patchwork Sent: Tuesday, March 8, 2022 3:40 AM To: Honnappa Nagarahalli Subject: [dpdk] Patch notification: 2 patches updated Hello, The following patches (submitted by you) have been updated in Patchwork: * dpdk: [v3,1/2] examples/l3fwd: use single set of variables throughout the code - http://patches.dpdk.org/project/dpdk/patch/20220223060931.2661907-1-honnappa.nagaraha...@arm.com/ - for: DPDK was: New now: Accepted * dpdk: [v3,2/2] examples/l3fwd: make RX and TX queue size configurable - http://patches.dpdk.org/project/dpdk/patch/20220223060931.2661907-2-honnappa.nagaraha...@arm.com/ - for: DPDK was: New now: Accepted This email is a notification only - you do not need to respond. Happy patchworking. -- This is an automated mail sent by the Patchwork system at patches.dpdk.org. To stop receiving these notifications, edit your mail settings at: http://patches.dpdk.org/mail/
Re: FW: [dpdk] Patch notification: 2 patches updated
On 09/03/2022 16:36, Honnappa Nagarahalli wrote: Hi Christian/Luca, It was discussed in the mailing list to merge this into stable releases, however the 'Cc' to stable was missing. Can you please make sure this is going into stable releases? Kevin is adding it to 21.11 LTS. In case it helps, here they are for 21.11 [0]. There may have been some context issues, but there's so many patches I don't remember :-) Note, if you are cherry picking from main branch, i think the title of one was change from below. Kevin. [0] https://github.com/kevintraynor/dpdk-stable/commit/b466dfc9a5e3c486c43421a5a072bae7a777e720 https://github.com/kevintraynor/dpdk-stable/commit/ae6979cd9c13e47704da99cbdcb14ad5a38fb4ff Thanks, Honnappa -Original Message- From: DPDK patchwork Sent: Tuesday, March 8, 2022 3:40 AM To: Honnappa Nagarahalli Subject: [dpdk] Patch notification: 2 patches updated Hello, The following patches (submitted by you) have been updated in Patchwork: * dpdk: [v3,1/2] examples/l3fwd: use single set of variables throughout the code - http://patches.dpdk.org/project/dpdk/patch/20220223060931.2661907-1-honnappa.nagaraha...@arm.com/ - for: DPDK was: New now: Accepted * dpdk: [v3,2/2] examples/l3fwd: make RX and TX queue size configurable - http://patches.dpdk.org/project/dpdk/patch/20220223060931.2661907-2-honnappa.nagaraha...@arm.com/ - for: DPDK was: New now: Accepted This email is a notification only - you do not need to respond. Happy patchworking. -- This is an automated mail sent by the Patchwork system at patches.dpdk.org. To stop receiving these notifications, edit your mail settings at: http://patches.dpdk.org/mail/
Re: [PATCH] net/mlx5: set correct CPU socket ID for mlx5_rxq_ctrl
Hi,
On 3/9/2022 2:50 AM, Slava Ovsiienko wrote:
Hi, Thinh
Thank you for the patch, the code looks OK to me, but commit message is not
compliant:
- it should contain "fix" keyword in the title, like this:
"net/mlx5: fix CPU socket ID for Rx queue creation"
- could you, please, make problem description less personal and less wordy?
"The default CPU socket ID was used while creating the Rx queue and this
caused
creation failure in case if hardware was not resided on the default socket.
The patch sets the correct CPU socket ID for the mlx5_rxq_ctrl before
calling the mlx5_rxq_create_devx_rq_resources() which eventually calls
mlx5_devx_rq_create() with correct CPU socket ID."
- please add tags:
Cc: sta...@dpdk.org
Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
With best regards,
Slava
I'll resubmit the patch with suggestions above.
Regards,
Thinh
-Original Message-
From: Raslan Darawsheh
Sent: Tuesday, March 8, 2022 14:25
To: Dmitry Kozlyuk
Cc: Thinh Tran ; dev@dpdk.org;
d...@linux.vnet.ibm.com
Subject: RE: [PATCH] net/mlx5: set correct CPU socket ID for mlx5_rxq_ctrl
-Original Message-
From: Dmitry Kozlyuk
Sent: Tuesday, March 8, 2022 2:23 PM
To: Raslan Darawsheh
Cc: Thinh Tran ; dev@dpdk.org;
d...@linux.vnet.ibm.com
Subject: Re: [PATCH] net/mlx5: set correct CPU socket ID for
mlx5_rxq_ctrl
Hi Raslan,
Missing:
Fixes tag:
Fixes: 5ceb3a02b000 ("net/mlx5: move Rx queue DevX resource")
Cc: xuemi...@nvidia.com
I believe the bug originates from my earlier commit, not Xueming's one:
Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
Yes I think you are correct, my mistake 😊
Kindest regards,
Raslan Darawsheh
[PATCH v4] crypto/ipsec_mb: fix coverity issue
This patch removes coverity defect CID 375828:
Untrusted value as argument (TAINTED_SCALAR)
Coverity issue: CID 375828
Fixes: ceb863938708 ("crypto/aesni_gcm: support all truncated digest sizes")
Signed-off-by: Piotr Bronowski
Cc: sta...@dpdk.org
---
v4: commit message corrected
---
drivers/crypto/ipsec_mb/pmd_aesni_gcm.c | 8 ++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
b/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
index e5ad629fe5..2c033c6f28 100644
--- a/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
+++ b/drivers/crypto/ipsec_mb/pmd_aesni_gcm.c
@@ -96,7 +96,9 @@ aesni_gcm_session_configure(IMB_MGR *mb_mgr, void *session,
sess->iv.length = auth_xform->auth.iv.length;
key_length = auth_xform->auth.key.length;
key = auth_xform->auth.key.data;
- sess->req_digest_length = auth_xform->auth.digest_length;
+ sess->req_digest_length =
+ RTE_MIN(auth_xform->auth.digest_length,
+ DIGEST_LENGTH_MAX);
break;
case IPSEC_MB_OP_AEAD_AUTHENTICATED_ENCRYPT:
case IPSEC_MB_OP_AEAD_AUTHENTICATED_DECRYPT:
@@ -116,7 +118,9 @@ aesni_gcm_session_configure(IMB_MGR *mb_mgr, void *session,
key_length = aead_xform->aead.key.length;
key = aead_xform->aead.key.data;
sess->aad_length = aead_xform->aead.aad_length;
- sess->req_digest_length = aead_xform->aead.digest_length;
+ sess->req_digest_length =
+ RTE_MIN(aead_xform->aead.digest_length,
+ DIGEST_LENGTH_MAX);
break;
default:
IPSEC_MB_LOG(
--
2.30.2
--
Intel Research and Development Ireland Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263
This e-mail and any attachments may contain confidential material for the sole
use of the intended recipient(s). Any review or distribution by others is
strictly prohibited. If you are not the intended recipient, please contact the
sender and delete all copies.
Re: [PATCH v2 2/2] build: hide local symbols in shared libraries
09/03/2022 11:58, Kevin Traynor:
> Hi Thomas,
>
> On 08/03/2022 14:24, Thomas Monjalon wrote:
> > The symbols which are not listed in the version script
> > are exported by default.
> > Adding a local section with a wildcard make non-listed functions
> > and variables as hidden, as it should be in all version.map files.
