Re: [dpdk-dev] [PATCH v3 1/4] vhost: prevent features to be changed while device is running
On Wed, Dec 06, 2017 at 10:20:45AM +0100, Maxime Coquelin wrote:
> As section 2.2 of the Virtio spec states about features
> negotiation:
> "During device initialization, the driver reads this and tells
> the device the subset that it accepts. The only way to
> renegotiate is to reset the device."
>
> This patch implements a check to prevent illegal features change
> while the device is running.
>
> One exception is the VHOST_F_LOG_ALL feature bit, which is enabled
> when live-migration is initiated. But this feature is not negotiated
> with the Virtio driver, but directly with the Vhost master.
>
> Signed-off-by: Maxime Coquelin
> ---
> lib/librte_vhost/vhost_user.c | 17 -
> 1 file changed, 16 insertions(+), 1 deletion(-)
>
> diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
> index f4c7ce462..2d86c0ca8 100644
> --- a/lib/librte_vhost/vhost_user.c
> +++ b/lib/librte_vhost/vhost_user.c
> @@ -183,7 +183,22 @@ vhost_user_set_features(struct virtio_net *dev, uint64_t
> features)
> return -1;
> }
>
> - if ((dev->flags & VIRTIO_DEV_RUNNING) && dev->features != features) {
> + if (dev->features == features)
> + return 0;
> +
We couldn't return directly when dev->features == features.
Otherwise, if the features provided by virtio driver is 0,
dev->vhost_hlen won't get a chance to be initialized.
Best regards,
Tiwei Bie
Re: [dpdk-dev] A question about GRO neighbor packet matching
Hi all,
> -Original Message-
> From: Stephen Hemminger [mailto:step...@networkplumber.org]
> Sent: Thursday, December 7, 2017 9:02 AM
> To: Ananyev, Konstantin
> Cc: Ilya Matveychikov ; dev@dpdk.org; Hu, Jiayu
>
> Subject: Re: [dpdk-dev] A question about GRO neighbor packet matching
>
> On Thu, 7 Dec 2017 00:19:46 +
> "Ananyev, Konstantin" wrote:
>
> > > -Original Message-
> > > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Stephen
> Hemminger
> > > Sent: Wednesday, December 6, 2017 11:16 PM
> > > To: Ilya Matveychikov
> > > Cc: dev@dpdk.org; Hu, Jiayu
> > > Subject: Re: [dpdk-dev] A question about GRO neighbor packet matching
> > >
> > > On Wed, 6 Dec 2017 22:38:12 +0400
> > > Ilya Matveychikov wrote:
> > >
> > > > > On Dec 6, 2017, at 10:12 PM, Stephen Hemminger
> wrote:
> > > > >
> > > > > On Wed, 6 Dec 2017 18:02:21 +0400
> > > > > Ilya Matveychikov wrote:
> > > > >
> > > > >> Hello all,
> > > > >>
> > > > >>
> > > > >> My question is about neighbor packet matching algorithm for TCP. Is
> it
> > > > >> correct to expect that IP packets should have continuous ID
> enumeration
> > > > >> (i.e. iph-next.id = iph-prev.id + 1)?
> > > > >
> > > > >
> > > > > No.
> > > > >
> > > > >> ~~~
> > > > >> lib/librte_gro/gro_tcp4.c:check_seq_option()
> > > > >> ...
> > > > >> /* check if the two packets are neighbors */
> > > > >> tcp_dl0 = pkt0->pkt_len - pkt0->l2_len - pkt0->l3_len -
> tcp_hl0;
> > > > >> if ((sent_seq == (item->sent_seq + tcp_dl0)) &&
> > > > >> (ip_id == (item->ip_id + 1)))
> > > > >> /* append the new packet */
> > > > >> return 1;
> > > > >> else if (((sent_seq + tcp_dl) == item->sent_seq) &&
> > > > >> ((ip_id + item->nb_merged) == item->ip_id))
> > > > >> /* pre-pend the new packet */
> > > > >> return -1;
> > > > >> else
> > > > >> return 0;
> > > > >> ~~~
> > > > >>
> > > > >> As per RFC791:
> > > > >>
> > > > >> Identification: 16 bits
> > > > >>
> > > > >>An identifying value assigned by the sender to aid in assembling
> > > > >> the
> > > > >>fragments of a datagram.
> > > > >
> > > > > The IP header id is meaningless in most TCP sessions.
> > > > > Good TCP implementations use PMTU discovery which sets the Don't
> Fragment bit.
> > > > > With DF, the IP id is unused (since no fragmentation).
> > > > > Many implementations just send 0 since generating unique IP id
> requires an
> > > > > atomic operation which is potential bottleneck.
> > > >
> > > > So, is my question correct and the code is wrong?
> > > >
> > >
> > > Yes. This code is wrong on several areas.
> > > * The ip_id on TCP flows is irrelevant.
@Stephen and @Konstantin:
In the latest linux, its GRO supports two kinds of IP ID: fixed or incremental.
You can see the commit 1530545ed64b42e87acb43c0c16401bd1ebae6bf.
It uses "skb->is_atomic" to reflect if the IP ID is ignored. Linux GRO only
checks
IP ID for the packets which are non-atomic (is_atomic is 0), and these packets
use
incremental IP ID. Others, which are atomic, use fixed IP ID and Linux doesn't
check
their IP ID.
You can see the codes in tcp_offload.c:
if (NAPI_GRO_CB(p)->flush_id != 1 || NAPI_GRO_CB(p)->count != 1 ||
!NAPI_GRO_CB(p)->is_atomic)
flush |= NAPI_GRO_CB(p)->flush_id;
In af_inet.c, is_atomic is set:
NAPI_GRO_CB(skb)->is_atomic = !!(iph->frag_off & htons(IP_DF));
I haven't figured out which kind of packets are set to is_atomic in Linux.
Maybe Linux has followed RFC 6864. I need to investigate further.
Especially, we plan to support tunneled GRO. The outer IP ID will encounter
the same issue.
If you have any suggestions, that will be highly appreciated.
> > > * packet should only be merged if TCP flags are the same.
@Stephen, we do check TCP flags when decide if two packets can be merged.
Thanks,
Jiayu
> > >
> > >
> > > The author should look at Linux net/ipv4/tcp_offload.c
> >
> > As I remember, linux GRO implementation *does* require that IP IDs
> > of the merging packets to be continuous.
> >
> > net/ipv4/af_inet.c:
> > static struct sk_buff **inet_gro_receive(struct sk_buff **head,
> > struct sk_buff *skb)
> > {
> > ...
> > id = ntohl(*(__be32 *)&iph->id);
> > flush = (u16)((ntohl(*(__be32 *)iph) ^ skb_gro_len(skb)) | (id &
> ~IP_DF));
> > id >>= 16;
> >
> > ...
> >
> > NAPI_GRO_CB(p)->flush_id =
> > ((u16)(ntohs(iph2->id) + NAPI_GRO_CB(p)->count)
> ^ id);
> > NAPI_GRO_CB(p)->flush |= flush;
> >
> >
> > And then at net/ipv4/tcp_offload.c:
> > struct sk_buff **tcp_gro_receive(struct sk_buff **head, struct sk_buff *skb)
> > {
> > ...
> > /* Include the IP ID check below from the inner most IP hdr */
> > flush = NAPI_GRO_CB(p)->flush | NAPI_GRO_CB(p)->flush_id;
> > ...
> > if (flush
Re: [dpdk-dev] [PATCH v3 1/4] vhost: prevent features to be changed while device is running
On 12/07/2017 09:08 AM, Tiwei Bie wrote:
On Wed, Dec 06, 2017 at 10:20:45AM +0100, Maxime Coquelin wrote:
As section 2.2 of the Virtio spec states about features
negotiation:
"During device initialization, the driver reads this and tells
the device the subset that it accepts. The only way to
renegotiate is to reset the device."
This patch implements a check to prevent illegal features change
while the device is running.
One exception is the VHOST_F_LOG_ALL feature bit, which is enabled
when live-migration is initiated. But this feature is not negotiated
with the Virtio driver, but directly with the Vhost master.
Signed-off-by: Maxime Coquelin
---
lib/librte_vhost/vhost_user.c | 17 -
1 file changed, 16 insertions(+), 1 deletion(-)
diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
index f4c7ce462..2d86c0ca8 100644
--- a/lib/librte_vhost/vhost_user.c
+++ b/lib/librte_vhost/vhost_user.c
@@ -183,7 +183,22 @@ vhost_user_set_features(struct virtio_net *dev, uint64_t
features)
return -1;
}
- if ((dev->flags & VIRTIO_DEV_RUNNING) && dev->features != features) {
+ if (dev->features == features)
+ return 0;
+
We couldn't return directly when dev->features == features.
Otherwise, if the features provided by virtio driver is 0,
dev->vhost_hlen won't get a chance to be initialized.
Good catch.
Either we do :
if ((dev->features == features) && dev->vhost_len)
return 0;
Or we could initialize dev->vhost_len to sizeof(struct virtio_net_hdr)
at alloc time.
I prefer the former, what do you think?
Thanks,
Maxime
Best regards,
Tiwei Bie
[dpdk-dev] [PATCH v5 0/2] net/i40e: support input set configuration
The patchset adds support RSS/FDIR/FDIR flexible payload input set configuration for some pctype. v5 changes: - Add DPDK version for new private API. v4 changes: - Change testpmd command token and print info. v3 changes: - Add support reset RSS/FDIR/FDIR flexible payload input set for some pctype. v2 changes: - Use 'static inline' to replace 'inline'. - Add support get status for some filed index of input set. Beilei Xing (2): net/i40e: support input set configuration app/testpmd: add configuration for input set app/test-pmd/cmdline.c| 237 ++ drivers/net/i40e/rte_pmd_i40e.c | 141 ++ drivers/net/i40e/rte_pmd_i40e.h | 138 + drivers/net/i40e/rte_pmd_i40e_version.map | 10 ++ 4 files changed, 526 insertions(+) -- 2.5.5
[dpdk-dev] [PATCH v5 1/2] net/i40e: support input set configuration
This patch supports getting/setting input set info for
RSS/FDIR/FDIR flexible payload.
Also add some helper functions for input set configuration.
Signed-off-by: Beilei Xing
---
drivers/net/i40e/rte_pmd_i40e.c | 141 ++
drivers/net/i40e/rte_pmd_i40e.h | 138 +
drivers/net/i40e/rte_pmd_i40e_version.map | 10 +++
3 files changed, 289 insertions(+)
diff --git a/drivers/net/i40e/rte_pmd_i40e.c b/drivers/net/i40e/rte_pmd_i40e.c
index aeb92af..1f95f91 100644
--- a/drivers/net/i40e/rte_pmd_i40e.c
+++ b/drivers/net/i40e/rte_pmd_i40e.c
@@ -2985,3 +2985,144 @@ int rte_pmd_i40e_flow_add_del_packet_template(
return i40e_flow_add_del_fdir_filter(dev, &filter_conf, add);
}
+
+int
+rte_pmd_i40e_inset_get(uint16_t port, uint8_t pctype,
+ struct rte_pmd_i40e_inset *inset,
+ enum rte_pmd_i40e_inset_type inset_type)
+{
+ struct rte_eth_dev *dev;
+ struct i40e_hw *hw;
+ uint64_t inset_reg;
+ uint32_t mask_reg[2];
+ int i;
+
+ RTE_ETH_VALID_PORTID_OR_ERR_RET(port, -ENODEV);
+
+ dev = &rte_eth_devices[port];
+
+ if (!is_i40e_supported(dev))
+ return -ENOTSUP;
+
+ if (pctype > 63)
+ return -EINVAL;
+
+ hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+ memset(inset, 0, sizeof(struct rte_pmd_i40e_inset));
+
+ switch (inset_type) {
+ case INSET_HASH:
+ /* Get input set */
+ inset_reg =
+ i40e_read_rx_ctl(hw, I40E_GLQF_HASH_INSET(1, pctype));
+ inset_reg <<= I40E_32_BIT_WIDTH;
+ inset_reg |=
+ i40e_read_rx_ctl(hw, I40E_GLQF_HASH_INSET(0, pctype));
+ /* Get field mask */
+ mask_reg[0] =
+ i40e_read_rx_ctl(hw, I40E_GLQF_HASH_MSK(0, pctype));
+ mask_reg[1] =
+ i40e_read_rx_ctl(hw, I40E_GLQF_HASH_MSK(1, pctype));
+ break;
+ case INSET_FDIR:
+ inset_reg =
+ i40e_read_rx_ctl(hw, I40E_PRTQF_FD_INSET(pctype, 1));
+ inset_reg <<= I40E_32_BIT_WIDTH;
+ inset_reg |=
+ i40e_read_rx_ctl(hw, I40E_PRTQF_FD_INSET(pctype, 0));
+ mask_reg[0] =
+ i40e_read_rx_ctl(hw, I40E_GLQF_FD_MSK(0, pctype));
+ mask_reg[1] =
+ i40e_read_rx_ctl(hw, I40E_GLQF_FD_MSK(1, pctype));
+ break;
+ case INSET_FDIR_FLX:
+ inset_reg =
+ i40e_read_rx_ctl(hw, I40E_PRTQF_FD_FLXINSET(pctype));
+ mask_reg[0] =
+ i40e_read_rx_ctl(hw, I40E_PRTQF_FD_MSK(pctype, 0));
+ mask_reg[1] =
+ i40e_read_rx_ctl(hw, I40E_PRTQF_FD_MSK(pctype, 1));
+ break;
+ default:
+ PMD_DRV_LOG(ERR, "Unsupported input set type.");
+ return -EINVAL;
+ }
+
+ inset->inset = inset_reg;
+
+ for (i = 0; i < 2; i++) {
+ inset->mask[i].field_idx = ((mask_reg[i] >> 16) & 0x3F);
+ inset->mask[i].mask = mask_reg[i] & 0x;
+ }
+
+ return 0;
+}
+
+int
+rte_pmd_i40e_inset_set(uint16_t port, uint8_t pctype,
+ struct rte_pmd_i40e_inset *inset,
+ enum rte_pmd_i40e_inset_type inset_type)
+{
+ struct rte_eth_dev *dev;
+ struct i40e_hw *hw;
+ uint64_t inset_reg;
+ uint32_t mask_reg[2];
+ int i;
+
+ RTE_ETH_VALID_PORTID_OR_ERR_RET(port, -ENODEV);
+
+ dev = &rte_eth_devices[port];
+
+ if (!is_i40e_supported(dev))
+ return -ENOTSUP;
+
+ if (pctype > 63)
+ return -EINVAL;
+
+ hw = I40E_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+
+ /* Clear mask first */
+ for (i = 0; i < 2; i++)
+ i40e_check_write_reg(hw, I40E_GLQF_FD_MSK(i, pctype), 0);
+
+ inset_reg = inset->inset;
+ for (i = 0; i < 2; i++)
+ mask_reg[i] = (inset->mask[i].field_idx << 16) |
+ inset->mask[i].mask;
+
+ switch (inset_type) {
+ case INSET_HASH:
+ i40e_check_write_reg(hw, I40E_GLQF_HASH_INSET(0, pctype),
+(uint32_t)(inset_reg & UINT32_MAX));
+ i40e_check_write_reg(hw, I40E_GLQF_HASH_INSET(1, pctype),
+(uint32_t)((inset_reg >>
+ I40E_32_BIT_WIDTH) & UINT32_MAX));
+ for (i = 0; i < 2; i++)
+ i40e_check_write_reg(hw, I40E_GLQF_HASH_MSK(i, pctype),
+mask_reg[i]);
+ break;
+ case INSET_FDIR:
+ i40e_check_write_reg(hw, I40E_PRTQF_FD_INSET(pctype, 0),
+
[dpdk-dev] [PATCH v5 2/2] app/testpmd: add configuration for input set
This patch adds command to configure input set for
RSS/flow director/flow director flexible payload.
Signed-off-by: Beilei Xing
---
app/test-pmd/cmdline.c | 237 +
1 file changed, 237 insertions(+)
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index f71d963..79fdd0b 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -864,6 +864,15 @@ static void cmd_help_long_parsed(void *parsed_result,
"port config (port_id) pctype mapping update"
" (pctype_id_0[,pctype_id_1]*) (flow_type_id)\n"
"Update a flow type to pctype mapping item on a
port\n\n"
+
+ "port config (port_id) pctype (pctype_id) hash_inset|"
+ "fdir_inset|fdir_flx_inset get|set|clear field\n"
+ " (field_idx)\n"
+ "Configure RSS|FDIR|FDIR_FLX input set for some
pctype\n\n"
+
+ "port config (port_id) pctype (pctype_id) hash_inset|"
+ "fdir_inset|fdir_flx_inset clear all"
+ "Clear RSS|FDIR|FDIR_FLX input set completely for
some pctype\n\n"
);
}
@@ -14652,6 +14661,232 @@ cmdline_parse_inst_t cmd_ddp_get_list = {
},
};
+/* Configure input set */
+struct cmd_cfg_input_set_result {
+ cmdline_fixed_string_t port;
+ cmdline_fixed_string_t cfg;
+ portid_t port_id;
+ cmdline_fixed_string_t pctype;
+ uint8_t pctype_id;
+ cmdline_fixed_string_t inset_type;
+ cmdline_fixed_string_t opt;
+ cmdline_fixed_string_t field;
+ uint8_t field_idx;
+};
+
+static void
+cmd_cfg_input_set_parsed(
+ void *parsed_result,
+ __attribute__((unused)) struct cmdline *cl,
+ __attribute__((unused)) void *data)
+{
+ struct cmd_cfg_input_set_result *res = parsed_result;
+#ifdef RTE_LIBRTE_I40E_PMD
+ enum rte_pmd_i40e_inset_type inset_type = INSET_NONE;
+ struct rte_pmd_i40e_inset inset;
+ int ret;
+#endif
+
+ if (res->port_id > nb_ports) {
+ printf("Invalid port, range is [0, %d]\n", nb_ports - 1);
+ return;
+ }
+
+ if (!all_ports_stopped()) {
+ printf("Please stop all ports first\n");
+ return;
+ }
+
+#ifdef RTE_LIBRTE_I40E_PMD
+ if (!strcmp(res->inset_type, "hash_inset"))
+ inset_type = INSET_HASH;
+ else if (!strcmp(res->inset_type, "fdir_inset"))
+ inset_type = INSET_FDIR;
+ else if (!strcmp(res->inset_type, "fdir_flx_inset"))
+ inset_type = INSET_FDIR_FLX;
+ ret = rte_pmd_i40e_inset_get(res->port_id, res->pctype_id,
+&inset, inset_type);
+ if (ret) {
+ printf("Failed to get input set.\n");
+ return;
+ }
+
+ if (!strcmp(res->opt, "get")) {
+ ret = rte_pmd_i40e_inset_field_get(inset.inset,
+ res->field_idx);
+ if (ret)
+ printf("Field index %d is enabled.\n", res->field_idx);
+ else
+ printf("Field index %d is disabled.\n", res->field_idx);
+ return;
+ } else if (!strcmp(res->opt, "set"))
+ ret = rte_pmd_i40e_inset_field_set(&inset.inset,
+ res->field_idx);
+ else if (!strcmp(res->opt, "clear"))
+ ret = rte_pmd_i40e_inset_field_clear(&inset.inset,
+res->field_idx);
+ if (ret) {
+ printf("Failed to configure input set field.\n");
+ return;
+ }
+
+ ret = rte_pmd_i40e_inset_set(res->port_id, res->pctype_id,
+&inset, inset_type);
+ if (ret) {
+ printf("Failed to set input set.\n");
+ return;
+ }
+
+#endif
+}
+
+cmdline_parse_token_string_t cmd_cfg_input_set_port =
+ TOKEN_STRING_INITIALIZER(struct cmd_cfg_input_set_result,
+port, "port");
+cmdline_parse_token_string_t cmd_cfg_input_set_cfg =
+ TOKEN_STRING_INITIALIZER(struct cmd_cfg_input_set_result,
+cfg, "config");
+cmdline_parse_token_num_t cmd_cfg_input_set_port_id =
+ TOKEN_NUM_INITIALIZER(struct cmd_cfg_input_set_result,
+ port_id, UINT16);
+cmdline_parse_token_string_t cmd_cfg_input_set_pctype =
+ TOKEN_STRING_INITIALIZER(struct cmd_cfg_input_set_result,
+pctype, "pctype");
+cmdline_parse_token_num_t cmd_cfg_input_set_pctype_id =
+ TOKEN_NUM_INITIALIZER(struct cmd_cfg_input_set_result,
+ pctype_id, UINT8);
+cmdline_parse_token_string_t cmd_cfg_input_set_inset_type =
+ TOKEN
Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq flushing
On 12/07/2017 06:30 AM, Tiwei Bie wrote:
The vector Rx will be broken if backend has consumed all
the descs in the avail ring before the device is started.
Because in current implementation, vector Rx will return
immediately without refilling the avail ring if the used
ring is empty. So we have to refill the avail ring after
flushing the elements in the used ring for vector Rx.
Besides, vector Rx has a different ring layout assumption
and mbuf management. So we need to handle it differently.
Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start")
Cc: sta...@dpdk.org
Reported-by: Antonio Fischetti
Signed-off-by: Tiwei Bie
---
drivers/net/virtio/virtio_ethdev.c | 2 +-
drivers/net/virtio/virtqueue.c | 31 ---
drivers/net/virtio/virtqueue.h | 2 +-
3 files changed, 26 insertions(+), 9 deletions(-)
Reviewed-by: Maxime Coquelin
Thanks,
Maxime
Re: [dpdk-dev] [PATCH 2/5] net/virtio: fix typo in LRO support
On 12/07/2017 06:30 AM, Tiwei Bie wrote:
Fixes: 86d59b21468a ("net/virtio: support LRO")
Fixes: ec9f3d122a58 ("net/virtio: revert not claiming LRO support")
Cc: sta...@dpdk.org
Signed-off-by: Tiwei Bie
---
drivers/net/virtio/virtio_ethdev.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 64a0cc608..9caa133c9 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1754,7 +1754,7 @@ virtio_dev_configure(struct rte_eth_dev *dev)
if (rxmode->enable_lro &&
(!vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
- !vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4))) {
+ !vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6))) {
PMD_DRV_LOG(ERR,
"Large Receive Offload not available on this host");
return -ENOTSUP;
Good catch!
Reviewed-by: Maxime Coquelin
Thanks,
Maxime
Re: [dpdk-dev] [PATCH 3/5] net/virtio: remove a redundant macro definition for ctrl vq
On 12/07/2017 06:30 AM, Tiwei Bie wrote:
VIRTIO_NET_CTRL_MAC_ADDR_SET is defined two times in
virtqueue.h, the second one is obviously not wanted.
Signed-off-by: Tiwei Bie
---
drivers/net/virtio/virtqueue.h | 2 --
1 file changed, 2 deletions(-)
diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
index ab466c2db..eaf9de13e 100644
--- a/drivers/net/virtio/virtqueue.h
+++ b/drivers/net/virtio/virtqueue.h
@@ -226,8 +226,6 @@ struct virtqueue {
#define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MIN1
#define VIRTIO_NET_CTRL_MQ_VQ_PAIRS_MAX0x8000
-#define VIRTIO_NET_CTRL_MAC_ADDR_SET 1
-
/**
* This is the first element of the scatter-gather list. If you don't
* specify GSO or CSUM features, you can simply ignore the header.
Reviewed-by: Maxime Coquelin
Thanks,
Maxime
Re: [dpdk-dev] [PATCH 4/5] net/virtio: remove redundant macro definitions for vector Rx
On 12/07/2017 06:30 AM, Tiwei Bie wrote: RTE_VIRTIO_VPMD_RX_BURST and RTE_VIRTIO_VPMD_RX_REARM_THRESH have been defined and used in virtio_rxtx_simple.h, but are defined agained in virtio_rxtx_simple_*.c. It just happens to s/agained/again/ work. So remove the redundant definitions from the *.c files. Signed-off-by: Tiwei Bie --- drivers/net/virtio/virtio_rxtx_simple_neon.c | 2 -- drivers/net/virtio/virtio_rxtx_simple_sse.c | 2 -- 2 files changed, 4 deletions(-) Apart from above typo: Reviewed-by: Maxime Coquelin Thanks, Maxime
Re: [dpdk-dev] [PATCH v3 1/2] net/e1000: move RSS to flow API
Hi, Zhao Wei
Please correct build error show in
http://dpdk.org/ml/archives/test-report/2017-November/035129.html
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Wei Zhao
> Sent: Friday, November 24, 2017 4:05 PM
> To: dev@dpdk.org
> Cc: Zhao1, Wei
> Subject: [dpdk-dev] [PATCH v3 1/2] net/e1000: move RSS to flow API
>
> Rte_flow actually defined to include RSS, but till now, RSS is out of
> rte_flow.
> This patch is to move igb existing RSS to rte_flow.
>
> Signed-off-by: Wei Zhao
> ---
> drivers/net/e1000/e1000_ethdev.h | 20 +
> drivers/net/e1000/igb_ethdev.c | 17 +
> drivers/net/e1000/igb_flow.c | 160
> +++
> drivers/net/e1000/igb_rxtx.c | 61 +++
> 4 files changed, 258 insertions(+)
>
> diff --git a/drivers/net/e1000/e1000_ethdev.h
> b/drivers/net/e1000/e1000_ethdev.h
> index 5668910..0731766 100644
> --- a/drivers/net/e1000/e1000_ethdev.h
> +++ b/drivers/net/e1000/e1000_ethdev.h
> @@ -257,6 +257,12 @@ struct igb_ethertype_filter {
> uint32_t etqf;
> };
>
> +struct igb_rte_flow_rss_conf {
> + struct rte_eth_rss_conf rss_conf; /**< RSS parameters. */
> + uint16_t num; /**< Number of entries in queue[]. */
> + uint16_t queue[IGB_MAX_RX_QUEUE_NUM]; /**< Queues indices to
> use. */
> +};
> +
> /*
> * Structure to store filters'info.
> */
> @@ -274,6 +280,8 @@ struct e1000_filter_info {
> struct e1000_2tuple_filter_list twotuple_list;
> /* store the SYN filter info */
> uint32_t syn_info;
> + /* store the rss filter info */
> + struct igb_rte_flow_rss_conf rss_info;
> };
>
> /*
> @@ -342,6 +350,12 @@ struct igb_flex_filter_ele {
> struct rte_eth_flex_filter filter_info; };
>
> +/* rss filter list structure */
> +struct igb_rss_conf_ele {
> + TAILQ_ENTRY(igb_rss_conf_ele) entries;
> + struct igb_rte_flow_rss_conf filter_info; };
> +
> /* igb_flow memory list structure */
> struct igb_flow_mem {
> TAILQ_ENTRY(igb_flow_mem) entries;
> @@ -357,6 +371,8 @@ TAILQ_HEAD(igb_syn_filter_list,
> igb_eth_syn_filter_ele); struct igb_syn_filter_list igb_filter_syn_list;
> TAILQ_HEAD(igb_flex_filter_list, igb_flex_filter_ele); struct
> igb_flex_filter_list igb_filter_flex_list;
> +TAILQ_HEAD(igb_rss_filter_list, igb_rss_conf_ele); struct
> +igb_rss_filter_list igb_filter_rss_list;
> TAILQ_HEAD(igb_flow_mem_list, igb_flow_mem); struct
> igb_flow_mem_list igb_flow_list;
>
> @@ -500,4 +516,8 @@ int eth_igb_syn_filter_set(struct rte_eth_dev *dev,
> int eth_igb_add_del_flex_filter(struct rte_eth_dev *dev,
> struct rte_eth_flex_filter *filter,
> bool add);
> +int igb_config_rss_filter(struct rte_eth_dev *dev,
> + struct igb_rte_flow_rss_conf *conf,
> + bool add);
> +
> #endif /* _E1000_ETHDEV_H_ */
> diff --git a/drivers/net/e1000/igb_ethdev.c
> b/drivers/net/e1000/igb_ethdev.c index fdc139f..275fa02 100644
> --- a/drivers/net/e1000/igb_ethdev.c
> +++ b/drivers/net/e1000/igb_ethdev.c
> @@ -948,6 +948,7 @@ eth_igb_dev_init(struct rte_eth_dev *eth_dev)
> TAILQ_INIT(&igb_filter_ethertype_list);
> TAILQ_INIT(&igb_filter_syn_list);
> TAILQ_INIT(&igb_filter_flex_list);
> + TAILQ_INIT(&igb_filter_rss_list);
> TAILQ_INIT(&igb_flow_list);
>
> return 0;
> @@ -1007,6 +1008,10 @@ eth_igb_dev_uninit(struct rte_eth_dev *eth_dev)
> memset(filter_info->ethertype_filters, 0,
> E1000_MAX_ETQF_FILTERS * sizeof(struct igb_ethertype_filter));
>
> + /* clear the rss filter info */
> + memset(&filter_info->rss_info, 0,
> + sizeof(struct igb_rte_flow_rss_conf));
> +
> /* remove all ntuple filters of the device */
> igb_ntuple_filter_uninit(eth_dev);
>
> @@ -5628,6 +5633,17 @@ igb_flex_filter_restore(struct rte_eth_dev *dev)
> }
> }
>
> +/* restore rss filter */
> +static inline void
> +igb_rss_filter_restore(struct rte_eth_dev *dev) {
> + struct e1000_filter_info *filter_info =
> + E1000_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
> +
> + if (filter_info->rss_info.num)
> + igb_config_rss_filter(dev, &filter_info->rss_info, TRUE); }
> +
> /* restore all types filter */
> static int
> igb_filter_restore(struct rte_eth_dev *dev) @@ -5636,6 +5652,7 @@
> igb_filter_restore(struct rte_eth_dev *dev)
> igb_ethertype_filter_restore(dev);
> igb_syn_filter_restore(dev);
> igb_flex_filter_restore(dev);
> + igb_rss_filter_restore(dev);
>
> return 0;
> }
> diff --git a/drivers/net/e1000/igb_flow.c b/drivers/net/e1000/igb_flow.c
> index 22bad26..bf5cfac 100644
> --- a/drivers/net/e1000/igb_flow.c
> +++ b/drivers/net/e1000/igb_flow.c
> @@ -1295,6 +1295,101 @@ igb_parse_flex_filter(struct rte_eth_dev *dev,
> return 0;
> }
>
> +static int
> +igb_parse_rss_filter(struct rte_eth_dev *dev,
> + con
Re: [dpdk-dev] [PATCH v3 2/2] net/ixgbe: move RSS to flow API
Hi, Zhao Wei
Please correct build error show in
http://dpdk.org/ml/archives/test-report/2017-November/035130.html
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Wei Zhao
> Sent: Friday, November 24, 2017 4:05 PM
> To: dev@dpdk.org
> Cc: Zhao1, Wei
> Subject: [dpdk-dev] [PATCH v3 2/2] net/ixgbe: move RSS to flow API
>
> Rte_flow actually defined to include RSS, but till now, RSS is out of
> rte_flow.
> This patch is to move ixgbe existing RSS to rte_flow.
>
> Signed-off-by: Wei Zhao
> ---
> drivers/net/ixgbe/ixgbe_ethdev.c | 13 +++
> drivers/net/ixgbe/ixgbe_ethdev.h | 10 +++
> drivers/net/ixgbe/ixgbe_flow.c | 165
> +++
> drivers/net/ixgbe/ixgbe_rxtx.c | 65 +++
> 4 files changed, 253 insertions(+)
>
> diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c
> b/drivers/net/ixgbe/ixgbe_ethdev.c
> index ff19a56..4960650 100644
> --- a/drivers/net/ixgbe/ixgbe_ethdev.c
> +++ b/drivers/net/ixgbe/ixgbe_ethdev.c
> @@ -8339,6 +8339,18 @@ ixgbe_l2_tn_filter_restore(struct rte_eth_dev
> *dev)
> }
> }
>
> +/* restore rss filter */
> +static inline void
> +ixgbe_rss_filter_restore(struct rte_eth_dev *dev) {
> + struct ixgbe_filter_info *filter_info =
> + IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data->dev_private);
> +
> + if (filter_info->rss_info.num)
> + ixgbe_config_rss_filter(dev,
> + &filter_info->rss_info, TRUE);
> +}
> +
> static int
> ixgbe_filter_restore(struct rte_eth_dev *dev) { @@ -8347,6 +8359,7 @@
> ixgbe_filter_restore(struct rte_eth_dev *dev)
> ixgbe_syn_filter_restore(dev);
> ixgbe_fdir_filter_restore(dev);
> ixgbe_l2_tn_filter_restore(dev);
> + ixgbe_rss_filter_restore(dev);
>
> return 0;
> }
> diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h
> b/drivers/net/ixgbe/ixgbe_ethdev.h
> index 51ddcfd..4af79b4 100644
> --- a/drivers/net/ixgbe/ixgbe_ethdev.h
> +++ b/drivers/net/ixgbe/ixgbe_ethdev.h
> @@ -224,6 +224,12 @@ struct ixgbe_hw_fdir_info {
> bool mask_added; /* If already got mask from consistent filter */ };
>
> +struct ixgbe_rte_flow_rss_conf {
> + struct rte_eth_rss_conf rss_conf; /**< RSS parameters. */
> + uint16_t num; /**< Number of entries in queue[]. */
> + uint16_t queue[IXGBE_MAX_RX_QUEUE_NUM]; /**< Queues indices to
> use. */
> +};
> +
> /* structure for interrupt relative data */ struct ixgbe_interrupt {
> uint32_t flags;
> @@ -340,6 +346,8 @@ struct ixgbe_filter_info {
> struct ixgbe_5tuple_filter_list fivetuple_list;
> /* store the SYN filter info */
> uint32_t syn_info;
> + /* store the rss filter info */
> + struct ixgbe_rte_flow_rss_conf rss_info;
> };
>
> struct ixgbe_l2_tn_key {
> @@ -719,6 +727,8 @@ void ixgbe_tm_conf_init(struct rte_eth_dev *dev);
> void ixgbe_tm_conf_uninit(struct rte_eth_dev *dev); int
> ixgbe_set_queue_rate_limit(struct rte_eth_dev *dev, uint16_t queue_idx,
> uint16_t tx_rate);
> +int ixgbe_config_rss_filter(struct rte_eth_dev *dev,
> + struct ixgbe_rte_flow_rss_conf *conf, bool add);
>
> static inline int
> ixgbe_ethertype_filter_lookup(struct ixgbe_filter_info *filter_info, diff
> --git
> a/drivers/net/ixgbe/ixgbe_flow.c b/drivers/net/ixgbe/ixgbe_flow.c index
> 19c2d47..8f964cf 100644
> --- a/drivers/net/ixgbe/ixgbe_flow.c
> +++ b/drivers/net/ixgbe/ixgbe_flow.c
> @@ -103,6 +103,11 @@ struct ixgbe_eth_l2_tunnel_conf_ele {
> TAILQ_ENTRY(ixgbe_eth_l2_tunnel_conf_ele) entries;
> struct rte_eth_l2_tunnel_conf filter_info; };
> +/* rss filter list structure */
> +struct ixgbe_rss_conf_ele {
> + TAILQ_ENTRY(ixgbe_rss_conf_ele) entries;
> + struct ixgbe_rte_flow_rss_conf filter_info; };
> /* ixgbe_flow memory list structure */
> struct ixgbe_flow_mem {
> TAILQ_ENTRY(ixgbe_flow_mem) entries;
> @@ -114,6 +119,7 @@ TAILQ_HEAD(ixgbe_ethertype_filter_list,
> ixgbe_ethertype_filter_ele); TAILQ_HEAD(ixgbe_syn_filter_list,
> ixgbe_eth_syn_filter_ele); TAILQ_HEAD(ixgbe_fdir_rule_filter_list,
> ixgbe_fdir_rule_ele); TAILQ_HEAD(ixgbe_l2_tunnel_filter_list,
> ixgbe_eth_l2_tunnel_conf_ele);
> +TAILQ_HEAD(ixgbe_rss_filter_list, ixgbe_rss_conf_ele);
> TAILQ_HEAD(ixgbe_flow_mem_list, ixgbe_flow_mem);
>
> static struct ixgbe_ntuple_filter_list filter_ntuple_list; @@ -121,6 +127,7
> @@ static struct ixgbe_ethertype_filter_list filter_ethertype_list; static
> struct ixgbe_syn_filter_list filter_syn_list; static struct
> ixgbe_fdir_rule_filter_list filter_fdir_list; static struct
> ixgbe_l2_tunnel_filter_list filter_l2_tunnel_list;
> +static struct ixgbe_rss_filter_list filter_rss_list;
> static struct ixgbe_flow_mem_list ixgbe_flow_list;
>
> /**
> @@ -2700,6 +2707,109 @@ ixgbe_parse_fdir_filter(struct rte_eth_dev
> *dev,
> return ret;
> }
>
> +static int
> +ixgbe_parse_rss_filter(struct rte_eth_dev *dev,
> + const struct rte_flow_attr *att
Re: [dpdk-dev] [PATCH 5/5] net/virtio: squeeze repeated blank lines
On 12/07/2017 06:30 AM, Tiwei Bie wrote:
Squeeze repeated blank lines with below scripts:
for i in $(find . -name "*.[ch]"); do \
cat -s $i > /tmp/x && mv /tmp/x $i; done
Signed-off-by: Tiwei Bie
---
drivers/net/virtio/virtio_ethdev.c | 1 -
drivers/net/virtio/virtio_logs.h | 1 -
drivers/net/virtio/virtio_pci.c| 1 -
drivers/net/virtio/virtio_pci.h| 2 --
drivers/net/virtio/virtio_rxtx.c | 2 --
5 files changed, 7 deletions(-)
I don't really like such cleaning patches.
It does not bring added value, but can create conflicts later when
backporting or rebasing.
Maxime
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index 9caa133c9..aa6d494ca 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1801,7 +1801,6 @@ virtio_dev_configure(struct rte_eth_dev *dev)
return 0;
}
-
static int
virtio_dev_start(struct rte_eth_dev *dev)
{
diff --git a/drivers/net/virtio/virtio_logs.h b/drivers/net/virtio/virtio_logs.h
index 90a79eaa5..6ce3149d5 100644
--- a/drivers/net/virtio/virtio_logs.h
+++ b/drivers/net/virtio/virtio_logs.h
@@ -59,7 +59,6 @@
#define PMD_TX_LOG(level, fmt, args...) do { } while(0)
#endif
-
#ifdef RTE_LIBRTE_VIRTIO_DEBUG_DRIVER
#define PMD_DRV_LOG(level, fmt, args...) \
RTE_LOG(level, PMD, "%s(): " fmt "\n", __func__, ## args)
diff --git a/drivers/net/virtio/virtio_pci.c b/drivers/net/virtio/virtio_pci.c
index 9574498fb..684ef560c 100644
--- a/drivers/net/virtio/virtio_pci.c
+++ b/drivers/net/virtio/virtio_pci.c
@@ -481,7 +481,6 @@ const struct virtio_pci_ops modern_ops = {
.notify_queue = modern_notify_queue,
};
-
void
vtpci_read_dev_config(struct virtio_hw *hw, size_t offset,
void *dst, int length)
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index 3c5ce66ce..e42962a99 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -274,7 +274,6 @@ struct virtio_hw {
struct virtqueue **vqs;
};
-
/*
* While virtio_hw is stored in shared memory, this structure stores
* some infos that may vary in the multiple process model locally.
@@ -290,7 +289,6 @@ struct virtio_hw_internal {
extern struct virtio_hw_internal virtio_hw_internal[RTE_MAX_ETHPORTS];
-
/*
* This structure is just a reference to read
* net device specific config space; it just a chodu structure
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index 390c137c8..7ccd1da34 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -66,7 +66,6 @@
#define VIRTIO_DUMP_PACKET(m, len) do { } while (0)
#endif
-
#define VIRTIO_SIMPLE_FLAGS ((uint32_t)ETH_TXQ_FLAGS_NOMULTSEGS | \
ETH_TXQ_FLAGS_NOOFFLOADS)
@@ -175,7 +174,6 @@ virtio_xmit_cleanup(struct virtqueue *vq, uint16_t num)
}
}
-
static inline int
virtqueue_enqueue_recv_refill(struct virtqueue *vq, struct rte_mbuf *cookie)
{
Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq flushing
Thanks Tiwei for working on this, I'll give it a try soon.
Antonio
> -Original Message-
> From: Maxime Coquelin [mailto:maxime.coque...@redhat.com]
> Sent: Thursday, December 7, 2017 9:15 AM
> To: Bie, Tiwei ; y...@fridaylinux.org; dev@dpdk.org
> Cc: Fischetti, Antonio ; sta...@dpdk.org
> Subject: Re: [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq
> flushing
>
>
>
> On 12/07/2017 06:30 AM, Tiwei Bie wrote:
> > The vector Rx will be broken if backend has consumed all
> > the descs in the avail ring before the device is started.
> > Because in current implementation, vector Rx will return
> > immediately without refilling the avail ring if the used
> > ring is empty. So we have to refill the avail ring after
> > flushing the elements in the used ring for vector Rx.
> >
> > Besides, vector Rx has a different ring layout assumption
> > and mbuf management. So we need to handle it differently.
> >
> > Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start")
> > Cc: sta...@dpdk.org
> >
> > Reported-by: Antonio Fischetti
> > Signed-off-by: Tiwei Bie
> > ---
> > drivers/net/virtio/virtio_ethdev.c | 2 +-
> > drivers/net/virtio/virtqueue.c | 31
> ---
> > drivers/net/virtio/virtqueue.h | 2 +-
> > 3 files changed, 26 insertions(+), 9 deletions(-)
>
> Reviewed-by: Maxime Coquelin
>
> Thanks,
> Maxime
Re: [dpdk-dev] [PATCH] vhost_user: protect active rings from async ring changes
> -Original Message-
> From: Victor Kaplansky [mailto:vkapl...@redhat.com]
> Sent: Wednesday, December 6, 2017 9:56 PM
> To: dev@dpdk.org; y...@fridaylinux.org; Bie, Tiwei; Tan, Jianfeng;
> vkapl...@redhat.com
> Cc: sta...@dpdk.org; jfrei...@redhat.com; Maxime Coquelin
> Subject: [PATCH] vhost_user: protect active rings from async ring changes
>
> When performing live migration or memory hot-plugging,
> the changes to the device and vrings made by message handler
> done independently from vring usage by PMD threads.
>
> This causes for example segfauls during live-migration
segfauls ->segfaults?
> with MQ enable, but in general virtually any request
> sent by qemu changing the state of device can cause
> problems.
>
> These patches fixes all above issues by adding a spinlock
> to every vring and requiring message handler to start operation
> only after ensuring that all PMD threads related to the divece
Another typo: divece.
> are out of critical section accessing the vring data.
>
> Each vring has its own lock in order to not create contention
> between PMD threads of different vrings and to prevent
> performance degradation by scaling queue pair number.
Also wonder how much overhead it brings.
Instead of locking each vring, can we just, waiting a while (10us for example)
after call destroy_device() callback so that every PMD thread has enough time
to skip out the criterial area?
> ---
> lib/librte_vhost/vhost.h | 1 +
> lib/librte_vhost/vhost_user.c | 44
> +++
> lib/librte_vhost/virtio_net.c | 8
> 3 files changed, 53 insertions(+)
>
> diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
> index 1cc81c17..812aeccd 100644
> --- a/lib/librte_vhost/vhost.h
> +++ b/lib/librte_vhost/vhost.h
> @@ -137,6 +137,7 @@ struct vhost_virtqueue {
> TAILQ_HEAD(, vhost_iotlb_entry) iotlb_list;
> int iotlb_cache_nr;
> TAILQ_HEAD(, vhost_iotlb_entry) iotlb_pending_list;
> +rte_spinlock_t active_lock;
> } __rte_cache_aligned;
>
> /* Old kernels have no such macros defined */
> diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
> index f4c7ce46..02a3a7d3 100644
> --- a/lib/librte_vhost/vhost_user.c
> +++ b/lib/librte_vhost/vhost_user.c
> @@ -1190,6 +1190,46 @@ vhost_user_check_and_alloc_queue_pair(struct
> virtio_net *dev, VhostUserMsg *msg)
> return alloc_vring_queue(dev, vring_idx);
> }
>
> +static void
> +vhost_user_lock_all_queue_pairs(struct virtio_net *dev)
> +{
> + unsigned int i = 0;
> + unsigned int vq_num = 0;
> +
> + while (vq_num < dev->nr_vring) {
> + struct vhost_virtqueue *vq = dev->virtqueue[i];
> +
> + if (!vq) {
> + i++;
> + continue;
> + }
> +
> + rte_spinlock_lock(&vq->active_lock);
> + vq_num++;
> + i++;
> + }
> +}
> +
> +static void
> +vhost_user_unlock_all_queue_pairs(struct virtio_net *dev)
> +{
> + unsigned int i = 0;
> + unsigned int vq_num = 0;
> +
> + while (vq_num < dev->nr_vring) {
> + struct vhost_virtqueue *vq = dev->virtqueue[i];
> +
> + if (!vq) {
> + i++;
> + continue;
> + }
> +
> + rte_spinlock_unlock(&vq->active_lock);
> + vq_num++;
> + i++;
> + }
> +}
> +
> int
> vhost_user_msg_handler(int vid, int fd)
> {
> @@ -1241,6 +1281,8 @@ vhost_user_msg_handler(int vid, int fd)
> return -1;
> }
>
> + vhost_user_lock_all_queue_pairs(dev);
> +
> switch (msg.request.master) {
> case VHOST_USER_GET_FEATURES:
> msg.payload.u64 = vhost_user_get_features(dev);
> @@ -1342,6 +1384,8 @@ vhost_user_msg_handler(int vid, int fd)
>
> }
>
> + vhost_user_unlock_all_queue_pairs(dev);
> +
> if (msg.flags & VHOST_USER_NEED_REPLY) {
> msg.payload.u64 = !!ret;
> msg.size = sizeof(msg.payload.u64);
> diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
> index 6fee16e5..de7e38bb 100644
> --- a/lib/librte_vhost/virtio_net.c
> +++ b/lib/librte_vhost/virtio_net.c
> @@ -44,6 +44,7 @@
> #include
> #include
> #include
> +#include
>
> #include "iotlb.h"
> #include "vhost.h"
> @@ -1180,9 +1181,12 @@ rte_vhost_dequeue_burst(int vid, uint16_t
> queue_id,
> }
>
> vq = dev->virtqueue[queue_id];
> +
Remove above blank line.
> if (unlikely(vq->enabled == 0))
> return 0;
>
> + rte_spinlock_lock(&vq->active_lock);
> +
> vq->batch_copy_nb_elems = 0;
>
> if (dev->features & (1ULL << VIRTIO_F_IOMMU_PLATFORM))
> @@ -1240,6 +1244,7 @@ rte_vhost_dequeue_burst(int vid, uint16_t
> queue_id,
> if (rarp_mbuf == NULL) {
> RTE_LOG(ERR, VHOST_DATA,
> "Failed to allocate
Re: [dpdk-dev] [PATCH v2 2/2] examples/ipsec-secgw: add target queues in flow actions
Hi Nelio,
On 12/04/2017 07:41 PM, Nelio Laranjeiro wrote:
Mellanox INNOVA NIC needs to have final target queue actions to perform
inline crypto.
Signed-off-by: Nelio Laranjeiro
---
Changes in v2:
* Test the rule by PASSTHRU/RSS/QUEUE and apply the first one validated.
---
examples/ipsec-secgw/ipsec.c | 81
examples/ipsec-secgw/ipsec.h | 2 +-
2 files changed, 76 insertions(+), 7 deletions(-)
diff --git a/examples/ipsec-secgw/ipsec.c b/examples/ipsec-secgw/ipsec.c
index 17bd7620d..f8823fb94 100644
--- a/examples/ipsec-secgw/ipsec.c
+++ b/examples/ipsec-secgw/ipsec.c
@@ -142,6 +142,7 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct ipsec_sa
*sa)
rte_eth_dev_get_sec_ctx(
sa->portid);
const struct rte_security_capability *sec_cap;
+ int ret = 0;
sa->sec_session = rte_security_session_create(ctx,
&sess_conf, ipsec_ctx->session_pool);
@@ -173,6 +174,10 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct
ipsec_sa *sa)
return -1;
}
+ sa->attr.egress = (sa->direction ==
+ RTE_SECURITY_IPSEC_SA_DIR_EGRESS);
+ sa->attr.ingress = (sa->direction ==
+ RTE_SECURITY_IPSEC_SA_DIR_INGRESS);
sa->ol_flags = sec_cap->ol_flags;
sa->security_ctx = ctx;
sa->pattern[0].type = RTE_FLOW_ITEM_TYPE_ETH;
@@ -201,15 +206,79 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct
ipsec_sa *sa)
sa->action[0].type = RTE_FLOW_ACTION_TYPE_SECURITY;
sa->action[0].conf = sa->sec_session;
- sa->action[1].type = RTE_FLOW_ACTION_TYPE_END;
-
- sa->attr.egress = (sa->direction ==
- RTE_SECURITY_IPSEC_SA_DIR_EGRESS);
- sa->attr.ingress = (sa->direction ==
- RTE_SECURITY_IPSEC_SA_DIR_INGRESS);
+ if (sa->attr.ingress) {
+ uint8_t rss_key[40];
+ struct rte_eth_rss_conf rss_conf = {
+ .rss_key = rss_key,
+ .rss_key_len = 40,
+ };
+ struct rte_eth_dev *eth_dev;
+ union {
+ struct rte_flow_action_rss rss;
+ struct {
+ const struct rte_eth_rss_conf *rss_conf;
+ uint16_t num;
+ uint16_t queue[RTE_MAX_QUEUES_PER_PORT];
+ } local;
+ } action_rss;
+ unsigned int i;
+ unsigned int j;
+
+ sa->action[2].type = RTE_FLOW_ACTION_TYPE_END;
+ /*
+* Try implicitly PASSTHRU, it can also be
+* explicit.
+*/
May be we can get rid of this check. You can do the check with RSS and
then QUEUE. That should be fine. SECURITY is terminating on Cavium
hardware, but according to the spec it is a non-terminating meta action.
We can stick to that. For Cavium hardware the PMD will give success to
SECURITY+QUEUE. That should resolve the issue.
+ sa->action[1].type = RTE_FLOW_ACTION_TYPE_END;
+ ret = rte_flow_validate(sa->portid, &sa->attr,
+ sa->pattern, sa->action,
+ &err);
+ if (!ret)
+ goto flow_create;
+ /* Try RSS. */
+ sa->action[1].type = RTE_FLOW_ACTION_TYPE_RSS;
+ sa->action[1].conf = &action_rss;
+ eth_dev = ctx->device;
+ rte_eth_dev_rss_hash_conf_get(sa->portid,
+ &rss_conf);
+ for (i = 0, j = 0;
+i < eth_dev->data->nb_rx_queues; ++i)
+ if (eth_dev->data->rx_queues[i])
+ action_rss.local.queue[j++] = i;
+ action_rss.local.num = j;
+ action_rss.loc
Re: [dpdk-dev] [RFC v2] lib: add compressdev API
> -Original Message-
> From: Trahe, Fiona [mailto:fiona.tr...@intel.com]
> Sent: 24 November 2017 22:26
> To: dev@dpdk.org; Verma, Shally
> Cc: Challa, Mahipal ; Athreya, Narayana
> Prasad ;
> pablo.de.lara.gua...@intel.com; fiona.tr...@intel.com
> Subject: [RFC v2] lib: add compressdev API
>
> compressdev API
>
> Signed-off-by: Trahe, Fiona
> ---
//snip//
> +unsigned int
> +rte_compressdev_get_header_session_size(void)
> +{
> + /*
> + * Header contains pointers to the private data
> + * of all registered drivers
> + */
> + return (sizeof(void *) * nb_drivers);
> +}
> +
> +unsigned int
> +rte_compressdev_get_private_session_size(uint8_t dev_id)
> +{
> + struct rte_compressdev *dev;
> + unsigned int header_size = sizeof(void *) * nb_drivers;
> + unsigned int priv_sess_size;
> +
> + if (!rte_compressdev_pmd_is_valid_dev(dev_id))
> + return 0;
> +
> + dev = rte_compressdev_pmd_get_dev(dev_id);
> +
> + if (*dev->dev_ops->session_get_size == NULL)
> + return 0;
> +
> + priv_sess_size = (*dev->dev_ops->session_get_size)(dev);
> +
> + /*
> + * If size is less than session header size,
> + * return the latter, as this guarantees that
> + * sessionless operations will work
> + */
[Shally] believe this comment need an edit
> + if (priv_sess_size < header_size)
> + return header_size;
> +
> + return priv_sess_size;
[Shally] This doesn't return header_size inclusive which is fine as per API
definition. So should application call
rte_compressdev_get_header_session_size() in case it want to know header_size
overhead per session and allocate pool with elt_size = sess_header_size +
dev_priv_sz?
> +
> +}
//snip//
Thanks
Shally
Re: [dpdk-dev] [PATCH] vhost_user: protect active rings from async ring changes
On 12/07/2017 10:33 AM, Tan, Jianfeng wrote: -Original Message- From: Victor Kaplansky [mailto:vkapl...@redhat.com] Sent: Wednesday, December 6, 2017 9:56 PM To: dev@dpdk.org; y...@fridaylinux.org; Bie, Tiwei; Tan, Jianfeng; vkapl...@redhat.com Cc: sta...@dpdk.org; jfrei...@redhat.com; Maxime Coquelin Subject: [PATCH] vhost_user: protect active rings from async ring changes When performing live migration or memory hot-plugging, the changes to the device and vrings made by message handler done independently from vring usage by PMD threads. This causes for example segfauls during live-migration segfauls ->segfaults? with MQ enable, but in general virtually any request sent by qemu changing the state of device can cause problems. These patches fixes all above issues by adding a spinlock to every vring and requiring message handler to start operation only after ensuring that all PMD threads related to the divece Another typo: divece. are out of critical section accessing the vring data. Each vring has its own lock in order to not create contention between PMD threads of different vrings and to prevent performance degradation by scaling queue pair number. Also wonder how much overhead it brings. Instead of locking each vring, can we just, waiting a while (10us for example) after call destroy_device() callback so that every PMD thread has enough time to skip out the criterial area? No, because we are not destroying the device when it is needed. Actually, once destroy_device() is called, it is likely that the application has taken care the ring aren't being processed anymore before returning from the callback (This is at least the case with Vhost PMD). The reason we need the lock is to protect PMD threads from the handling of some vhost-user protocol requests. For example SET_MEM_TABLE in the case of memory hotplug, or SET_LOG_BASE in case of multiqueue, which is sent for every queue pair and results in unmapping/remapping the logging area. Maxime
[dpdk-dev] [PATCH 01/2] vfio: expose clear group function for internal usages
other vfio based module e.g. fslmc will also need to use
the clear_group call.
So, exposing it and renaming it to *rte_vfio_clear_group*
Signed-off-by: Hemant Agrawal
---
lib/librte_eal/linuxapp/eal/eal_vfio.c | 18 +-
lib/librte_eal/linuxapp/eal/eal_vfio.h | 2 +-
lib/librte_eal/linuxapp/eal/eal_vfio_mp_sync.c | 2 +-
lib/librte_eal/rte_eal_version.map | 7 +++
4 files changed, 18 insertions(+), 11 deletions(-)
diff --git a/lib/librte_eal/linuxapp/eal/eal_vfio.c
b/lib/librte_eal/linuxapp/eal/eal_vfio.c
index 58f0123..76184d1 100644
--- a/lib/librte_eal/linuxapp/eal/eal_vfio.c
+++ b/lib/librte_eal/linuxapp/eal/eal_vfio.c
@@ -226,7 +226,7 @@ vfio_group_device_count(int vfio_group_fd)
}
int
-clear_group(int vfio_group_fd)
+rte_vfio_clear_group(int vfio_group_fd)
{
int i;
int socket_fd, ret;
@@ -329,12 +329,12 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
RTE_LOG(ERR, EAL, " %s cannot get group status, "
"error %i (%s)\n", dev_addr, errno,
strerror(errno));
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
} else if (!(group_status.flags & VFIO_GROUP_FLAGS_VIABLE)) {
RTE_LOG(ERR, EAL, " %s VFIO group is not viable!\n", dev_addr);
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
@@ -348,7 +348,7 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
RTE_LOG(ERR, EAL, " %s cannot add VFIO group to
container, "
"error %i (%s)\n", dev_addr, errno,
strerror(errno));
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
@@ -370,7 +370,7 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
" %s failed to select IOMMU type\n",
dev_addr);
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
ret = t->dma_map_func(vfio_cfg.vfio_container_fd);
@@ -379,7 +379,7 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
" %s DMA remapping failed, error %i
(%s)\n",
dev_addr, errno, strerror(errno));
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
}
@@ -395,7 +395,7 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
RTE_LOG(WARNING, EAL, "Getting a vfio_dev_fd for %s failed\n",
dev_addr);
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
@@ -407,7 +407,7 @@ rte_vfio_setup_device(const char *sysfs_base, const char
*dev_addr,
strerror(errno));
close(*vfio_dev_fd);
close(vfio_group_fd);
- clear_group(vfio_group_fd);
+ rte_vfio_clear_group(vfio_group_fd);
return -1;
}
vfio_group_device_get(vfio_group_fd);
@@ -467,7 +467,7 @@ rte_vfio_release_device(const char *sysfs_base, const char
*dev_addr,
return -1;
}
- if (clear_group(vfio_group_fd) < 0) {
+ if (rte_vfio_clear_group(vfio_group_fd) < 0) {
RTE_LOG(INFO, EAL, "Error when clearing group for %s\n",
dev_addr);
return -1;
diff --git a/lib/librte_eal/linuxapp/eal/eal_vfio.h
b/lib/librte_eal/linuxapp/eal/eal_vfio.h
index ba7892b..12e38cc 100644
--- a/lib/librte_eal/linuxapp/eal/eal_vfio.h
+++ b/lib/librte_eal/linuxapp/eal/eal_vfio.h
@@ -179,7 +179,7 @@ vfio_get_group_fd(int iommu_group_no);
/* remove group fd from internal VFIO group fd array */
int
-clear_group(int vfio_group_fd);
+rte_vfio_clear_group(int vfio_group_fd);
int vfio_mp_sync_setup(void);
diff --git a/lib/librte_eal/linuxapp/eal/eal_vfio_mp_sync.c
b/lib/librte_eal/linuxapp/eal/eal_vfio_mp_sync.c
index b53ed7e..e32f2e3 100644
--- a/lib/librte_eal/linuxapp/eal/eal_vf
[dpdk-dev] [PATCH 02/2] bus/fslmc: clear the vfio group on error
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/fslmc_vfio.c | 4
1 file changed, 4 insertions(+)
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 7831201..25c87ad 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -660,12 +660,14 @@ fslmc_vfio_setup_group(void)
if (ret) {
FSLMC_VFIO_LOG(ERR, "VFIO error getting group status");
close(vfio_group.fd);
+ rte_vfio_clear_group(vfio_group.fd);
return ret;
}
if (!(status.flags & VFIO_GROUP_FLAGS_VIABLE)) {
FSLMC_VFIO_LOG(ERR, "VFIO group not viable");
close(vfio_group.fd);
+ rte_vfio_clear_group(vfio_group.fd);
return -EPERM;
}
/* Since Group is VIABLE, Store the groupid */
@@ -680,6 +682,7 @@ fslmc_vfio_setup_group(void)
"Error connecting container with groupid %d",
groupid);
close(vfio_group.fd);
+ rte_vfio_clear_group(vfio_group.fd);
return ret;
}
}
@@ -690,6 +693,7 @@ fslmc_vfio_setup_group(void)
FSLMC_VFIO_LOG(ERR, "Error getting device %s fd from group %d",
g_container, vfio_group.groupid);
close(vfio_group.fd);
+ rte_vfio_clear_group(vfio_group.fd);
return ret;
}
container_device_fd = ret;
--
2.7.4
Re: [dpdk-dev] [PATCH v3 1/4] vhost: prevent features to be changed while device is running
On Thu, Dec 07, 2017 at 09:39:06AM +0100, Maxime Coquelin wrote:
> On 12/07/2017 09:08 AM, Tiwei Bie wrote:
> > On Wed, Dec 06, 2017 at 10:20:45AM +0100, Maxime Coquelin wrote:
> > > As section 2.2 of the Virtio spec states about features
> > > negotiation:
> > > "During device initialization, the driver reads this and tells
> > > the device the subset that it accepts. The only way to
> > > renegotiate is to reset the device."
> > >
> > > This patch implements a check to prevent illegal features change
> > > while the device is running.
> > >
> > > One exception is the VHOST_F_LOG_ALL feature bit, which is enabled
> > > when live-migration is initiated. But this feature is not negotiated
> > > with the Virtio driver, but directly with the Vhost master.
> > >
> > > Signed-off-by: Maxime Coquelin
> > > ---
> > > lib/librte_vhost/vhost_user.c | 17 -
> > > 1 file changed, 16 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c
> > > index f4c7ce462..2d86c0ca8 100644
> > > --- a/lib/librte_vhost/vhost_user.c
> > > +++ b/lib/librte_vhost/vhost_user.c
> > > @@ -183,7 +183,22 @@ vhost_user_set_features(struct virtio_net *dev,
> > > uint64_t features)
> > > return -1;
> > > }
> > > - if ((dev->flags & VIRTIO_DEV_RUNNING) && dev->features != features) {
> > > + if (dev->features == features)
> > > + return 0;
> > > +
> >
> > We couldn't return directly when dev->features == features.
> > Otherwise, if the features provided by virtio driver is 0,
> > dev->vhost_hlen won't get a chance to be initialized.
>
> Good catch.
>
> Either we do :
> if ((dev->features == features) && dev->vhost_len)
> return 0;
>
> Or we could initialize dev->vhost_len to sizeof(struct virtio_net_hdr)
> at alloc time.
>
> I prefer the former, what do you think?
>
I prefer to give other code (e.g. LOG code) a chance
to run. So maybe we could remove the "fast return" and
check whether the features are changed when calling
dev->notify_ops->features_changed()? Or return only
when device is running and features are not changed?
Best regards,
Tiwei Bie
Re: [dpdk-dev] [PATCH 5/5] net/virtio: squeeze repeated blank lines
On Thu, Dec 07, 2017 at 10:22:43AM +0100, Maxime Coquelin wrote: > On 12/07/2017 06:30 AM, Tiwei Bie wrote: > > Squeeze repeated blank lines with below scripts: > > > > for i in $(find . -name "*.[ch]"); do \ > > cat -s $i > /tmp/x && mv /tmp/x $i; done > > > > Signed-off-by: Tiwei Bie > > --- > > drivers/net/virtio/virtio_ethdev.c | 1 - > > drivers/net/virtio/virtio_logs.h | 1 - > > drivers/net/virtio/virtio_pci.c| 1 - > > drivers/net/virtio/virtio_pci.h| 2 -- > > drivers/net/virtio/virtio_rxtx.c | 2 -- > > 5 files changed, 7 deletions(-) > > > I don't really like such cleaning patches. > It does not bring added value, but can create conflicts later when > backporting or rebasing. > Okay. I'll drop this patch in the next version. Thanks! Best regards, Tiwei Bie
Re: [dpdk-dev] [PATCH 4/5] net/virtio: remove redundant macro definitions for vector Rx
On Thu, Dec 07, 2017 at 10:18:14AM +0100, Maxime Coquelin wrote: > On 12/07/2017 06:30 AM, Tiwei Bie wrote: > > RTE_VIRTIO_VPMD_RX_BURST and RTE_VIRTIO_VPMD_RX_REARM_THRESH > > have been defined and used in virtio_rxtx_simple.h, but are > > defined agained in virtio_rxtx_simple_*.c. It just happens to > s/agained/again/ > Thanks! Best regards, Tiwei Bie > > work. So remove the redundant definitions from the *.c files. > > > > Signed-off-by: Tiwei Bie > > --- > > drivers/net/virtio/virtio_rxtx_simple_neon.c | 2 -- > > drivers/net/virtio/virtio_rxtx_simple_sse.c | 2 -- > > 2 files changed, 4 deletions(-) > > Apart from above typo: > > Reviewed-by: Maxime Coquelin > > Thanks, > Maxime
Re: [dpdk-dev] [PATCH v3 0/8] improve mlx4 Tx performance
On Wed, Dec 06, 2017 at 05:57:48PM +, Matan Azrad wrote: > This series improves mlx4 Tx performance and fix and clean some Tx code. > 1. 10% MPPS improvement for 1 queue, 1 core, 64B packets, txonly mode. > 2. 20% MPPS improvement for 1 queue, 1 core, 32B*4(segs) packets, txonly mode. > > V2: > Add missed function descriptions. > Accurate descriptions. > Change Tx descriptor alignment to be like Rx. > Move mlx4_fill_tx_data_seg to mlx4_rxtx.c and use rte_be32_t for byte count. > Change remain_size type to uin32_t. > Poisoning with memset. > > V3: > Accurate descriptions. > Fix poisoning from v2. For the remaining patches in the series: Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
Re: [dpdk-dev] [PATCH v2 2/2] examples/ipsec-secgw: add target queues in flow actions
Hi Anoob,
On Thu, Dec 07, 2017 at 03:17:40PM +0530, Anoob wrote:
> Hi Nelio,
>
>
> On 12/04/2017 07:41 PM, Nelio Laranjeiro wrote:
> > Mellanox INNOVA NIC needs to have final target queue actions to perform
> > inline crypto.
> >
> > Signed-off-by: Nelio Laranjeiro
> >
> > ---
> >
> > Changes in v2:
> >
> > * Test the rule by PASSTHRU/RSS/QUEUE and apply the first one validated.
> > ---
> > examples/ipsec-secgw/ipsec.c | 81
> >
> > examples/ipsec-secgw/ipsec.h | 2 +-
> > 2 files changed, 76 insertions(+), 7 deletions(-)
> >
> > diff --git a/examples/ipsec-secgw/ipsec.c b/examples/ipsec-secgw/ipsec.c
> > index 17bd7620d..f8823fb94 100644
> > --- a/examples/ipsec-secgw/ipsec.c
> > +++ b/examples/ipsec-secgw/ipsec.c
> > @@ -142,6 +142,7 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct
> > ipsec_sa *sa)
> > rte_eth_dev_get_sec_ctx(
> > sa->portid);
> > const struct rte_security_capability *sec_cap;
> > + int ret = 0;
> > sa->sec_session = rte_security_session_create(ctx,
> > &sess_conf, ipsec_ctx->session_pool);
> > @@ -173,6 +174,10 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct
> > ipsec_sa *sa)
> > return -1;
> > }
> > + sa->attr.egress = (sa->direction ==
> > + RTE_SECURITY_IPSEC_SA_DIR_EGRESS);
> > + sa->attr.ingress = (sa->direction ==
> > + RTE_SECURITY_IPSEC_SA_DIR_INGRESS);
> > sa->ol_flags = sec_cap->ol_flags;
> > sa->security_ctx = ctx;
> > sa->pattern[0].type = RTE_FLOW_ITEM_TYPE_ETH;
> > @@ -201,15 +206,79 @@ create_session(struct ipsec_ctx *ipsec_ctx, struct
> > ipsec_sa *sa)
> > sa->action[0].type = RTE_FLOW_ACTION_TYPE_SECURITY;
> > sa->action[0].conf = sa->sec_session;
> > - sa->action[1].type = RTE_FLOW_ACTION_TYPE_END;
> > -
> > - sa->attr.egress = (sa->direction ==
> > - RTE_SECURITY_IPSEC_SA_DIR_EGRESS);
> > - sa->attr.ingress = (sa->direction ==
> > - RTE_SECURITY_IPSEC_SA_DIR_INGRESS);
> > + if (sa->attr.ingress) {
> > + uint8_t rss_key[40];
> > + struct rte_eth_rss_conf rss_conf = {
> > + .rss_key = rss_key,
> > + .rss_key_len = 40,
> > + };
> > + struct rte_eth_dev *eth_dev;
> > + union {
> > + struct rte_flow_action_rss rss;
> > + struct {
> > + const struct rte_eth_rss_conf *rss_conf;
> > + uint16_t num;
> > + uint16_t queue[RTE_MAX_QUEUES_PER_PORT];
> > + } local;
> > + } action_rss;
> > + unsigned int i;
> > + unsigned int j;
> > +
> > + sa->action[2].type = RTE_FLOW_ACTION_TYPE_END;
> > + /*
> > +* Try implicitly PASSTHRU, it can also be
> > +* explicit.
> > +*/
> May be we can get rid of this check. You can do the check with RSS and then
> QUEUE. That should be fine. SECURITY is terminating on Cavium hardware, but
> according to the spec it is a non-terminating meta action. We can stick to
> that. For Cavium hardware the PMD will give success to SECURITY+QUEUE. That
> should resolve the issue.
I'll remove it in a v3, I will send it tomorrow to let a little more
time for other people to review.
Thanks,
--
Nélio Laranjeiro
6WIND
[dpdk-dev] document for rte_flow
Hi all, Thanks to the document Zhao Wei-san sent to me the other day, I could evaluate the L2 switching performance with Generic flow API, with "Intel X552 NIC (10G)". Now I am very interested in what happens with the performance with my Intel "XL710 NIC (40G)" and "Intel XXV710 (25G)". So my question is that, Is there the same rte_flow document for these NICs just as we as ixgbe NICs that is attached to this email as a word file? If there is one. I really would like you to send me one. Thanks in advance, Tetsuro Nakamura
Re: [dpdk-dev] [PATCH] vhost: support UDP Fragmentation Offload
Hi Maxime,
> -Original Message-
> From: Maxime Coquelin [mailto:maxime.coque...@redhat.com]
> Sent: Wednesday, December 6, 2017 4:37 PM
> To: Hu, Jiayu ; dev@dpdk.org
> Cc: y...@fridaylinux.org; Tan, Jianfeng
> Subject: Re: [dpdk-dev] [PATCH] vhost: support UDP Fragmentation Offload
>
>
>
> On 11/21/2017 07:56 AM, Jiayu Hu wrote:
> > In virtio, UDP Fragmentation Offload (UFO) includes two parts: host UFO
> > and guest UFO. Guest UFO means the frontend can receive large UDP
> packets,
> > and host UFO means the backend can receive large UDP packets. This patch
> > supports host UFO and guest UFO for vhost-user.
> >
> > Signed-off-by: Jiayu Hu
> > ---
> > lib/librte_mbuf/rte_mbuf.h| 7 +++
> > lib/librte_vhost/vhost.h | 2 ++
> > lib/librte_vhost/virtio_net.c | 10 ++
> > 3 files changed, 19 insertions(+)
> >
> > diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> > index ce8a05d..3d8cfc9 100644
> > --- a/lib/librte_mbuf/rte_mbuf.h
> > +++ b/lib/librte_mbuf/rte_mbuf.h
> > @@ -209,6 +209,13 @@ extern "C" {
> > /* add new TX flags here */
> >
> > /**
> > + * UDP Fragmentation Offload flag. This flag is used for enabling UDP
> > + * fragmentation in SW or in HW. When use UFO, mbuf->tso_segsz is used
> > + * to store the MSS of UDP fragments.
> > + */
> > +#define PKT_TX_UDP_SEG (1ULL << 42)
> > +
> > +/**
> >* Request security offload processing on the TX packet.
> >*/
> > #define PKT_TX_SEC_OFFLOAD(1ULL << 43)
> > diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
> > index 1cc81c1..fc109ef 100644
> > --- a/lib/librte_vhost/vhost.h
> > +++ b/lib/librte_vhost/vhost.h
> > @@ -206,10 +206,12 @@ struct vhost_msg {
> > (1ULL <<
> VHOST_USER_F_PROTOCOL_FEATURES) | \
> > (1ULL << VIRTIO_NET_F_HOST_TSO4) | \
> > (1ULL << VIRTIO_NET_F_HOST_TSO6) | \
> > + (1ULL << VIRTIO_NET_F_HOST_UFO) | \
> > (1ULL << VIRTIO_NET_F_CSUM)| \
> > (1ULL << VIRTIO_NET_F_GUEST_CSUM) | \
> > (1ULL << VIRTIO_NET_F_GUEST_TSO4) | \
> > (1ULL << VIRTIO_NET_F_GUEST_TSO6) | \
> > + (1ULL << VIRTIO_NET_F_GUEST_UFO) | \
>
> I actually have the same question as for GSO.
> Dos it impact performance as it seems enabled by default in QEMU?
When enable host/guest UFO, the frontend and the backend can communicate
via large UDP packets, thus reducing the number of packets to be processed.
It's similar with TSO for virtio. Therefore, I think host/guest UFO can bring
performance gains.