> >
> > These are the changes done in the shared libraries:
> > - DF .text Base auxiliary_add_device
> > - DF .text Base auxiliary_dev_exists
> > - DF .text Base auxiliary_dev_iterate
> > - DF .text Base auxiliary_insert_device
> > - DF .text Base auxiliary_is_ignored_device
> > - DF .text Base auxiliary_match
> > - DF .text Base auxiliary_on_scan
> > - DF .text Base auxiliary_scan
> > - DO .bss Base auxiliary_bus_logtype
> > - DO .data Base auxiliary_bus
> > - DO .bss Base gpu_logtype
> >
> > There is no impact on regexdev library.
> >
> > Because these local symbols were exported as non-internal
> > in DPDK 21.11, any change in these functions would break the ABI.
> > Exception rules are added for these experimental libraries,
> > so the ABI check will skip them until the next ABI version.
> >
> > A check is added to avoid such miss in future.
> >
> > Fixes: 1afce3086cf4 ("bus/auxiliary: introduce auxiliary bus")
> > Fixes: 8b8036a66e3d ("gpudev: introduce GPU device class library")
> > Cc: sta...@dpdk.org
> >
>
> If I take this 2/2 for 21.11.1, then I also need to backport [0] so I
> won't have errors for common_mlx5.
>
> Any problem with taking both?
>
> [0]
> commit c2e3059a10f2389b791d5d485fe71e666984c193
> Author: Michael Baum
> Date: Fri Feb 25 01:25:06 2022 +0200
>
> common/mlx5: consider local functions as internal
I think that's fine to backport this as well.
RE: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
Hi Thomas,
> -Original Message-
> From: Zhang, Roy Fan
> Sent: Wednesday, March 9, 2022 11:01 AM
> To: De Lara Guarch, Pablo ; Nicolau, Radu
> ; Buckley, Daniel M
> Cc: dev@dpdk.org; sta...@dpdk.org
> Subject: RE: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
>
> > -Original Message-
> > From: De Lara Guarch, Pablo
> > Sent: Wednesday, March 9, 2022 10:45 AM
> > To: Nicolau, Radu ; Buckley, Daniel M
> > ; Zhang, Roy Fan
> > Cc: dev@dpdk.org; De Lara Guarch, Pablo
> > ; sta...@dpdk.org
> > Subject: [PATCH] crypto/ipsec_mb: fix GMAC parameters setting
> >
> > AES-GMAC requires plaintext length to be 0 when using AES-GCM, so only
> > AAD data is used.
> >
> > Fixes: a501609ea646 ("crypto/ipsec_mb: fix length and offset
> > settings")
> > Cc: pablo.de.lara.gua...@intel.com
> > Cc: sta...@dpdk.org
> >
> > Signed-off-by: Pablo de Lara
> > ---
> > drivers/crypto/ipsec_mb/pmd_aesni_mb.c | 16 +---
> > 1 file changed, 13 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> > b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> > index c974886032..afa0b6e3a4 100644
> > --- a/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> > +++ b/drivers/crypto/ipsec_mb/pmd_aesni_mb.c
> > @@ -1296,9 +1296,9 @@ set_mb_job_params(IMB_JOB *job, struct
> > ipsec_mb_qp *qp,
> > op->sym->aead.data.offset;
> > job->msg_len_to_hash_in_bytes =
> > op->sym->aead.data.length;
> > - } else {
> > - job->msg_len_to_cipher_in_bytes = 0;
> > + } else { /* AES-GMAC only, only AAD used */
> > job->msg_len_to_hash_in_bytes = 0;
> > + job->hash_start_src_offset_in_bytes = 0;
> > }
> >
> > job->iv = rte_crypto_op_ctod_offset(op, uint8_t *, @@ -1378,8
> > +1378,18 @@ set_mb_job_params(IMB_JOB *job, struct ipsec_mb_qp *qp,
> > job->msg_len_to_cipher_in_bits =
> > op->sym->cipher.data.length;
> > break;
> > - case IMB_CIPHER_CCM:
> > case IMB_CIPHER_GCM:
> > + if (session->cipher.mode == IMB_CIPHER_NULL) {
> > + /* AES-GMAC only (only AAD used) */
> > + job->msg_len_to_cipher_in_bytes = 0;
> > + job->cipher_start_src_offset_in_bytes = 0;
> > + } else {
> > + job->cipher_start_src_offset_in_bytes =
> > + op->sym->aead.data.offset;
> > + job->msg_len_to_cipher_in_bytes = op->sym-
> > >aead.data.length;
> > + }
> > + break;
> > + case IMB_CIPHER_CCM:
> > case IMB_CIPHER_CHACHA20_POLY1305:
> > job->cipher_start_src_offset_in_bytes =
> > op->sym->aead.data.offset;
> > --
> > 2.25.1
> Acked-by: Fan Zhang Thanks Pablo!
Would it be possible to merge this fix for this release?
A previous commit merged in this release has introduced a bug,
so it would be great if this fix can make it into 22.03.
Thanks,
Pablo
Re: [EXT] Re: [PATCH] examples/l3fwd: resolve stack buffer overflow issue
09/03/2022 16:24, Rahul Bhansali:
> Hi Thomas,
>
> From: Thomas Monjalon
> > 11/01/2022 13:50, Rahul Bhansali:
> > > /* copy rest of the packets into the TX buffer. */
> > > len = num - n;
> > > + if (len == 0)
> > > + goto exit;
> > > +
> >
> > I don't understand how it can fix something.
> > There is already "while (j < len)" with j and len being 0, the loop should
> > not be
> > effective in this case.
>
> This Switch will execute Case statement first even before considering the
> while condition or anything else before case statement. While condition will
> be executed only after all switch cases are executed.
I don't know this construct. Is it part of the C standard?
We learn something everyday :)
> Hence in case of len = 0 and n > 28, it is throwing stack buffer overflow
> error.
>
> Below is sample code to simulate the while loop behavior inside switch.
> Checked it for both x86 and arm64.
> https://godbolt.org/z/4Kecqbsde
>
> >
> > > j = 0;
> > > switch (len % FWDSTEP) {
> > > while (j < len) {
RE: [PATCH] net/mlx5: fix the port match in sample flow
> -Original Message-
> From: Jiawei(Jonny) Wang
> Sent: Monday, March 7, 2022 4:11
> To: Slava Ovsiienko ; Matan Azrad
> ; Ori Kam
> Cc: dev@dpdk.org; Raslan Darawsheh ;
> sta...@dpdk.org
> Subject: [PATCH] net/mlx5: fix the port match in sample flow
>
> If there are an explicit port match and sample action in the same flow,
> mlx5 PMD pushes the explicit port match in the prefix subflow, and uses the
> tag item match in the suffix subflow.
>
> The explicit port match was translated into source vport match so the sample
> suffix subflow lost this match after flow split.
>
> This patch copies the explicit port match to the sample suffix subflow, and
> the latter gets the correct source vport value in the flow matcher.
>
> Fixes: b4c0ddbfcc58 ("net/mlx5: split sample flow into two sub-flows")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Jiawei Wang
Acked-by: Viacheslav Ovsiienko
[PATCH v2] net/mlx5: fix CPU socket ID for Rx queue creation
The default CPU socket ID was used while creating the Rx queue and this caused
creation failure in case if hardware was not resided on the default socket.