> How do you test it?
We launch testpmd with two vhost-user ports, which connect to two VMs in one
server. When launch qemu, we enable host and guest UFO with command
"host_ufo=on,
guest_ufo=on,csum=on". Then we run iperf in the two VMs to send large UDP
packets.
If you use "show port xstats all" in testpmd, you can see the vhost-user port
can receive
large UDP packets, which means host_ufo works. Additionally, if the forwarded
large UDP
packets can be received by the VM, it means guest_ufo also works.
Thanks,
Jiayu
>
> Thanks,
> Maxime
>
> > (1ULL << VIRTIO_RING_F_INDIRECT_DESC)
> | \
> > (1ULL << VIRTIO_NET_F_MTU) | \
> > (1ULL << VIRTIO_F_IOMMU_PLATFORM))
> > diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
> > index 6fee16e..3a3a0ad 100644
> > --- a/lib/librte_vhost/virtio_net.c
> > +++ b/lib/librte_vhost/virtio_net.c
> > @@ -188,6 +188,11 @@ virtio_enqueue_offload(struct rte_mbuf *m_buf,
> struct virtio_net_hdr *net_hdr)
> > net_hdr->gso_size = m_buf->tso_segsz;
> > net_hdr->hdr_len = m_buf->l2_len + m_buf->l3_len
> > + m_buf->l4_len;
> > + } else if (m_buf->ol_flags & PKT_TX_UDP_SEG) {
> > + net_hdr->gso_type = VIRTIO_NET_HDR_GSO_UDP;
> > + net_hdr->gso_size = m_buf->tso_segsz;
> > + net_hdr->hdr_len = m_buf->l2_len + m_buf->l3_len +
> > + m_buf->l4_len;
> > } else {
> > ASSIGN_UNLESS_EQUAL(net_hdr->gso_type, 0);
> > ASSIGN_UNLESS_EQUAL(net_hdr->gso_size, 0);
> > @@ -834,6 +839,11 @@ vhost_dequeue_offload(struct virtio_net_hdr
> *hdr, struct rte_mbuf *m)
> > m->tso_segsz = hdr->gso_size;
> > m->l4_len = (tcp_hdr->data_off & 0xf0) >> 2;
> > break;
> > + case VIRTIO_NET_HDR_GSO_UDP:
> > + m->ol_flags |= PKT_TX_UDP_SEG;
> > + m->tso_segsz = hdr->gso_size;
> > + m->l4_len = sizeof(struct udp_hdr);
> > + break;
> > default:
> > RTE_LOG(WARNING, VHOST_DATA,
> > "unsuppo
Re: [dpdk-dev] [PATCH 1/3] eal: update legacy modules dynamic logs regex
On Wed, Nov 22, 2017 at 02:58:04PM +0530, Pavan Nikhilesh wrote:
> Update legacy log types regex strings used for registering dynamic logs.
>
> Signed-off-by: Pavan Nikhilesh
> ---
> Note:
> This patchset is based on patch set
> http://dpdk.org/dev/patchwork/patch/31443/
> followed by ml discussion
> http://dpdk.org/ml/archives/dev/2017-November/081953.html
>
> lib/librte_eal/common/eal_common_log.c | 39
> +-
> 1 file changed, 20 insertions(+), 19 deletions(-)
>
> diff --git a/lib/librte_eal/common/eal_common_log.c
> b/lib/librte_eal/common/eal_common_log.c
> index e894b75ec..fa9ac7247 100644
> --- a/lib/librte_eal/common/eal_common_log.c
> +++ b/lib/librte_eal/common/eal_common_log.c
> @@ -219,26 +219,27 @@ struct logtype {
> };
>
> static const struct logtype logtype_strings[] = {
> - {RTE_LOGTYPE_EAL,"eal"},
> - {RTE_LOGTYPE_MALLOC, "malloc"},
> - {RTE_LOGTYPE_RING, "ring"},
> - {RTE_LOGTYPE_MEMPOOL,"mempool"},
> - {RTE_LOGTYPE_TIMER, "timer"},
> + {RTE_LOGTYPE_EAL,"lib.eal"},
> + {RTE_LOGTYPE_MALLOC, "lib.malloc"},
> + {RTE_LOGTYPE_RING, "lib.ring"},
> + {RTE_LOGTYPE_MEMPOOL,"lib.mempool"},
> + {RTE_LOGTYPE_TIMER, "lib.timer"},
[...]
I agree it's much better to have the library prefixed by "lib.".
Reviewed-by: Olivier Matz
Re: [dpdk-dev] [PATCH 2/3] eal: update default log levels
On Wed, Nov 22, 2017 at 02:58:05PM +0530, Pavan Nikhilesh wrote: > Use global default loglevel to DEBUG(8) and dynamic default loglevel > to INFO(7). > > Signed-off-by: Pavan Nikhilesh Reviewed-by: Olivier Matz
Re: [dpdk-dev] [PATCH 3/3] logs: remove log level config option
On Wed, Nov 22, 2017 at 02:58:06PM +0530, Pavan Nikhilesh wrote:
> Remove RTE_LOG_LEVEL config option, use existing RTE_LOG_DP_LEVEL config
> option for controlling datapath log level.
> RTE_LOG_LEVEL is no longer needed as dynamic logging can be used to
> control global and module specific log levels.
>
> Signed-off-by: Pavan Nikhilesh
Tome, removing the RTE_LOG_LEVEL goes in the good direction (less
compile-time options).
[...]
> --- a/drivers/net/ena/base/ena_plat_dpdk.h
> +++ b/drivers/net/ena/base/ena_plat_dpdk.h
> @@ -96,7 +96,7 @@ typedef uint64_t dma_addr_t;
> #define ENA_GET_SYSTEM_USECS()
> \
> (rte_get_timer_cycles() * US_PER_S / rte_get_timer_hz())
>
> -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> #define ENA_ASSERT(cond, format, arg...) \
> do {\
> if (unlikely(!(cond))) {\
> diff --git a/drivers/net/sfc/sfc_debug.h b/drivers/net/sfc/sfc_debug.h
> index 92eba9c38..3f9ccf1e6 100644
> --- a/drivers/net/sfc/sfc_debug.h
> +++ b/drivers/net/sfc/sfc_debug.h
> @@ -35,7 +35,7 @@
> #include
>
> #ifdef RTE_LIBRTE_SFC_EFX_DEBUG
> -/* Avoid dependency from RTE_LOG_LEVEL to be able to enable debug check
> +/* Avoid dependency from RTE_LOG_DP_LEVEL to be able to enable debug check
> * in the driver only.
> */
> #define SFC_ASSERT(exp) RTE_VERIFY(exp)
> diff --git a/examples/l3fwd-acl/main.c b/examples/l3fwd-acl/main.c
> index e50b1a1a8..eb78decdd 100644
> --- a/examples/l3fwd-acl/main.c
> +++ b/examples/l3fwd-acl/main.c
> @@ -68,7 +68,7 @@
> #include
> #include
>
> -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> #define L3FWDACL_DEBUG
> #endif
> #define DO_RFC_1812_CHECKS
> diff --git a/test/test/test.h b/test/test/test.h
> index 08ffe949c..8fdb3045e 100644
> --- a/test/test/test.h
> +++ b/test/test/test.h
> @@ -204,7 +204,7 @@ struct unit_test_case {
>
> #define TEST_CASES_END() { NULL, NULL, NULL, NULL, 0 }
>
> -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> #define TEST_HEXDUMP(file, title, buf, len) rte_hexdump(file, title, buf,
> len)
> #else
> #define TEST_HEXDUMP(file, title, buf, len) do {} while (0)
For drivers, it looks correct to replace RTE_LOG_LEVEL by
RTE_LOG_DP_LEVEL, from what I see it's about dataplane logs.
For l3fwd, I'm less sure, but it could make sense too.
For test, I think we should replace TEST_HEXDUMP() by a function
that checks the runtime log level instead of relying on RTE_LOG_DP_LEVEL.
I can submit a patch for this.
Olivier
Re: [dpdk-dev] 17.08.1 patches review and test
On Mon, Dec 04, 2017 at 08:40:03PM +, Patil, Harish wrote: > > > -Original Message- > From: dev on behalf of Yuanhan Liu > > Date: Monday, November 27, 2017 at 4:21 AM > To: dpdk stable > Cc: "dev@dpdk.org" , "Xu, Qian Q" > Subject: [dpdk-dev] 17.08.1 patches review and test > > >Hi all, > > > >Here is a list of patches targeted for stable release 17.08.1. Please > >help review and test. The planned date for the final release is 7th, > >Dec. Before that, please shout if anyone has objections with these > >patches being applied. > > > >These patches are located at branch 17.08 of dpdk-stable repo: > >http://dpdk.org/browse/dpdk-stable/ > > > >Thanks. > > > >--yliu > > > >--- > >Aaron Conole (1): > > net/enic: fix assignment > > > >Ajit Khaparde (28): > > net/bnxt: fix HWRM macros and locking > > net/bnxt: use 64-bits of address for VLAN table > > net/bnxt: fix an issue with group id calculation > > net/bnxt: fix calculation of number of pools > > net/bnxt: handle multi queue mode properly > > net/bnxt: fix Rx handling and buffer allocation logic > > net/bnxt: fix an issue with broadcast traffic > > net/bnxt: fix usage of VMDq flags > > net/bnxt: set checksum offload flags correctly > > net/bnxt: update status of Rx IP/L4 CKSUM > > net/bnxt: fix config RSS update > > net/bnxt: set the hash key size > > net/bnxt: fix per queue stats display in xstats > > net/bnxt: fix interrupt handler > > net/bnxt: fix number of MAC addresses for VMDq > > net/bnxt: fix the association of a MACVLAN per VNIC > > net/bnxt: fix Tx offload capability > > net/bnxt: fix Rx offload capability > > net/bnxt: handle Rx multi queue creation properly > > net/bnxt: remove redundant code parsing pool map > > net/bnxt: fix a bit shift operation > > net/bnxt: fix a potential null pointer dereference > > net/bnxt: fix a potential null pointer dereference > > net/bnxt: fix a pointer deref before null check > > net/bnxt: fix an unused value > > net/bnxt: check VLANs from pool map only for VMDq > > net/bnxt: do not set hash type unnecessarily > > net/bnxt: fix VLAN spoof configuration > > > >Akhil Goyal (2): > > test/crypto: fix dpaa2 sec macros and definitions > > net/dpaa2: set queues after reconfiguration > > > >Alejandro Lucero (2): > > net/nfp: fix RSS > > net/nfp: fix Rx interrupt when multiqueue > > > >Alok Makhariya (2): > > crypto/dpaa2_sec: remove ICV memset on decryption side > > crypto/dpaa2_sec: add check for segmented buffer > > > >Anatoly Burakov (1): > > vfio: fix secondary process initialization > > > >Andrey Chilikin (1): > > net/i40e: fix flexible payload configuration > > > >Aviad Yehezkel (4): > > examples/ipsec-secgw: fix crypto device mapping > > examples/ipsec-secgw: fix session creation > > examples/ipsec-secgw: fix AAD length setting > > app/testpmd: fix build without ixgbe and bnxt PMDs > > > >Beilei Xing (1): > > net/i40e: fix VF device stop issue > > > >Chas Williams (1): > > net/vmxnet3: fix memory leak when releasing queues > > > >Congwen Zhang (1): > > net/cxgbe: fix memory leak > > > >Daniel Mrzyglod (3): > > net/virtio: fix untrusted scalar value > > app/testpmd: fix DDP package filesize detection > > net/bonding: fix default aggregator mode to stable > > > >David Harton (2): > > net/vmxnet3: fix MAC address set > > net/i40e: fix i40evf MAC filter table > > > >Ferruh Yigit (4): > > ethdev: fix ABI version > > ethdev: revert use port name from device structure > > igb_uio: remove device reset in open > > net/qede: fix icc build > > > >Gaetan Rivet (1): > > net/failsafe: fix errno set on command execution > > > >Gowrishankar Muthukrishnan (1): > > net/bonding: support bifurcated driver in eal > > > >Guduri Prathyusha (2): > > examples/l3fwd: fix NEON instructions > > examples/l3fwd: fix aliasing in port grouping > > > >Harish Patil (2): > > net/qede: fix supported packet types > > net/qede: fix to re-enable LRO during device start > > > >Hemant Agrawal (3): > > net/dpaa2: fix the Tx handling of non HW pool bufs > > examples/l2fwd-crypto: fix uninitialized errno value > > app/crypto-perf: fix uninitialized errno value > > > >Ian Stokes (1): > > cryptodev: fix build with -Ofast > > > >Ivan Malov (2): > > net/sfc: specify correct scale table size on Rx start > > net/sfc: fix unused variable in RSS-agnostic build > > > >Jacek Piasecki (1): > > examples/vhost_scsi: fix product id string termination > > > >Jasvinder Singh (1): > > examples/qos_sched: fix uninitialized config > > > >Jerin Jacob (1): > > timer: use 64-bit specific code on more platforms > > > >Jianbo Liu (1): > > net/i40e: fix Rx packets number for NEON > > > >Jiayu Hu (1): > > gro
Re: [dpdk-dev] [PATCH 3/3] logs: remove log level config option
Hi Oliver,
Thanks for the review.
On Thu, Dec 07, 2017 at 02:21:21PM +0100, Olivier MATZ wrote:
> On Wed, Nov 22, 2017 at 02:58:06PM +0530, Pavan Nikhilesh wrote:
> > Remove RTE_LOG_LEVEL config option, use existing RTE_LOG_DP_LEVEL config
> > option for controlling datapath log level.
> > RTE_LOG_LEVEL is no longer needed as dynamic logging can be used to
> > control global and module specific log levels.
> >
> > Signed-off-by: Pavan Nikhilesh
>
> Tome, removing the RTE_LOG_LEVEL goes in the good direction (less
> compile-time options).
>
> [...]
>
> > --- a/drivers/net/ena/base/ena_plat_dpdk.h
> > +++ b/drivers/net/ena/base/ena_plat_dpdk.h
> > @@ -96,7 +96,7 @@ typedef uint64_t dma_addr_t;
> > #define ENA_GET_SYSTEM_USECS()
> > \
> > (rte_get_timer_cycles() * US_PER_S / rte_get_timer_hz())
> >
> > -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> > +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> > #define ENA_ASSERT(cond, format, arg...) \
> > do {\
> > if (unlikely(!(cond))) {\
> > diff --git a/drivers/net/sfc/sfc_debug.h b/drivers/net/sfc/sfc_debug.h
> > index 92eba9c38..3f9ccf1e6 100644
> > --- a/drivers/net/sfc/sfc_debug.h
> > +++ b/drivers/net/sfc/sfc_debug.h
> > @@ -35,7 +35,7 @@
> > #include
> >
> > #ifdef RTE_LIBRTE_SFC_EFX_DEBUG
> > -/* Avoid dependency from RTE_LOG_LEVEL to be able to enable debug check
> > +/* Avoid dependency from RTE_LOG_DP_LEVEL to be able to enable debug check
> > * in the driver only.
> > */
> > #define SFC_ASSERT(exp)RTE_VERIFY(exp)
> > diff --git a/examples/l3fwd-acl/main.c b/examples/l3fwd-acl/main.c
> > index e50b1a1a8..eb78decdd 100644
> > --- a/examples/l3fwd-acl/main.c
> > +++ b/examples/l3fwd-acl/main.c
> > @@ -68,7 +68,7 @@
> > #include
> > #include
> >
> > -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> > +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> > #define L3FWDACL_DEBUG
> > #endif
> > #define DO_RFC_1812_CHECKS
> > diff --git a/test/test/test.h b/test/test/test.h
> > index 08ffe949c..8fdb3045e 100644
> > --- a/test/test/test.h
> > +++ b/test/test/test.h
> > @@ -204,7 +204,7 @@ struct unit_test_case {
> >
> > #define TEST_CASES_END() { NULL, NULL, NULL, NULL, 0 }
> >
> > -#if RTE_LOG_LEVEL >= RTE_LOG_DEBUG
> > +#if RTE_LOG_DP_LEVEL >= RTE_LOG_DEBUG
> > #define TEST_HEXDUMP(file, title, buf, len) rte_hexdump(file, title, buf,
> > len)
> > #else
> > #define TEST_HEXDUMP(file, title, buf, len) do {} while (0)
>
> For drivers, it looks correct to replace RTE_LOG_LEVEL by
> RTE_LOG_DP_LEVEL, from what I see it's about dataplane logs.
>
> For l3fwd, I'm less sure, but it could make sense too.
>
> For test, I think we should replace TEST_HEXDUMP() by a function
> that checks the runtime log level instead of relying on RTE_LOG_DP_LEVEL.
> I can submit a patch for this.
Agreed.
Pavan
>
> Olivier
Re: [dpdk-dev] [PATCH v2 06/10] net/virtio: fix queue setup consistency
Hi Tiwei,
On Wed, Dec 06, 2017 at 01:25:29PM +0800, Tiwei Bie wrote:
> Hi Maxime and Olivier:
>
> On Thu, Sep 07, 2017 at 02:13:43PM +0200, Olivier Matz wrote:
> [...]
> > diff --git a/drivers/net/virtio/virtio_ethdev.c
> > b/drivers/net/virtio/virtio_ethdev.c
> > index 8eee3ff80..c7888f103 100644
> > --- a/drivers/net/virtio/virtio_ethdev.c
> > +++ b/drivers/net/virtio/virtio_ethdev.c
> > @@ -1737,6 +1737,19 @@ virtio_dev_start(struct rte_eth_dev *dev)
> > struct virtnet_rx *rxvq;
> > struct virtnet_tx *txvq __rte_unused;
> > struct virtio_hw *hw = dev->data->dev_private;
> > + int ret;
> > +
> > + /* Finish the initialization of the queues */
> > + for (i = 0; i < dev->data->nb_rx_queues; i++) {
> > + ret = virtio_dev_rx_queue_setup_finish(dev, i);
> > + if (ret < 0)
> > + return ret;
> > + }
>
> I'm trying to fix an issue [1] reported by Antonio. And during
> the debugging, I found that vector Rx of virtio PMD has been
> broken (when doing port stop/start) since below two patches were
> applied:
>
> 25bf7a0b0936 ("vhost: make error handling consistent in Rx path")
>-- needed on the Tx side (testpmd/vhost-pmd in below test)
> efc83a1e7fc3 ("net/virtio: fix queue setup consistency")
>-- needed on the Rx side (testpmd/virtio-user in below test)
Just to be sure I understand properly: each of these 2 patches
break a different part your test case?
I tried to reproduce your test case (the working case first):
- on 0c4f909c17 (the commit before the efc83a1e7fc3)
- without the patch disabling mergeable Rx
No packet is received. Am I doing something wrong? Please see the
log:
cd /root/dpdk.org
git checkout -b test 0c4f909c17
rm -rf build && make config T=x86_64-native-linuxapp-gcc && make -j32
insmod build/kmod/igb_uio.ko
echo 1000 >
/sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages
echo 1000 >
/sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages
mkdir -p /mnt/huge
mount -t hugetlbfs none /mnt/huge
# term 1: testpmd/vhost-pmd
/root/dpdk.org/build/app/testpmd -l 1,2 \
--socket-mem 512,512 \
--file-prefix=vhost \
--no-pci \
--vdev=net_vhost0,iface=/tmp/socket-0,queues=1 \
-- \
--port-topology=chained \
-i \
--nb-cores=1
# term 2: virtio-user
/root/dpdk.org/build/app/testpmd -l 5,6 \
--socket-mem 512,512 \
--file-prefix=virtio-user \
--no-pci \
--vdev=net_virtio_user0,path=/tmp/socket-0 \
-- \
--port-topology=chained \
-i \
--nb-cores=1 \
--disable-hw-vlan \
--txqflags=0xf01
testpmd> set fwd rxonly
testpmd> start
# back to term1: vhost
testpmd> set burst 1
testpmd> set fwd rxonly
testpmd> start tx_first 1
testpmd> stop
Result on term1:
-- Forward statistics for port 0 --
RX-packets: 0 RX-dropped: 0 RX-total: 0
TX-packets: 0 TX-dropped: 1 TX-total: 1
+++ Accumulated forward statistics for all ports+++
RX-packets: 0 RX-dropped: 0 RX-total: 0
TX-packets: 0 TX-dropped: 1 TX-total: 1
Olivier
Re: [dpdk-dev] [PATCH] lib/cmdline: support backspace key
Hi Xueming,
On Wed, Nov 15, 2017 at 11:51:56PM +0800, Xueming Li wrote:
> Support windows putty "\b"(Ctrl-H) backspace key.
>
> Signed-off-by: Xueming Li
> ---
> lib/librte_cmdline/cmdline_rdline.c | 1 +
> lib/librte_cmdline/cmdline_vt100.c | 1 +
> lib/librte_cmdline/cmdline_vt100.h | 1 +
> 3 files changed, 3 insertions(+)
>
> diff --git a/lib/librte_cmdline/cmdline_rdline.c
> b/lib/librte_cmdline/cmdline_rdline.c
> index 1ef2258d0..167657f4b 100644
> --- a/lib/librte_cmdline/cmdline_rdline.c
> +++ b/lib/librte_cmdline/cmdline_rdline.c
> @@ -331,6 +331,7 @@ rdline_char_in(struct rdline *rdl, char c)
>
> /* delete 1 char from the left */
> case CMDLINE_KEY_BKSPACE:
> + case CMDLINE_KEY_BKSPACE1:
> if(!cirbuf_del_tail_safe(&rdl->left)) {
> rdline_puts(rdl, vt100_bs);
> display_right_buffer(rdl, 1);
I would have used CMDLINE_KEY_BKSPACE2 instead of CMDLINE_KEY_BKSPACE1 :)
> diff --git a/lib/librte_cmdline/cmdline_vt100.c
> b/lib/librte_cmdline/cmdline_vt100.c
> index a253e8b6c..bb317507e 100644
> --- a/lib/librte_cmdline/cmdline_vt100.c
> +++ b/lib/librte_cmdline/cmdline_vt100.c
> @@ -95,6 +95,7 @@ const char *cmdline_vt100_commands[] = {
> "\020",
> "\016",
> "\033\144",
> + "\b",
> };
>
> void
You can reuse vt100_bs instead of '\b', which is defined in
cmdline_vt100.h (note that '\010' == '\b')
Thanks
Olivier
[dpdk-dev] [VMXNET3]: device activation failed, while hotplug a vmxnet3 nic
Hi,
I hit an interesting issue recently, while I try to hotplug a vmxnet3 nic to a
vm.
Here is the environment,
ESXi 6.0
DPDK-16.07
Guest linux kernel 3.10.20
VM is running in ESXi6.0, I use Vsphere Client to Edit setting, add a vmxnet3
nic to this running vm (hotplug). Device is failed to start.
In vmxnet3_dev_start(),
/* Activate device by register write */
VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_ACTIVATE_DEV);
status = VMXNET3_READ_BAR1_REG(hw, VMXNET3_REG_CMD);
if (status != 0) {
PMD_INIT_LOG(ERR, "Device activation: UNSUCCESSFUL");
return -1;
}
Also, I have these error log:
In /var/log/vmkernel.log
…
2017-12-07T14:00:37.524Z cpu18:32859)Net: 3649: disconnected client
from port 0x20005f9
2017-12-07T14:00:38.559Z cpu19:4357909)Vmxnet3: 12840: invalid intr index 56
for tq 0, only 1 intr allocated
…
In /vmfs/volumes/57073409-d705da7d-8b7c-44a8422585c2/KFC_8373/vmware.log
…
2017-12-07T14:00:38.559Z| vcpu-0| I120: VMXNET3 user: failed to
activate Ethernet1, status: 0xbad0001
..
I also notice these llinks,
https://kb.vmware.com/s/article/2136932
“ESXi 5.5 Update 3 and 6.0 permits only a maximum Rx Ring #2 size of 2048”
https://mail-archive.com/dev@dpdk.org/msg14401.html“Vmxnet3
activation of device fails in DPDK1.7”
in my envirment, rx desc num is 1024, as same as tx desc num. I’m not sure if
it is the root case.
For more information, vmxnet3 nics are OK while VM is cold boot. ( nic is not
hotplugged )
Anyone met this issue before? It seems like interrupt problem, but I m not sure
it is vmxnet3 pmd driver issue or igb uio issue.
I’m glad to have your reply.
Best regards,
XU Ke
Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory
On Wed, Nov 15, 2017 at 11:54:02PM +0800, Xueming Li wrote: > Initialize binary result memory before parsing to avoid garbage in > parsing result. > > Signed-off-by: Xueming Li > --- > lib/librte_cmdline/cmdline_parse.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/lib/librte_cmdline/cmdline_parse.c > b/lib/librte_cmdline/cmdline_parse.c > index 3e12ee54f..9124758f1 100644 > --- a/lib/librte_cmdline/cmdline_parse.c > +++ b/lib/librte_cmdline/cmdline_parse.c > @@ -267,6 +267,8 @@ cmdline_parse(struct cmdline *cl, const char * buf) > if (!cl || !buf) > return CMDLINE_PARSE_BAD_ARGS; > > + memset(tmp_result.buf, 0, sizeof(tmp_result.buf)); > + > ctx = cl->ctx; > > /* Did you see an issue (a bug or a crash) without the memset()? Or is it to avoid filling unused fields in the parsed struct? I'm not sure if your patch is enough: cmdline_parse() calls match_inst() for each registered command. If a command partially matches (only the first tokens), the buffer is modified. So the next one will start with a dirty buffer. I suggest to put the memset() in match_inst() instead. Something like this: if (resbuf != NULL) memset(resbuf, 0, resbuf_size); It will reset the buffer before using it. Olivier
[dpdk-dev] [PATCH v1] lib/cmdline: support backspace key
Support windows putty "\b"(Ctrl-H) backspace key.
Signed-off-by: Xueming Li
---
lib/librte_cmdline/cmdline_rdline.c | 1 +
lib/librte_cmdline/cmdline_vt100.c | 1 +
lib/librte_cmdline/cmdline_vt100.h | 1 +
3 files changed, 3 insertions(+)
diff --git a/lib/librte_cmdline/cmdline_rdline.c
b/lib/librte_cmdline/cmdline_rdline.c
index 1ef2258d0..a1aa88c96 100644
--- a/lib/librte_cmdline/cmdline_rdline.c
+++ b/lib/librte_cmdline/cmdline_rdline.c
@@ -331,6 +331,7 @@ rdline_char_in(struct rdline *rdl, char c)
/* delete 1 char from the left */
case CMDLINE_KEY_BKSPACE:
+ case CMDLINE_KEY_BKSPACE2:
if(!cirbuf_del_tail_safe(&rdl->left)) {
rdline_puts(rdl, vt100_bs);
display_right_buffer(rdl, 1);
diff --git a/lib/librte_cmdline/cmdline_vt100.c
b/lib/librte_cmdline/cmdline_vt100.c
index a253e8b6c..c7b8e60bd 100644
--- a/lib/librte_cmdline/cmdline_vt100.c
+++ b/lib/librte_cmdline/cmdline_vt100.c
@@ -95,6 +95,7 @@ const char *cmdline_vt100_commands[] = {
"\020",
"\016",
"\033\144",
+ vt100_bs,
};
void
diff --git a/lib/librte_cmdline/cmdline_vt100.h
b/lib/librte_cmdline/cmdline_vt100.h
index 963add8df..b94b24e8c 100644
--- a/lib/librte_cmdline/cmdline_vt100.h
+++ b/lib/librte_cmdline/cmdline_vt100.h
@@ -117,6 +117,7 @@ extern "C" {
#define CMDLINE_KEY_CTRL_P 23
#define CMDLINE_KEY_CTRL_N 24
#define CMDLINE_KEY_META_D 25
+#define CMDLINE_KEY_BKSPACE2 26
extern const char *cmdline_vt100_commands[];
--
2.13.3
[dpdk-dev] [PATCH] net/szedata2: fix check of mmap return value
Fixes: 9eddbdb4b094 ("szedata2: support link state operations")
Cc: sta...@dpdk.org
Signed-off-by: Matej Vido
---
drivers/net/szedata2/rte_eth_szedata2.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/szedata2/rte_eth_szedata2.c
b/drivers/net/szedata2/rte_eth_szedata2.c
index 74f151c..af0580a 100644
--- a/drivers/net/szedata2/rte_eth_szedata2.c
+++ b/drivers/net/szedata2/rte_eth_szedata2.c
@@ -1553,7 +1553,7 @@ struct pmd_internals {
pci_dev->mem_resource[PCI_RESOURCE_NUMBER].len,
PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
close(fd);
- if (pci_resource_ptr == NULL) {
+ if (pci_resource_ptr == MAP_FAILED) {
RTE_LOG(ERR, PMD, "Could not mmap file %s (fd = %d)\n",
rsc_filename, fd);
return -EINVAL;
--
1.8.4
Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory
Hi Olivier, > -Original Message- > From: Olivier MATZ [mailto:olivier.m...@6wind.com] > Sent: Thursday, December 7, 2017 10:48 PM > To: Xueming(Steven) Li > Cc: dev@dpdk.org > Subject: Re: [PATCH] lib/cmdline: init parse result memeory > > On Wed, Nov 15, 2017 at 11:54:02PM +0800, Xueming Li wrote: > > Initialize binary result memory before parsing to avoid garbage in > > parsing result. > > > > Signed-off-by: Xueming Li > > --- > > lib/librte_cmdline/cmdline_parse.c | 2 ++ > > 1 file changed, 2 insertions(+) > > > > diff --git a/lib/librte_cmdline/cmdline_parse.c > > b/lib/librte_cmdline/cmdline_parse.c > > index 3e12ee54f..9124758f1 100644 > > --- a/lib/librte_cmdline/cmdline_parse.c > > +++ b/lib/librte_cmdline/cmdline_parse.c > > @@ -267,6 +267,8 @@ cmdline_parse(struct cmdline *cl, const char * buf) > > if (!cl || !buf) > > return CMDLINE_PARSE_BAD_ARGS; > > > > + memset(tmp_result.buf, 0, sizeof(tmp_result.buf)); > > + > > ctx = cl->ctx; > > > > /* > > > Did you see an issue (a bug or a crash) without the memset()? > Or is it to avoid filling unused fields in the parsed struct? Yes, I'm using same struct for some similar commands, have to avoid filling unused fields. > > I'm not sure if your patch is enough: cmdline_parse() calls match_inst() > for each registered command. If a command partially matches (only the > first tokens), the buffer is modified. So the next one will start with a > dirty buffer. > > I suggest to put the memset() in match_inst() instead. Something like this: > > if (resbuf != NULL) > memset(resbuf, 0, resbuf_size); > > > It will reset the buffer before using it. It was there for performance concern, since the buffer could be tainted, I'll upload a new version according to your suggestion, appreciate your suggestion. > > Olivier
Re: [dpdk-dev] [PATCH v1] lib/cmdline: support backspace key
On Thu, Dec 07, 2017 at 10:52:02PM +0800, Xueming Li wrote: > Support windows putty "\b"(Ctrl-H) backspace key. > > Signed-off-by: Xueming Li Acked-by: Olivier Matz
Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory
> -Original Message- > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Xueming(Steven) Li > Sent: Thursday, December 7, 2017 11:05 PM > To: Olivier MATZ > Cc: dev@dpdk.org > Subject: Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory > > Hi Olivier, > > > -Original Message- > > From: Olivier MATZ [mailto:olivier.m...@6wind.com] > > Sent: Thursday, December 7, 2017 10:48 PM > > To: Xueming(Steven) Li > > Cc: dev@dpdk.org > > Subject: Re: [PATCH] lib/cmdline: init parse result memeory > > > > On Wed, Nov 15, 2017 at 11:54:02PM +0800, Xueming Li wrote: > > > Initialize binary result memory before parsing to avoid garbage in > > > parsing result. > > > > > > Signed-off-by: Xueming Li > > > --- > > > lib/librte_cmdline/cmdline_parse.c | 2 ++ > > > 1 file changed, 2 insertions(+) > > > > > > diff --git a/lib/librte_cmdline/cmdline_parse.c > > > b/lib/librte_cmdline/cmdline_parse.c > > > index 3e12ee54f..9124758f1 100644 > > > --- a/lib/librte_cmdline/cmdline_parse.c > > > +++ b/lib/librte_cmdline/cmdline_parse.c > > > @@ -267,6 +267,8 @@ cmdline_parse(struct cmdline *cl, const char * buf) > > > if (!cl || !buf) > > > return CMDLINE_PARSE_BAD_ARGS; > > > > > > + memset(tmp_result.buf, 0, sizeof(tmp_result.buf)); > > > + > > > ctx = cl->ctx; > > > > > > /* > > > > > > Did you see an issue (a bug or a crash) without the memset()? > > Or is it to avoid filling unused fields in the parsed struct? > Yes, I'm using same struct for some similar commands, have to avoid > filling unused fields. > > > > > I'm not sure if your patch is enough: cmdline_parse() calls > > match_inst() for each registered command. If a command partially > > matches (only the first tokens), the buffer is modified. So the next > > one will start with a dirty buffer. > > > > I suggest to put the memset() in match_inst() instead. Something like > this: > > > > if (resbuf != NULL) > > memset(resbuf, 0, resbuf_size); > > > > > > It will reset the buffer before using it. > It was there for performance concern, since the buffer could be tainted, > I'll upload a new version according to your suggestion, appreciate your > suggestion. Rte_flow CLI seems to be relying on modified buffer, add Adrien: testpmd> flow create 0 ingress pattern eth / ipv4 / udp / vxlan / end actions rss queues 1 2 end level 1 / mark id 0x123456 / end Caught error type 2 (flow rule (handle)): no valid action > > > > > Olivier
[dpdk-dev] [dpdk-announce] DPDK 17.08.1 released
Hi all, Here is a new stable release: http://fast.dpdk.org/rel/dpdk-17.08.1.tar.xz The git tree is at: http://dpdk.org/browse/dpdk-stable/ Thanks. --yliu --- app/test-crypto-perf/cperf_options_parsing.c | 20 +- app/test-crypto-perf/cperf_test_vector_parsing.c | 55 +++ app/test-crypto-perf/cperf_test_verify.c | 5 + app/test-pmd/cmdline.c | 4 +- app/test-pmd/config.c | 53 ++- app/test-pmd/parameters.c | 18 +- app/test-pmd/testpmd.c | 22 +- config/common_base | 1 + doc/guides/nics/features/enic.ini | 1 + doc/guides/rel_notes/release_17_08.rst | 201 drivers/crypto/aesni_gcm/aesni_gcm_pmd.c | 2 +- drivers/crypto/aesni_gcm/aesni_gcm_pmd_ops.c | 2 +- drivers/crypto/aesni_mb/rte_aesni_mb_pmd.c | 7 +- drivers/crypto/aesni_mb/rte_aesni_mb_pmd_ops.c | 4 +- drivers/crypto/armv8/rte_armv8_pmd_ops.c | 2 +- drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c| 11 +- drivers/crypto/kasumi/rte_kasumi_pmd_ops.c | 2 +- drivers/crypto/null/null_crypto_pmd_ops.c | 2 +- drivers/crypto/openssl/rte_openssl_pmd.c | 15 +- drivers/crypto/openssl/rte_openssl_pmd_ops.c | 2 +- drivers/crypto/qat/qat_crypto_capabilities.h | 24 +- drivers/crypto/snow3g/rte_snow3g_pmd_ops.c | 2 +- drivers/crypto/zuc/rte_zuc_pmd_ops.c | 2 +- drivers/net/ark/ark_ethdev.c | 2 +- drivers/net/bnxt/bnxt.h| 4 +- drivers/net/bnxt/bnxt_cpr.c| 2 + drivers/net/bnxt/bnxt_ethdev.c | 41 +- drivers/net/bnxt/bnxt_hwrm.c | 525 + drivers/net/bnxt/bnxt_hwrm.h | 2 +- drivers/net/bnxt/bnxt_irq.c| 11 +- drivers/net/bnxt/bnxt_ring.c | 4 +- drivers/net/bnxt/bnxt_rxq.c| 199 drivers/net/bnxt/bnxt_rxr.c| 47 +- drivers/net/bnxt/bnxt_rxr.h| 16 + drivers/net/bnxt/bnxt_stats.c | 6 +- drivers/net/bnxt/bnxt_txr.c| 32 +- drivers/net/bnxt/bnxt_txr.h| 21 + drivers/net/bnxt/rte_pmd_bnxt.c| 18 +- drivers/net/bonding/rte_eth_bond_api.c | 9 +- drivers/net/bonding/rte_eth_bond_args.c| 35 +- drivers/net/bonding/rte_eth_bond_pmd.c | 46 +- drivers/net/cxgbe/base/t4_hw.c | 3 + drivers/net/dpaa2/dpaa2_ethdev.c | 4 +- drivers/net/dpaa2/dpaa2_rxtx.c | 41 +- drivers/net/e1000/igb_ethdev.c | 18 +- drivers/net/enic/base/vnic_dev.c | 10 +- drivers/net/enic/enic_ethdev.c | 43 ++ drivers/net/enic/enic_main.c | 13 +- drivers/net/enic/enic_rxtx.c | 25 +- drivers/net/failsafe/failsafe_args.c | 32 +- drivers/net/failsafe/failsafe_eal.c| 28 +- drivers/net/failsafe/failsafe_ether.c | 1 + drivers/net/failsafe/failsafe_ops.c| 118 - drivers/net/failsafe/failsafe_private.h| 9 +- drivers/net/failsafe/failsafe_rxtx.c | 3 +- drivers/net/i40e/i40e_ethdev.c | 117 +++-- drivers/net/i40e/i40e_ethdev.h | 4 +- drivers/net/i40e/i40e_ethdev_vf.c | 151 +- drivers/net/i40e/i40e_pf.c | 19 +- drivers/net/i40e/i40e_rxtx.c | 110 +++-- drivers/net/i40e/i40e_rxtx_vec_neon.c | 19 +- drivers/net/i40e/i40e_tm.c | 102 ++-- drivers/net/ixgbe/ixgbe_ethdev.c | 69 ++- drivers/net/ixgbe/ixgbe_fdir.c | 3 +- drivers/net/ixgbe/ixgbe_flow.c | 29 +- drivers/net/ixgbe/ixgbe_rxtx.c | 41 +- drivers/net/ixgbe/ixgbe_tm.c | 91 ++-- drivers/net/kni/rte_eth_kni.c | 2 - drivers/net/liquidio/lio_ethdev.c | 1 + drivers/net/mlx5/mlx5.c| 10 +- drivers/net/mlx5/mlx5_ethdev.c | 28 +- drivers/net/mlx5/mlx5_rxq.c| 6 +- drivers/net/mlx5/mlx5_rxtx.c | 68 +-- drivers/net/mlx5/mlx5_rxtx.h | 2 +- drivers/net/mlx5/mlx5_rxtx_vec_sse.c | 21 +- drivers/net/mlx5/mlx5_stats.c | 11 +- drivers/net/nfp/nfp_net.c | 25 +- drivers/net/pcap/rte_eth_pcap.c| 2 + drivers/net/qede/Mak
Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq flushing
On 12/07/2017 05:30 AM, Tiwei Bie wrote:
> The vector Rx will be broken if backend has consumed all
> the descs in the avail ring before the device is started.
> Because in current implementation, vector Rx will return
> immediately without refilling the avail ring if the used
> ring is empty. So we have to refill the avail ring after
> flushing the elements in the used ring for vector Rx.
>
> Besides, vector Rx has a different ring layout assumption
> and mbuf management. So we need to handle it differently.
>
Hi Tiwei, does the issue only occur with the vector rx? How about if the
simple rx is used because VIRTIO_NET_F_MRG_RXBUF is set?
Kevin.
> Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start")
> Cc: sta...@dpdk.org
>
> Reported-by: Antonio Fischetti
> Signed-off-by: Tiwei Bie
> ---
> drivers/net/virtio/virtio_ethdev.c | 2 +-
> drivers/net/virtio/virtqueue.c | 31 ---
> drivers/net/virtio/virtqueue.h | 2 +-
> 3 files changed, 26 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_ethdev.c
> b/drivers/net/virtio/virtio_ethdev.c
> index e0328f61d..64a0cc608 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -1860,7 +1860,7 @@ virtio_dev_start(struct rte_eth_dev *dev)
> for (i = 0; i < dev->data->nb_rx_queues; i++) {
> rxvq = dev->data->rx_queues[i];
> /* Flush the old packets */
> - virtqueue_flush(rxvq->vq);
> + virtqueue_rxvq_flush(rxvq->vq);
> virtqueue_notify(rxvq->vq);
> }
>
> diff --git a/drivers/net/virtio/virtqueue.c b/drivers/net/virtio/virtqueue.c
> index c3a536f8a..696d0e4a4 100644
> --- a/drivers/net/virtio/virtqueue.c
> +++ b/drivers/net/virtio/virtqueue.c
> @@ -37,6 +37,7 @@
> #include "virtqueue.h"
> #include "virtio_logs.h"
> #include "virtio_pci.h"
> +#include "virtio_rxtx_simple.h"
>
> /*
> * Two types of mbuf to be cleaned:
> @@ -62,8 +63,10 @@ virtqueue_detatch_unused(struct virtqueue *vq)
>
> /* Flush the elements in the used ring. */
> void
> -virtqueue_flush(struct virtqueue *vq)
> +virtqueue_rxvq_flush(struct virtqueue *vq)
> {
> + struct virtnet_rx *rxq = &vq->rxq;
> + struct virtio_hw *hw = vq->hw;
> struct vring_used_elem *uep;
> struct vq_desc_extra *dxp;
> uint16_t used_idx, desc_idx;
> @@ -74,13 +77,27 @@ virtqueue_flush(struct virtqueue *vq)
> for (i = 0; i < nb_used; i++) {
> used_idx = vq->vq_used_cons_idx & (vq->vq_nentries - 1);
> uep = &vq->vq_ring.used->ring[used_idx];
> - desc_idx = (uint16_t)uep->id;
> - dxp = &vq->vq_descx[desc_idx];
> - if (dxp->cookie != NULL) {
> - rte_pktmbuf_free(dxp->cookie);
> - dxp->cookie = NULL;
> + if (hw->use_simple_rx) {
> + desc_idx = used_idx;
> + rte_pktmbuf_free(vq->sw_ring[desc_idx]);
> + vq->vq_free_cnt++;
> + } else {
> + desc_idx = (uint16_t)uep->id;
> + dxp = &vq->vq_descx[desc_idx];
> + if (dxp->cookie != NULL) {
> + rte_pktmbuf_free(dxp->cookie);
> + dxp->cookie = NULL;
> + }
> + vq_ring_free_chain(vq, desc_idx);
> }
> vq->vq_used_cons_idx++;
> - vq_ring_free_chain(vq, desc_idx);
> + }
> +
> + if (hw->use_simple_rx) {
> + while (vq->vq_free_cnt >= RTE_VIRTIO_VPMD_RX_REARM_THRESH) {
> + virtio_rxq_rearm_vec(rxq);
> + if (virtqueue_kick_prepare(vq))
> + virtqueue_notify(vq);
> + }
> }
> }
> diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h
> index 2305d91a4..ab466c2db 100644
> --- a/drivers/net/virtio/virtqueue.h
> +++ b/drivers/net/virtio/virtqueue.h
> @@ -304,7 +304,7 @@ void virtqueue_dump(struct virtqueue *vq);
> struct rte_mbuf *virtqueue_detatch_unused(struct virtqueue *vq);
>
> /* Flush the elements in the used ring. */
> -void virtqueue_flush(struct virtqueue *vq);
> +void virtqueue_rxvq_flush(struct virtqueue *vq);
>
> static inline int
> virtqueue_full(const struct virtqueue *vq)
>
Re: [dpdk-dev] Max rnc/dec flow for coletocreek
Including dpdk_dev On 6 Dec 2017 10:50 pm, "Pankaj Joshi" wrote: > Hello All, > I want to know that how many maximum no of flow/rule can be installed on > coletocreek hardware. > > Regards, > Pankaj Joshi > >
Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory
On Thu, Dec 07, 2017 at 03:35:24PM +, Xueming(Steven) Li wrote:
>
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Xueming(Steven) Li
> > Sent: Thursday, December 7, 2017 11:05 PM
> > To: Olivier MATZ
> > Cc: dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH] lib/cmdline: init parse result memeory
> >
> > Hi Olivier,
> >
> > > -Original Message-
> > > From: Olivier MATZ [mailto:olivier.m...@6wind.com]
> > > Sent: Thursday, December 7, 2017 10:48 PM
> > > To: Xueming(Steven) Li
> > > Cc: dev@dpdk.org
> > > Subject: Re: [PATCH] lib/cmdline: init parse result memeory
> > >
> > > On Wed, Nov 15, 2017 at 11:54:02PM +0800, Xueming Li wrote:
> > > > Initialize binary result memory before parsing to avoid garbage in
> > > > parsing result.
> > > >
> > > > Signed-off-by: Xueming Li
> > > > ---
> > > > lib/librte_cmdline/cmdline_parse.c | 2 ++
> > > > 1 file changed, 2 insertions(+)
> > > >
> > > > diff --git a/lib/librte_cmdline/cmdline_parse.c
> > > > b/lib/librte_cmdline/cmdline_parse.c
> > > > index 3e12ee54f..9124758f1 100644
> > > > --- a/lib/librte_cmdline/cmdline_parse.c
> > > > +++ b/lib/librte_cmdline/cmdline_parse.c
> > > > @@ -267,6 +267,8 @@ cmdline_parse(struct cmdline *cl, const char * buf)
> > > > if (!cl || !buf)
> > > > return CMDLINE_PARSE_BAD_ARGS;
> > > >
> > > > + memset(tmp_result.buf, 0, sizeof(tmp_result.buf));
> > > > +
> > > > ctx = cl->ctx;
> > > >
> > > > /*
> > >
> > >
> > > Did you see an issue (a bug or a crash) without the memset()?
> > > Or is it to avoid filling unused fields in the parsed struct?
> > Yes, I'm using same struct for some similar commands, have to avoid
> > filling unused fields.
> >
> > >
> > > I'm not sure if your patch is enough: cmdline_parse() calls
> > > match_inst() for each registered command. If a command partially
> > > matches (only the first tokens), the buffer is modified. So the next
> > > one will start with a dirty buffer.
> > >
> > > I suggest to put the memset() in match_inst() instead. Something like
> > this:
> > >
> > > if (resbuf != NULL)
> > > memset(resbuf, 0, resbuf_size);
> > >
> > >
> > > It will reset the buffer before using it.
> > It was there for performance concern, since the buffer could be tainted,
> > I'll upload a new version according to your suggestion, appreciate your
> > suggestion.
> Rte_flow CLI seems to be relying on modified buffer, add Adrien:
> testpmd> flow create 0 ingress pattern eth / ipv4 / udp / vxlan / end
> actions rss queues 1 2 end level 1 / mark id 0x123456 / end
> Caught error type 2 (flow rule (handle)): no valid action
While the flow command relies on the contents of this buffer when parsing
tokens, it doesn't expect it to be maintained across match_inst() calls.
In fact another issue prevents Olivier's suggestion from working.
Commit 9b3fbb051d2e ("cmdline: fix parsing") is supposed prevent further
match_inst() calls after the right inst is found but doesn't break
cmdline_parse()'s loop; subsequent iterations clear the result buffer.
Here's a suggestion to try:
diff --git a/lib/librte_cmdline/cmdline_parse.c
b/lib/librte_cmdline/cmdline_parse.c
index 3e12ee5..e967410 100644
--- a/lib/librte_cmdline/cmdline_parse.c
+++ b/lib/librte_cmdline/cmdline_parse.c
@@ -338,8 +338,8 @@ cmdline_parse(struct cmdline *cl, const char * buf)
err = CMDLINE_PARSE_AMBIGUOUS;
f=NULL;
debug_printf("Ambiguous cmd\n");
- break;
}
+ break;
}
}
--
Adrien Mazarguil
6WIND
Re: [dpdk-dev] [PATCH 1/2] event/sw: fix queue memory leak and multi-link bug
> From: Eads, Gage
> Sent: Thursday, November 30, 2017 3:09 AM
> To: dev@dpdk.org
> Cc: jerin.ja...@caviumnetworks.com; Van Haaren, Harry
> ; Richardson, Bruce
> ; hemant.agra...@nxp.com; nipun.gu...@nxp.com;
> santosh.shu...@caviumnetworks.com; pbhagavat...@caviumnetworks.com
> Subject: [PATCH 1/2] event/sw: fix queue memory leak and multi-link bug
>
> This commit reinitializes a queue before it is reconfigured, such that
> reorder buffer memory is not leaked.
>
> This bug masked a few other problems, which this commit corrects as well:
> - sw_port_link() allowed a port to link to a queue twice, such that the
> port could then successfully unlink the queue twice. Now the link
> function checks whether a port is already linked to the queue, and if so
> returns success but doesn't assign the a port a second slot in the
> queue's cq map.
> - test_eventdev.c's test_eventdev_unlink() was unlinking a queue twice
> from the same port, and expecting the second unlink to succeed. Now the
> test unlinks, links, then unlinks again.
> - test_eventdev.c's test_eventdev_link_get() was linking a single queue but
> expecting the unlink function to return nb_queues (where nb_queues > 1).
> The test now checks for a return value of 1.
>
> Fixes: 5ffb2f142d95 ("event/sw: support event queues")
> Fixes: 371a688fc159 ("event/sw: support linking queues to ports")
> Fixes: f8f9d233ea0e ("test/eventdev: add unit tests")
>
> Signed-off-by: Gage Eads
Looks good to me,
Acked-by: Harry van Haaren
Re: [dpdk-dev] [PATCH 2/2] event/sw: use dynamically-sized IQs
> From: Eads, Gage
> Sent: Thursday, November 30, 2017 3:09 AM
> To: dev@dpdk.org
> Cc: jerin.ja...@caviumnetworks.com; Van Haaren, Harry
> ; Richardson, Bruce
> ; hemant.agra...@nxp.com; nipun.gu...@nxp.com;
> santosh.shu...@caviumnetworks.com; pbhagavat...@caviumnetworks.com
> Subject: [PATCH 2/2] event/sw: use dynamically-sized IQs
>
> This commit introduces dynamically-sized IQs, by switching the underlying
> data structure from a fixed-size ring to a linked list of queue 'chunks.'
> This has a number of benefits:
> - Certain corner cases were observed in which all of a pipeline's flows
> could be pinned to one port for extended periods, effectively turning a
> multi-core pipeline into single-core one. This was caused by an event
> producer having a larger new_event_threshold than the IQ depth, and
> injecting large numbers of packets that are ultimately backpressured in a
> worker's rx_ring, causing those packets' flows to be scheduled to that
> port.
> The dynamically sized IQ does not have this problem because each IQ can
> grow large enough to store all the system's events, such that
> backpressure will not reach the worker_ring.
> - Slight performance improvement (~1-2%) in high throughput scenarios,
> tested with eventdev_pipeline_sw_pmd.
>
> This implementation has a small increase in the queue storage memory
> footprint (~70KB). This commit also removes the iq_size xstat, which no
> longer applies to this implementation.
>
> Signed-off-by: Gage Eads
Some review notes below - but nothing that needs changing.
Acked-by: Harry van Haaren
> diff --git a/drivers/event/sw/sw_evdev.c b/drivers/event/sw/sw_evdev.c
> - for (i = 0; i < SW_IQS_MAX; i++) {
> - snprintf(buf, sizeof(buf), "q_%u_iq_%d", idx, i);
> - qid->iq[i] = iq_ring_create(buf, socket_id);
> - if (!qid->iq[i]) {
> - SW_LOG_DBG("ring create failed");
> - goto cleanup;
> - }
> - }
> + for (i = 0; i < SW_IQS_MAX; i++)
> + iq_init(sw, &qid->iq[i]);
Review notes;
- It looks like error checking is being lost above, but in reality the
iq_init() cannot fail in this new scheme, provided that the pool of chunks is
provisioned correctly at configure() time (see provisioning below).
No action required here - just noting this error-checking change.
> + /* Number of chunks sized for worst-case spread of events across IQs */
> + num_chunks = ((SW_INFLIGHT_EVENTS_TOTAL/SW_EVS_PER_Q_CHUNK)+1) +
> + sw->qid_count*SW_IQS_MAX*2;
To verify the logic here;
((TOTAL / CHUNK_SIZE) + 1) is the main body of chunks.
QID * IQ * 2 is there so each QID can hold a chunk, and a next pointer, for
each priority of IQ.
add these two together, is the total number of chunks. Check :)
Re: [dpdk-dev] [PATCH] net/szedata2: fix check of mmap return value
On 12/7/2017 6:54 AM, Matej Vido wrote:
> Fixes: 9eddbdb4b094 ("szedata2: support link state operations")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Matej Vido
Applied to dpdk-next-net/master, thanks.
Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq flushing
Hi Tiwei,
I've tested this patch on my 2 test cases described
in the previous threads
http://www.dpdk.org/ml/archives/dev/2017-November/081839.html
http://www.dpdk.org/ml/archives/dev/2017-December/082801.html
1. testpmd on the host and testpmd in the guest
2. OvS-DPDK on the host and testpmd in the guest
and it works fine.
I'm using DPDK v17.11 + this patch and I see traffic is now
flowing through the VM.
Tested-by: Antonio Fischetti
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Fischetti, Antonio
> Sent: Thursday, December 7, 2017 9:30 AM
> To: Maxime Coquelin ; Bie, Tiwei
> ; y...@fridaylinux.org; dev@dpdk.org
> Cc: sta...@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break
> caused by rxq flushing
>
> Thanks Tiwei for working on this, I'll give it a try soon.
>
> Antonio
>
> > -Original Message-
> > From: Maxime Coquelin [mailto:maxime.coque...@redhat.com]
> > Sent: Thursday, December 7, 2017 9:15 AM
> > To: Bie, Tiwei ; y...@fridaylinux.org;
> dev@dpdk.org
> > Cc: Fischetti, Antonio ; sta...@dpdk.org
> > Subject: Re: [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq
> > flushing
> >
> >
> >
> > On 12/07/2017 06:30 AM, Tiwei Bie wrote:
> > > The vector Rx will be broken if backend has consumed all
> > > the descs in the avail ring before the device is started.
> > > Because in current implementation, vector Rx will return
> > > immediately without refilling the avail ring if the used
> > > ring is empty. So we have to refill the avail ring after
> > > flushing the elements in the used ring for vector Rx.
> > >
> > > Besides, vector Rx has a different ring layout assumption
> > > and mbuf management. So we need to handle it differently.
> > >
> > > Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start")
> > > Cc: sta...@dpdk.org
> > >
> > > Reported-by: Antonio Fischetti
> > > Signed-off-by: Tiwei Bie
> > > ---
> > > drivers/net/virtio/virtio_ethdev.c | 2 +-
> > > drivers/net/virtio/virtqueue.c | 31 --
> --
> > ---
> > > drivers/net/virtio/virtqueue.h | 2 +-
> > > 3 files changed, 26 insertions(+), 9 deletions(-)
> >
> > Reviewed-by: Maxime Coquelin
> >
> > Thanks,
> > Maxime
Re: [dpdk-dev] [PATCH 4/5] app/testpmd: add configuration for udp port tunnel type
> -Original Message- > From: Ferruh Yigit [mailto:ferruh.yi...@intel.com] > Sent: Thursday, December 7, 2017 6:09 AM > To: Mody, Rasesh ; dev@dpdk.org > Cc: Shaikh, Shahed ; Dept-Eng DPDK Dev engdpdk...@cavium.com> > Subject: Re: [dpdk-dev] [PATCH 4/5] app/testpmd: add configuration for udp > port tunnel type > > On 11/24/2017 12:35 PM, Rasesh Mody wrote: > > From: Shahed Shaikh > > > > Replace rx_vxlan_port command with rx_tunnel_udp_port to support both > > VXLAN and GENEVE udp ports. > > Also updates tunnel_filter command to accept "geneve" argument, can you > please separate to another patch. > > And to prevent these patches hold PMD patches, you can send a new version of > the patchset splitting qede PMD patches into their own patchset. Sure. We'll send separate patchset for qede PMD. > > > > > Signed-off-by: Shahed Shaikh > > --- > > app/test-pmd/cmdline.c | 28 +++- > > 1 file changed, 19 insertions(+), 9 deletions(-) > > > > diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c index > > f71d963..4b5a8cd 100644 > > --- a/app/test-pmd/cmdline.c > > +++ b/app/test-pmd/cmdline.c > > @@ -402,11 +402,11 @@ static void cmd_help_long_parsed(void > *parsed_result, > > "imac-tenid|imac|omac-imac-tenid|oip|iip) (tenant_id) > (queue_id)\n" > > " remove a tunnel filter of a port.\n\n" > > > > - "rx_vxlan_port add (udp_port) (port_id)\n" > > - "Add an UDP port for VXLAN packet filter on a > port\n\n" > > + "rx_tunnel_udp_port add vxlan|geneve (udp_port) > (port_id)\n" > > Not sure about "rx_tunnel_udp_port" command. > > What do you think something like: > "port config (port_id) udp_tunnel_port add|rm vxlan|geneve (udp_port)" > > to expand ""port config (port_id) ..." command instead of introducing a new > one? Makes sense. I'll update port_config command. Thanks, Shahed
Re: [dpdk-dev] [PATCH v1] net/tap: allow user mac to be passed as args
On 11/30/2017 11:49 AM, Vipin Varghese wrote:
> One of the uses cases from customer site is use TAP PMD to intake
> user specific MAC address during probe. This allows applications
> make use of interfaces with desired MAC.
>
> Extending MAC argumentinfrastructure for tap PMD; we pass custom
> MAC address in string format (sample - 11:22:33:44:55:66).
Overall lgtm, please check a few comments below.
>
> Signed-off-by: Vipin Varghese
> ---
> drivers/net/tap/rte_eth_tap.c | 56
> +++
> 1 file changed, 52 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/rte_eth_tap.c
> index 6b27679..0c53458 100644
> --- a/drivers/net/tap/rte_eth_tap.c
> +++ b/drivers/net/tap/rte_eth_tap.c
> @@ -81,6 +81,8 @@
> #define FLOWER_KERNEL_VERSION KERNEL_VERSION(4, 2, 0)
> #define FLOWER_VLAN_KERNEL_VERSION KERNEL_VERSION(4, 9, 0)
>
> +static unsigned char user_mac[ETHER_ADDR_LEN];
> +
> static struct rte_vdev_driver pmd_tap_drv;
>
> static const char *valid_arguments[] = {
> @@ -1291,13 +1293,20 @@ enum ioctl_mode {
> pmd->txq[i].fd = -1;
> }
>
> - if (fixed_mac_type) {
> + if (fixed_mac_type == 1) {
Instead of hardcoded type values 1 & 2, can you please use macros?
> /* fixed mac = 00:64:74:61:70: */
> static int iface_idx;
> char mac[ETHER_ADDR_LEN] = "\0dtap";
>
> mac[ETHER_ADDR_LEN - 1] = iface_idx++;
> rte_memcpy(&pmd->eth_addr, mac, ETHER_ADDR_LEN);
> + } else if (fixed_mac_type == 2) {
> + /* user mac is recieved */
s/recieved/received
> + RTE_LOG(INFO, PMD,
> + "Using user MAC (%02x:%02x:%02x:%02x:%02x:%02x)\n",
> + user_mac[0], user_mac[1], user_mac[2],
> + user_mac[3], user_mac[4], user_mac[5]);
> + rte_memcpy(&pmd->eth_addr, user_mac, ETHER_ADDR_LEN);
> } else {
> eth_random_addr((uint8_t *)&pmd->eth_addr);
> }
> @@ -1471,9 +1480,48 @@ enum ioctl_mode {
>const char *value,
>void *extra_args)
> {
> - if (value &&
> - !strncasecmp(ETH_TAP_MAC_FIXED, value, strlen(ETH_TAP_MAC_FIXED)))
> - *(int *)extra_args = 1;
> + char macTemp[20], *macByte = NULL;
> + unsigned int index = 0;
> +
> + if (value) {
> + if (!strncasecmp(ETH_TAP_MAC_FIXED, value,
> + strlen(ETH_TAP_MAC_FIXED))) {
> + *(int *)extra_args = 1;
> + } else {
> + RTE_LOG(INFO, PMD, "User shared MAC param (%s)\n",
> + value);
Is this log needs to be in INFO level, since there is already and info level log
when MAC set, what about making this debug?
> +
> + /* desired format aa:bb:cc:dd:ee:ff:11 */
Can you please update RTE_PMD_REGISTER_PARAM_STRING with new param, put expected
format info there as well?
> + if (strlen(value) == 17) {
> + strncpy(macTemp, value, 18);
> +
> + macByte = strtok(macTemp, ":");
> + while ((macByte != NULL) &&
> + (strspn(macByte,
> "0123456789ABCDEFabcdef")) &&
> + (strspn((macByte + 1),
> "0123456789ABCDEFabcdef")) &&
> + (strlen(macByte) == 2)) {
> + user_mac[index] = strtoul(macByte,
> NULL, 16);
> + macByte = strtok(NULL, ":");
> + index += 1;
> + }
I would extract the string to mac logic into its own function, but up to you.
> +
> + if (index != 6) {
> + RTE_LOG(ERR, PMD, " failure in MAC
> value (%s) at %u\n",
> + macByte, index + 1);
> + return -1;
And this is not related to this patch, but just as reminder, when a virtual
driver probe fails vdev bus stops probing with an error, so all remaining
virtual devices are not probed, in case one might want to fix ;)
> + }
> +
> + RTE_LOG(DEBUG, PMD, " User MAC (%s)
> considered\n",
"User defined MAC" ? And can you please add an port identifier, port_id or
device name or interface name, for the case there are multiple tap device to
identify which one get user defined MAC.
> + value);
> + *(int *)extra_args = 2;
> + } else {
> + RTE_LOG(ERR, PMD, " Mismatch on format for
> (%s)\n",
> + value);
> + return -1;
> + }
> +
[dpdk-dev] [PATCH 00/13] examples/eventdev: add capability based pipeline support
The eventdev_pipeline_sw_pmd supports a single pipeline type that follows a
strict Rx - Worker - Tx pipeline flow.
This patch-set aims to generalise pipeline configuration across event devices
based on their capabilities. The Rx, Tx and worker functionalities are
classified based on these capabilities.
This allows us to abstract the underlying event device and seamlessly
switch between pipeline models.
Pavan Nikhilesh (13):
examples/eventdev: add Rx adapter support
examples/eventdev: move common data into pipeline common
examples/eventdev: add framework for caps based pipeline
examples/eventdev: add generic worker pipeline
examples/eventdev: add ops to check cmdline args
examples/eventdev: add non burst mode generic worker
examples/eventdev: add thread safe Tx worker pipeline
examples/eventdev: add burst for thread safe pipeline
examples/eventdev: add all type queue option
examples/eventdev: add single stage pipeline worker
examples/eventdev: add atq single stage pipeline worker
examples/eventdev_pipeline_sw_pmd: rename example
doc: update example eventdev_pipeline
...v_pipeline_sw_pmd.rst => eventdev_pipeline.rst} | 6 +-
doc/guides/sample_app_ug/index.rst | 2 +-
examples/Makefile | 2 +-
.../Makefile | 4 +-
.../main.c | 527 ++
examples/eventdev_pipeline/pipeline_common.h | 183 +
.../eventdev_pipeline/pipeline_worker_generic.c| 575 +++
examples/eventdev_pipeline/pipeline_worker_tx.c| 804 +
8 files changed, 1615 insertions(+), 488 deletions(-)
rename doc/guides/sample_app_ug/{eventdev_pipeline_sw_pmd.rst =>
eventdev_pipeline.rst} (97%)
rename examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/Makefile (95%)
rename examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/main.c (51%)
create mode 100644 examples/eventdev_pipeline/pipeline_common.h
create mode 100644 examples/eventdev_pipeline/pipeline_worker_generic.c
create mode 100644 examples/eventdev_pipeline/pipeline_worker_tx.c
--
2.14.1
[dpdk-dev] [PATCH 01/13] examples/eventdev: add Rx adapter support
Use event Rx adapter for packets Rx instead of explicit producer logic.
Use service run iter function for granular control instead of using
dedicated service lcore.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/main.c | 149 +--
1 file changed, 82 insertions(+), 67 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index 5f431d87d..bb87c9544 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -46,6 +46,7 @@
#include
#include
#include
+#include
#include
#define MAX_NUM_STAGES 8
@@ -78,6 +79,7 @@ struct fastpath_data {
uint32_t tx_lock;
uint32_t sched_lock;
uint32_t evdev_service_id;
+ uint32_t rxadptr_service_id;
bool rx_single;
bool tx_single;
bool sched_single;
@@ -105,6 +107,7 @@ struct config_data {
unsigned int worker_cq_depth;
int16_t next_qid[MAX_NUM_STAGES+2];
int16_t qid[MAX_NUM_STAGES];
+ uint8_t rx_adapter_id;
};
static struct config_data cdata = {
@@ -193,53 +196,12 @@ consumer(void)
return 0;
}
-static int
-producer(void)
-{
- static uint8_t eth_port;
- struct rte_mbuf *mbufs[BATCH_SIZE+2];
- struct rte_event ev[BATCH_SIZE+2];
- uint32_t i, num_ports = prod_data.num_nic_ports;
- int32_t qid = prod_data.qid;
- uint8_t dev_id = prod_data.dev_id;
- uint8_t port_id = prod_data.port_id;
- uint32_t prio_idx = 0;
-
- const uint16_t nb_rx = rte_eth_rx_burst(eth_port, 0, mbufs, BATCH_SIZE);
- if (++eth_port == num_ports)
- eth_port = 0;
- if (nb_rx == 0) {
- rte_pause();
- return 0;
- }
-
- for (i = 0; i < nb_rx; i++) {
- ev[i].flow_id = mbufs[i]->hash.rss;
- ev[i].op = RTE_EVENT_OP_NEW;
- ev[i].sched_type = cdata.queue_type;
- ev[i].queue_id = qid;
- ev[i].event_type = RTE_EVENT_TYPE_ETHDEV;
- ev[i].sub_event_type = 0;
- ev[i].priority = RTE_EVENT_DEV_PRIORITY_NORMAL;
- ev[i].mbuf = mbufs[i];
- RTE_SET_USED(prio_idx);
- }
-
- const int nb_tx = rte_event_enqueue_burst(dev_id, port_id, ev, nb_rx);
- if (nb_tx != nb_rx) {
- for (i = nb_tx; i < nb_rx; i++)
- rte_pktmbuf_free(mbufs[i]);
- }
-
- return 0;
-}
-
static inline void
schedule_devices(unsigned int lcore_id)
{
if (fdata->rx_core[lcore_id] && (fdata->rx_single ||
rte_atomic32_cmpset(&(fdata->rx_lock), 0, 1))) {
- producer();
+ rte_service_run_iter_on_app_lcore(fdata->rxadptr_service_id, 1);
rte_atomic32_clear((rte_atomic32_t *)&(fdata->rx_lock));
}
@@ -553,6 +515,79 @@ parse_app_args(int argc, char **argv)
}
}
+static inline void
+init_rx_adapter(uint16_t nb_ports)
+{
+ int i;
+ int ret;
+ uint8_t evdev_id = 0;
+ uint8_t port_needed = 0;
+ struct rte_event_dev_info dev_info;
+
+ ret = rte_event_dev_info_get(evdev_id, &dev_info);
+
+ struct rte_event_port_conf rx_p_conf = {
+ .dequeue_depth = 8,
+ .enqueue_depth = 8,
+ .new_event_threshold = 1200,
+ };
+
+ if (rx_p_conf.dequeue_depth > dev_info.max_event_port_dequeue_depth)
+ rx_p_conf.dequeue_depth = dev_info.max_event_port_dequeue_depth;
+ if (rx_p_conf.enqueue_depth > dev_info.max_event_port_enqueue_depth)
+ rx_p_conf.enqueue_depth = dev_info.max_event_port_enqueue_depth;
+
+ ret = rte_event_eth_rx_adapter_create(cdata.rx_adapter_id, evdev_id,
+ &rx_p_conf);
+ if (ret)
+ rte_exit(EXIT_FAILURE, "failed to create rx adapter[%d]",
+ cdata.rx_adapter_id);
+
+ struct rte_event_eth_rx_adapter_queue_conf queue_conf = {
+ .ev.sched_type = cdata.queue_type,
+ .ev.queue_id = cdata.qid[0],
+ };
+
+ for (i = 0; i < nb_ports; i++) {
+ uint32_t cap;
+
+ ret = rte_event_eth_rx_adapter_caps_get(evdev_id, i, &cap);
+ if (ret)
+ rte_exit(EXIT_FAILURE,
+ "failed to get event rx adapter "
+ "capabilities");
+ /* Producer needs port. */
+ port_needed |= !(cap &
+ RTE_EVENT_ETH_RX_ADAPTER_CAP_INTERNAL_PORT);
+
+ ret = rte_event_eth_rx_adapter_queue_add(cdata.rx_adapter_id, i,
+ -1, &queue_conf);
+ if (ret)
+ rte_exit(EXIT_FAILURE,
+ "Failed to add queues to Rx adapter");
+ }
+
+ if (
[dpdk-dev] [PATCH 02/13] examples/eventdev: move common data into pipeline common
Move common structures and functions into pipeline_common.h so that they
can be used by different kinds of pipelines.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/main.c | 77 +--
.../eventdev_pipeline_sw_pmd/pipeline_common.h | 109 +
2 files changed, 112 insertions(+), 74 deletions(-)
create mode 100644 examples/eventdev_pipeline_sw_pmd/pipeline_common.h
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index bb87c9544..c9702fddd 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -2,6 +2,7 @@
* BSD LICENSE
*
* Copyright(c) 2016-2017 Intel Corporation. All rights reserved.
+ * Copyright 2016 Cavium, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
@@ -35,82 +36,10 @@
#include
#include
#include
-#include
-
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-
-#define MAX_NUM_STAGES 8
-#define BATCH_SIZE 16
-#define MAX_NUM_CORE 64
-
-struct prod_data {
- uint8_t dev_id;
- uint8_t port_id;
- int32_t qid;
- unsigned int num_nic_ports;
-} __rte_cache_aligned;
-
-struct cons_data {
- uint8_t dev_id;
- uint8_t port_id;
-} __rte_cache_aligned;
-
-static struct prod_data prod_data;
-static struct cons_data cons_data;
-
-struct worker_data {
- uint8_t dev_id;
- uint8_t port_id;
-} __rte_cache_aligned;
-
-struct fastpath_data {
- volatile int done;
- uint32_t rx_lock;
- uint32_t tx_lock;
- uint32_t sched_lock;
- uint32_t evdev_service_id;
- uint32_t rxadptr_service_id;
- bool rx_single;
- bool tx_single;
- bool sched_single;
- unsigned int rx_core[MAX_NUM_CORE];
- unsigned int tx_core[MAX_NUM_CORE];
- unsigned int sched_core[MAX_NUM_CORE];
- unsigned int worker_core[MAX_NUM_CORE];
- struct rte_eth_dev_tx_buffer *tx_buf[RTE_MAX_ETHPORTS];
-};
-static struct fastpath_data *fdata;
-
-struct config_data {
- unsigned int active_cores;
- unsigned int num_workers;
- int64_t num_packets;
- unsigned int num_fids;
- int queue_type;
- int worker_cycles;
- int enable_queue_priorities;
- int quiet;
- int dump_dev;
- int dump_dev_signal;
- unsigned int num_stages;
- unsigned int worker_cq_depth;
- int16_t next_qid[MAX_NUM_STAGES+2];
- int16_t qid[MAX_NUM_STAGES];
- uint8_t rx_adapter_id;
-};
+#include "pipeline_common.h"
-static struct config_data cdata = {
+struct config_data cdata = {
.num_packets = (1L << 25), /* do ~32M packets */
.num_fids = 512,
.queue_type = RTE_SCHED_TYPE_ATOMIC,
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
new file mode 100644
index 0..938e155d3
--- /dev/null
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
@@ -0,0 +1,109 @@
+/*
+ * BSD LICENSE
+ *
+ * Copyright 2016 Intel Corporation.
+ * Copyright 2016 Cavium, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Cavium, Inc nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include
+
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#include
+#i
[dpdk-dev] [PATCH 03/13] examples/eventdev: add framework for caps based pipeline
Add framework to support capability based pipeline.