The patch sets the correct CPU socket ID for the mlx5_rxq_ctrl before
calling the mlx5_rxq_create_devx_rq_resources() which eventually calls
mlx5_devx_rq_create() with correct CPU socket ID.
v2: change patch description, no code change
Fixes: bc5bee028ebc ("net/mlx5: create drop queue using DevX")
Cc: sta...@dpdk.org
Signed-off-by: Thinh Tran
Reviewed-by: David Christensen
---
drivers/net/mlx5/mlx5_devx.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/mlx5/mlx5_devx.c b/drivers/net/mlx5/mlx5_devx.c
index af106bda50..5ab092a259 100644
--- a/drivers/net/mlx5/mlx5_devx.c
+++ b/drivers/net/mlx5/mlx5_devx.c
@@ -947,6 +947,8 @@ mlx5_rxq_devx_obj_drop_create(struct rte_eth_dev *dev)
rte_errno = ENOMEM;
goto error;
}
+ /* set the CPU socket ID where the rxq_ctrl was allocated */
+ rxq_ctrl->socket = socket_id;
rxq_obj->rxq_ctrl = rxq_ctrl;
rxq_ctrl->is_hairpin = false;
rxq_ctrl->sh = priv->sh;
--
2.27.0
[PATCH] net/af_xdp: fix custom program loading with multiple queues
When the PMD is configured to load a custom XDP program, it sets
XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD flag to prevent libbpf from
loading its default XDP program. However, when queue_count is set to
greater than 1, this flag is only set for the first XSK socket but not
for subsequent XSK sockets. This causes XSK socket creation failure.
This commit ensures that XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD flag is
set for all XSK socket creations when custom XDP program is being used.
Fixes: 01fa83c94d7e ("net/af_xdp: workaround custom program loading")
Signed-off-by: Junxiao Shi
---
drivers/net/af_xdp/rte_eth_af_xdp.c | 23 ---
1 file changed, 12 insertions(+), 11 deletions(-)
diff --git a/drivers/net/af_xdp/rte_eth_af_xdp.c
b/drivers/net/af_xdp/rte_eth_af_xdp.c
index 65479138d3..9920f49870 100644
--- a/drivers/net/af_xdp/rte_eth_af_xdp.c
+++ b/drivers/net/af_xdp/rte_eth_af_xdp.c
@@ -1307,18 +1307,19 @@ xsk_configure(struct pmd_internals *internals, struct
pkt_rx_queue *rxq,
cfg.bind_flags |= XDP_USE_NEED_WAKEUP;
#endif
- if (strnlen(internals->prog_path, PATH_MAX) &&
- !internals->custom_prog_configured) {
- ret = load_custom_xdp_prog(internals->prog_path,
- internals->if_index,
- &internals->map);
- if (ret) {
- AF_XDP_LOG(ERR, "Failed to load custom XDP program
%s\n",
- internals->prog_path);
- goto out_umem;
+ if (strnlen(internals->prog_path, PATH_MAX)) {
+ if (!internals->custom_prog_configured) {
+ ret = load_custom_xdp_prog(internals->prog_path,
+ internals->if_index,
+ &internals->map);
+ if (ret) {
+ AF_XDP_LOG(ERR, "Failed to load custom XDP
program %s\n",
+ internals->prog_path);
+ goto out_umem;
+ }
+ internals->custom_prog_configured = 1;
}
- internals->custom_prog_configured = 1;
- cfg.libbpf_flags = XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD;
+ cfg.libbpf_flags |= XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD;
}
if (internals->shared_umem)
--
2.17.1
[PATCH] app/test-regex: fix incorrect number of matches
Depending on number of jobs specified on command line, part of the
data buffer may not get searched, resulting in incorrect number of
matches being reported.
Additional change to ensure the "All Matches" summary outputs the
correct match start locations in the supplied data buffer.
Fixes: de06137cb295 ("app/regex: add RegEx test application")
Cc: yuva...@mellanox.com
Cc: sta...@dpdk.org
Signed-off-by: Gerry Gribbon
Acked-by: Ori Kam
---
app/test-regex/main.c | 38 --
1 file changed, 28 insertions(+), 10 deletions(-)
diff --git a/app/test-regex/main.c b/app/test-regex/main.c
index ab8a3e56e7..7c014b2210 100644
--- a/app/test-regex/main.c
+++ b/app/test-regex/main.c
@@ -382,10 +382,13 @@ run_regex(void *args)
char *data_buf = rgxc->data_buf;
long data_len = rgxc->data_len;
long job_len = rgxc->job_len;
-
+ long remainder;
+ long act_job_len = 0;
+ bool last_job = false;
char *buf = NULL;
uint32_t actual_jobs = 0;
uint32_t i;
+ uint32_t job_id;
uint16_t qp_id;
uint16_t dev_id = 0;
uint8_t nb_matches;
@@ -456,9 +459,16 @@ run_regex(void *args)
/* Assign each mbuf with the data to handle. */
actual_jobs = 0;
pos = 0;
+ remainder = data_len % nb_jobs;
+
/* Allocate the jobs and assign each job with an mbuf. */
for (i = 0; (pos < data_len) && (i < nb_jobs) ; i++) {
- long act_job_len = RTE_MIN(job_len, data_len - pos);
+ act_job_len = RTE_MIN(job_len, data_len - pos);
+
+ if (i == (nb_jobs - 1)) {
+ last_job = true;
+ act_job_len += remainder;
+ }
ops[i] = rte_malloc(NULL, sizeof(*ops[0]) +
nb_max_matches *
@@ -478,7 +488,12 @@ run_regex(void *args)
if (ops[i]->mbuf) {
rte_pktmbuf_attach_extbuf(ops[i]->mbuf,
&buf[pos], 0, act_job_len, &shinfo);
- ops[i]->mbuf->data_len = job_len;
+
+ if (!last_job)
+ ops[i]->mbuf->data_len =
job_len;
+ else
+ ops[i]->mbuf->data_len =
act_job_len;
+
ops[i]->mbuf->pkt_len = act_job_len;
}
}
@@ -506,6 +521,9 @@ run_regex(void *args)
qp = &qps[qp_id];
qp->total_enqueue = 0;
qp->total_dequeue = 0;
+ /* Re-set user id after dequeue to match data in mbuf.
*/
+ for (job_id = 0 ; job_id < nb_jobs; job_id++)
+ qp->ops[job_id]->user_id = job_id;
}
do {
update = false;
@@ -551,10 +569,10 @@ run_regex(void *args)
for (qp_id = 0; qp_id < nb_qps; qp_id++) {
qp = &qps[qp_id];
time = (long double)qp->cycles / rte_get_timer_hz();
- printf("Core=%u QP=%u Job=%ld Bytes Time=%Lf sec Perf=%Lf "
+ printf("Core=%u QP=%u Job=%ld Bytes Last Job=%ld Bytes Time=%Lf
sec Perf=%Lf "
"Gbps\n", rte_lcore_id(), qp_id + qp_id_base,
- job_len, time,
- (((double)actual_jobs * job_len * nb_iterations * 8)
+ job_len, act_job_len, time,
+ (((double)data_len * nb_iterations * 8)
/ time) / 10.0);
}
@@ -587,10 +605,10 @@ run_regex(void *args)
qp->total_matches += nb_matches;
match = qp->ops[d_ind % actual_jobs]->matches;
for (i = 0; i < nb_matches; i++) {
- printf("start = %ld, len = %d, rule = %d\n",
- match->start_offset +
- d_ind * job_len,
- match->len, match->rule_id);
+ printf("start = %d, len = %d, rule = %d\n",
+ match->start_offset +
+ (int)(qp->ops[d_ind %
actual_jobs]->user_id * job_len),
+ match->len, match->rule_id);
match++;
}
}
--
2.25.1
RE: [PATCH] net/cnxk: fix error when compiled for x86
Hi maintainers,
Who can give acked-by for this patch?