Based on the capability of event device and probed ethernet devices the
optimal pipeline configuration can be chosen.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/pipeline_common.h | 11 +++
1 file changed, 11 insertions(+)
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
index 938e155d3..5219a4e85 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
@@ -67,6 +67,16 @@ struct worker_data {
uint8_t port_id;
} __rte_cache_aligned;
+struct setup_data {
+ int (*worker_loop)(void *);
+ int (*consumer_loop)(void);
+ void (*schedule_loop)(unsigned int);
+ int (*eventdev_setup)(struct prod_data *, struct cons_data *,
+ struct worker_data *);
+ void (*rx_adapter_setup)(uint16_t nb_ports);
+ void (*opt_check)(void);
+};
+
struct fastpath_data {
volatile int done;
uint32_t rx_lock;
@@ -82,6 +92,7 @@ struct fastpath_data {
unsigned int sched_core[MAX_NUM_CORE];
unsigned int worker_core[MAX_NUM_CORE];
struct rte_eth_dev_tx_buffer *tx_buf[RTE_MAX_ETHPORTS];
+ struct setup_data cap;
} __rte_cache_aligned;
struct config_data {
--
2.14.1
[dpdk-dev] [PATCH 04/13] examples/eventdev: add generic worker pipeline
Rename existing pipeline as generic worker pipeline.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/Makefile | 1 +
examples/eventdev_pipeline_sw_pmd/main.c | 432 +
.../eventdev_pipeline_sw_pmd/pipeline_common.h | 56 +++
.../pipeline_worker_generic.c | 383 ++
4 files changed, 455 insertions(+), 417 deletions(-)
create mode 100644 examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
diff --git a/examples/eventdev_pipeline_sw_pmd/Makefile
b/examples/eventdev_pipeline_sw_pmd/Makefile
index de4e22c88..5e30556fb 100644
--- a/examples/eventdev_pipeline_sw_pmd/Makefile
+++ b/examples/eventdev_pipeline_sw_pmd/Makefile
@@ -42,6 +42,7 @@ APP = eventdev_pipeline_sw_pmd
# all source are stored in SRCS-y
SRCS-y := main.c
+SRCS-y += pipeline_worker_generic.c
CFLAGS += -O3
CFLAGS += $(WERROR_FLAGS)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index c9702fddd..bd53acf76 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -69,169 +69,6 @@ eth_tx_buffer_retry(struct rte_mbuf **pkts, uint16_t unsent,
} while (_sent != unsent);
}
-static int
-consumer(void)
-{
- const uint64_t freq_khz = rte_get_timer_hz() / 1000;
- struct rte_event packets[BATCH_SIZE];
-
- static uint64_t received;
- static uint64_t last_pkts;
- static uint64_t last_time;
- static uint64_t start_time;
- unsigned int i, j;
- uint8_t dev_id = cons_data.dev_id;
- uint8_t port_id = cons_data.port_id;
-
- uint16_t n = rte_event_dequeue_burst(dev_id, port_id,
- packets, RTE_DIM(packets), 0);
-
- if (n == 0) {
- for (j = 0; j < rte_eth_dev_count(); j++)
- rte_eth_tx_buffer_flush(j, 0, fdata->tx_buf[j]);
- return 0;
- }
- if (start_time == 0)
- last_time = start_time = rte_get_timer_cycles();
-
- received += n;
- for (i = 0; i < n; i++) {
- uint8_t outport = packets[i].mbuf->port;
- rte_eth_tx_buffer(outport, 0, fdata->tx_buf[outport],
- packets[i].mbuf);
- }
-
- /* Print out mpps every 1<22 packets */
- if (!cdata.quiet && received >= last_pkts + (1<<22)) {
- const uint64_t now = rte_get_timer_cycles();
- const uint64_t total_ms = (now - start_time) / freq_khz;
- const uint64_t delta_ms = (now - last_time) / freq_khz;
- uint64_t delta_pkts = received - last_pkts;
-
- printf("# consumer RX=%"PRIu64", time %"PRIu64 "ms, "
- "avg %.3f mpps [current %.3f mpps]\n",
- received,
- total_ms,
- received / (total_ms * 1000.0),
- delta_pkts / (delta_ms * 1000.0));
- last_pkts = received;
- last_time = now;
- }
-
- cdata.num_packets -= n;
- if (cdata.num_packets <= 0)
- fdata->done = 1;
-
- return 0;
-}
-
-static inline void
-schedule_devices(unsigned int lcore_id)
-{
- if (fdata->rx_core[lcore_id] && (fdata->rx_single ||
- rte_atomic32_cmpset(&(fdata->rx_lock), 0, 1))) {
- rte_service_run_iter_on_app_lcore(fdata->rxadptr_service_id, 1);
- rte_atomic32_clear((rte_atomic32_t *)&(fdata->rx_lock));
- }
-
- if (fdata->sched_core[lcore_id] && (fdata->sched_single ||
- rte_atomic32_cmpset(&(fdata->sched_lock), 0, 1))) {
- rte_service_run_iter_on_app_lcore(fdata->evdev_service_id, 1);
- if (cdata.dump_dev_signal) {
- rte_event_dev_dump(0, stdout);
- cdata.dump_dev_signal = 0;
- }
- rte_atomic32_clear((rte_atomic32_t *)&(fdata->sched_lock));
- }
-
- if (fdata->tx_core[lcore_id] && (fdata->tx_single ||
- rte_atomic32_cmpset(&(fdata->tx_lock), 0, 1))) {
- consumer();
- rte_atomic32_clear((rte_atomic32_t *)&(fdata->tx_lock));
- }
-}
-
-static inline void
-work(struct rte_mbuf *m)
-{
- struct ether_hdr *eth;
- struct ether_addr addr;
-
- /* change mac addresses on packet (to use mbuf data) */
- /*
-* FIXME Swap mac address properly and also handle the
-* case for both odd and even number of stages that the
-* addresses end up the same at the end of the pipeline
-*/
- eth = rte_pktmbuf_mtod(m, struct ether_hdr *);
- ether_addr_copy(ð->d_addr, &addr);
- ether_addr_copy(&addr, ð->d_addr);
-
- /* do a number of cycles of work per packet */
- volatile uint64_t start_tsc = rte_rdtsc();
- while (rte_rdtsc() < start_tsc + cdata.w
[dpdk-dev] [PATCH 05/13] examples/eventdev: add ops to check cmdline args
Each eventdev pipeline needs to allow different cmdline args combination
based on pipeline type.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/main.c | 16 +++--
.../eventdev_pipeline_sw_pmd/pipeline_common.h | 4 +++
.../pipeline_worker_generic.c | 40 ++
3 files changed, 49 insertions(+), 11 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index bd53acf76..2e80841d0 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -254,17 +254,11 @@ parse_app_args(int argc, char **argv)
}
}
- if (worker_lcore_mask == 0 || rx_lcore_mask == 0 ||
- sched_lcore_mask == 0 || tx_lcore_mask == 0) {
- printf("Core part of pipeline was not assigned any cores. "
- "This will stall the pipeline, please check core masks "
- "(use -h for details on setting core masks):\n"
- "\trx: %"PRIu64"\n\ttx: %"PRIu64"\n\tsched: %"PRIu64
- "\n\tworkers: %"PRIu64"\n",
- rx_lcore_mask, tx_lcore_mask, sched_lcore_mask,
- worker_lcore_mask);
- rte_exit(-1, "Fix core masks\n");
- }
+ cdata.worker_lcore_mask = worker_lcore_mask;
+ cdata.sched_lcore_mask = sched_lcore_mask;
+ cdata.rx_lcore_mask = rx_lcore_mask;
+ cdata.tx_lcore_mask = tx_lcore_mask;
+
if (cdata.num_stages == 0 || cdata.num_stages > MAX_NUM_STAGES)
usage();
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
index 0f3426a3a..a5837c99b 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
@@ -111,6 +111,10 @@ struct config_data {
int16_t next_qid[MAX_NUM_STAGES+2];
int16_t qid[MAX_NUM_STAGES];
uint8_t rx_adapter_id;
+ uint64_t worker_lcore_mask;
+ uint64_t rx_lcore_mask;
+ uint64_t tx_lcore_mask;
+ uint64_t sched_lcore_mask;
};
struct port_link {
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
index 032a4f2d2..a72b7b2f9 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
@@ -370,6 +370,45 @@ init_rx_adapter(uint16_t nb_ports)
cdata.rx_adapter_id);
}
+static void
+generic_opt_check(void)
+{
+ int i;
+ int ret;
+ uint32_t cap = 0;
+ uint8_t rx_needed = 0;
+ struct rte_event_dev_info eventdev_info;
+
+ memset(&eventdev_info, 0, sizeof(struct rte_event_dev_info));
+ rte_event_dev_info_get(0, &eventdev_info);
+
+ for (i = 0; i < rte_eth_dev_count(); i++) {
+ ret = rte_event_eth_rx_adapter_caps_get(0, i, &cap);
+ if (ret)
+ rte_exit(EXIT_FAILURE,
+ "failed to get event rx adapter "
+ "capabilities");
+ rx_needed |=
+ !(cap & RTE_EVENT_ETH_RX_ADAPTER_CAP_INTERNAL_PORT);
+ }
+
+ if (cdata.worker_lcore_mask == 0 ||
+ (rx_needed && cdata.rx_lcore_mask == 0) ||
+ cdata.tx_lcore_mask == 0 || (cdata.sched_lcore_mask == 0
+ && !(eventdev_info.event_dev_cap &
+ RTE_EVENT_DEV_CAP_DISTRIBUTED_SCHED))) {
+ printf("Core part of pipeline was not assigned any cores. "
+ "This will stall the pipeline, please check core masks "
+ "(use -h for details on setting core masks):\n"
+ "\trx: %"PRIu64"\n\ttx: %"PRIu64"\n\tsched: %"PRIu64
+ "\n\tworkers: %"PRIu64"\n",
+ cdata.rx_lcore_mask, cdata.tx_lcore_mask,
+ cdata.sched_lcore_mask,
+ cdata.worker_lcore_mask);
+ rte_exit(-1, "Fix core masks\n");
+ }
+}
+
void
set_worker_generic_setup_data(struct setup_data *caps, bool burst)
{
@@ -377,6 +416,7 @@ set_worker_generic_setup_data(struct setup_data *caps, bool
burst)
caps->consumer_loop = consumer_burst;
caps->worker_loop = worker_generic_burst;
+ caps->opt_check = generic_opt_check;
caps->rx_adapter_setup = init_rx_adapter;
caps->schedule_loop = schedule_devices;
caps->eventdev_setup = setup_eventdev_cw;
--
2.14.1
[dpdk-dev] [PATCH 06/13] examples/eventdev: add non burst mode generic worker
Currently, worker uses burst dequeue and burst enqueue to forward events.
Add a non burst mode based on the event dev capabilities.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/main.c | 12 +-
.../pipeline_worker_generic.c | 153 -
2 files changed, 160 insertions(+), 5 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index 2e80841d0..153467893 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -383,8 +383,16 @@ static void
do_capability_setup(uint16_t nb_ethdev, uint8_t eventdev_id)
{
RTE_SET_USED(nb_ethdev);
- RTE_SET_USED(eventdev_id);
- set_worker_generic_setup_data(&fdata->cap, 1);
+ uint8_t burst = 0;
+
+ struct rte_event_dev_info eventdev_info;
+ memset(&eventdev_info, 0, sizeof(struct rte_event_dev_info));
+
+ rte_event_dev_info_get(eventdev_id, &eventdev_info);
+ burst = eventdev_info.event_dev_cap & RTE_EVENT_DEV_CAP_BURST_MODE ? 1 :
+ 0;
+
+ set_worker_generic_setup_data(&fdata->cap, burst);
}
static void
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
index a72b7b2f9..5998aae95 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
@@ -1,5 +1,91 @@
+/*
+ * BSD LICENSE
+ *
+ * Copyright 2016 Intel Corporation.
+ * Copyright 2016 Cavium, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Cavium, Inc nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
#include "pipeline_common.h"
+static __rte_always_inline int
+worker_generic(void *arg)
+{
+ struct rte_event ev;
+
+ struct worker_data *data = (struct worker_data *)arg;
+ uint8_t dev_id = data->dev_id;
+ uint8_t port_id = data->port_id;
+ size_t sent = 0, received = 0;
+ unsigned int lcore_id = rte_lcore_id();
+
+ while (!fdata->done) {
+
+ if (fdata->cap.schedule_loop)
+ fdata->cap.schedule_loop(lcore_id);
+
+ if (!fdata->worker_core[lcore_id]) {
+ rte_pause();
+ continue;
+ }
+
+ const uint16_t nb_rx = rte_event_dequeue_burst(dev_id, port_id,
+ &ev, 1, 0);
+
+ if (nb_rx == 0) {
+ rte_pause();
+ continue;
+ }
+ received++;
+
+ /* The first worker stage does classification */
+ if (ev.queue_id == cdata.qid[0])
+ ev.flow_id = ev.mbuf->hash.rss
+ % cdata.num_fids;
+
+ ev.queue_id = cdata.next_qid[ev.queue_id];
+ ev.op = RTE_EVENT_OP_FORWARD;
+ ev.sched_type = cdata.queue_type;
+
+ work(ev.mbuf);
+
+ while (rte_event_enqueue_burst(dev_id, port_id, &ev, 1) != 1)
+ rte_pause();
+ sent++;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu TX=%zu\n",
+ rte_lcore_id(), received, sent);
+
+ return 0;
+}
+
static int
worker_generic_burst(void *arg)
{
@@ -60,6 +146,63 @@ worker_generic_burst(void *arg)
[dpdk-dev] [PATCH 07/13] examples/eventdev: add thread safe Tx worker pipeline
Add worker pipeline when Tx is multi thread safe.
Probe Ethernet dev capabilities and select it it is supported.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/Makefile | 1 +
examples/eventdev_pipeline_sw_pmd/main.c | 18 +-
.../eventdev_pipeline_sw_pmd/pipeline_common.h | 2 +
.../eventdev_pipeline_sw_pmd/pipeline_worker_tx.c | 433 +
4 files changed, 452 insertions(+), 2 deletions(-)
create mode 100644 examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
diff --git a/examples/eventdev_pipeline_sw_pmd/Makefile
b/examples/eventdev_pipeline_sw_pmd/Makefile
index 5e30556fb..59ee9840a 100644
--- a/examples/eventdev_pipeline_sw_pmd/Makefile
+++ b/examples/eventdev_pipeline_sw_pmd/Makefile
@@ -43,6 +43,7 @@ APP = eventdev_pipeline_sw_pmd
# all source are stored in SRCS-y
SRCS-y := main.c
SRCS-y += pipeline_worker_generic.c
+SRCS-y += pipeline_worker_tx.c
CFLAGS += -O3
CFLAGS += $(WERROR_FLAGS)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index 153467893..3be981c15 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -382,9 +382,20 @@ init_ports(unsigned int num_ports)
static void
do_capability_setup(uint16_t nb_ethdev, uint8_t eventdev_id)
{
- RTE_SET_USED(nb_ethdev);
+ int i;
+ uint8_t mt_unsafe = 0;
uint8_t burst = 0;
+ for (i = 0; i < nb_ethdev; i++) {
+ struct rte_eth_dev_info dev_info;
+ memset(&dev_info, 0, sizeof(struct rte_eth_dev_info));
+
+ rte_eth_dev_info_get(i, &dev_info);
+ /* Check if it is safe ask worker to tx. */
+ mt_unsafe |= !(dev_info.tx_offload_capa &
+ DEV_TX_OFFLOAD_MT_LOCKFREE);
+ }
+
struct rte_event_dev_info eventdev_info;
memset(&eventdev_info, 0, sizeof(struct rte_event_dev_info));
@@ -392,7 +403,10 @@ do_capability_setup(uint16_t nb_ethdev, uint8_t
eventdev_id)
burst = eventdev_info.event_dev_cap & RTE_EVENT_DEV_CAP_BURST_MODE ? 1 :
0;
- set_worker_generic_setup_data(&fdata->cap, burst);
+ if (mt_unsafe)
+ set_worker_generic_setup_data(&fdata->cap, burst);
+ else
+ set_worker_tx_setup_data(&fdata->cap, burst);
}
static void
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
index a5837c99b..0b27d1eb0 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
@@ -108,6 +108,7 @@ struct config_data {
int dump_dev_signal;
unsigned int num_stages;
unsigned int worker_cq_depth;
+ unsigned int rx_stride;
int16_t next_qid[MAX_NUM_STAGES+2];
int16_t qid[MAX_NUM_STAGES];
uint8_t rx_adapter_id;
@@ -178,3 +179,4 @@ schedule_devices(unsigned int lcore_id)
}
void set_worker_generic_setup_data(struct setup_data *caps, bool burst);
+void set_worker_tx_setup_data(struct setup_data *caps, bool burst);
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
new file mode 100644
index 0..31b7d8936
--- /dev/null
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
@@ -0,0 +1,433 @@
+/*
+ * BSD LICENSE
+ *
+ * Copyright 2016 Cavium, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Cavium, Inc nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY
[dpdk-dev] [PATCH 08/13] examples/eventdev: add burst for thread safe pipeline
Add burst mode worker pipeline when Tx is multi thread safe.
Signed-off-by: Pavan Nikhilesh
---
.../eventdev_pipeline_sw_pmd/pipeline_worker_tx.c | 78 +-
1 file changed, 76 insertions(+), 2 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
index 31b7d8936..a824f1f49 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
@@ -48,6 +48,19 @@ worker_event_enqueue(const uint8_t dev, const uint8_t port,
rte_pause();
}
+static __rte_always_inline void
+worker_event_enqueue_burst(const uint8_t dev, const uint8_t port,
+ struct rte_event *ev, const uint16_t nb_rx)
+{
+ uint16_t enq;
+
+ enq = rte_event_enqueue_burst(dev, port, ev, nb_rx);
+ while (enq < nb_rx) {
+ enq += rte_event_enqueue_burst(dev, port,
+ ev + enq, nb_rx - enq);
+ }
+}
+
static __rte_always_inline void
worker_tx_pkt(struct rte_mbuf *mbuf)
{
@@ -106,6 +119,65 @@ worker_do_tx(void *arg)
return 0;
}
+static int
+worker_do_tx_burst(void *arg)
+{
+ struct rte_event ev[BATCH_SIZE];
+
+ struct worker_data *data = (struct worker_data *)arg;
+ uint8_t dev = data->dev_id;
+ uint8_t port = data->port_id;
+ uint8_t lst_qid = cdata.num_stages - 1;
+ size_t fwd = 0, received = 0, tx = 0;
+
+ while (!fdata->done) {
+ uint16_t i;
+ const uint16_t nb_rx = rte_event_dequeue_burst(dev, port,
+ ev, BATCH_SIZE, 0);
+
+ if (nb_rx == 0) {
+ rte_pause();
+ continue;
+ }
+ received += nb_rx;
+
+ for (i = 0; i < nb_rx; i++) {
+ const uint8_t cq_id = ev[i].queue_id % cdata.num_stages;
+
+ if (cq_id >= lst_qid) {
+ if (ev[i].sched_type ==
+ RTE_SCHED_TYPE_ATOMIC) {
+ worker_tx_pkt(ev[i].mbuf);
+ tx++;
+ ev[i].op = RTE_EVENT_OP_RELEASE;
+ continue;
+ }
+ ev[i].queue_id = (cq_id == lst_qid) ?
+ cdata.next_qid[ev[i].queue_id] :
+ ev[i].queue_id;
+
+ worker_fwd_event(&ev[i],
+ RTE_SCHED_TYPE_ATOMIC);
+ } else {
+ ev[i].queue_id = cdata.next_qid[
+ ev[i].queue_id];
+ worker_fwd_event(&ev[i],
+ cdata.queue_type);
+ }
+ work(ev[i].mbuf);
+ }
+ worker_event_enqueue_burst(dev, port, ev, nb_rx);
+
+ fwd += nb_rx;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu FWD=%zu TX=%zu\n",
+ rte_lcore_id(), received, fwd, tx);
+
+ return 0;
+}
+
static int
setup_eventdev_w(struct prod_data *prod_data,
struct cons_data *cons_data,
@@ -422,8 +494,10 @@ opt_check(void)
void
set_worker_tx_setup_data(struct setup_data *caps, bool burst)
{
- RTE_SET_USED(burst);
- caps->worker_loop = worker_do_tx;
+ if (burst)
+ caps->worker_loop = worker_do_tx_burst;
+ if (!burst)
+ caps->worker_loop = worker_do_tx;
caps->opt_check = opt_check;
caps->consumer_loop = NULL;
--
2.14.1
[dpdk-dev] [PATCH 09/13] examples/eventdev: add all type queue option
Added configurable option to make queue type as all type queues i.e.
RTE_EVENT_QUEUE_CFG_ALL_TYPES based on event dev capability
RTE_EVENT_DEV_CAP_QUEUE_ALL_TYPES.
This can be enabled by supplying '-a' as a cmdline argument.
Signed-off-by: Pavan Nikhilesh
---
examples/eventdev_pipeline_sw_pmd/main.c | 7 +-
.../eventdev_pipeline_sw_pmd/pipeline_common.h | 1 +
.../pipeline_worker_generic.c | 5 +
.../eventdev_pipeline_sw_pmd/pipeline_worker_tx.c | 134 +++--
4 files changed, 139 insertions(+), 8 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline_sw_pmd/main.c
index 3be981c15..289f7204d 100644
--- a/examples/eventdev_pipeline_sw_pmd/main.c
+++ b/examples/eventdev_pipeline_sw_pmd/main.c
@@ -149,6 +149,7 @@ static struct option long_options[] = {
{"parallel", no_argument, 0, 'p'},
{"ordered", no_argument, 0, 'o'},
{"quiet", no_argument, 0, 'q'},
+ {"use-atq", no_argument, 0, 'a'},
{"dump", no_argument, 0, 'D'},
{0, 0, 0, 0}
};
@@ -172,6 +173,7 @@ usage(void)
" -o, --orderedUse ordered scheduling\n"
" -p, --parallel Use parallel scheduling\n"
" -q, --quiet Minimize printed output\n"
+ " -a, --use-atqUse all type queues\n"
" -D, --dump Print detailed statistics
before exit"
"\n";
fprintf(stderr, "%s", usage_str);
@@ -192,7 +194,7 @@ parse_app_args(int argc, char **argv)
int i;
for (;;) {
- c = getopt_long(argc, argv, "r:t:e:c:w:n:f:s:poPqDW:",
+ c = getopt_long(argc, argv, "r:t:e:c:w:n:f:s:paoPqDW:",
long_options, &option_index);
if (c == -1)
break;
@@ -225,6 +227,9 @@ parse_app_args(int argc, char **argv)
case 'p':
cdata.queue_type = RTE_SCHED_TYPE_PARALLEL;
break;
+ case 'a':
+ cdata.all_type_queues = 1;
+ break;
case 'q':
cdata.quiet = 1;
break;
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
index 0b27d1eb0..62755f6d0 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
@@ -106,6 +106,7 @@ struct config_data {
int quiet;
int dump_dev;
int dump_dev_signal;
+ int all_type_queues;
unsigned int num_stages;
unsigned int worker_cq_depth;
unsigned int rx_stride;
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
index 5998aae95..908d64c87 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
@@ -525,6 +525,11 @@ generic_opt_check(void)
memset(&eventdev_info, 0, sizeof(struct rte_event_dev_info));
rte_event_dev_info_get(0, &eventdev_info);
+ if (cdata.all_type_queues && !(eventdev_info.event_dev_cap &
+ RTE_EVENT_DEV_CAP_QUEUE_ALL_TYPES))
+ rte_exit(EXIT_FAILURE,
+ "Event dev doesn't support all type queues\n");
+
for (i = 0; i < rte_eth_dev_count(); i++) {
ret = rte_event_eth_rx_adapter_caps_get(0, i, &cap);
if (ret)
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
index a824f1f49..e25a06027 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
@@ -119,6 +119,51 @@ worker_do_tx(void *arg)
return 0;
}
+static int
+worker_do_tx_atq(void *arg)
+{
+ struct rte_event ev;
+
+ struct worker_data *data = (struct worker_data *)arg;
+ const uint8_t dev = data->dev_id;
+ const uint8_t port = data->port_id;
+ const uint8_t lst_qid = cdata.num_stages - 1;
+ size_t fwd = 0, received = 0, tx = 0;
+
+ while (!fdata->done) {
+
+ if (!rte_event_dequeue_burst(dev, port, &ev, 1, 0)) {
+ rte_pause();
+ continue;
+ }
+
+ received++;
+ const uint8_t cq_id = ev.queue_id % cdata.num_stages;
+
+ if (cq_id == lst_qid) {
+ if (ev.sched_type == RTE_SCHED_TYPE_ATOMIC) {
+ worker_tx_pkt(ev.mbuf);
+ tx++;
+ continue;
+ }
+ worker_fwd_event(&ev, RTE_
[dpdk-dev] [PATCH 11/13] examples/eventdev: add atq single stage pipeline worker
Add optimized eventdev pipeline when ethdev supports thread safe Tx,
number of configured stages is one and all type queue option is enabled.
Signed-off-by: Pavan Nikhilesh
---
.../eventdev_pipeline_sw_pmd/pipeline_worker_tx.c | 88 +-
1 file changed, 86 insertions(+), 2 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
index 15df21b7e..24ffd431f 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
@@ -106,6 +106,41 @@ worker_do_tx_single(void *arg)
return 0;
}
+static int
+worker_do_tx_single_atq(void *arg)
+{
+ struct worker_data *data = (struct worker_data *)arg;
+ const uint8_t dev = data->dev_id;
+ const uint8_t port = data->port_id;
+ size_t fwd = 0, received = 0, tx = 0;
+ struct rte_event ev;
+
+ while (!fdata->done) {
+
+ if (!rte_event_dequeue_burst(dev, port, &ev, 1, 0)) {
+ rte_pause();
+ continue;
+ }
+
+ received++;
+
+ if (ev.sched_type == RTE_SCHED_TYPE_ATOMIC) {
+ worker_tx_pkt(ev.mbuf);
+ tx++;
+ continue;
+ }
+ work(ev.mbuf);
+ worker_fwd_event(&ev, RTE_SCHED_TYPE_ATOMIC);
+ worker_event_enqueue(dev, port, &ev);
+ fwd++;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu FWD=%zu TX=%zu\n",
+ rte_lcore_id(), received, fwd, tx);
+ return 0;
+}
+
static int
worker_do_tx_single_burst(void *arg)
{
@@ -153,6 +188,51 @@ worker_do_tx_single_burst(void *arg)
return 0;
}
+static int
+worker_do_tx_single_burst_atq(void *arg)
+{
+ struct rte_event ev[BATCH_SIZE + 1];
+
+ struct worker_data *data = (struct worker_data *)arg;
+ const uint8_t dev = data->dev_id;
+ const uint8_t port = data->port_id;
+ size_t fwd = 0, received = 0, tx = 0;
+
+ while (!fdata->done) {
+ uint16_t i;
+ uint16_t nb_rx = rte_event_dequeue_burst(dev, port, ev,
+ BATCH_SIZE, 0);
+
+ if (!nb_rx) {
+ rte_pause();
+ continue;
+ }
+
+ received += nb_rx;
+
+ for (i = 0; i < nb_rx; i++) {
+ rte_prefetch0(ev[i + 1].mbuf);
+ if (ev[i].sched_type == RTE_SCHED_TYPE_ATOMIC) {
+
+ worker_tx_pkt(ev[i].mbuf);
+ ev[i].op = RTE_EVENT_OP_RELEASE;
+ tx++;
+ } else
+ worker_fwd_event(&ev[i],
+ RTE_SCHED_TYPE_ATOMIC);
+ work(ev[i].mbuf);
+ }
+
+ worker_event_enqueue_burst(dev, port, ev, nb_rx);
+ fwd += nb_rx;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu FWD=%zu TX=%zu\n",
+ rte_lcore_id(), received, fwd, tx);
+ return 0;
+}
+
/* Multi stage Pipeline Workers */
static int
@@ -697,9 +777,13 @@ set_worker_tx_setup_data(struct setup_data *caps, bool
burst)
uint8_t atq = cdata.all_type_queues ? 1 : 0;
if (cdata.num_stages == 1) {
- if (burst)
+ if (burst && atq)
+ caps->worker_loop = worker_do_tx_single_burst_atq;
+ if (burst && !atq)
caps->worker_loop = worker_do_tx_single_burst;
- if (!burst)
+ if (!burst && atq)
+ caps->worker_loop = worker_do_tx_single_atq;
+ if (!burst && !atq)
caps->worker_loop = worker_do_tx_single;
} else {
if (burst && atq)
--
2.14.1
[dpdk-dev] [PATCH 10/13] examples/eventdev: add single stage pipeline worker
Add optimized eventdev pipeline when ethdev supports thread safe Tx
and number of configured stages is one.
Signed-off-by: Pavan Nikhilesh
---
.../eventdev_pipeline_sw_pmd/pipeline_worker_tx.c | 109 +++--
1 file changed, 101 insertions(+), 8 deletions(-)
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
index e25a06027..15df21b7e 100644
--- a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
+++ b/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
@@ -68,6 +68,91 @@ worker_tx_pkt(struct rte_mbuf *mbuf)
rte_pause();
}
+/* Single stage pipeline workers */
+
+static int
+worker_do_tx_single(void *arg)
+{
+ struct worker_data *data = (struct worker_data *)arg;
+ const uint8_t dev = data->dev_id;
+ const uint8_t port = data->port_id;
+ size_t fwd = 0, received = 0, tx = 0;
+ struct rte_event ev;
+
+ while (!fdata->done) {
+
+ if (!rte_event_dequeue_burst(dev, port, &ev, 1, 0)) {
+ rte_pause();
+ continue;
+ }
+
+ received++;
+
+ if (ev.sched_type == RTE_SCHED_TYPE_ATOMIC) {
+ worker_tx_pkt(ev.mbuf);
+ tx++;
+ continue;
+ }
+ work(ev.mbuf);
+ ev.queue_id++;
+ worker_fwd_event(&ev, RTE_SCHED_TYPE_ATOMIC);
+ worker_event_enqueue(dev, port, &ev);
+ fwd++;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu FWD=%zu TX=%zu\n",
+ rte_lcore_id(), received, fwd, tx);
+ return 0;
+}
+
+static int
+worker_do_tx_single_burst(void *arg)
+{
+ struct rte_event ev[BATCH_SIZE + 1];
+
+ struct worker_data *data = (struct worker_data *)arg;
+ const uint8_t dev = data->dev_id;
+ const uint8_t port = data->port_id;
+ size_t fwd = 0, received = 0, tx = 0;
+
+ while (!fdata->done) {
+ uint16_t i;
+ uint16_t nb_rx = rte_event_dequeue_burst(dev, port, ev,
+ BATCH_SIZE, 0);
+
+ if (!nb_rx) {
+ rte_pause();
+ continue;
+ }
+ received += nb_rx;
+
+ for (i = 0; i < nb_rx; i++) {
+ rte_prefetch0(ev[i + 1].mbuf);
+ if (ev[i].sched_type == RTE_SCHED_TYPE_ATOMIC) {
+
+ worker_tx_pkt(ev[i].mbuf);
+ ev[i].op = RTE_EVENT_OP_RELEASE;
+ tx++;
+
+ } else {
+ ev[i].queue_id++;
+ worker_fwd_event(&ev[i],
+ RTE_SCHED_TYPE_ATOMIC);
+ }
+ work(ev[i].mbuf);
+ }
+
+ worker_event_enqueue_burst(dev, port, ev, nb_rx);
+ fwd += nb_rx;
+ }
+
+ if (!cdata.quiet)
+ printf(" worker %u thread done. RX=%zu FWD=%zu TX=%zu\n",
+ rte_lcore_id(), received, fwd, tx);
+ return 0;
+}
+
/* Multi stage Pipeline Workers */
static int
@@ -265,6 +350,7 @@ worker_do_tx_burst_atq(void *arg)
worker_fwd_event(&ev[i],
cdata.queue_type);
}
+ work(ev[i].mbuf);
}
worker_event_enqueue_burst(dev, port, ev, nb_rx);
@@ -610,14 +696,21 @@ set_worker_tx_setup_data(struct setup_data *caps, bool
burst)
{
uint8_t atq = cdata.all_type_queues ? 1 : 0;
- if (burst && atq)
- caps->worker_loop = worker_do_tx_burst_atq;
- if (burst && !atq)
- caps->worker_loop = worker_do_tx_burst;
- if (!burst && atq)
- caps->worker_loop = worker_do_tx_atq;
- if (!burst && !atq)
- caps->worker_loop = worker_do_tx;
+ if (cdata.num_stages == 1) {
+ if (burst)
+ caps->worker_loop = worker_do_tx_single_burst;
+ if (!burst)
+ caps->worker_loop = worker_do_tx_single;
+ } else {
+ if (burst && atq)
+ caps->worker_loop = worker_do_tx_burst_atq;
+ if (burst && !atq)
+ caps->worker_loop = worker_do_tx_burst;
+ if (!burst && atq)
+ caps->worker_loop = worker_do_tx_atq;
+ if (!burst && !atq)
+ caps->worker_loop = worker_do_tx;
+ }
caps->opt_check = opt_check;
caps->consumer_loop = NULL;
--
2.14.1
[dpdk-dev] [PATCH 12/13] examples/eventdev_pipeline_sw_pmd: rename example
Rename eventdev_pipeline_sw_pmd to eventdev_pipeline as it is no longer
specific underlying event device.
Signed-off-by: Pavan Nikhilesh
---
examples/Makefile | 2 +-
examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/Makefile | 2 +-
examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/main.c | 0
.../{eventdev_pipeline_sw_pmd => eventdev_pipeline}/pipeline_common.h | 0
.../pipeline_worker_generic.c | 0
.../pipeline_worker_tx.c| 0
6 files changed, 2 insertions(+), 2 deletions(-)
rename examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/Makefile (98%)
rename examples/{eventdev_pipeline_sw_pmd => eventdev_pipeline}/main.c (100%)
rename examples/{eventdev_pipeline_sw_pmd =>
eventdev_pipeline}/pipeline_common.h (100%)
rename examples/{eventdev_pipeline_sw_pmd =>
eventdev_pipeline}/pipeline_worker_generic.c (100%)
rename examples/{eventdev_pipeline_sw_pmd =>
eventdev_pipeline}/pipeline_worker_tx.c (100%)
diff --git a/examples/Makefile b/examples/Makefile
index 9f7974a19..a35434d74 100644
--- a/examples/Makefile
+++ b/examples/Makefile
@@ -102,6 +102,6 @@ $(info vm_power_manager requires libvirt >= 0.9.3)
endif
endif
-DIRS-y += eventdev_pipeline_sw_pmd
+DIRS-y += eventdev_pipeline
include $(RTE_SDK)/mk/rte.extsubdir.mk
diff --git a/examples/eventdev_pipeline_sw_pmd/Makefile
b/examples/eventdev_pipeline/Makefile
similarity index 98%
rename from examples/eventdev_pipeline_sw_pmd/Makefile
rename to examples/eventdev_pipeline/Makefile
index 59ee9840a..893220d34 100644
--- a/examples/eventdev_pipeline_sw_pmd/Makefile
+++ b/examples/eventdev_pipeline/Makefile
@@ -38,7 +38,7 @@ RTE_TARGET ?= x86_64-native-linuxapp-gcc
include $(RTE_SDK)/mk/rte.vars.mk
# binary name
-APP = eventdev_pipeline_sw_pmd
+APP = eventdev_pipeline
# all source are stored in SRCS-y
SRCS-y := main.c
diff --git a/examples/eventdev_pipeline_sw_pmd/main.c
b/examples/eventdev_pipeline/main.c
similarity index 100%
rename from examples/eventdev_pipeline_sw_pmd/main.c
rename to examples/eventdev_pipeline/main.c
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_common.h
b/examples/eventdev_pipeline/pipeline_common.h
similarity index 100%
rename from examples/eventdev_pipeline_sw_pmd/pipeline_common.h
rename to examples/eventdev_pipeline/pipeline_common.h
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
b/examples/eventdev_pipeline/pipeline_worker_generic.c
similarity index 100%
rename from examples/eventdev_pipeline_sw_pmd/pipeline_worker_generic.c
rename to examples/eventdev_pipeline/pipeline_worker_generic.c
diff --git a/examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
b/examples/eventdev_pipeline/pipeline_worker_tx.c
similarity index 100%
rename from examples/eventdev_pipeline_sw_pmd/pipeline_worker_tx.c
rename to examples/eventdev_pipeline/pipeline_worker_tx.c
--
2.14.1
[dpdk-dev] [PATCH 13/13] doc: update example eventdev_pipeline
Removed eventdev sw pmd specific information in document, renamed the
document from eventdev_pipeline_sw_pmd to eventdev_pipeline.