May I know this patch can be merged into 22.03 or not?
> -Original Message-
> From: Rakesh Kudurumalla
> Sent: 2022年3月4日 22:24
> To: Nithin Dabilpuram ; Kiran Kumar K
> ; Sunil Kumar Kori ; Satha
> Rao
> Cc: dev@dpdk.org; jer...@marvell.com; Rakesh Kudurumalla
> ; sta...@dpdk.org; Yigit, Ferruh
>
> Subject: [PATCH] net/cnxk: fix error when compiled for x86
>
> fix error when compiled for x86 platform when compiled with optimization
> flag enabled
> error: ‘str’ may be used uninitialized in this function
> error: ‘frag_ptr’ may be used uninitialized in this function
>
> Bugzilla ID: 939
> Fixes: b526599020ef ("net/cnxk: fix build with GCC 12")
> Cc: sta...@dpdk.org
>
> Reported-by: Ferruh Yigit
> Signed-off-by: Rakesh Kudurumalla
> ---
[PATCH] net/mlx5: fix the implicit tag insertion in sample flow
The flow with sample action was split into two sub-flows,
and the implicit tag action with unique id was added in the prefix
sub-flow, the suffix sub-flow used the tag item to match with that
unique id, and the implicit set tag action was inserted next to
the sample action.
While there's either PUSH VLAN action or ENCAP action preceding the
sample action, implicit set tag action was added after PUSH VLAN or
ENCAP actions, causing flow creation failure due to rdma-core
does not support this action order.
This patch ensures the implicit set tag action is inserted before
either PUSH VLAN or encap action (if any) in the prefix sub-flow.
Fixes: 6a951567c159 ("net/mlx5: support E-Switch mirroring and jump in one
flow")
Cc: sta...@dpdk.org
Signed-off-by: Jiawei Wang
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5_flow.c | 63 ++--
1 file changed, 61 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
index a690e2d337..d26454a1b1 100644
--- a/drivers/net/mlx5/mlx5_flow.c
+++ b/drivers/net/mlx5/mlx5_flow.c
@@ -5831,8 +5831,9 @@ flow_sample_split_prep(struct rte_eth_dev *dev,
struct mlx5_rte_flow_item_tag *tag_mask;
struct rte_flow_action_jump *jump_action;
uint32_t tag_id = 0;
- int index;
int append_index = 0;
+ int set_tag_idx = -1;
+ int index;
int ret;
if (sample_action_pos < 0)
@@ -5841,6 +5842,52 @@ flow_sample_split_prep(struct rte_eth_dev *dev,
NULL, "invalid position of sample "
"action in list");
/* Prepare the actions for prefix and suffix flow. */
+ if (add_tag) {
+ /* Update the new added tag action index preceding
+* the PUSH_VLAN or ENCAP action.
+*/
+ const struct rte_flow_action_raw_encap *raw_encap;
+ const struct rte_flow_action *action = actions;
+ int encap_idx;
+ int action_idx = 0;
+ int raw_decap_idx = -1;
+ int push_vlan_idx = -1;
+ for (; action->type != RTE_FLOW_ACTION_TYPE_END; action++) {
+ switch (action->type) {
+ case RTE_FLOW_ACTION_TYPE_RAW_DECAP:
+ raw_decap_idx = action_idx;
+ break;
+ case RTE_FLOW_ACTION_TYPE_RAW_ENCAP:
+ raw_encap = action->conf;
+ if (raw_encap->size >
+ MLX5_ENCAPSULATION_DECISION_SIZE) {
+ encap_idx = raw_decap_idx != -1 ?
+ raw_decap_idx : action_idx;
+ if (encap_idx < sample_action_pos &&
+ push_vlan_idx == -1)
+ set_tag_idx = encap_idx;
+ }
+ break;
+ case RTE_FLOW_ACTION_TYPE_VXLAN_ENCAP:
+ case RTE_FLOW_ACTION_TYPE_NVGRE_ENCAP:
+ encap_idx = action_idx;
+ if (encap_idx < sample_action_pos &&
+ push_vlan_idx == -1)
+ set_tag_idx = encap_idx;
+ break;
+ case RTE_FLOW_ACTION_TYPE_OF_PUSH_VLAN:
+ case RTE_FLOW_ACTION_TYPE_OF_SET_VLAN_VID:
+ push_vlan_idx = action_idx;
+ if (push_vlan_idx < sample_action_pos)
+ set_tag_idx = action_idx;
+ break;
+ default:
+ break;
+ }
+ action_idx++;
+ }
+ }
+ /* Prepare the actions for prefix and suffix flow. */
if (qrss_action_pos >= 0 && qrss_action_pos < sample_action_pos) {
index = qrss_action_pos;
/* Put the preceding the Queue/RSS action into prefix flow. */
@@ -5857,6 +5904,14 @@ flow_sample_split_prep(struct rte_eth_dev *dev,
memcpy(actions_sfx, actions + qrss_action_pos,
sizeof(struct rte_flow_action));
actions_sfx++;
+ } else if (add_tag && set_tag_idx >= 0) {
+ if (set_tag_idx > 0)
+ memcpy(actions_pre, actions,
+ sizeof(struct rte_flow_action) * set_tag_idx);
+ memcpy(actions_pre + set_tag_idx + 1, actions + set_tag_idx,
+ sizeof(struct rte_flow_action) *
+ (sample_action_pos - set_tag_idx));
+ in
Re: [PATCH] usertools: add check for IOMMU support in dpdk-devbind
I agree with your suggestion. I would add new flag for noiommu and
submit v2 for the patch.
On Tue, Mar 8, 2022 at 5:57 PM Bruce Richardson
wrote:
>
> On Tue, Mar 08, 2022 at 05:49:01PM +0500, Fidaullah Noonari wrote:
> > binding with vfio driver, when IOMMU is disabled, causes program to crash.
> > this patch checks for IOMMU support, if it is disabled, changes vfio
> > into unsafe noiommu mode and prints a warning message.
> >
> > Signed-off-by: Fidaullah Noonari
> > ---
> > usertools/dpdk-devbind.py | 25 +
> > 1 file changed, 25 insertions(+)
> >
> > diff --git a/usertools/dpdk-devbind.py b/usertools/dpdk-devbind.py
> > index ace4627218..f42eead0c4 100755
> > --- a/usertools/dpdk-devbind.py
> > +++ b/usertools/dpdk-devbind.py
> > @@ -466,6 +466,26 @@ def unbind_all(dev_list, force=False):
> > unbind_one(d, force)
> >
> >
> > +def check_iommu():
> > +"""Check if IOMMU is enabled on system"""
> > +if len(os.listdir("/sys/class/iommu")) != 0:
> > +return True
> > +print("Warning: IOMMU support disabled")
> > +return False
> > +
> > +
> > +def enable_noiommu_mode():
> > +"""Enables the noiommu mode for vfio drivers"""
> > +filename = "/sys/module/vfio/parameters/enable_unsafe_noiommu_mode"
> > +try:
> > +f = open(filename, "w")
> > +except:
> > +sys.exit("Error: unable to enable noiommu mode, could not open
> > '%s'" % filename)
> > +f.write("1")
> > +f.close()
> > +print("Warning: enabling unsafe no IOMMU mode for vfio drivers")
> > +
> > +
> > def bind_all(dev_list, driver, force=False):
> > """Bind method, takes a list of device locations"""
> > global devices
> > @@ -492,6 +512,11 @@ def bind_all(dev_list, driver, force=False):
> > except ValueError as ex:
> > sys.exit(ex)
> >
> > +# check for IOMMU support
> > +if not check_iommu():
> > +if driver == "vfio-pci":
> > +enable_noiommu_mode()
> > +
> > for d in dev_list:
> > bind_one(d, driver, force)
> >
> I like the idea of doing this from the script, but I would rather it not
> happen by default, since no-iommu reduces the security of the system. In
> the normal case, I think it would be best if the script printed an error
> message asking the user to enable iommu. We can also then add a flag to
> allow the user to explicitly ask for no-iommu mode and take the risk.