Signed-off-by: Pavan Nikhilesh
---
.../{eventdev_pipeline_sw_pmd.rst => eventdev_pipeline.rst} | 6 +++---
doc/guides/sample_app_ug/index.rst | 2 +-
2 files changed, 4 insertions(+), 4 deletions(-)
rename doc/guides/sample_app_ug/{eventdev_pipeline_sw_pmd.rst =>
eventdev_pipeline.rst} (97%)
diff --git a/doc/guides/sample_app_ug/eventdev_pipeline_sw_pmd.rst
b/doc/guides/sample_app_ug/eventdev_pipeline.rst
similarity index 97%
rename from doc/guides/sample_app_ug/eventdev_pipeline_sw_pmd.rst
rename to doc/guides/sample_app_ug/eventdev_pipeline.rst
index 01a5f9b21..ff6d2f0b0 100644
--- a/doc/guides/sample_app_ug/eventdev_pipeline_sw_pmd.rst
+++ b/doc/guides/sample_app_ug/eventdev_pipeline.rst
@@ -29,8 +29,8 @@
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-Eventdev Pipeline SW PMD Sample Application
-===
+Eventdev Pipeline Sample Application
+
The eventdev pipeline sample application is a sample app that demonstrates
the usage of the eventdev API using the software PMD. It shows how an
@@ -74,7 +74,7 @@ these settings is shown below:
.. code-block:: console
-./build/eventdev_pipeline_sw_pmd --vdev event_sw0 -- -r1 -t1 -e4 -w FF00
-s4 -n0 -c32 -W1000 -D
+./build/eventdev_pipeline --vdev event_sw0 -- -r1 -t1 -e4 -w FF00 -s4 -n0
-c32 -W1000 -D
The application has some sanity checking built-in, so if there is a function
(eg; the RX core) which doesn't have a cpu core mask assigned, the application
diff --git a/doc/guides/sample_app_ug/index.rst
b/doc/guides/sample_app_ug/index.rst
index db68ef765..6fcdeb0fb 100644
--- a/doc/guides/sample_app_ug/index.rst
+++ b/doc/guides/sample_app_ug/index.rst
@@ -74,7 +74,7 @@ Sample Applications User Guides
netmap_compatibility
ip_pipeline
test_pipeline
-eventdev_pipeline_sw_pmd
+eventdev_pipeline
dist_app
vm_power_management
tep_termination
--
2.14.1
Re: [dpdk-dev] [PATCH 1/2] net/softnic: enable flow classification function
On 11/30/2017 12:08 PM, Jasvinder Singh wrote: > Enables flow classification on softnic rx path so that proceding > functions of the packet processing pipeline such as metering and > policing could be implemented. > > Example: Create "soft" port for "hard" port ":02:00.1", > enable the Flow Classification (FC) feature with default > settings: > --vdev 'net_softnic0,hard_name=:02:00.1,soft_fc=on' > > Signed-off-by: Jasvinder Singh <...> > +#define PMD_PARAM_SOFT_FC"soft_fc" > +#define PMD_PARAM_SOFT_FC_NB_RULES "soft_fc_nb_rules" > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_SIZE "soft_fc_flow_key_size" > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_OFFSET"soft_fc_flow_key_offset" > +#define PMD_PARAM_SOFT_FC_FLOW_KEY_MASK "soft_fc_flow_key_mask" > + > #define PMD_PARAM_HARD_NAME "hard_name" > #define PMD_PARAM_HARD_TX_QUEUE_ID "hard_tx_queue_id" I wouldn't mention if you won't be already sending a new version but can you please align them :) <...> > @@ -60,7 +60,9 @@ _LDLIBS-y += -L$(RTE_SDK_BIN)/lib > # > _LDLIBS-$(CONFIG_RTE_LIBRTE_FLOW_CLASSIFY) += -lrte_flow_classify > _LDLIBS-$(CONFIG_RTE_LIBRTE_PIPELINE) += -lrte_pipeline > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)+= --whole-archive > _LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE) += -lrte_table > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)+= --no-whole-archive I think it is better to move rte_table between whole-archive flags, but I vaguely remember Thomas preferred this in the past, so adding him for comment.
Re: [dpdk-dev] [PATCH 0/2] net/softnic: add flow classification support
On 11/30/2017 12:08 PM, Jasvinder Singh wrote: > This patchset extends the softnic device by implementing > the software fallback for flow classification as defined > using Flow APIs. When feature is enabled, softnic classifies > and executes actions on the received packets based on flow > rules specified using rte_flow.h. > > The motivation behind this feature is to implement metering > and policing function in softnic which requires each flow to > be classified at the preceding stage of the packet processing > pipeline. Another aspect is to support classification of the > greater number of flows (For e.g. 64K) which is not feasible > in some hardware due to resource limitation such as memory, etc. > > This feature, inherently, uses the abstraction provided by dpdk > librte_table library to create hash table for classification, > and allows users to validate, add and delete flow rules. Current > version does not implement all functions and will be completed > in the next version. > > Jasvinder Singh (2): > net/softnic: add flow classification support > net/softnic: add flow classification ops Hi Jasvinder, I was about to ask adding documentation for "software fallback for flow classification" but it seems there is no documentation at all for softnic, it seems we missed it, and softnic is a complex PMD which requires some documentation I believe. Can you please add documentation including new flow classification fallback info? And I am getting some build errors [1][2]. Thanks, ferruh [1] .../dpdk/drivers/net/softnic/rte_eth_softnic_fc.c:558:3: error: implicit conversion from enumeration type 'enum rte_tm_error_type' to different enumeration type 'enum rte_flow_error_type' [-Werror,-Wenum-conversion] RTE_TM_ERROR_TYPE_UNSPECIFIED, ^ .../dpdk/drivers/net/softnic/rte_eth_softnic_fc.c:577:3: error: implicit conversion from enumeration type 'enum rte_tm_error_type' to different enumeration type 'enum rte_flow_error_type' [-Werror,-Wenum-conversion] RTE_TM_ERROR_TYPE_UNSPECIFIED, ^ .../dpdk/drivers/net/softnic/rte_eth_softnic_fc.c:596:3: error: implicit conversion from enumeration type 'enum rte_tm_error_type' to different enumeration type 'enum rte_flow_error_type' [-Werror,-Wenum-conversion] RTE_TM_ERROR_TYPE_UNSPECIFIED, ^ [2] .../dpdk/drivers/net/softnic/rte_eth_softnic_fc.c: In function ‘pmd_flow_create’: .../dpdk/drivers/net/softnic/rte_eth_softnic_fc.c:584:2: error: argument 1 null where non-null expected [-Werror=nonnull] memcpy((void *)f->pattern, (const void *)pattern, sizeof(*f->pattern)); ^~~
[dpdk-dev] [PATCH v3 1/5] bbdev: librte_bbdev library
- BBDEV library files - BBDEV is tagged as EXPERIMENTAL - Makefiles and configuration macros definition - The bbdev framework and the 'null' driver are enabled by default - The bbdev test framework is enabled by default - Release Notes of the initial version and MAINTAINERS Signed-off-by: Amr Mokhtar --- MAINTAINERS| 11 + config/common_base | 18 + doc/guides/rel_notes/release_18_02.rst | 10 + lib/Makefile |3 + lib/librte_bbdev/Makefile | 56 ++ lib/librte_bbdev/rte_bbdev.c | 1095 lib/librte_bbdev/rte_bbdev.h | 741 + lib/librte_bbdev/rte_bbdev_op.h| 532 lib/librte_bbdev/rte_bbdev_op_ldpc.h | 545 lib/librte_bbdev/rte_bbdev_pmd.h | 223 +++ lib/librte_bbdev/rte_bbdev_version.map | 37 ++ mk/rte.app.mk | 13 + 12 files changed, 3284 insertions(+) create mode 100644 lib/librte_bbdev/Makefile create mode 100644 lib/librte_bbdev/rte_bbdev.c create mode 100644 lib/librte_bbdev/rte_bbdev.h create mode 100644 lib/librte_bbdev/rte_bbdev_op.h create mode 100644 lib/librte_bbdev/rte_bbdev_op_ldpc.h create mode 100644 lib/librte_bbdev/rte_bbdev_pmd.h create mode 100644 lib/librte_bbdev/rte_bbdev_version.map diff --git a/MAINTAINERS b/MAINTAINERS index f0baeb4..696f048 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -272,6 +272,17 @@ F: lib/librte_cryptodev/ F: test/test/test_cryptodev* F: examples/l2fwd-crypto/ +BBDEV API - EXPERIMENTAL +M: Amr Mokhtar +F: lib/librte_bbdev/ +F: drivers/bbdev/ +F: app/test-bbdev +F: examples/bbdev_app/ +F: doc/guides/bbdevs/ +F: doc/guides/prog_guide/bbdev.rst +F: doc/guides/sample_app_ug/bbdev_app.rst +F: doc/guides/tools/testbbdev.rst + Security API - EXPERIMENTAL M: Akhil Goyal M: Declan Doherty diff --git a/config/common_base b/config/common_base index e74febe..5243e0f 100644 --- a/config/common_base +++ b/config/common_base @@ -593,6 +593,24 @@ CONFIG_RTE_LIBRTE_PMD_SW_EVENTDEV_DEBUG=n CONFIG_RTE_LIBRTE_PMD_OCTEONTX_SSOVF=y CONFIG_RTE_LIBRTE_PMD_OCTEONTX_SSOVF_DEBUG=n +# Compile generic wireless base band device library +# EXPERIMENTAL: API may change without prior notice +# +CONFIG_RTE_LIBRTE_BBDEV=y +CONFIG_RTE_LIBRTE_BBDEV_DEBUG=n +CONFIG_RTE_BBDEV_MAX_DEVS=128 +CONFIG_RTE_BBDEV_NAME_MAX_LEN=64 + +# +# Compile PMD for NULL bbdev device +# +CONFIG_RTE_LIBRTE_PMD_BBDEV_NULL=y + +# +# Compile PMD for turbo software bbdev device +# +CONFIG_RTE_LIBRTE_PMD_BBDEV_TURBO_SW=n + # # Compile librte_ring # diff --git a/doc/guides/rel_notes/release_18_02.rst b/doc/guides/rel_notes/release_18_02.rst index 24b67bb..3e5a8de 100644 --- a/doc/guides/rel_notes/release_18_02.rst +++ b/doc/guides/rel_notes/release_18_02.rst @@ -41,6 +41,16 @@ New Features Also, make sure to start the actual text at the margin. = +* **Added Wireless Base Band Device (bbdev).** + + The Wireless Baseband Device library is an acceleration abstraction + framework for 3gpp Layer 1 processing functions that provides a common + programming interface for seamless opeartion on integrated or discrete + hardware accelerators or using optimized software libraries for signal + processing. + The current release only supports 4G Turbo Code FEC function. + + See the :doc:`../prog_guide/bbdev` programmer's guide for more details. API Changes --- diff --git a/lib/Makefile b/lib/Makefile index dc4e8df..8e5bd7c 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -58,6 +58,9 @@ DEPDIRS-librte_security += librte_ether DEPDIRS-librte_security += librte_cryptodev DIRS-$(CONFIG_RTE_LIBRTE_EVENTDEV) += librte_eventdev DEPDIRS-librte_eventdev := librte_eal librte_ring librte_ether librte_hash +DIRS-$(CONFIG_RTE_LIBRTE_BBDEV) += librte_bbdev +DEPDIRS-librte_bbdev := librte_eal librte_mempool librte_ring librte_mbuf +DEPDIRS-librte_bbdev += librte_kvargs DIRS-$(CONFIG_RTE_LIBRTE_VHOST) += librte_vhost DEPDIRS-librte_vhost := librte_eal librte_mempool librte_mbuf librte_ether DIRS-$(CONFIG_RTE_LIBRTE_HASH) += librte_hash diff --git a/lib/librte_bbdev/Makefile b/lib/librte_bbdev/Makefile new file mode 100644 index 000..8d59564 --- /dev/null +++ b/lib/librte_bbdev/Makefile @@ -0,0 +1,56 @@ +# BSD LICENSE +# +# Copyright(c) 2017 Intel Corporation. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials p
[dpdk-dev] [PATCH v3 2/5] bbdev: PMD drivers (null/turbo_sw)
- bbdev 'null' PMD enabled by default - bbdev 'turbo_sw' PMD disabled by default - 'turbo_sw' requires the external FLEXRAN SDK libraries Signed-off-by: Amr Mokhtar --- drivers/Makefile |2 + drivers/bbdev/Makefile | 41 + drivers/bbdev/null/Makefile| 51 + drivers/bbdev/null/bbdev_null.c| 385 ++ drivers/bbdev/null/rte_pmd_bbdev_null_version.map |3 + drivers/bbdev/turbo_sw/Makefile| 68 ++ drivers/bbdev/turbo_sw/bbdev_turbo_software.c | 1226 .../turbo_sw/rte_pmd_bbdev_turbo_sw_version.map|3 + 8 files changed, 1779 insertions(+) create mode 100644 drivers/bbdev/Makefile create mode 100644 drivers/bbdev/null/Makefile create mode 100644 drivers/bbdev/null/bbdev_null.c create mode 100644 drivers/bbdev/null/rte_pmd_bbdev_null_version.map create mode 100644 drivers/bbdev/turbo_sw/Makefile create mode 100644 drivers/bbdev/turbo_sw/bbdev_turbo_software.c create mode 100644 drivers/bbdev/turbo_sw/rte_pmd_bbdev_turbo_sw_version.map diff --git a/drivers/Makefile b/drivers/Makefile index db0cd76..eb9a1f8 100644 --- a/drivers/Makefile +++ b/drivers/Makefile @@ -40,5 +40,7 @@ DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += crypto DEPDIRS-crypto := bus mempool DIRS-$(CONFIG_RTE_LIBRTE_EVENTDEV) += event DEPDIRS-event := bus mempool net +DIRS-$(CONFIG_RTE_LIBRTE_BBDEV) += bbdev +DEPDIRS-bbdev := bus mempool include $(RTE_SDK)/mk/rte.subdir.mk diff --git a/drivers/bbdev/Makefile b/drivers/bbdev/Makefile new file mode 100644 index 000..c70cd06 --- /dev/null +++ b/drivers/bbdev/Makefile @@ -0,0 +1,41 @@ +# BSD LICENSE +# +# Copyright(c) 2017 Intel Corporation. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +include $(RTE_SDK)/mk/rte.vars.mk + +core-libs := librte_eal librte_mbuf librte_mempool librte_ring +core-libs += librte_bbdev librte_kvargs librte_cfgfile + +DIRS-$(CONFIG_RTE_LIBRTE_PMD_BBDEV_NULL) += null +DEPDIRS-null = $(core-libs) +DIRS-$(CONFIG_RTE_LIBRTE_PMD_BBDEV_TURBO_SW) += turbo_sw +DEPDIRS-turbo_sw = $(core-libs) + +include $(RTE_SDK)/mk/rte.subdir.mk diff --git a/drivers/bbdev/null/Makefile b/drivers/bbdev/null/Makefile new file mode 100644 index 000..2b51b02 --- /dev/null +++ b/drivers/bbdev/null/Makefile @@ -0,0 +1,51 @@ +# BSD LICENSE +# +# Copyright(c) 2017 Intel Corporation. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE
[dpdk-dev] [PATCH v3 4/5] bbdev: sample app
- Sample application performing a loop-back over ethernet using a bbbdev device - A packet is received on an Rx ethdev port -> enqueued for baseband operation -> dequeued -> looped-back to a Tx ethdev port - 'Turbo_sw' PMD must be enabled for the app to be functional Signed-off-by: Amr Mokhtar --- examples/Makefile |1 + examples/bbdev_app/Makefile | 50 ++ examples/bbdev_app/main.c | 1452 +++ 3 files changed, 1503 insertions(+) create mode 100644 examples/bbdev_app/Makefile create mode 100644 examples/bbdev_app/main.c diff --git a/examples/Makefile b/examples/Makefile index 9f7974a..26bf256 100644 --- a/examples/Makefile +++ b/examples/Makefile @@ -61,6 +61,7 @@ ifneq ($(PQOS_INSTALL_PATH),) DIRS-y += l2fwd-cat endif DIRS-$(CONFIG_RTE_LIBRTE_CRYPTODEV) += l2fwd-crypto +DIRS-$(CONFIG_RTE_LIBRTE_BBDEV) += bbdev_app DIRS-$(CONFIG_RTE_LIBRTE_JOBSTATS) += l2fwd-jobstats DIRS-y += l2fwd-keepalive DIRS-y += l2fwd-keepalive/ka-agent diff --git a/examples/bbdev_app/Makefile b/examples/bbdev_app/Makefile new file mode 100644 index 000..1583251 --- /dev/null +++ b/examples/bbdev_app/Makefile @@ -0,0 +1,50 @@ +# BSD LICENSE +# +# Copyright(c) 2017 Intel Corporation. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overridden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +# binary name +APP = bbdev + +# all source are stored in SRCS-y +SRCS-y := main.c + +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/examples/bbdev_app/main.c b/examples/bbdev_app/main.c new file mode 100644 index 000..9343e95 --- /dev/null +++ b/examples/bbdev_app/main.c @@ -0,0 +1,1452 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2017 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHER
[dpdk-dev] [PATCH v3 5/5] bbdev: documentation
- Wireless Baseband Device Library Programmer’s Guide - test-bbdev User Guide - BBDEV Sample Application User Guides - Baseband Device Drivers Guides - Doxygen API Signed-off-by: Amr Mokhtar --- doc/api/doxy-api-index.md | 1 + doc/api/doxy-api.conf | 1 + doc/guides/bbdevs/index.rst| 40 +++ doc/guides/bbdevs/null.rst | 77 + doc/guides/bbdevs/turbo_sw.rst | 175 ++ doc/guides/index.rst | 1 + doc/guides/prog_guide/bbdev.rst| 613 + doc/guides/prog_guide/index.rst| 1 + doc/guides/sample_app_ug/bbdev_app.rst | 160 + doc/guides/sample_app_ug/index.rst | 1 + doc/guides/tools/index.rst | 1 + doc/guides/tools/testbbdev.rst | 600 12 files changed, 1671 insertions(+) create mode 100644 doc/guides/bbdevs/index.rst create mode 100644 doc/guides/bbdevs/null.rst create mode 100644 doc/guides/bbdevs/turbo_sw.rst create mode 100644 doc/guides/prog_guide/bbdev.rst create mode 100644 doc/guides/sample_app_ug/bbdev_app.rst create mode 100644 doc/guides/tools/testbbdev.rst diff --git a/doc/api/doxy-api-index.md b/doc/api/doxy-api-index.md index 3492702..8d7ff89 100644 --- a/doc/api/doxy-api-index.md +++ b/doc/api/doxy-api-index.md @@ -50,6 +50,7 @@ The public API headers are grouped by topics: [bitrate](@ref rte_bitrate.h), [latency](@ref rte_latencystats.h), [devargs](@ref rte_devargs.h), + [bbdev] (@ref rte_bbdev.h), [PCI](@ref rte_pci.h) - **device specific**: diff --git a/doc/api/doxy-api.conf b/doc/api/doxy-api.conf index b2cbe94..241cae3 100644 --- a/doc/api/doxy-api.conf +++ b/doc/api/doxy-api.conf @@ -39,6 +39,7 @@ INPUT = doc/api/doxy-api-index.md \ lib/librte_eal/common/include \ lib/librte_eal/common/include/generic \ lib/librte_acl \ + lib/librte_bbdev \ lib/librte_bitratestats \ lib/librte_cfgfile \ lib/librte_cmdline \ diff --git a/doc/guides/bbdevs/index.rst b/doc/guides/bbdevs/index.rst new file mode 100644 index 000..c9aa1b0 --- /dev/null +++ b/doc/guides/bbdevs/index.rst @@ -0,0 +1,40 @@ +.. + BSD LICENSE + + Copyright(c) 2017 Intel Corporation. All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in + the documentation and/or other materials provided with the + distribution. + * Neither the name of Intel Corporation nor the names of its + contributors may be used to endorse or promote products derived + from this software without specific prior written permission. + + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +Baseband Device Drivers +=== + +.. toctree:: +:maxdepth: 2 +:numbered: + +null +turbo_sw diff --git a/doc/guides/bbdevs/null.rst b/doc/guides/bbdevs/null.rst new file mode 100644 index 000..0f40232 --- /dev/null +++ b/doc/guides/bbdevs/null.rst @@ -0,0 +1,77 @@ +.. + BSD LICENSE + + Copyright(c) 2017 Intel Corporation. All rights reserved. + + Redistribution and use in source and binary forms, with or without + modification, are permitted provided that the following conditions + are met: + + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in + the documentation and/or other materials provided with the + distribution.
[dpdk-dev] [PATCH v3 0/5] Wireless Baseband Device (bbdev)
Hello everyone, A refreshed v3 patch of the Wireless Baseband Device (bbdev) is enclosed. Addressing the feedback received from the community on the application interface in the RFC and v1 patch with an enhanced SW Turbo driver. v3: * Cleaner Turbo Code operation interface * Enahnced SW Turbo PMD (turbo_sw) * Removed pci & vdev dependency from bbdev library interface * Updated download instructions for Intel FlexRAN SDK library v2: * Split the functionality of rte_bbdev_configure() into smaller portions -> rte_bbdev_setup_queues() and rte_bbdev_enale_intr() * Split rte_bbdev_enqueue() -> rte_bbdev_enc_enqueue() and rte_bbdev_dec_enqueue() * Split rte_bbdev_dequeue() -> rte_bbdev_enc_dequeue() and rte_bbdev_dec_dequeue() * Removed attached flag until hotplug is properly supported in DPDK * More details on the installation of FlexRAN SDK libraries in accordance with Turbo_sw PMD * Minor build fixes for other targets: bsdapp-gcc, bsdapp-clang and linuxapp-clang. * Better-organized patchwork http://dpdk.org/dev/patchwork/patch/30498/ http://dpdk.org/dev/patchwork/patch/30499/ http://dpdk.org/dev/patchwork/patch/30500/ http://dpdk.org/dev/patchwork/patch/30501/ http://dpdk.org/dev/patchwork/patch/30502/ v1: * Initial release of BBDEV library. * Support Turbo Code FEC with two virtual devices (vdev): - Null Turbo PMD - Turbo_sw PMD * A complete Test suite for Turbo Encode/Decode and None operations * Test Vectors parsing and testing functionality * Sample App for a looped-back bbdev with ethdev * Documentation in rst format for all new components http://dpdk.org/dev/patchwork/patch/29447/ http://dpdk.org/dev/patchwork/patch/29448/ http://dpdk.org/dev/patchwork/patch/29450/ http://dpdk.org/dev/patchwork/patch/29449/ http://dpdk.org/dev/patchwork/patch/29452/ http://dpdk.org/dev/patchwork/patch/29451/ RFC: [1] http://dpdk.org/ml/archives/dev/2017-August/073585.html [2] http://dpdk.org/ml/archives/dev/2017-August/073584.html Amr Mokhtar (5): bbdev: librte_bbdev library bbdev: PMD drivers (null/turbo_sw) bbdev: test applications bbdev: sample app bbdev: documentation MAINTAINERS| 11 + app/Makefile |4 + app/test-bbdev/Makefile| 53 + app/test-bbdev/main.c | 353 app/test-bbdev/main.h | 148 ++ app/test-bbdev/test-bbdev.py | 139 ++ app/test-bbdev/test_bbdev.c| 1406 + app/test-bbdev/test_bbdev_perf.c | 2193 app/test-bbdev/test_bbdev_vector.c | 963 + app/test-bbdev/test_bbdev_vector.h | 99 + app/test-bbdev/test_vectors/bbdev_vector_null.data | 32 + .../test_vectors/bbdev_vector_td_default.data | 81 + .../test_vectors/bbdev_vector_te_default.data | 60 + config/common_base | 18 + doc/api/doxy-api-index.md |1 + doc/api/doxy-api.conf |1 + doc/guides/bbdevs/index.rst| 40 + doc/guides/bbdevs/null.rst | 77 + doc/guides/bbdevs/turbo_sw.rst | 175 ++ doc/guides/index.rst |1 + doc/guides/prog_guide/bbdev.rst| 613 ++ doc/guides/prog_guide/index.rst|1 + doc/guides/rel_notes/release_18_02.rst | 10 + doc/guides/sample_app_ug/bbdev_app.rst | 160 ++ doc/guides/sample_app_ug/index.rst |1 + doc/guides/tools/index.rst |1 + doc/guides/tools/testbbdev.rst | 600 ++ drivers/Makefile |2 + drivers/bbdev/Makefile | 41 + drivers/bbdev/null/Makefile| 51 + drivers/bbdev/null/bbdev_null.c| 385 drivers/bbdev/null/rte_pmd_bbdev_null_version.map |3 + drivers/bbdev/turbo_sw/Makefile| 68 + drivers/bbdev/turbo_sw/bbdev_turbo_software.c | 1226 +++ .../turbo_sw/rte_pmd_bbdev_turbo_sw_version.map|3 + examples/Makefile |1 + examples/bbdev_app/Makefile| 50 + examples/bbdev_app/main.c | 1452 + lib/Makefile |3 + lib/librte_bbdev/Makefile | 56 + lib/librte_bbdev/rte_bbdev.c | 1095 ++ lib/librte_bbdev/rte_bbdev.h | 741 +++ lib/librte_bbdev/rte_bbdev_op.h| 532 + lib/librte_bbdev/rte_bbdev_op_ldpc.h | 545 + lib/librte_bbdev/rte_bbdev_pmd.h | 223 ++ lib/librte_bbdev/rte_bbdev_
[dpdk-dev] [PATCH] net/vmxnet3: convert to 3-Clause BSD License
On behalf of the DPDK Technical board, Hemant Agrawal observed that the DPDK project's Intellectual Property Policy (http://dpdk.org/about/charter) requires 3-Clause BSD license or an exception approval. However, two vmxnet3 source files have 2-Clause BSD license. This patch modifies those licenses to 3-Clause BSD license. Reported-by: Hemant Agrawal Acked-by: Cheryl Houser Acked-by: Bharat Mota Signed-off-by: Shrikrishna Khare --- drivers/net/vmxnet3/base/upt1_defs.h| 22 +- drivers/net/vmxnet3/base/vmxnet3_defs.h | 22 +- 2 files changed, 2 insertions(+), 42 deletions(-) diff --git a/drivers/net/vmxnet3/base/upt1_defs.h b/drivers/net/vmxnet3/base/upt1_defs.h index d9144e3..cf9141b 100644 --- a/drivers/net/vmxnet3/base/upt1_defs.h +++ b/drivers/net/vmxnet3/base/upt1_defs.h @@ -1,27 +1,7 @@ /* * Copyright (C) 2007 VMware, Inc. All rights reserved. * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - *notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - *notice, this list of conditions and the following disclaimer in the - *documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. + * SPDX-License-Identifier:BSD-3-Clause * */ diff --git a/drivers/net/vmxnet3/base/vmxnet3_defs.h b/drivers/net/vmxnet3/base/vmxnet3_defs.h index bfa9622..a455e27 100644 --- a/drivers/net/vmxnet3/base/vmxnet3_defs.h +++ b/drivers/net/vmxnet3/base/vmxnet3_defs.h @@ -1,27 +1,7 @@ /* * Copyright (C) 2007 VMware, Inc. All rights reserved. * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * - * 1. Redistributions of source code must retain the above copyright - *notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - *notice, this list of conditions and the following disclaimer in the - *documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. + * SPDX-License-Identifier:BSD-3-Clause * */ -- 2.6.2
Re: [dpdk-dev] [PATCH v3] bnxt: Support time_sync related APIs
On Mon, Dec 4, 2017 at 11:26 PM, Somnath Kotur wrote: > Implemented the 'time_sync' related APIs for supporting the PTP protocol. > > Signed-off-by: Somnath Kotur > Acked-by: Ajit Khaparde > --- > v3 Changes > - Initialize the time counters correctly during timesync_enable > - Use correct flag to check for PTP in Rx completion > v2 Changes > - Updated bnxt.ini to announce 'timesync' feature > - Added code to correctly detect PTP in the pkt and set corresponding > ol_flags > for PTP in Rx path >
Re: [dpdk-dev] [PATCH 1/2] net/softnic: enable flow classification function
07/12/2017 22:02, Ferruh Yigit: > On 11/30/2017 12:08 PM, Jasvinder Singh wrote: > > _LDLIBS-$(CONFIG_RTE_LIBRTE_FLOW_CLASSIFY) += -lrte_flow_classify > > _LDLIBS-$(CONFIG_RTE_LIBRTE_PIPELINE) += -lrte_pipeline > > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)+= --whole-archive > > _LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE) += -lrte_table > > +_LDLIBS-$(CONFIG_RTE_LIBRTE_TABLE)+= --no-whole-archive > > I think it is better to move rte_table between whole-archive flags, but I > vaguely remember Thomas preferred this in the past, so adding him for comment. Yes it is good to keep the order, and insert whole-archive flags where necessary. Ideally we should not require these flags. Just checking: you need them for static link of libs while dynamically linking the PMDs, right?
Re: [dpdk-dev] [PATCH v3] bnxt: Support time_sync related APIs
On 12/7/2017 2:04 PM, Ajit Khaparde wrote: > On Mon, Dec 4, 2017 at 11:26 PM, Somnath Kotur > wrote: > >> Implemented the 'time_sync' related APIs for supporting the PTP protocol. >> >> Signed-off-by: Somnath Kotur >> > Acked-by: > Ajit Khaparde "net/bnxt: support timesync" Applied to dpdk-next-net/master, thanks.
Re: [dpdk-dev] [PATCH] net/vmxnet3: convert to 3-Clause BSD License
On 12/7/2017 1:59 PM, Shrikrishna Khare wrote: > On behalf of the DPDK Technical board, Hemant Agrawal observed that the > DPDK project's Intellectual Property Policy (http://dpdk.org/about/charter) > requires 3-Clause BSD license or an exception approval. However, two > vmxnet3 source files have 2-Clause BSD license. > > This patch modifies those licenses to 3-Clause BSD license. > > Reported-by: Hemant Agrawal > Acked-by: Cheryl Houser > Acked-by: Bharat Mota > Signed-off-by: Shrikrishna Khare Applied to dpdk-next-net/master, thanks.
Re: [dpdk-dev] [PATCH] net/vmxnet3: convert to 3-Clause BSD License
On 12/7/2017 3:27 PM, Ferruh Yigit wrote: > On 12/7/2017 1:59 PM, Shrikrishna Khare wrote: >> On behalf of the DPDK Technical board, Hemant Agrawal observed that the >> DPDK project's Intellectual Property Policy (http://dpdk.org/about/charter) >> requires 3-Clause BSD license or an exception approval. However, two >> vmxnet3 source files have 2-Clause BSD license. >> >> This patch modifies those licenses to 3-Clause BSD license. >> >> Reported-by: Hemant Agrawal >> Acked-by: Cheryl Houser >> Acked-by: Bharat Mota >> Signed-off-by: Shrikrishna Khare > > Applied to dpdk-next-net/master, thanks. Hi Hemant, Just to clarify, we want license header update for all licenses including existing BSD-3 ones, right? Not just for wrong/different license types. Thanks, ferruh
Re: [dpdk-dev] [PATCH v2 1/2] Introducing SPDX License Identifiers
On 12/1/2017 12:38 AM, Hemant Agrawal wrote: > The DPDK uses the Open Source BSD-3-Clause license for the core libraries > and drivers. The kernel components are naturally GPLv2 licensed. > > Many of the files in the DPDK source code contain the full text of the > applicable license. For example, most of the BSD-3-Clause files contain a > full copy of the BSD-3-Clause license text. > > Including big blocks of License headers in all files blows up the source > code with mostly redundant information. An additional problem is that even > the same licenses are referred to by a number of slightly varying text > blocks (full, abbreviated, different indentation, line wrapping and/or > white space, with obsolete address information, ...) which makes validation > and automatic processing a nightmare. > > To make this easier, DPDK is adpoting the use of a single line reference to > Unique License Identifiers in source files as defined by the Linux > Foundation's SPDX project [1]. > > Adding license information in this fashion, rather than adding full license > text, can be more efficient for developers; decreases errors; and improves > automated detection of licenses. The current set of valid, predefined SPDX > identifiers is set forth on the SPDX License List[2] > at https://spdx.org/licenses/. > > For example, to label a file as subject to the BSD-3-Clause license, > the following text would be used: > > Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER > SPDX-License-Identifier:BSD-3-Clause > > To label a file as GPL-2.0 (e.g., for code that runs in the kernel), the > following text would be used: > > Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER > SPDX-License-Identifier:GPL-2.0 > > To label a file as dual-licensed with BSD-3-Clause and GPL-2.0 (e.g., for > code that is shared between the kernel and userspace), the following text > would be used: > > Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER > SPDX-License-Identifier:BSD-3-Clause OR GPL-2.0 > > To label a file as dual-licensed with BSD-3-Clause and LGPL-2.1 (e.g., for > code that is shared between the kernel and userspace), the following text > would be used: > > Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER > SPDX-License-Identifier:BSD-3-Clause OR LGPL-2.1 > > Note: Any new file contributions in DPDK shall adhere to the above scheme. > It is also being recommended to replace the existing license text in the > code with SPDX-License-Identifiers. > > Note 2: DPDK currently adhere to it's IP policies[3]. Any exception to this > shall be approved by DPDK tech board and DPDK Governing Board. Steps for > any exception approval: > 1. Mention the appropriate license identifier form SPDX. If the license is >not listed in SPDX Licenses. It is the submitters responsibiliity to get >it first listed. > 2. Get the required approval from the DPDK Technical Board. Technical board >may advise the author to check alternate means first. If no other >alternatives are found and the merit of the contributions are important >for DPDK's mission, it may decide on such exception with two-thirds vote >of the members. > 3. Technical board then approach Governing board for such limited approval >for the given contribution only. > > Any approvals shall be documented in "Licenses/exceptions.txt" with record > dates. > > Note 3: Projects like U-boot have been been using SPDX License Idenfiers > successfully [2]. They have been referered in implementing SPDX based > guidelines in DPDK. > > Note 4: From the legal point of view, this patch is supposed to be only a > change to the textual representation of the license information, but in no > way any change to the actual license terms. With this patch applied, all > files will still be licensed under the same terms they were before. > > Signed-off-by: Hemant Agrawal > Acked-by: Stephen Hemminger > --- > LICENSE.GPL | 339 > LICENSE.LGPL| 502 > > Licenses/Exceptions.txt | 12 + > Licenses/README | 82 ++ > Licenses/bsd-3-clause.txt | 9 + > Licenses/gpl-2.0.txt| 339 > Licenses/lgpl-2.1.txt | 502 > Hi Hemant, Are new gpl-2.0.txt & lgpl-2.1.txt files identical with old LICENSE.GPL & LICENSE.LGPL? If so, does it make sense to make commit with "git mv" so that patch won't contain all text resulting smaller patch, and it will highlight that nothing changed but moved/renamed? Thanks, ferruh <>
Re: [dpdk-dev] [PATCH v2 1/2] ethdev: add GENEVE flow pattern item
On 12/4/2017 6:03 AM, Adrien Mazarguil wrote: > On Fri, Dec 01, 2017 at 10:43:15AM +, Andrew Rybchenko wrote: >> From: Roman Zhukov >> >> Add new pattern item RTE_FLOW_ITEM_TYPE_GENEVE in flow API. >> Add default mask for the item. >> >> Signed-off-by: Roman Zhukov >> Signed-off-by: Andrew Rybchenko > > Just one remaining nit, please see below. While you could address it through > a subsequent patch, you might as well send v3 directly to make things easier > for Ferruh. > > Otherwise it's all good, thanks. May I take this as (for series): Acked-by: Adrien Mazarguil > >> --- >> doc/guides/prog_guide/rte_flow.rst | 12 >> lib/librte_ether/rte_flow.c| 1 + >> lib/librte_ether/rte_flow.h| 30 ++ >> 3 files changed, 43 insertions(+) >> >> diff --git a/doc/guides/prog_guide/rte_flow.rst >> b/doc/guides/prog_guide/rte_flow.rst >> index d158be5..5b8f9c5 100644 >> --- a/doc/guides/prog_guide/rte_flow.rst >> +++ b/doc/guides/prog_guide/rte_flow.rst >> @@ -980,6 +980,18 @@ Matches an ESP header. >> - ``hdr``: ESP header definition (``rte_esp.h``). >> - Default ``mask`` matches SPI only. >> >> +Item: ``GENEVE`` >> +^^^ > > Missing "^" under title, this may trigger a warning during documentation > generation. I can add that one missing '^' while applying, no patch required.