>
> /Bruce
[PATCH] doc: mention MPRQ incompatibility with external buffers
Multi-Packet Rx queue uses PMD-managed buffers to store packets. These buffers are externally attached to user mbufs. This conflicts with the feature that allows using user-managed externally attached buffers in an application. Add the corresponding limitation to MLX5 documentation that MPRQ and external data buffers cannot be used together. Signed-off-by: Alexander Kozyrev --- doc/guides/nics/mlx5.rst | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst index a734d10d3f..9c633f1c07 100644 --- a/doc/guides/nics/mlx5.rst +++ b/doc/guides/nics/mlx5.rst @@ -260,7 +260,9 @@ Limitations ol_flags. As the mempool for the external buffer is managed by PMD, all the Rx mbufs must be freed before the device is closed. Otherwise, the mempool of the external buffers will be freed by PMD and the application which still - holds the external buffers may be corrupted. + holds the external buffers may be corrupted. User-managed mempools with + external pinned data buffers cannot be used in conjunction with MPRQ + since packets may be already attached to PMD-managed external buffers. - If Multi-Packet Rx queue is configured (``mprq_en``) and Rx CQE compression is enabled (``rxq_cqe_comp_en``) at the same time, RSS hash result is not fully -- 2.18.2
[PATCH] app/testpmd: fix flex item input link parsing
Testpmd reads flex item configuration from a JSON file.
Flex item input link description is stored in testpmd
flow item format. For example, `eth type is 0x0800`.
The item description is placed into a general testpmd CLI
flow rule command template and parsed to convert string into
flow item object.
The patch adds the `actions` section to the flow rule template.
Cc: sta...@dpdk.org
Fixes: 59f3a8acbcdb ("app/testpmd: add flex item commands")
Signed-off-by: Gregory Etelson
---
app/test-pmd/cmd_flex_item.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/app/test-pmd/cmd_flex_item.c b/app/test-pmd/cmd_flex_item.c
index 9050825a81..78a89c0f8a 100644
--- a/app/test-pmd/cmd_flex_item.c
+++ b/app/test-pmd/cmd_flex_item.c
@@ -133,7 +133,8 @@ flex_link_item_parse(const char *src, struct rte_flow_item
*item)
struct rte_flow_item *pattern;
struct rte_flow_action *actions;
- sprintf(flow_rule, "flow create 0 pattern %s / end", src);
+ sprintf(flow_rule,
+ "flow create 0 pattern %s / end actions drop / end", src);
src = flow_rule;
ret = flow_parse(src, (void *)data, sizeof(data),
&attr, &pattern, &actions);
--
2.35.1
[RFC,v3 0/2] vhost: support async dequeue data path
From: Xuan Ding The presence of asynchronous path allows applications to offload memory copies to DMA engine, so as to save CPU cycles and improve the copy performance. This patch set is a draft implementation for split ring in vhost async dequeue data path. The code is based on latest enqueue changes [1]. This patch set is a new design and implementation of [2]. Since dmadev was introduced in DPDK 21.11, to simplify application logics, this patch integrates dmadev in vhost. With dmadev integrated, vhost supports M:N mapping between vrings and DMA virtual channels. Specifically, one vring can use multiple different DMA channels and one DMA channel can be shared by multiple vrings at the same time. A new asynchronous dequeue function is introduced: 1) rte_vhost_async_try_dequeue_burst(int vid, uint16_t queue_id, struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count, int *nr_inflight, uint16_t dma_id, uint16_t vchan_id) Receive packets from the guest and offloads copies to DMA virtual channel. [1] https://mails.dpdk.org/archives/dev/2022-February/234555.html [2] https://mails.dpdk.org/archives/dev/2021-September/218591.html RFC v2->v3: * rebase to latest DPDK version RFC v1 -> v2: * fix one bug in example * rename vchan to vchan_id * check if dma_id and vchan_id valid * rework all the logs to new standard Xuan Ding (2): vhost: support async dequeue for split ring examples/vhost: support async dequeue data path doc/guides/sample_app_ug/vhost.rst | 9 +- examples/vhost/main.c | 292 +++-- examples/vhost/main.h | 35 +- examples/vhost/virtio_net.c| 16 +- lib/vhost/rte_vhost_async.h| 37 ++- lib/vhost/version.map | 1 + lib/vhost/vhost.h | 1 + lib/vhost/virtio_net.c | 504 + 8 files changed, 795 insertions(+), 100 deletions(-) -- 2.17.1
[RFC,v3 1/2] vhost: support async dequeue for split ring
From: Xuan Ding
This patch implements asynchronous dequeue data path for vhost split
ring, with dmadev library integrated.
Signed-off-by: Xuan Ding
Signed-off-by: Yuan Wang
---
lib/vhost/rte_vhost_async.h | 37 ++-
lib/vhost/version.map | 1 +
lib/vhost/vhost.h | 1 +
lib/vhost/virtio_net.c | 504
4 files changed, 541 insertions(+), 2 deletions(-)
diff --git a/lib/vhost/rte_vhost_async.h b/lib/vhost/rte_vhost_async.h
index f1293c6a9d..b6ab0b06a2 100644
--- a/lib/vhost/rte_vhost_async.h
+++ b/lib/vhost/rte_vhost_async.h
@@ -155,9 +155,9 @@ int rte_vhost_async_get_inflight(int vid, uint16_t
queue_id);
* @param count
* Size of the packet array
* @param dma_id
- * the identifier of DMA device
+ * The identifier of DMA device
* @param vchan_id
- * the identifier of virtual DMA channel
+ * The identifier of virtual DMA channel
* @return
* Number of packets returned
*/
@@ -187,6 +187,39 @@ uint16_t rte_vhost_clear_queue_thread_unsafe(int vid,
uint16_t queue_id,
__rte_experimental
int rte_vhost_async_dma_configure(int16_t dma_id, uint16_t vchan_id);
+/**
+ * This function tries to receive packets from the guest with offloading
+ * copies to the async channel. The packets that are transfer completed
+ * are returned in "pkts". The other packets that their copies are submitted to
+ * the async channel but not completed are called "in-flight packets".
+ * This function will not return in-flight packets until their copies are
+ * completed by the async channel.