Re: [dpdk-dev] [PATCH 1/2] net/octeontx: add channel to port id mapping
On 11/28/2017 6:58 AM, Pavan Nikhilesh wrote:
> The channel to port id map is used by event octeontx to map the received
> wqe to the respective ethdev port.
>
> Signed-off-by: Pavan Nikhilesh
<...>
> @@ -52,12 +52,18 @@
> #define OCTEONTX_VDEV_NR_PORT_ARG("nr_port")
> #define OCTEONTX_MAX_NAME_LEN32
>
> +#define OCTEONTX_MAX_BGX_PORTS 4
> +#define OCTEONTX_MAX_LMAC_PER_BGX4
> +
> static inline struct octeontx_nic *
> octeontx_pmd_priv(struct rte_eth_dev *dev)
> {
> return dev->data->dev_private;
> }
>
> +uint16_t __rte_cache_aligned
> +octeontx_pchan_map[OCTEONTX_MAX_BGX_PORTS][OCTEONTX_MAX_LMAC_PER_BGX];
defining global variable in header is generally not good a idea, is there a
reason why not variable defined in octeontx_ethdev.c and exported here, so that
both octeontx ethdev and eventdev can use it?
btw, is build time dependency between octeontx ethdev and eventdev documented
somewhere?
[dpdk-dev] dpdk native tcp/ip stack (ans) performance reach 11.78Mpps
hi, Share dpdk tcp/ip stack performance here: CPU Intel(R) Xeon(R) CPU E5-2683 v3 @ 2.00GHz NIC 02:00.0 Ethernet controller: Intel Corporation 82599ES 10-Gigabit SFI/SFP+ Network Connection (rev 01) 02:00.1 Ethernet controller: Intel Corporation 82599ES 10-Gigabit SFI/SFP+ Network Connection (rev 01) | routing forwarding performance | | (one lcore) | |--| | Packet size(byte)| Throughput(Mpps) | |--| | 64 | 11.78 | |--| | 128 | Line Rate| |--| If add 1000 route entity, the performance isn't reduced. The performance increases as core increases. For detail steps, please refer to doc(https://github.com/ansyun/dpdk-ans/blob/master/doc/guides/ans_performance_report.pdf) -- Best Regards, ANS-DEV-TEAM
Re: [dpdk-dev] [PATCH v6] net/i40e: determine number of queues per VF during run time
> -Original Message-
> From: Ananyev, Konstantin
> Sent: Sunday, December 3, 2017 7:20 PM
> To: Dai, Wei ; Wu, Jingjing ;
> Xing, Beilei
> Cc: dev@dpdk.org
> Subject: RE: [PATCH v6] net/i40e: determine number of queues per VF during
> run time
>
> Hi Wei,
>
> > -Original Message-
> > From: Dai, Wei
> > Sent: Monday, November 27, 2017 8:09 AM
> > To: Wu, Jingjing ; Xing, Beilei
> > ; Ananyev, Konstantin
> >
> > Cc: dev@dpdk.org; Dai, Wei
> > Subject: [PATCH v6] net/i40e: determine number of queues per VF during
> > run time
> >
> > Without this patch, the number of queues per i40e VF is defined as 4
> > by CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF=4 in
> config/common_base.
> > It is fixed value determined in building time and can't be changed
> > during run time.
> > With this patch, the number of queues per i40e VF can be determinated
> > during run time. For example, if the PCI address of an i40e VF is
Here it should be 'if the PCI address of an i40e PF is'
I will correct it in my v7 patch.
> > :bb.cc, with the EAL parameter -w :bb.cc,queue-num-per-vf=8,
> > the number of queues per VF is 8.
> > If there is no "queue-num-per-vf" setting in EAL parameters, it is 4
> > by default as before. And if the value after the "queue-num-per-vf"
> > is invalid, it is set as 4 forcibly. The valid values include 1, 2, 4,
> > 8, 16 .
> >
> > Signed-off-by: Wei Dai
> >
> > ---
> > v6:
> > fix a small bug when detecting end character of strtoul
> > v5:
> > fix git log message and WARNING of coding stype
> > v4:
> > use rte_kvargs instead of pervious parsing function;
> > use malloc/free instead of rte_zmalloc/rte_free.
> > v3:
> > fix WARNING of coding style issues from checkpa...@dpdk.org
> > v2:
> > fix WARNING of coding style issues from checkpa...@dpdk.org
> > ---
> > config/common_base | 1 -
> > drivers/net/i40e/i40e_ethdev.c | 67
> > --
> > 2 files changed, 65 insertions(+), 3 deletions(-)
> >
> > + if (errno != 0 || end == value || *end != 0) {
> > + PMD_DRV_LOG(WARNING, "Wrong VF queue number = %s, Now it
> is "
> > + "kept the value = %hu", value, pf->vf_nb_qp_max);
> > + return -(EINVAL);
> > + }
> > +
> > + if (num <= 16 && rte_is_power_of_2(num))
>
> As a nit - better to use some macro instead of '16' here.
> Apart from that - looks good to me.
> Acked-by: Konstantin Ananyev
Thanks for your Ack, Konstantin.
I will use the macro I40E_MAX_QP_NUM_PER_VF instead of 16 in my v7 patch.
This macro is already defined as 16 in i40e_ethdev.h
>
> > + pf->vf_nb_qp_max = (uint16_t)num;
> > + else
> > + /* here return 0 to make next valid same argument work */
> > + PMD_DRV_LOG(WARNING, "Wrong VF queue number = %lu, it
> must be "
> > + "power of 2 and equal or less than 16 !, Now it is "
> > + "kept the value = %hu", num, pf->vf_nb_qp_max);
> > +
> > + return 0;
> > +}
[dpdk-dev] [PATCH v7] net/i40e: determine number of queues per VF during run time
Without this patch, the number of queues per i40e VF is defined as 4
by CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF=4 in config/common_base.
It is fixed value determined in building time and can't be changed
during run time.
With this patch, the number of queues per i40e VF can be determinated
during run time. For example, if the PCI address of an i40e PF is
:bb.cc, with the EAL parameter -w :bb.cc,queue-num-per-vf=8 ,
the number of queues per VF created from this PF is 8.
If there is no "queue-num-per-vf" setting in EAL parameters, it is 4
by default as before. And if the value after the "queue-num-per-vf"
is invalid, it is set as 4 forcibly. The valid values include 1, 2, 4,
8, 16 .
Signed-off-by: Wei Dai
Acked-by: Konstantin Ananyev
---
v7:
use the macro instead of natural number
correct git log message as the EAL parameter is only valid for PF
v6:
fix a small bug when detecting end character of strtoul
v5:
fix git log message and WARNING of coding stype
v4:
use rte_kvargs instead of pervious parsing function;
use malloc/free instead of rte_zmalloc/rte_free.
v3:
fix WARNING of coding style issues from checkpa...@dpdk.org
v2:
fix WARNING of coding style issues from checkpa...@dpdk.org
---
config/common_base | 1 -
drivers/net/i40e/i40e_ethdev.c | 67 --
2 files changed, 65 insertions(+), 3 deletions(-)
diff --git a/config/common_base b/config/common_base
index e74febe..4e20389 100644
--- a/config/common_base
+++ b/config/common_base
@@ -208,7 +208,6 @@ CONFIG_RTE_LIBRTE_I40E_RX_ALLOW_BULK_ALLOC=y
CONFIG_RTE_LIBRTE_I40E_INC_VECTOR=y
CONFIG_RTE_LIBRTE_I40E_16BYTE_RX_DESC=n
CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_PF=64
-CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF=4
CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VM=4
# interval up to 8160 us, aligned to 2 (or default value)
CONFIG_RTE_LIBRTE_I40E_ITR_INTERVAL=-1
diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
index 811cc9f..9295e1b 100644
--- a/drivers/net/i40e/i40e_ethdev.c
+++ b/drivers/net/i40e/i40e_ethdev.c
@@ -3971,6 +3971,67 @@ i40e_get_cap(struct i40e_hw *hw)
return ret;
}
+#define RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF 4
+#define QUEUE_NUM_PER_VF_ARG "queue-num-per-vf"
+static int i40e_pf_parse_vf_queue_number_handler(const char *key,
+ const char *value,
+ void *opaque)
+{
+ struct i40e_pf *pf;
+ unsigned long num;
+ char *end;
+
+ pf = (struct i40e_pf *)opaque;
+ RTE_SET_USED(key);
+
+ errno = 0;
+ num = strtoul(value, &end, 0);
+ if (errno != 0 || end == value || *end != 0) {
+ PMD_DRV_LOG(WARNING, "Wrong VF queue number = %s, Now it is "
+ "kept the value = %hu", value, pf->vf_nb_qp_max);
+ return -(EINVAL);
+ }
+
+ if (num <= I40E_MAX_QP_NUM_PER_VF && rte_is_power_of_2(num))
+ pf->vf_nb_qp_max = (uint16_t)num;
+ else
+ /* here return 0 to make next valid same argument work */
+ PMD_DRV_LOG(WARNING, "Wrong VF queue number = %lu, it must be "
+ "power of 2 and equal or less than 16 !, Now it is "
+ "kept the value = %hu", num, pf->vf_nb_qp_max);
+
+ return 0;
+}
+
+static int i40e_pf_config_vf_rxq_number(struct rte_eth_dev *dev)
+{
+ static const char * const valid_keys[] = {QUEUE_NUM_PER_VF_ARG, ""};
+ struct i40e_pf *pf = I40E_DEV_PRIVATE_TO_PF(dev->data->dev_private);
+ struct rte_kvargs *kvlist;
+
+ /* set default queue number per VF as 4 */
+ pf->vf_nb_qp_max = RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF;
+
+ if (dev->device->devargs == NULL)
+ return 0;
+
+ kvlist = rte_kvargs_parse(dev->device->devargs->args, valid_keys);
+ if (kvlist == NULL)
+ return -(EINVAL);
+
+ if (rte_kvargs_count(kvlist, QUEUE_NUM_PER_VF_ARG) > 1)
+ PMD_DRV_LOG(WARNING, "More than one argument \"%s\" and only "
+ "the first invalid or last valid one is used !",
+ QUEUE_NUM_PER_VF_ARG);
+
+ rte_kvargs_process(kvlist, QUEUE_NUM_PER_VF_ARG,
+ i40e_pf_parse_vf_queue_number_handler, pf);
+
+ rte_kvargs_free(kvlist);
+
+ return 0;
+}
+
static int
i40e_pf_parameter_init(struct rte_eth_dev *dev)
{
@@ -3983,6 +4044,9 @@ i40e_pf_parameter_init(struct rte_eth_dev *dev)
PMD_INIT_LOG(ERR, "HW configuration doesn't support SRIOV");
return -EINVAL;
}
+
+ i40e_pf_config_vf_rxq_number(dev);
+
/* Add the parameter init for LFC */
pf->fc_conf.pause_time = I40E_DEFAULT_PAUSE_TIME;
pf->fc_conf.high_water[I40E_MAX_TRAFFIC_CLASS] =
I40E_DEFAULT_HIGH_WATER;
@@ -3992,7 +4056,6 @@ i40e_pf_parameter_init(struct rte_eth_dev *dev)
pf->max_num_vsi = hw->f
Re: [dpdk-dev] [PATCH] vhost_user: protect active rings from async ring changes
> -Original Message- > From: Maxime Coquelin [mailto:maxime.coque...@redhat.com] > Sent: Thursday, December 7, 2017 6:02 PM > To: Tan, Jianfeng; Victor Kaplansky; dev@dpdk.org; y...@fridaylinux.org; Bie, > Tiwei > Cc: sta...@dpdk.org; jfrei...@redhat.com > Subject: Re: [PATCH] vhost_user: protect active rings from async ring > changes > > > > On 12/07/2017 10:33 AM, Tan, Jianfeng wrote: > > > > > >> -Original Message- > >> From: Victor Kaplansky [mailto:vkapl...@redhat.com] > >> Sent: Wednesday, December 6, 2017 9:56 PM > >> To: dev@dpdk.org; y...@fridaylinux.org; Bie, Tiwei; Tan, Jianfeng; > >> vkapl...@redhat.com > >> Cc: sta...@dpdk.org; jfrei...@redhat.com; Maxime Coquelin > >> Subject: [PATCH] vhost_user: protect active rings from async ring changes > >> > >> When performing live migration or memory hot-plugging, > >> the changes to the device and vrings made by message handler > >> done independently from vring usage by PMD threads. > >> > >> This causes for example segfauls during live-migration > > > > segfauls ->segfaults? > > > >> with MQ enable, but in general virtually any request > >> sent by qemu changing the state of device can cause > >> problems. > >> > >> These patches fixes all above issues by adding a spinlock > >> to every vring and requiring message handler to start operation > >> only after ensuring that all PMD threads related to the divece > > > > Another typo: divece. > > > >> are out of critical section accessing the vring data. > >> > >> Each vring has its own lock in order to not create contention > >> between PMD threads of different vrings and to prevent > >> performance degradation by scaling queue pair number. > > > > Also wonder how much overhead it brings. > > > > Instead of locking each vring, can we just, waiting a while (10us for > > example) > after call destroy_device() callback so that every PMD thread has enough > time to skip out the criterial area? > > No, because we are not destroying the device when it is needed. > Actually, once destroy_device() is called, it is likely that the > application has taken care the ring aren't being processed anymore > before returning from the callback (This is at least the case with Vhost > PMD). OK, I did not put it right way as there are multiple cases above: migration and memory hot plug. Let me try again: Whenever a vhost thread handles a message affecting PMD threads, (like SET_MEM_TABLE, GET_VRING_BASE, etc) we can remove the dev flag - VIRTIO_DEV_RUNNING, and wait for a while so that PMD threads skip out of those criterial area. After message handling, reset the flag - VIRTIO_DEV_RUNNING. I suppose it can work, basing on an assumption that PMD threads work in polling mode and can skip criterial area quickly and inevitably. > > The reason we need the lock is to protect PMD threads from the handling > of some vhost-user protocol requests. > For example SET_MEM_TABLE in the case of memory hotplug, or > SET_LOG_BASE > in case of multiqueue, which is sent for every queue pair and results in > unmapping/remapping the logging area. Yes, I understand how it fails. Thanks, Jianfeng > > Maxime
Re: [dpdk-dev] [PATCH v2 06/10] net/virtio: fix queue setup consistency
Hi Olivier,
On Thu, Dec 07, 2017 at 03:14:44PM +0100, Olivier MATZ wrote:
> On Wed, Dec 06, 2017 at 01:25:29PM +0800, Tiwei Bie wrote:
> > Hi Maxime and Olivier:
> >
> > On Thu, Sep 07, 2017 at 02:13:43PM +0200, Olivier Matz wrote:
> > [...]
> > > diff --git a/drivers/net/virtio/virtio_ethdev.c
> > > b/drivers/net/virtio/virtio_ethdev.c
> > > index 8eee3ff80..c7888f103 100644
> > > --- a/drivers/net/virtio/virtio_ethdev.c
> > > +++ b/drivers/net/virtio/virtio_ethdev.c
> > > @@ -1737,6 +1737,19 @@ virtio_dev_start(struct rte_eth_dev *dev)
> > > struct virtnet_rx *rxvq;
> > > struct virtnet_tx *txvq __rte_unused;
> > > struct virtio_hw *hw = dev->data->dev_private;
> > > + int ret;
> > > +
> > > + /* Finish the initialization of the queues */
> > > + for (i = 0; i < dev->data->nb_rx_queues; i++) {
> > > + ret = virtio_dev_rx_queue_setup_finish(dev, i);
> > > + if (ret < 0)
> > > + return ret;
> > > + }
> >
> > I'm trying to fix an issue [1] reported by Antonio. And during
> > the debugging, I found that vector Rx of virtio PMD has been
> > broken (when doing port stop/start) since below two patches were
> > applied:
> >
> > 25bf7a0b0936 ("vhost: make error handling consistent in Rx path")
> >-- needed on the Tx side (testpmd/vhost-pmd in below test)
> > efc83a1e7fc3 ("net/virtio: fix queue setup consistency")
> >-- needed on the Rx side (testpmd/virtio-user in below test)
>
> Just to be sure I understand properly: each of these 2 patches
> break a different part your test case?
>
Thank you for looking into this! ;-)
I mean the above test case won't pass when we have both
of them applied. And the first patch changes the Tx side,
and the second one changes the Rx side.
I haven't done thorough analysis on the first patch, so
I'm not sure what would be affected in the non-mergeable
Rx and vector Rx of virtio-PMD after changing the error
handling in vhost.
But I think there is something wrong with this patch (i.e.
the second patch). From my understanding, it seems that
virtio_rxq_rearm_vec() has an assumption that each time
it's called, the starting 'desc_idx' should be multiple
times of RTE_VIRTIO_VPMD_RX_REARM_THRESH (or 0). After
introducing virtio_dev_rx_queue_setup_finish() in device
start, the rxq will be fully refilled no matter where
the 'desc_idx' is after a device stop/start. And it could
break such assumption.
> I tried to reproduce your test case (the working case first):
> - on 0c4f909c17 (the commit before the efc83a1e7fc3)
> - without the patch disabling mergeable Rx
>
> No packet is received. Am I doing something wrong? Please see the
> log:
>
>cd /root/dpdk.org
>git checkout -b test 0c4f909c17
>rm -rf build && make config T=x86_64-native-linuxapp-gcc && make -j32
>insmod build/kmod/igb_uio.ko
>echo 1000 >
> /sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages
>echo 1000 >
> /sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages
Sorry, I forgot to mention that, 1G hugepage is required
to use virtio-user (2M hugepage won't work). For more
details about it, you could refer to the "Limitations"
section in below doc:
http://dpdk.org/doc/guides/howto/virtio_user_for_container_networking.html#limitations
Best regards,
Tiwei Bie
Re: [dpdk-dev] [PATCH v3 2/2] net/ixgbe: move RSS to flow API
Hi, daiwei
I build the code with ICC for second time, it can pass build and there
is no build error.
It seems the icc report is not credible.
> -Original Message-
> From: Dai, Wei
> Sent: Thursday, December 7, 2017 5:20 PM
> To: Zhao1, Wei ; dev@dpdk.org
> Cc: Zhao1, Wei
> Subject: RE: [dpdk-dev] [PATCH v3 2/2] net/ixgbe: move RSS to flow API
>
> Hi, Zhao Wei
> Please correct build error show in http://dpdk.org/ml/archives/test-
> report/2017-November/035130.html
>
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Wei Zhao
> > Sent: Friday, November 24, 2017 4:05 PM
> > To: dev@dpdk.org
> > Cc: Zhao1, Wei
> > Subject: [dpdk-dev] [PATCH v3 2/2] net/ixgbe: move RSS to flow API
> >
> > Rte_flow actually defined to include RSS, but till now, RSS is out of
> > rte_flow.
> > This patch is to move ixgbe existing RSS to rte_flow.
> >
> > Signed-off-by: Wei Zhao
> > ---
> > drivers/net/ixgbe/ixgbe_ethdev.c | 13 +++
> > drivers/net/ixgbe/ixgbe_ethdev.h | 10 +++
> > drivers/net/ixgbe/ixgbe_flow.c | 165
> > +++
> > drivers/net/ixgbe/ixgbe_rxtx.c | 65 +++
> > 4 files changed, 253 insertions(+)
> >
> > diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c
> > b/drivers/net/ixgbe/ixgbe_ethdev.c
> > index ff19a56..4960650 100644
> > --- a/drivers/net/ixgbe/ixgbe_ethdev.c
> > +++ b/drivers/net/ixgbe/ixgbe_ethdev.c
> > @@ -8339,6 +8339,18 @@ ixgbe_l2_tn_filter_restore(struct rte_eth_dev
> > *dev)
> > }
> > }
> >
> > +/* restore rss filter */
> > +static inline void
> > +ixgbe_rss_filter_restore(struct rte_eth_dev *dev) {
> > + struct ixgbe_filter_info *filter_info =
> > + IXGBE_DEV_PRIVATE_TO_FILTER_INFO(dev->data-
> >dev_private);
> > +
> > + if (filter_info->rss_info.num)
> > + ixgbe_config_rss_filter(dev,
> > + &filter_info->rss_info, TRUE);
> > +}
> > +
> > static int
> > ixgbe_filter_restore(struct rte_eth_dev *dev) { @@ -8347,6 +8359,7
> > @@ ixgbe_filter_restore(struct rte_eth_dev *dev)
> > ixgbe_syn_filter_restore(dev);
> > ixgbe_fdir_filter_restore(dev);
> > ixgbe_l2_tn_filter_restore(dev);
> > + ixgbe_rss_filter_restore(dev);
> >
> > return 0;
> > }
> > diff --git a/drivers/net/ixgbe/ixgbe_ethdev.h
> > b/drivers/net/ixgbe/ixgbe_ethdev.h
> > index 51ddcfd..4af79b4 100644
> > --- a/drivers/net/ixgbe/ixgbe_ethdev.h
> > +++ b/drivers/net/ixgbe/ixgbe_ethdev.h
> > @@ -224,6 +224,12 @@ struct ixgbe_hw_fdir_info {
> > bool mask_added; /* If already got mask from consistent filter */
> > };
> >
> > +struct ixgbe_rte_flow_rss_conf {
> > + struct rte_eth_rss_conf rss_conf; /**< RSS parameters. */
> > + uint16_t num; /**< Number of entries in queue[]. */
> > + uint16_t queue[IXGBE_MAX_RX_QUEUE_NUM]; /**< Queues
> indices to
> > use. */
> > +};
> > +
> > /* structure for interrupt relative data */ struct ixgbe_interrupt {
> > uint32_t flags;
> > @@ -340,6 +346,8 @@ struct ixgbe_filter_info {
> > struct ixgbe_5tuple_filter_list fivetuple_list;
> > /* store the SYN filter info */
> > uint32_t syn_info;
> > + /* store the rss filter info */
> > + struct ixgbe_rte_flow_rss_conf rss_info;
> > };
> >
> > struct ixgbe_l2_tn_key {
> > @@ -719,6 +727,8 @@ void ixgbe_tm_conf_init(struct rte_eth_dev *dev);
> > void ixgbe_tm_conf_uninit(struct rte_eth_dev *dev); int
> > ixgbe_set_queue_rate_limit(struct rte_eth_dev *dev, uint16_t queue_idx,
> >uint16_t tx_rate);
> > +int ixgbe_config_rss_filter(struct rte_eth_dev *dev,
> > + struct ixgbe_rte_flow_rss_conf *conf, bool add);
> >
> > static inline int
> > ixgbe_ethertype_filter_lookup(struct ixgbe_filter_info *filter_info,
> > diff --git a/drivers/net/ixgbe/ixgbe_flow.c
> > b/drivers/net/ixgbe/ixgbe_flow.c index 19c2d47..8f964cf 100644
> > --- a/drivers/net/ixgbe/ixgbe_flow.c
> > +++ b/drivers/net/ixgbe/ixgbe_flow.c
> > @@ -103,6 +103,11 @@ struct ixgbe_eth_l2_tunnel_conf_ele {
> > TAILQ_ENTRY(ixgbe_eth_l2_tunnel_conf_ele) entries;
> > struct rte_eth_l2_tunnel_conf filter_info; };
> > +/* rss filter list structure */
> > +struct ixgbe_rss_conf_ele {
> > + TAILQ_ENTRY(ixgbe_rss_conf_ele) entries;
> > + struct ixgbe_rte_flow_rss_conf filter_info; };
> > /* ixgbe_flow memory list structure */ struct ixgbe_flow_mem {
> > TAILQ_ENTRY(ixgbe_flow_mem) entries; @@ -114,6 +119,7 @@
> > TAILQ_HEAD(ixgbe_ethertype_filter_list,
> > ixgbe_ethertype_filter_ele); TAILQ_HEAD(ixgbe_syn_filter_list,
> > ixgbe_eth_syn_filter_ele); TAILQ_HEAD(ixgbe_fdir_rule_filter_list,
> > ixgbe_fdir_rule_ele); TAILQ_HEAD(ixgbe_l2_tunnel_filter_list,
> > ixgbe_eth_l2_tunnel_conf_ele);
> > +TAILQ_HEAD(ixgbe_rss_filter_list, ixgbe_rss_conf_ele);
> > TAILQ_HEAD(ixgbe_flow_mem_list, ixgbe_flow_mem);
> >
> > static struct ixgbe_ntuple_filter_list filter_ntuple_list; @@ -121,6
> > +127,7 @@ static struct ixgbe_ethertype_filter_list
> > filter_ethertype_li
Re: [dpdk-dev] [PATCH v3 1/2] net/e1000: move RSS to flow API
Hi, daiwei
I build the code with ICC for second time just now, it can pass build
and there is no build error.
It seems the icc report is not credible.
> -Original Message-
> From: Dai, Wei
> Sent: Thursday, December 7, 2017 5:19 PM
> To: Zhao1, Wei ; dev@dpdk.org
> Cc: Zhao1, Wei
> Subject: RE: [dpdk-dev] [PATCH v3 1/2] net/e1000: move RSS to flow API
>
> Hi, Zhao Wei
> Please correct build error show in http://dpdk.org/ml/archives/test-
> report/2017-November/035129.html
>
>
> > -Original Message-
> > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Wei Zhao
> > Sent: Friday, November 24, 2017 4:05 PM
> > To: dev@dpdk.org
> > Cc: Zhao1, Wei
> > Subject: [dpdk-dev] [PATCH v3 1/2] net/e1000: move RSS to flow API
> >
> > Rte_flow actually defined to include RSS, but till now, RSS is out of
> > rte_flow.
> > This patch is to move igb existing RSS to rte_flow.
> >
> > Signed-off-by: Wei Zhao
> > ---
> > drivers/net/e1000/e1000_ethdev.h | 20 +
> > drivers/net/e1000/igb_ethdev.c | 17 +
> > drivers/net/e1000/igb_flow.c | 160
> > +++
> > drivers/net/e1000/igb_rxtx.c | 61 +++
> > 4 files changed, 258 insertions(+)
> >
> > diff --git a/drivers/net/e1000/e1000_ethdev.h
> > b/drivers/net/e1000/e1000_ethdev.h
> > index 5668910..0731766 100644
> > --- a/drivers/net/e1000/e1000_ethdev.h
> > +++ b/drivers/net/e1000/e1000_ethdev.h
> > @@ -257,6 +257,12 @@ struct igb_ethertype_filter {
> > uint32_t etqf;
> > };
> >
> > +struct igb_rte_flow_rss_conf {
> > + struct rte_eth_rss_conf rss_conf; /**< RSS parameters. */
> > + uint16_t num; /**< Number of entries in queue[]. */
> > + uint16_t queue[IGB_MAX_RX_QUEUE_NUM]; /**< Queues indices
> to
> > use. */
> > +};
> > +
> > /*
> > * Structure to store filters'info.
> > */
> > @@ -274,6 +280,8 @@ struct e1000_filter_info {
> > struct e1000_2tuple_filter_list twotuple_list;
> > /* store the SYN filter info */
> > uint32_t syn_info;
> > + /* store the rss filter info */
> > + struct igb_rte_flow_rss_conf rss_info;
> > };
> >
> > /*
> > @@ -342,6 +350,12 @@ struct igb_flex_filter_ele {
> > struct rte_eth_flex_filter filter_info; };
> >
> > +/* rss filter list structure */
> > +struct igb_rss_conf_ele {
> > + TAILQ_ENTRY(igb_rss_conf_ele) entries;
> > + struct igb_rte_flow_rss_conf filter_info; };
> > +
> > /* igb_flow memory list structure */
> > struct igb_flow_mem {
> > TAILQ_ENTRY(igb_flow_mem) entries;
> > @@ -357,6 +371,8 @@ TAILQ_HEAD(igb_syn_filter_list,
> > igb_eth_syn_filter_ele); struct igb_syn_filter_list
> > igb_filter_syn_list; TAILQ_HEAD(igb_flex_filter_list,
> > igb_flex_filter_ele); struct igb_flex_filter_list
> > igb_filter_flex_list;
> > +TAILQ_HEAD(igb_rss_filter_list, igb_rss_conf_ele); struct
> > +igb_rss_filter_list igb_filter_rss_list;
> > TAILQ_HEAD(igb_flow_mem_list, igb_flow_mem); struct
> > igb_flow_mem_list igb_flow_list;
> >
> > @@ -500,4 +516,8 @@ int eth_igb_syn_filter_set(struct rte_eth_dev
> > *dev, int eth_igb_add_del_flex_filter(struct rte_eth_dev *dev,
> > struct rte_eth_flex_filter *filter,
> > bool add);
> > +int igb_config_rss_filter(struct rte_eth_dev *dev,
> > + struct igb_rte_flow_rss_conf *conf,
> > + bool add);
> > +
> > #endif /* _E1000_ETHDEV_H_ */
> > diff --git a/drivers/net/e1000/igb_ethdev.c
> > b/drivers/net/e1000/igb_ethdev.c index fdc139f..275fa02 100644
> > --- a/drivers/net/e1000/igb_ethdev.c
> > +++ b/drivers/net/e1000/igb_ethdev.c
> > @@ -948,6 +948,7 @@ eth_igb_dev_init(struct rte_eth_dev *eth_dev)
> > TAILQ_INIT(&igb_filter_ethertype_list);
> > TAILQ_INIT(&igb_filter_syn_list);
> > TAILQ_INIT(&igb_filter_flex_list);
> > + TAILQ_INIT(&igb_filter_rss_list);
> > TAILQ_INIT(&igb_flow_list);
> >
> > return 0;
> > @@ -1007,6 +1008,10 @@ eth_igb_dev_uninit(struct rte_eth_dev
> *eth_dev)
> > memset(filter_info->ethertype_filters, 0,
> > E1000_MAX_ETQF_FILTERS * sizeof(struct
> igb_ethertype_filter));
> >
> > + /* clear the rss filter info */
> > + memset(&filter_info->rss_info, 0,
> > + sizeof(struct igb_rte_flow_rss_conf));
> > +
> > /* remove all ntuple filters of the device */
> > igb_ntuple_filter_uninit(eth_dev);
> >
> > @@ -5628,6 +5633,17 @@ igb_flex_filter_restore(struct rte_eth_dev *dev)
> > }
> > }
> >
> > +/* restore rss filter */
> > +static inline void
> > +igb_rss_filter_restore(struct rte_eth_dev *dev) {
> > + struct e1000_filter_info *filter_info =
> > + E1000_DEV_PRIVATE_TO_FILTER_INFO(dev->data-
> >dev_private);
> > +
> > + if (filter_info->rss_info.num)
> > + igb_config_rss_filter(dev, &filter_info->rss_info, TRUE); }
> > +
> > /* restore all types filter */
> > static int
> > igb_filter_restore(struct rte_eth_dev *dev) @@ -5636,6 +5652,7 @@
> > igb_filter_restore(struct rte_eth_dev *de
Re: [dpdk-dev] [PATCH 1/5] net/virtio: fix vector Rx break caused by rxq flushing
On Thu, Dec 07, 2017 at 04:00:57PM +, Kevin Traynor wrote: > On 12/07/2017 05:30 AM, Tiwei Bie wrote: > > The vector Rx will be broken if backend has consumed all > > the descs in the avail ring before the device is started. > > Because in current implementation, vector Rx will return > > immediately without refilling the avail ring if the used > > ring is empty. So we have to refill the avail ring after > > flushing the elements in the used ring for vector Rx. > > > > Besides, vector Rx has a different ring layout assumption > > and mbuf management. So we need to handle it differently. > > > > Hi Tiwei, does the issue only occur with the vector rx? How about if the > simple rx is used because VIRTIO_NET_F_MRG_RXBUF is set? Hi Kevin, Yes, you are right! The issue only occurs with the vector Rx. The vector Rx (i.e. the simple Rx) won't be used if VIRTIO_NET_F_MRG_RXBUF is set. Best regards, Tiwei Bie
Re: [dpdk-dev] [PATCH v2 1/2] Introducing SPDX License Identifiers
On 12/8/2017 5:16 AM, Ferruh Yigit wrote: On 12/1/2017 12:38 AM, Hemant Agrawal wrote: The DPDK uses the Open Source BSD-3-Clause license for the core libraries and drivers. The kernel components are naturally GPLv2 licensed. Many of the files in the DPDK source code contain the full text of the applicable license. For example, most of the BSD-3-Clause files contain a full copy of the BSD-3-Clause license text. Including big blocks of License headers in all files blows up the source code with mostly redundant information. An additional problem is that even the same licenses are referred to by a number of slightly varying text blocks (full, abbreviated, different indentation, line wrapping and/or white space, with obsolete address information, ...) which makes validation and automatic processing a nightmare. To make this easier, DPDK is adpoting the use of a single line reference to Unique License Identifiers in source files as defined by the Linux Foundation's SPDX project [1]. Adding license information in this fashion, rather than adding full license text, can be more efficient for developers; decreases errors; and improves automated detection of licenses. The current set of valid, predefined SPDX identifiers is set forth on the SPDX License List[2] at https://spdx.org/licenses/. For example, to label a file as subject to the BSD-3-Clause license, the following text would be used: Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER SPDX-License-Identifier:BSD-3-Clause To label a file as GPL-2.0 (e.g., for code that runs in the kernel), the following text would be used: Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER SPDX-License-Identifier:GPL-2.0 To label a file as dual-licensed with BSD-3-Clause and GPL-2.0 (e.g., for code that is shared between the kernel and userspace), the following text would be used: Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER SPDX-License-Identifier:BSD-3-Clause OR GPL-2.0 To label a file as dual-licensed with BSD-3-Clause and LGPL-2.1 (e.g., for code that is shared between the kernel and userspace), the following text would be used: Copyright (C) [YEAR] NAME-OF-COPYRIGHT-HOLDER SPDX-License-Identifier:BSD-3-Clause OR LGPL-2.1 Note: Any new file contributions in DPDK shall adhere to the above scheme. It is also being recommended to replace the existing license text in the code with SPDX-License-Identifiers. Note 2: DPDK currently adhere to it's IP policies[3]. Any exception to this shall be approved by DPDK tech board and DPDK Governing Board. Steps for any exception approval: 1. Mention the appropriate license identifier form SPDX. If the license is not listed in SPDX Licenses. It is the submitters responsibiliity to get it first listed. 2. Get the required approval from the DPDK Technical Board. Technical board may advise the author to check alternate means first. If no other alternatives are found and the merit of the contributions are important for DPDK's mission, it may decide on such exception with two-thirds vote of the members. 3. Technical board then approach Governing board for such limited approval for the given contribution only. Any approvals shall be documented in "Licenses/exceptions.txt" with record dates. Note 3: Projects like U-boot have been been using SPDX License Idenfiers successfully [2]. They have been referered in implementing SPDX based guidelines in DPDK. Note 4: From the legal point of view, this patch is supposed to be only a change to the textual representation of the license information, but in no way any change to the actual license terms. With this patch applied, all files will still be licensed under the same terms they were before. Signed-off-by: Hemant Agrawal Acked-by: Stephen Hemminger --- LICENSE.GPL | 339 LICENSE.LGPL| 502 Licenses/Exceptions.txt | 12 + Licenses/README | 82 ++ Licenses/bsd-3-clause.txt | 9 + Licenses/gpl-2.0.txt| 339 Licenses/lgpl-2.1.txt | 502 Hi Hemant, Are new gpl-2.0.txt & lgpl-2.1.txt files identical with old LICENSE.GPL & LICENSE.LGPL? If so, does it make sense to make commit with "git mv" so that patch won't contain all text resulting smaller patch, and it will highlight that nothing changed but moved/renamed? Thanks, ferruh <> Yes. I missed to notice that. I will fix it.