+ *
+ * @param vid
+ * ID of vhost device to dequeue data
+ * @param queue_id
+ * ID of virtqueue to dequeue data
+ * @param mbuf_pool
+ * Mbuf_pool where host mbuf is allocated
+ * @param pkts
+ * Blank array to keep successfully dequeued packets
+ * @param count
+ * Size of the packet array
+ * @param nr_inflight
+ * The amount of in-flight packets. If error occurred, its value is set to -1.
+ * @param dma_id
+ * The identifier of DMA device
+ * @param vchan_id
+ * The identifier of virtual DMA channel
+ * @return
+ * Number of successfully dequeued packets
+ */
+__rte_experimental
+uint16_t
+rte_vhost_async_try_dequeue_burst(int vid, uint16_t queue_id,
+ struct rte_mempool *mbuf_pool, struct rte_mbuf **pkts, uint16_t count,
+ int *nr_inflight, uint16_t dma_id, uint16_t vchan_id);
+
#ifdef __cplusplus
}
#endif
diff --git a/lib/vhost/version.map b/lib/vhost/version.map
index 0a66c5840c..968d6d4290 100644
--- a/lib/vhost/version.map
+++ b/lib/vhost/version.map
@@ -87,6 +87,7 @@ EXPERIMENTAL {
# added in 22.03
rte_vhost_async_dma_configure;
+ rte_vhost_async_try_dequeue_burst;
};
INTERNAL {
diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index a9edc271aa..3799d41089 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -178,6 +178,7 @@ extern struct async_dma_info
dma_copy_track[RTE_DMADEV_DEFAULT_MAX];
*/
struct async_inflight_info {
struct rte_mbuf *mbuf;
+ struct virtio_net_hdr nethdr;
uint16_t descs; /* num of descs inflight */
uint16_t nr_buffers; /* num of buffers inflight for packed ring */
};
diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c
index 5f432b0d77..3816caca79 100644
--- a/lib/vhost/virtio_net.c
+++ b/lib/vhost/virtio_net.c
@@ -3141,3 +3141,507 @@ rte_vhost_dequeue_burst(int vid, uint16_t queue_id,
return count;
}
+
+static __rte_always_inline int
+async_desc_to_mbuf_seg(struct virtio_net *dev, struct vhost_virtqueue *vq,
+ struct rte_mbuf *m, uint32_t mbuf_offset,
+ uint64_t buf_iova, uint32_t cpy_len)
+{
+ struct vhost_async *async = vq->async;
+ uint64_t mapped_len;
+ uint32_t buf_offset = 0;
+ void *host_iova;
+
+ while (cpy_len) {
+ host_iova = (void *)(uintptr_t)gpa_to_first_hpa(dev,
+ buf_iova + buf_offset, cpy_len,
+ &mapped_len);
+ if (unlikely(!host_iova)) {
+ VHOST_LOG_DATA(ERR, "(%s) %s: failed to get
host_iova.\n",
+ dev->ifname, __func__);
+ return -1;
+ }
+
+ if (unlikely(async_iter_add_iovec(dev, async, host_iova,
+ (void *)(uintptr_t)rte_pktmbuf_iova_offset(m,
mbuf_offset),
+ (size_t)mapped_len)))
+ return -1;
+
+ cpy_len -= (uint32_t)mapped_len;
+ mbuf_offset += (uint32_t)mapped_len;
+ buf_offset += (uint32_t)mapped_len;
+ }
+
+ return 0;
+}
+
+static __rte_always_inline int
+async_desc_to_mbuf(struct virtio_net *dev, struct vhost_virtqueue *vq,
+ struct buf_vector *buf_vec, uint16_t nr_vec,
+ struct rte_mbuf *m, struct rte_mempool *mbuf_pool,
+ struct virtio_ne
[RFC,v3 2/2] examples/vhost: support async dequeue data path
From: Xuan Ding
This patch adds the use case for async dequeue API. Vswitch can
leverage DMA device to accelerate vhost async dequeue path.
Signed-off-by: Wenwu Ma
Signed-off-by: Yuan Wang
Signed-off-by: Xuan Ding
---
doc/guides/sample_app_ug/vhost.rst | 9 +-
examples/vhost/main.c | 292 -
examples/vhost/main.h | 35 +++-
examples/vhost/virtio_net.c| 16 +-
4 files changed, 254 insertions(+), 98 deletions(-)
diff --git a/doc/guides/sample_app_ug/vhost.rst
b/doc/guides/sample_app_ug/vhost.rst
index a6ce4bc8ac..09db965e70 100644
--- a/doc/guides/sample_app_ug/vhost.rst
+++ b/doc/guides/sample_app_ug/vhost.rst
@@ -169,9 +169,12 @@ demonstrates how to use the async vhost APIs. It's used in
combination with dmas
**--dmas**
This parameter is used to specify the assigned DMA device of a vhost device.
Async vhost-user net driver will be used if --dmas is set. For example
---dmas [txd0@00:04.0,txd1@00:04.1] means use DMA channel 00:04.0 for vhost
-device 0 enqueue operation and use DMA channel 00:04.1 for vhost device 1
-enqueue operation.
+--dmas [txd0@00:04.0,txd1@00:04.1,rxd0@00:04.2,rxd1@00:04.3] means use
+DMA channel 00:04.0/00:04.2 for vhost device 0 enqueue/dequeue operation
+and use DMA channel 00:04.1/00:04.3 for vhost device 1 enqueue/dequeue
+operation. The index of the device corresponds to the socket file in order,
+that means vhost device 0 is created through the first socket file, vhost
+device 1 is created through the second socket file, and so on.
Common Issues
-
diff --git a/examples/vhost/main.c b/examples/vhost/main.c
index d94fabb060..d26e40ab73 100644
--- a/examples/vhost/main.c
+++ b/examples/vhost/main.c
@@ -63,6 +63,9 @@
#define DMA_RING_SIZE 4096
+#define ASYNC_ENQUEUE_VHOST 1
+#define ASYNC_DEQUEUE_VHOST 2
+
/* number of mbufs in all pools - if specified on command-line. */
static int total_num_mbufs = NUM_MBUFS_DEFAULT;
@@ -116,6 +119,8 @@ static uint32_t burst_rx_retry_num = BURST_RX_RETRIES;
static char *socket_files;
static int nb_sockets;
+static struct vhost_queue_ops vdev_queue_ops[RTE_MAX_VHOST_DEVICE];
+
/* empty VMDq configuration structure. Filled in programmatically */
static struct rte_eth_conf vmdq_conf_default = {
.rxmode = {
@@ -205,6 +210,18 @@ struct vhost_bufftable *vhost_txbuff[RTE_MAX_LCORE *
RTE_MAX_VHOST_DEVICE];
#define MBUF_TABLE_DRAIN_TSC ((rte_get_tsc_hz() + US_PER_S - 1) \
/ US_PER_S * BURST_TX_DRAIN_US)
+static int vid2socketid[RTE_MAX_VHOST_DEVICE];
+
+static uint32_t get_async_flag_by_socketid(int socketid)
+{
+ return dma_bind[socketid].async_flag;
+}
+
+static void init_vid2socketid_array(int vid, int socketid)
+{
+ vid2socketid[vid] = socketid;
+}
+
static inline bool
is_dma_configured(int16_t dev_id)
{
@@ -224,7 +241,7 @@ open_dma(const char *value)
char *addrs = input;
char *ptrs[2];
char *start, *end, *substr;
- int64_t vid;
+ int64_t socketid, vring_id;
struct rte_dma_info info;
struct rte_dma_conf dev_config = { .nb_vchans = 1 };
@@ -262,7 +279,9 @@ open_dma(const char *value)
while (i < args_nr) {
char *arg_temp = dma_arg[i];
+ char *txd, *rxd;
uint8_t sub_nr;
+ int async_flag;
sub_nr = rte_strsplit(arg_temp, strlen(arg_temp), ptrs, 2, '@');
if (sub_nr != 2) {
@@ -270,14 +289,23 @@ open_dma(const char *value)
goto out;
}
- start = strstr(ptrs[0], "txd");
- if (start == NULL) {
+ txd = strstr(ptrs[0], "txd");
+ rxd = strstr(ptrs[0], "rxd");
+ if (txd) {
+ start = txd;
+ vring_id = VIRTIO_RXQ;
+ async_flag = ASYNC_ENQUEUE_VHOST;
+ } else if (rxd) {
+ start = rxd;
+ vring_id = VIRTIO_TXQ;
+ async_flag = ASYNC_DEQUEUE_VHOST;
+ } else {
ret = -1;
goto out;
}
start += 3;
- vid = strtol(start, &end, 0);
+ socketid = strtol(start, &end, 0);
if (end == start) {
ret = -1;
goto out;
@@ -338,7 +366,8 @@ open_dma(const char *value)
dmas_id[dma_count++] = dev_id;
done:
- (dma_info + vid)->dmas[VIRTIO_RXQ].dev_id = dev_id;
+ (dma_info + socketid)->dmas[vring_id].dev_id = dev_id;
+ (dma_info + socketid)->async_flag |= async_flag;
i++;
}
out:
@@ -990,13 +1019,13 @@ complete_async_pkts(struct vhost_dev *vdev)
{
struct rte_mbuf *p_cpl[MAX_PKT_BURST];
uint16_t complete_count;
- int16_t dma_id = dma_
RE: [PATCH v2 2/2] app/testpmd:fix testpmd quit failure
> -Original Message-
> From: Wu, WenxuanX
> Sent: 2022年3月9日 11:07
> To: Yigit, Ferruh ; Li, Xiaoyun
> ;
> dev@dpdk.org
> Cc: sta...@dpdk.org
> Subject: RE: [PATCH v2 2/2] app/testpmd:fix testpmd quit failure
>
>
>
> > -Original Message-
> > From: Yigit, Ferruh
> > Sent: 2022年3月5日 0:16
> > To: Wu, WenxuanX ; Li, Xiaoyun
> > ; dev@dpdk.org
> > Cc: sta...@dpdk.org
> > Subject: Re: [PATCH v2 2/2] app/testpmd:fix testpmd quit failure
> >
> > On 3/3/2022 1:22 PM, Wu, WenxuanX wrote:
> >
> > moved down, please don't top post
> >
> > >> -Original Message-
> > >> From: Wu, WenxuanX
> > >> Sent: 2022年2月23日 19:33
> > >> To: Li, Xiaoyun ; Yigit, Ferruh
> > >> ; dev@dpdk.org
> > >> Cc: Wu, WenxuanX ; sta...@dpdk.org
> > >> Subject: [PATCH v2 2/2] app/testpmd:fix testpmd quit failure
> > >>
> > >> From: wenxuan wu
> > >>
> > >> When testpmd start ed with 1 pf and 2 vfs, testpmd quited while vfs
> > >> were still alive would result in failure. Root cause is that pf had
> > >> been released already but vfs were still accessing by func
> > >> rte_eth_dev_info_get, which would result in heap-free-after-use error.
> > >>
> > >> By quitting our ports in reverse order to avoid this.And the order
> > >> is guaranteed that vf are created after pfs.
> > >>
> > >> Fixes: d3a274ce9dee ("app/testpmd: handle SIGINT and SIGTERM")
> > >> Cc: sta...@dpdk.org
> > >>
> > >> Signed-off-by: wenxuan wu
> > >> ---
> > >> app/test-pmd/testpmd.c | 4 ++--
> > >> 1 file changed, 2 insertions(+), 2 deletions(-)
> > >>
> > >> diff --git a/app/test-pmd/testpmd.c b/app/test-pmd/testpmd.c index
> > >> e1da961311..698b6d8cc4 100644
> > >> --- a/app/test-pmd/testpmd.c
> > >> +++ b/app/test-pmd/testpmd.c
> > >> @@ -3384,12 +3384,12 @@ pmd_test_exit(void) #endif
> > >> if (ports != NULL) {
> > >> no_link_check = 1;
> > >> -RTE_ETH_FOREACH_DEV(pt_id) {
> > >> +RTE_ETH_FOREACH_DEV_REVERSE(pt_id) {
> > >> printf("\nStopping port %d...\n", pt_id);
> > >> fflush(stdout);
> > >> stop_port(pt_id);
> > >> }
> > >> -RTE_ETH_FOREACH_DEV(pt_id) {
> > >> +RTE_ETH_FOREACH_DEV_REVERSE(pt_id) {
> > >> printf("\nShutting down port %d...\n", pt_id);
> > >> fflush(stdout);
> > >> close_port(pt_id);
> > >> --
> > >> 2.25.1
> > >
> > >
> > > I found this meaning in DPDK testplan.
> > > Note that currently hot-plugging of representor ports is not
> > > supported so all
> > the required representors must be specified on the creation of the PF
> > or the trusted VF.
> > > When testpmd is started with pf and vf representors, the order of
> > representor is determined on creation. So it is guaranteed that ,pf is
> > beneath the vf representors, we implemented in a reverse way is
> > acceptable just at present, depends on when the hot-plugging of
> representor is supported.
> > >
> >
> > The patch mentions from PF and VFs, and now you are referring to port
> > representor.
> > Is the problem related to VF or port representor.
> >
> > For both, VF and port reporesentor should be closed before PF, that
> > part is OK. My comment is if reversing port id traverse will fix the
> > issue or do we need more complex solution.
> > Like have APIs to get VF and representor ports from a given port id,
> > and free them first.
The problem is that when I attempted to use a func like
get_representor_info(pid,info); I didn't found this func implemented by driver ,
so I can not get the type of port(VF or PF ) directly by
get_representor_info(pid,info),especially when the connected driver is i40e,
representor_info_get occurred below.
./drivers/net/mlx5/mlx5.c: .representor_info_get =
mlx5_representor_info_get,
./drivers/net/mlx5/mlx5.c: .representor_info_get =
mlx5_representor_info_get,
./drivers/net/mlx5/mlx5.c: .representor_info_get =
mlx5_representor_info_get,
./drivers/net/mlx5/mlx5.h:int mlx5_representor_info_get(struct rte_eth_dev *dev,
./drivers/net/mlx5/mlx5_ethdev.c:mlx5_representor_info_get(struct rte_eth_dev
*dev,
./drivers/net/sfc/sfc_ethdev.c:sfc_representor_info_get(struct rte_eth_dev *dev,
./drivers/net/sfc/sfc_ethdev.c: .representor_info_get =
sfc_representor_info_get,
./app/test-pmd/cmdline.c: ret =
rte_eth_representor_info_get(res->cmd_pid, NULL);
./app/test-pmd/cmdline.c: ret =
rte_eth_representor_info_get(res->cmd_pid, info);
./app/test-pmd/testpmd.c: ret = rte_eth_representor_info_get(pi,&info);
./lib/ethdev/version.map: rte_eth_representor_info_get;
./lib/ethdev/rte_ethdev.h:int rte_eth_representor_info_get(uint16_t port_id,
./lib/ethdev/ethdev_driver.h:typedef int (*eth_representor_info_get_t)(struct
rte_eth_dev *dev,
./lib/ethdev/ethdev_driver.h: eth_representor_info_get_t representor_info_get;
./lib/ethdev/ethdev_driver.
[PATCH] eal/windows: fix data race when creating threads
create lcore worker threads suspended and then subsequently resume to
allow &lcore_config[i].thread_id be stored before eal_thread_loop
execution.
Fixes: 53ffd9f080fc ("eal/windows: add minimum viable code")
Cc: anand.ra...@intel.com
Cc: sta...@dpdk.org
Signed-off-by: Tyler Retzlaff
---
lib/eal/windows/eal_thread.c | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/lib/eal/windows/eal_thread.c b/lib/eal/windows/eal_thread.c
index 54fa93f..ff84cb4 100644
--- a/lib/eal/windows/eal_thread.c
+++ b/lib/eal/windows/eal_thread.c
@@ -150,13 +150,18 @@
th = CreateThread(NULL, 0,
(LPTHREAD_START_ROUTINE)(ULONG_PTR)eal_thread_loop,
- NULL, 0, (LPDWORD)thread);
+ NULL, CREATE_SUSPENDED,
(LPDWORD)thread);
if (!th)
return -1;
SetPriorityClass(GetCurrentProcess(), NORMAL_PRIORITY_CLASS);
SetThreadPriority(th, THREAD_PRIORITY_NORMAL);
+ if (ResumeThread(th) == (DWORD)-1) {
+ (void)CloseHandle(th);
+ return -1;
+ }
+
return 0;
}
--
1.8.3.1
Re: [PATCH 1/3] net/qede: fix Tx callback completion routine
On Fri, Mar 4, 2022 at 5:38 PM Devendra Singh Rawat
wrote:
>
> Tx completion routine was first incrementing no. of free slots in Tx
> ring and then freeing corresponding mbufs in bulk. In some situations
> no. of mbufs freed were less than no. of Tx ring slots freed. This
> caused TX ring to get into an inconsistent state and ultimately
> application fails to transmit further traffic.
>
> The fix first updates Tx ring SW consumer index, then increments Tx ring
> free slot no. and finally frees the mbuf, this is done in a single
> iteration of loop.
>
> Fixes: 2c41740bf19e ("net/qede: get consumer index once")
> Fixes: 4996b959cde6 ("net/qede: free packets in bulk")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Devendra Singh Rawat
> Signed-off-by: Rasesh Mody
Updated the git commits as follows and applied series to
dpdk-next-net-mrvl/for-next-net. Thanks
--
net/qede: fix max Rx packet length calculation
Size of CRC is not added to max_rx_pktlen, due to this bigger sized
packets(size 1480, 1490 1500) are being dropped.
This fix adds RTE_ETHER_CRC_LEN to max_rx_pktlen.
Fixes: 1bb4a528c41f ("ethdev: fix max Rx packet length")
Cc: sta...@dpdk.org
Signed-off-by: Devendra Singh Rawat
Signed-off-by: Rasesh Mody
---
net/qede: fix Rx callback
qede_alloc_rx_bulk_mbufs() was trimming the number of requested
mbufs count to QEDE_MAX_BULK_ALLOC_COUNT.
The RX callback was ignorant of this trimming and it was always
resetting the number of empty Rx BD ring slots to 0.
This resulted in Rx BD ring getting into an inconsistent
state and ultimately the application fails to receive any traffic.
The fix trims the number of requested mbufs count before
making call to qede_alloc_rx_bulk_mbufs().
After qede_alloc_rx_bulk_mbufs() returns successfully, the
number of empty Rx BD ring slots are decremented by the
correct count.
Fixes: 8f2312474529 ("net/qede: fix performance bottleneck in Rx path")
Cc: sta...@dpdk.org
Signed-off-by: Devendra Singh Rawat
Signed-off-by: Rasesh Mody
--
net/qede: fix Tx callback completion routine
Tx completion routine was first incrementing the number of free slots in
Tx ring and then freeing corresponding mbufs in bulk.
In some situations, the number of mbufs freed were less than
number of Tx ring slots freed. This caused Tx ring to get into an
inconsistent state and ultimately application fails to transmit
further traffic.
The fix first updates Tx ring SW consumer index, then increments
Tx ring free slot number and finally frees the mbuf,
this is done in a single iteration of loop.
Fixes: 2c41740bf19e ("net/qede: get consumer index once")
Fixes: 4996b959cde6 ("net/qede: free packets in bulk")
Cc: sta...@dpdk.org
Signed-off-by: Devendra Singh Rawat
Signed-off-by: Rasesh Mody
---
> ---
> drivers/net/qede/qede_rxtx.c | 79 +++-
> 1 file changed, 33 insertions(+), 46 deletions(-)
>
> diff --git a/drivers/net/qede/qede_rxtx.c b/drivers/net/qede/qede_rxtx.c
> index 911bb1a260..0c52568180 100644
> --- a/drivers/net/qede/qede_rxtx.c
> +++ b/drivers/net/qede/qede_rxtx.c
> @@ -885,68 +885,55 @@ qede_tx_queue_start(struct rte_eth_dev *eth_dev,
> uint16_t tx_queue_id)
> }
>
> static inline void
> -qede_process_tx_compl(__rte_unused struct ecore_dev *edev,
> - struct qede_tx_queue *txq)
> +qede_free_tx_pkt(struct qede_tx_queue *txq)
> {
> - uint16_t hw_bd_cons;
> - uint16_t sw_tx_cons;
> - uint16_t remaining;
> - uint16_t mask;
> struct rte_mbuf *mbuf;
> uint16_t nb_segs;
> uint16_t idx;
> - uint16_t first_idx;
> -
> - rte_compiler_barrier();
> - rte_prefetch0(txq->hw_cons_ptr);
> - sw_tx_cons = ecore_chain_get_cons_idx(&txq->tx_pbl);
> - hw_bd_cons = rte_le_to_cpu_16(*txq->hw_cons_ptr);
> -#ifdef RTE_LIBRTE_QEDE_DEBUG_TX
> - PMD_TX_LOG(DEBUG, txq, "Tx Completions = %u\n",
> - abs(hw_bd_cons - sw_tx_cons));
> -#endif
> -
> - mask = NUM_TX_BDS(txq);
> - idx = txq->sw_tx_cons & mask;
>
> - remaining = hw_bd_cons - sw_tx_cons;
> - txq->nb_tx_avail += remaining;
> - first_idx = idx;
> -
> - while (remaining) {
> - mbuf = txq->sw_tx_ring[idx];
> - RTE_ASSERT(mbuf);
> + idx = TX_CONS(txq);
> + mbuf = txq->sw_tx_ring[idx];
> + if (mbuf) {
> nb_segs = mbuf->nb_segs;
> - remaining -= nb_segs;
> -
> - /* Prefetch the next mbuf. Note that at least the last 4 mbufs
> -* that are prefetched will not be used in the current call.
> -*/
> - rte_mbuf_prefetch_part1(txq->sw_tx_ring[(idx + 4) & mask]);
> - rte_mbuf_prefetch_part2(txq->sw_tx_ring[(idx + 4) & mask]);
> -
> PMD_TX_LOG(DEBUG, txq, "nb_segs to