Re: [dpdk-dev] [PATCH 3/6] doc: add dynamic logging to PMD todo list
On 11/21/2017 7:12 AM, Ferruh Yigit wrote:
To track modification:
c1b5fa94a46f ("eal: support dynamic log types")
Proposed deadline for PMDs is v18.08
Signed-off-by: Ferruh Yigit
---
doc/guides/nics/todo.rst | 11 +++
1 file changed, 11 insertions(+)
diff --git a/doc/guides/nics/todo.rst b/doc/guides/nics/todo.rst
index b408fc691..973379a3d 100644
--- a/doc/guides/nics/todo.rst
+++ b/doc/guides/nics/todo.rst
@@ -50,3 +50,14 @@ This is the list for tracking required PMD changes triggered
by library modifica
| | | thunderx, vhost, | |
| |
| | | vmxnet3 | |
| |
+---++--++---+
+ | dynamic logging | | af_packet, ark, avp, bnx2x, | v18.08 |
c1b5fa94a46f | Switch to dynamic logging |
+ | | | bnxt, bonding, cxgbe, dpaa, | |
| functions, remove static debug|
+ | | | dpaa2, e1000, ena, enic, | |
| config options. |
+ | | | failsafe, fm10k, ixgbe, | |
| |
+ | | | kni, liquidio, mlx4, mlx5, | |
| |
+ | | | mrvl, nfp, null, octeontx, | |
| |
+ | | | pcap, qede, ring, sfc, | |
| |
+ | | | softnic, szedata2, tap, | |
| |
+ | | | thunderx, vhost, virtio, | |
| |
+ | | | vmxnet3 | |
| |
+
+---++--++---+
One small correction - dpaa is already supporting dynamic logging.
It is a good idea to have todo item documented.
Series-acked-by: Hemant Agrawal
[dpdk-dev] [PATCH 00/14] DPAA2 PMD fixes and enhancements
This patch set contains some basic fixes, enhancements and optimzations for DPAA2 driver. Ashish Jain (1): net/dpaa2: align the frame size in MTU set Hemant Agrawal (8): bus/fslmc: fix the cplusplus macro closure drivers: change the deprecated memseg physaddr to iova bus/fslmc: add support for dynamic iova for DPAA2 devices net/dpaa2: link status check as driver flag bus/fslmc: expose platform soc value register bus/fslmc: add qman HW fq query count API net/dpaa2: add Rx queue count support net/dpaa2: add VLAN insert offload Nipun Gupta (5): bus/fslmc: add braces for pointers in macros net/dpaa2: add parse function for LX2 device net/dpaa2: optimize Rx path packet parsing net/dpaa2: optimize Tx path for best case net/dpaa2: prefetch the parse results from next fd drivers/bus/dpaa/rte_dpaa_bus.h | 6 +- drivers/bus/fslmc/Makefile| 3 +- drivers/bus/fslmc/fslmc_bus.c | 44 + drivers/bus/fslmc/fslmc_vfio.c| 5 +- drivers/bus/fslmc/portal/dpaa2_hw_dpio.c | 42 ++--- drivers/bus/fslmc/portal/dpaa2_hw_dpio.h | 3 + drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 74 +--- drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h | 30 drivers/bus/fslmc/qbman/qbman_debug.c | 66 +++ drivers/bus/fslmc/rte_bus_fslmc_version.map | 10 ++ drivers/bus/fslmc/rte_fslmc.h | 15 +- drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c | 1 + drivers/crypto/dpaa_sec/dpaa_sec.c| 8 +- drivers/net/dpaa2/base/dpaa2_hw_dpni_annot.h | 20 +++ drivers/net/dpaa2/dpaa2_ethdev.c | 66 +-- drivers/net/dpaa2/dpaa2_ethdev.h | 27 +++ drivers/net/dpaa2/dpaa2_rxtx.c| 207 +++--- 17 files changed, 524 insertions(+), 103 deletions(-) create mode 100644 drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h create mode 100644 drivers/bus/fslmc/qbman/qbman_debug.c -- 2.7.4
[dpdk-dev] [PATCH 02/14] drivers: change the deprecated memseg physaddr to iova
DPAA and DPAA2 drivers were using memseg physaddr, which
has been deprecated.
Fixes: 7ba49d39f14c ("mem: rename segment address from physical to IOVA")
Cc: Santosh Shukla
Cc: sta...@dpdk.org
Signed-off-by: Hemant Agrawal
---
drivers/bus/dpaa/rte_dpaa_bus.h | 6 +++---
drivers/bus/fslmc/fslmc_vfio.c | 2 +-
drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 8
drivers/crypto/dpaa_sec/dpaa_sec.c | 8
4 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/drivers/bus/dpaa/rte_dpaa_bus.h b/drivers/bus/dpaa/rte_dpaa_bus.h
index eafc944..bc933af 100644
--- a/drivers/bus/dpaa/rte_dpaa_bus.h
+++ b/drivers/bus/dpaa/rte_dpaa_bus.h
@@ -113,10 +113,10 @@ static inline void *rte_dpaa_mem_ptov(phys_addr_t paddr)
int i;
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr != NULL; i++) {
- if (paddr >= memseg[i].phys_addr && paddr <
- memseg[i].phys_addr + memseg[i].len)
+ if (paddr >= memseg[i].iova && paddr <
+ memseg[i].iova + memseg[i].len)
return (uint8_t *)(memseg[i].addr) +
- (paddr - memseg[i].phys_addr);
+ (paddr - memseg[i].iova);
}
return NULL;
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 25c87ad..e47215c 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -249,7 +249,7 @@ int rte_fslmc_vfio_dmamap(void)
dma_map.size = memseg[i].len;
dma_map.vaddr = memseg[i].addr_64;
#ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
- dma_map.iova = memseg[i].phys_addr;
+ dma_map.iova = memseg[i].iova;
#else
dma_map.iova = dma_map.vaddr;
#endif
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index c1b842f..ece1a7d 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -284,10 +284,10 @@ static void *dpaa2_mem_ptov(phys_addr_t paddr)
int i;
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
- if (paddr >= memseg[i].phys_addr &&
- (char *)paddr < (char *)memseg[i].phys_addr + memseg[i].len)
+ if (paddr >= memseg[i].iova &&
+ (char *)paddr < (char *)memseg[i].iova + memseg[i].len)
return (void *)(memseg[i].addr_64
- + (paddr - memseg[i].phys_addr));
+ + (paddr - memseg[i].iova));
}
return NULL;
}
@@ -301,7 +301,7 @@ static phys_addr_t dpaa2_mem_vtop(uint64_t vaddr)
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
if (vaddr >= memseg[i].addr_64 &&
vaddr < memseg[i].addr_64 + memseg[i].len)
- return memseg[i].phys_addr
+ return memseg[i].iova
+ (vaddr - memseg[i].addr_64);
}
return (phys_addr_t)(NULL);
diff --git a/drivers/crypto/dpaa_sec/dpaa_sec.c
b/drivers/crypto/dpaa_sec/dpaa_sec.c
index 16155b1..438dd3b 100644
--- a/drivers/crypto/dpaa_sec/dpaa_sec.c
+++ b/drivers/crypto/dpaa_sec/dpaa_sec.c
@@ -121,7 +121,7 @@ dpaa_mem_vtop(void *vaddr)
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
if (vaddr_64 >= memseg[i].addr_64 &&
vaddr_64 < memseg[i].addr_64 + memseg[i].len) {
- paddr = memseg[i].phys_addr +
+ paddr = memseg[i].iova +
(vaddr_64 - memseg[i].addr_64);
return (rte_iova_t)paddr;
@@ -137,10 +137,10 @@ dpaa_mem_ptov(rte_iova_t paddr)
int i;
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
- if (paddr >= memseg[i].phys_addr &&
- (char *)paddr < (char *)memseg[i].phys_addr + memseg[i].len)
+ if (paddr >= memseg[i].iova &&
+ (char *)paddr < (char *)memseg[i].iova + memseg[i].len)
return (void *)(memseg[i].addr_64 +
- (paddr - memseg[i].phys_addr));
+ (paddr - memseg[i].iova));
}
return NULL;
}
--
2.7.4
[dpdk-dev] [PATCH 04/14] net/dpaa2: link status check as driver flag
Signed-off-by: Hemant Agrawal
DPDK-809
---
drivers/bus/fslmc/rte_fslmc.h| 4
drivers/net/dpaa2/dpaa2_ethdev.c | 6 --
2 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/drivers/bus/fslmc/rte_fslmc.h b/drivers/bus/fslmc/rte_fslmc.h
index 0c7872d..fd52e2b 100644
--- a/drivers/bus/fslmc/rte_fslmc.h
+++ b/drivers/bus/fslmc/rte_fslmc.h
@@ -62,6 +62,10 @@ extern "C" {
#define FSLMC_OBJECT_MAX_LEN 32 /**< Length of each device on bus */
+
+/** Device driver supports link state interrupt */
+#define RTE_DPAA2_DRV_INTR_LSC 0x0008
+
/** Device driver supports IOVA as VA */
#define RTE_DPAA2_DRV_IOVA_AS_VA 0X0040
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 38de3d9..1cd302d 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -1872,7 +1872,6 @@ dpaa2_dev_init(struct rte_eth_dev *eth_dev)
}
eth_dev->dev_ops = &dpaa2_ethdev_ops;
- eth_dev->data->dev_flags |= RTE_ETH_DEV_INTR_LSC;
eth_dev->rx_pkt_burst = dpaa2_dev_prefetch_rx;
eth_dev->tx_pkt_burst = dpaa2_dev_tx;
@@ -1976,6 +1975,9 @@ rte_dpaa2_probe(struct rte_dpaa2_driver *dpaa2_drv,
dpaa2_dev->eth_dev = eth_dev;
eth_dev->data->rx_mbuf_alloc_failed = 0;
+ if (dpaa2_drv->drv_flags & RTE_DPAA2_DRV_INTR_LSC)
+ eth_dev->data->dev_flags |= RTE_ETH_DEV_INTR_LSC;
+
/* Invoke PMD device initialization function */
diag = dpaa2_dev_init(eth_dev);
if (diag == 0)
@@ -2003,7 +2005,7 @@ rte_dpaa2_remove(struct rte_dpaa2_device *dpaa2_dev)
}
static struct rte_dpaa2_driver rte_dpaa2_pmd = {
- .drv_flags = RTE_DPAA2_DRV_IOVA_AS_VA,
+ .drv_flags = RTE_DPAA2_DRV_INTR_LSC | RTE_DPAA2_DRV_IOVA_AS_VA,
.drv_type = DPAA2_ETH,
.probe = rte_dpaa2_probe,
.remove = rte_dpaa2_remove,
--
2.7.4
[dpdk-dev] [PATCH 01/14] bus/fslmc: fix the cplusplus macro closure
Fixes: 10f1614f36a6 ("drivers: refactor DPAA2 object definition")
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/rte_fslmc.h | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/bus/fslmc/rte_fslmc.h b/drivers/bus/fslmc/rte_fslmc.h
index 4c32db6..0814e69 100644
--- a/drivers/bus/fslmc/rte_fslmc.h
+++ b/drivers/bus/fslmc/rte_fslmc.h
@@ -175,10 +175,6 @@ static void dpaa2initfn_ ##nm(void) \
} \
RTE_PMD_EXPORT_NAME(nm, __COUNTER__)
-#ifdef __cplusplus
-}
-#endif
-
/**
* Register a DPAA2 MC Object driver.
*
@@ -198,4 +194,8 @@ static void dpaa2objinitfn_ ##nm(void) \
} \
RTE_PMD_EXPORT_NAME(nm, __COUNTER__)
+#ifdef __cplusplus
+}
+#endif
+
#endif /* _RTE_FSLMC_H_ */
--
2.7.4
[dpdk-dev] [PATCH 07/14] bus/fslmc: add qman HW fq query count API
This patch add support for rx query debug API.
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/Makefile| 3 +-
drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h | 30 +++
drivers/bus/fslmc/qbman/qbman_debug.c | 66 +++
drivers/bus/fslmc/rte_bus_fslmc_version.map | 2 +
4 files changed, 100 insertions(+), 1 deletion(-)
create mode 100644 drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h
create mode 100644 drivers/bus/fslmc/qbman/qbman_debug.c
diff --git a/drivers/bus/fslmc/Makefile b/drivers/bus/fslmc/Makefile
index c08b2af..7ab39cb 100644
--- a/drivers/bus/fslmc/Makefile
+++ b/drivers/bus/fslmc/Makefile
@@ -62,7 +62,8 @@ EXPORT_MAP := rte_bus_fslmc_version.map
LIBABIVER := 1
SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
-qbman/qbman_portal.c
+qbman/qbman_portal.c \
+qbman/qbman_debug.c
SRCS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += \
mc/dpmng.c \
diff --git a/drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h
b/drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h
new file mode 100644
index 000..072ad55
--- /dev/null
+++ b/drivers/bus/fslmc/qbman/include/fsl_qbman_debug.h
@@ -0,0 +1,30 @@
+/* Copyright (C) 2015 Freescale Semiconductor, Inc.
+ *
+ * SPDX-License-Identifier:BSD-3-Clause
+ */
+struct qbman_swp;
+
+struct qbman_fq_query_np_rslt {
+uint8_t verb;
+ uint8_t rslt;
+ uint8_t st1;
+ uint8_t st2;
+ uint8_t reserved[2];
+ uint16_t od1_sfdr;
+ uint16_t od2_sfdr;
+ uint16_t od3_sfdr;
+ uint16_t ra1_sfdr;
+ uint16_t ra2_sfdr;
+ uint32_t pfdr_hptr;
+ uint32_t pfdr_tptr;
+ uint32_t frm_cnt;
+ uint32_t byte_cnt;
+ uint16_t ics_surp;
+ uint8_t is;
+ uint8_t reserved2[29];
+};
+
+int qbman_fq_query_state(struct qbman_swp *s, uint32_t fqid,
+struct qbman_fq_query_np_rslt *r);
+uint32_t qbman_fq_state_frame_count(const struct qbman_fq_query_np_rslt *r);
+uint32_t qbman_fq_state_byte_count(const struct qbman_fq_query_np_rslt *r);
diff --git a/drivers/bus/fslmc/qbman/qbman_debug.c
b/drivers/bus/fslmc/qbman/qbman_debug.c
new file mode 100644
index 000..591673a
--- /dev/null
+++ b/drivers/bus/fslmc/qbman/qbman_debug.c
@@ -0,0 +1,66 @@
+/* Copyright (C) 2015 Freescale Semiconductor, Inc.
+ *
+ * SPDX-License-Identifier:BSD-3-Clause
+ */
+
+#include "compat.h"
+#include
+#include "qbman_portal.h"
+
+/* QBMan portal management command code */
+#define QBMAN_BP_QUERY0x32
+#define QBMAN_FQ_QUERY0x44
+#define QBMAN_FQ_QUERY_NP 0x45
+#define QBMAN_WQ_QUERY0x47
+#define QBMAN_CGR_QUERY 0x51
+#define QBMAN_WRED_QUERY 0x54
+#define QBMAN_CGR_STAT_QUERY 0x55
+#define QBMAN_CGR_STAT_QUERY_CLR 0x56
+
+struct qbman_fq_query_desc {
+ uint8_t verb;
+ uint8_t reserved[3];
+ uint32_t fqid;
+ uint8_t reserved2[57];
+};
+
+int qbman_fq_query_state(struct qbman_swp *s, uint32_t fqid,
+struct qbman_fq_query_np_rslt *r)
+{
+ struct qbman_fq_query_desc *p;
+
+ p = (struct qbman_fq_query_desc *)qbman_swp_mc_start(s);
+ if (!p)
+ return -EBUSY;
+
+ p->fqid = fqid;
+ *r = *(struct qbman_fq_query_np_rslt *)qbman_swp_mc_complete(s, p,
+ QBMAN_FQ_QUERY_NP);
+ if (!r) {
+ pr_err("qbman: Query FQID %d NP fields failed, no response\n",
+ fqid);
+ return -EIO;
+ }
+
+ /* Decode the outcome */
+ QBMAN_BUG_ON((r->verb & QBMAN_RESPONSE_VERB_MASK) != QBMAN_FQ_QUERY_NP);
+
+ /* Determine success or failure */
+ if (r->rslt != QBMAN_MC_RSLT_OK) {
+ pr_err("Query NP fields of FQID 0x%x failed, code=0x%02x\n",
+ fqid, r->rslt);
+ return -EIO;
+ }
+
+ return 0;
+}
+
+uint32_t qbman_fq_state_frame_count(const struct qbman_fq_query_np_rslt *r)
+{
+ return (r->frm_cnt & 0x00FF);
+}
+
+uint32_t qbman_fq_state_byte_count(const struct qbman_fq_query_np_rslt *r)
+{
+ return r->byte_cnt;
+}
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index f266d6d..f59fc67 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -95,5 +95,7 @@ DPDK_18.02 {
dpaa2_svr_family;
dpaa2_virt_mode;
+ qbman_fq_query_state;
+ qbman_fq_state_frame_count;
} DPDK_17.11;
--
2.7.4
[dpdk-dev] [PATCH 05/14] bus/fslmc: expose platform soc value register
This patch expose the dpaa2 soc platform family type.
This is required to make some soc variant specific
decision during configuration and runtime.
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/portal/dpaa2_hw_dpio.c| 42 +++--
drivers/bus/fslmc/portal/dpaa2_hw_dpio.h| 3 +++
drivers/bus/fslmc/rte_bus_fslmc_version.map | 1 +
drivers/net/dpaa2/dpaa2_ethdev.c| 17 ++--
4 files changed, 35 insertions(+), 28 deletions(-)
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
index f00070f..a98991a 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.c
@@ -76,6 +76,9 @@ static struct dpio_dev_list dpio_dev_list
= TAILQ_HEAD_INITIALIZER(dpio_dev_list); /*!< DPIO device list */
static uint32_t io_space_count;
+/* Variable to store DPAA2 platform type */
+uint32_t dpaa2_svr_family;
+
/*Stashing Macros default for LS208x*/
static int dpaa2_core_cluster_base = 0x04;
static int dpaa2_cluster_sz = 2;
@@ -265,26 +268,6 @@ static int
dpaa2_configure_stashing(struct dpaa2_dpio_dev *dpio_dev, int cpu_id)
{
int sdest, ret;
- static int first_time;
-
- /* find the SoC type for the first time */
- if (!first_time) {
- struct mc_soc_version mc_plat_info = {0};
-
- if (mc_get_soc_version(dpio_dev->dpio,
- CMD_PRI_LOW, &mc_plat_info)) {
- PMD_INIT_LOG(ERR, "\tmc_get_soc_version failed\n");
- } else if ((mc_plat_info.svr & 0x) == SVR_LS1080A) {
- dpaa2_core_cluster_base = 0x02;
- dpaa2_cluster_sz = 4;
- PMD_INIT_LOG(DEBUG, "\tLS108x (A53) Platform Detected");
- } else if ((mc_plat_info.svr & 0x) == SVR_LX2160A) {
- dpaa2_core_cluster_base = 0x00;
- dpaa2_cluster_sz = 2;
- PMD_INIT_LOG(DEBUG, "\tLX2160 Platform Detected");
- }
- first_time = 1;
- }
/* Set the Stashing Destination */
if (cpu_id < 0) {
@@ -499,6 +482,25 @@ dpaa2_create_dpio_device(int vdev_fd,
rte_free(dpio_dev);
}
+ /* find the SoC type for the first time */
+ if (!dpaa2_svr_family) {
+ struct mc_soc_version mc_plat_info = {0};
+
+ if (mc_get_soc_version(dpio_dev->dpio,
+ CMD_PRI_LOW, &mc_plat_info)) {
+ PMD_INIT_LOG(ERR, "\tmc_get_soc_version failed\n");
+ } else if ((mc_plat_info.svr & 0x) == SVR_LS1080A) {
+ dpaa2_core_cluster_base = 0x02;
+ dpaa2_cluster_sz = 4;
+ PMD_INIT_LOG(DEBUG, "\tLS108x (A53) Platform Detected");
+ } else if ((mc_plat_info.svr & 0x) == SVR_LX2160A) {
+ dpaa2_core_cluster_base = 0x00;
+ dpaa2_cluster_sz = 2;
+ PMD_INIT_LOG(DEBUG, "\tLX2160 Platform Detected");
+ }
+ dpaa2_svr_family = (mc_plat_info.svr & 0x);
+ }
+
TAILQ_INSERT_TAIL(&dpio_dev_list, dpio_dev, next);
RTE_LOG(DEBUG, PMD, "DPAA2: Added [dpio.%d]\n", object_id);
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
index e845340..a3240b2 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_dpio.h
@@ -54,6 +54,9 @@ RTE_DECLARE_PER_LCORE(struct dpaa2_io_portal_t, _dpaa2_io);
#define DPAA2_PER_LCORE_SEC_DPIO RTE_PER_LCORE(_dpaa2_io).sec_dpio_dev
#define DPAA2_PER_LCORE_SEC_PORTAL DPAA2_PER_LCORE_SEC_DPIO->sw_portal
+/* Variable to store DPAA2 platform type */
+extern uint32_t dpaa2_svr_family;
+
extern struct dpaa2_io_portal_t dpaa2_io_portal[RTE_MAX_LCORE];
struct dpaa2_dpio_dev *dpaa2_get_qbman_swp(int cpu_id);
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_version.map
b/drivers/bus/fslmc/rte_bus_fslmc_version.map
index a1e30d6..f266d6d 100644
--- a/drivers/bus/fslmc/rte_bus_fslmc_version.map
+++ b/drivers/bus/fslmc/rte_bus_fslmc_version.map
@@ -93,6 +93,7 @@ DPDK_17.11 {
DPDK_18.02 {
global:
+ dpaa2_svr_family;
dpaa2_virt_mode;
} DPDK_17.11;
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 1cd302d..577bd8f 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -418,7 +418,6 @@ dpaa2_dev_rx_queue_setup(struct rte_eth_dev *dev,
{
struct dpaa2_dev_priv *priv = dev->data->dev_private;
struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
- struct mc_soc_version mc_plat_info = {0};
struct dpaa2_queue *dpaa2_q;
struct dpni_queue cfg;
uint8_t options = 0;
@@ -450,18 +449,20 @@ dpaa2_dev_rx_que
[dpdk-dev] [PATCH 03/14] bus/fslmc: add support for dynamic iova for DPAA2 devices
This patch add support for dynamic iova detection for DPAA2
devices and use of virtual address in such cases.
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/fslmc_bus.c | 44 +
drivers/bus/fslmc/fslmc_vfio.c | 5 +++-
drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 15 --
drivers/bus/fslmc/rte_bus_fslmc_version.map | 7 +
drivers/bus/fslmc/rte_fslmc.h | 3 ++
drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c | 1 +
drivers/net/dpaa2/dpaa2_ethdev.c| 1 +
7 files changed, 73 insertions(+), 3 deletions(-)
diff --git a/drivers/bus/fslmc/fslmc_bus.c b/drivers/bus/fslmc/fslmc_bus.c
index 480857e..63c333a 100644
--- a/drivers/bus/fslmc/fslmc_bus.c
+++ b/drivers/bus/fslmc/fslmc_bus.c
@@ -51,6 +51,7 @@
#define VFIO_IOMMU_GROUP_PATH "/sys/kernel/iommu_groups"
struct rte_fslmc_bus rte_fslmc_bus;
+uint8_t dpaa2_virt_mode;
static void
cleanup_fslmc_device_list(void)
@@ -300,6 +301,9 @@ rte_fslmc_probe(void)
}
}
+ if (rte_eal_iova_mode() == RTE_IOVA_VA)
+ dpaa2_virt_mode = 1;
+
return 0;
}
@@ -347,11 +351,51 @@ rte_fslmc_driver_unregister(struct rte_dpaa2_driver
*driver)
}
/*
+ * All device has iova as va
+ */
+static inline int
+fslmc_all_device_support_iova(void)
+{
+ int ret = 0;
+ struct rte_dpaa2_device *dev;
+ struct rte_dpaa2_driver *drv;
+
+ TAILQ_FOREACH(dev, &rte_fslmc_bus.device_list, next) {
+ TAILQ_FOREACH(drv, &rte_fslmc_bus.driver_list, next) {
+ ret = rte_fslmc_match(drv, dev);
+ if (ret)
+ continue;
+ /* if the driver is not supporting IOVA */
+ if (!(drv->drv_flags & RTE_DPAA2_DRV_IOVA_AS_VA))
+ return 0;
+ }
+ }
+ return 1;
+}
+
+/*
* Get iommu class of DPAA2 devices on the bus.
*/
static enum rte_iova_mode
rte_dpaa2_get_iommu_class(void)
{
+ bool is_vfio_noiommu_enabled = 1;
+ bool has_iova_va;
+
+ if (TAILQ_EMPTY(&rte_fslmc_bus.device_list))
+ return RTE_IOVA_DC;
+
+ /* check if all devices on the bus support Virtual addressing or not */
+ has_iova_va = fslmc_all_device_support_iova();
+
+#ifdef VFIO_PRESENT
+ is_vfio_noiommu_enabled = rte_vfio_noiommu_is_enabled() == true ?
+ true : false;
+#endif
+
+ if (has_iova_va && !is_vfio_noiommu_enabled)
+ return RTE_IOVA_VA;
+
return RTE_IOVA_PA;
}
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index e47215c..c0709de 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -249,7 +249,10 @@ int rte_fslmc_vfio_dmamap(void)
dma_map.size = memseg[i].len;
dma_map.vaddr = memseg[i].addr_64;
#ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
- dma_map.iova = memseg[i].iova;
+ if (rte_eal_iova_mode() == RTE_IOVA_VA)
+ dma_map.iova = dma_map.vaddr;
+ else
+ dma_map.iova = memseg[i].iova;
#else
dma_map.iova = dma_map.vaddr;
#endif
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index ece1a7d..7937293 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -276,13 +276,19 @@ enum qbman_fd_format {
#define DPAA2_EQ_RESP_ALWAYS 1
#ifdef RTE_LIBRTE_DPAA2_USE_PHYS_IOVA
+extern uint8_t dpaa2_virt_mode;
static void *dpaa2_mem_ptov(phys_addr_t paddr) __attribute__((unused));
/* todo - this is costly, need to write a fast coversion routine */
static void *dpaa2_mem_ptov(phys_addr_t paddr)
{
- const struct rte_memseg *memseg = rte_eal_get_physmem_layout();
+ const struct rte_memseg *memseg;
int i;
+ if (dpaa2_virt_mode)
+ return (void *)paddr;
+
+ memseg = rte_eal_get_physmem_layout();
+
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
if (paddr >= memseg[i].iova &&
(char *)paddr < (char *)memseg[i].iova + memseg[i].len)
@@ -295,9 +301,14 @@ static void *dpaa2_mem_ptov(phys_addr_t paddr)
static phys_addr_t dpaa2_mem_vtop(uint64_t vaddr) __attribute__((unused));
static phys_addr_t dpaa2_mem_vtop(uint64_t vaddr)
{
- const struct rte_memseg *memseg = rte_eal_get_physmem_layout();
+ const struct rte_memseg *memseg;
int i;
+ if (dpaa2_virt_mode)
+ return vaddr;
+
+ memseg = rte_eal_get_physmem_layout();
+
for (i = 0; i < RTE_MAX_MEMSEG && memseg[i].addr_64 != 0; i++) {
if (vaddr >= memseg[i].addr_64 &&
vaddr < memseg[i].addr_64 + memseg[i].len)
diff --git a/drivers/bus/fslmc/rte_bus_fslmc_
[dpdk-dev] [PATCH 06/14] bus/fslmc: add braces for pointers in macros
From: Nipun Gupta
Signed-off-by: Nipun Gupta
---
drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 43 +
1 file changed, 22 insertions(+), 21 deletions(-)
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 7937293..a432b6f 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -199,56 +199,57 @@ enum qbman_fd_format {
};
/*Macros to define operations on FD*/
#define DPAA2_SET_FD_ADDR(fd, addr) do { \
- fd->simple.addr_lo = lower_32_bits((uint64_t)(addr)); \
- fd->simple.addr_hi = upper_32_bits((uint64_t)(addr)); \
+ (fd)->simple.addr_lo = lower_32_bits((uint64_t)(addr)); \
+ (fd)->simple.addr_hi = upper_32_bits((uint64_t)(addr)); \
} while (0)
-#define DPAA2_SET_FD_LEN(fd, length) (fd)->simple.len = length
+#define DPAA2_SET_FD_LEN(fd, length) ((fd)->simple.len = length)
#define DPAA2_SET_FD_BPID(fd, bpid)((fd)->simple.bpid_offset |= bpid)
-#define DPAA2_SET_FD_IVP(fd) ((fd->simple.bpid_offset |= 0x4000))
+#define DPAA2_SET_FD_IVP(fd) (((fd)->simple.bpid_offset |= 0x4000))
#define DPAA2_SET_FD_OFFSET(fd, offset)\
- ((fd->simple.bpid_offset |= (uint32_t)(offset) << 16))
-#define DPAA2_SET_FD_INTERNAL_JD(fd, len) fd->simple.frc = (0x8000 | (len))
-#define DPAA2_SET_FD_FRC(fd, frc) fd->simple.frc = frc
-#define DPAA2_RESET_FD_CTRL(fd)(fd)->simple.ctrl = 0
+ (((fd)->simple.bpid_offset |= (uint32_t)(offset) << 16))
+#define DPAA2_SET_FD_INTERNAL_JD(fd, len) \
+ ((fd)->simple.frc = (0x8000 | (len)))
+#define DPAA2_SET_FD_FRC(fd, frc) ((fd)->simple.frc = frc)
+#define DPAA2_RESET_FD_CTRL(fd) ((fd)->simple.ctrl = 0)
#defineDPAA2_SET_FD_ASAL(fd, asal) ((fd)->simple.ctrl |= (asal <<
16))
#define DPAA2_SET_FD_FLC(fd, addr) do { \
- fd->simple.flc_lo = lower_32_bits((uint64_t)(addr));\
- fd->simple.flc_hi = upper_32_bits((uint64_t)(addr));\
+ (fd)->simple.flc_lo = lower_32_bits((uint64_t)(addr)); \
+ (fd)->simple.flc_hi = upper_32_bits((uint64_t)(addr)); \
} while (0)
-#define DPAA2_SET_FLE_INTERNAL_JD(fle, len) (fle->frc = (0x8000 | (len)))
+#define DPAA2_SET_FLE_INTERNAL_JD(fle, len) ((fle)->frc = (0x8000 | (len)))
#define DPAA2_GET_FLE_ADDR(fle)\
- (uint64_t)uint64_t)(fle->addr_hi)) << 32) + fle->addr_lo)
+ (uint64_t)uint64_t)((fle)->addr_hi)) << 32) + (fle)->addr_lo)
#define DPAA2_SET_FLE_ADDR(fle, addr) do { \
- fle->addr_lo = lower_32_bits((uint64_t)addr); \
- fle->addr_hi = upper_32_bits((uint64_t)addr); \
+ (fle)->addr_lo = lower_32_bits((uint64_t)addr); \
+ (fle)->addr_hi = upper_32_bits((uint64_t)addr); \
} while (0)
#define DPAA2_GET_FLE_CTXT(fle)\
(uint64_t)uint64_t)((fle)->reserved[1])) << 32) + \
(fle)->reserved[0])
#define DPAA2_FLE_SAVE_CTXT(fle, addr) do { \
- fle->reserved[0] = lower_32_bits((uint64_t)addr); \
- fle->reserved[1] = upper_32_bits((uint64_t)addr); \
+ (fle)->reserved[0] = lower_32_bits((uint64_t)addr); \
+ (fle)->reserved[1] = upper_32_bits((uint64_t)addr); \
} while (0)
#define DPAA2_SET_FLE_OFFSET(fle, offset) \
((fle)->fin_bpid_offset |= (uint32_t)(offset) << 16)
#define DPAA2_SET_FLE_BPID(fle, bpid) ((fle)->fin_bpid_offset |=
(uint64_t)bpid)
#define DPAA2_GET_FLE_BPID(fle) ((fle)->fin_bpid_offset & 0x00ff)
-#define DPAA2_SET_FLE_FIN(fle) (fle->fin_bpid_offset |= (uint64_t)1 << 31)
+#define DPAA2_SET_FLE_FIN(fle) ((fle)->fin_bpid_offset |= (uint64_t)1 << 31)
#define DPAA2_SET_FLE_IVP(fle) (((fle)->fin_bpid_offset |= 0x4000))
#define DPAA2_SET_FD_COMPOUND_FMT(fd) \
- (fd->simple.bpid_offset |= (uint32_t)1 << 28)
+ ((fd)->simple.bpid_offset |= (uint32_t)1 << 28)
#define DPAA2_GET_FD_ADDR(fd) \
((uint64_t)uint64_t)((fd)->simple.addr_hi)) << 32) + (fd)->simple.addr_lo))
#define DPAA2_GET_FD_LEN(fd) ((fd)->simple.len)
#define DPAA2_GET_FD_BPID(fd) (((fd)->simple.bpid_offset & 0x3FFF))
-#define DPAA2_GET_FD_IVP(fd) ((fd->simple.bpid_offset & 0x4000) >> 14)
+#define DPAA2_GET_FD_IVP(fd) (((fd)->simple.bpid_offset & 0x4000) >> 14)
#define DPAA2_GET_FD_OFFSET(fd)(((fd)->simple.bpid_offset &
0x0FFF) >> 16)
#define DPAA2_GET_FLE_OFFSET(fle) (((fle)->fin_bpid_offset & 0x0FFF) >> 16)
-#define DPAA2_SET_FLE_SG_EXT(fle) (fle->fin_bpid_offset |= (uint64_t)1 << 29)
+#define DPAA2_SET_FLE_SG_EXT(fle) ((fle)->fin_bpid_offset |= (uint64_t)1 << 29)
#define DPAA2_IS_SET_FLE_SG_EXT(fle) \
- ((fle->fin_bpid_offset & ((uint64_t)1 << 29)) ? 1 : 0)
+ (((fle)->fin_bpid_offset & ((uint64_t)1 << 29)) ? 1 : 0)
#define DPAA2_INLINE_MBUF_FROM_BUF(buf, meta_data_size) \
[dpdk-dev] [PATCH 09/14] net/dpaa2: align the frame size in MTU set
From: Ashish Jain
Setting correct frame size in dpaa2_dev_mtu_set
api call. Also correcting the correct max frame size
setting in hardware while dev_configure for jumbo frames
Signed-off-by: Ashish Jain
---
drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 4 ++--
drivers/net/dpaa2/dpaa2_ethdev.c| 11 +++
2 files changed, 9 insertions(+), 6 deletions(-)
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index a432b6f..9f9ce0b 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -53,8 +53,8 @@
#define SVR_LS2088A 0x8709
#define SVR_LX2160A 0x8736
-#ifndef ETH_VLAN_HLEN
-#define ETH_VLAN_HLEN 4 /** < Vlan Header Length */
+#ifndef VLAN_TAG_SIZE
+#define VLAN_TAG_SIZE 4 /** < Vlan Header Length */
#endif
#define MAX_TX_RING_SLOTS 8
diff --git a/drivers/net/dpaa2/dpaa2_ethdev.c b/drivers/net/dpaa2/dpaa2_ethdev.c
index 3e89d8d..894c60e 100644
--- a/drivers/net/dpaa2/dpaa2_ethdev.c
+++ b/drivers/net/dpaa2/dpaa2_ethdev.c
@@ -347,8 +347,8 @@ dpaa2_eth_dev_configure(struct rte_eth_dev *dev)
if (eth_conf->rxmode.jumbo_frame == 1) {
if (eth_conf->rxmode.max_rx_pkt_len <= DPAA2_MAX_RX_PKT_LEN) {
- ret = dpaa2_dev_mtu_set(dev,
- eth_conf->rxmode.max_rx_pkt_len);
+ ret = dpni_set_max_frame_length(dpni, CMD_PRI_LOW,
+ priv->token, eth_conf->rxmode.max_rx_pkt_len);
if (ret) {
PMD_INIT_LOG(ERR,
"unable to set mtu. check
config\n");
@@ -1000,7 +1000,8 @@ dpaa2_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
int ret;
struct dpaa2_dev_priv *priv = dev->data->dev_private;
struct fsl_mc_io *dpni = (struct fsl_mc_io *)priv->hw;
- uint32_t frame_size = mtu + ETHER_HDR_LEN + ETHER_CRC_LEN;
+ uint32_t frame_size = mtu + ETHER_HDR_LEN + ETHER_CRC_LEN
+ + VLAN_TAG_SIZE;
PMD_INIT_FUNC_TRACE();
@@ -1018,11 +1019,13 @@ dpaa2_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
else
dev->data->dev_conf.rxmode.jumbo_frame = 0;
+ dev->data->dev_conf.rxmode.max_rx_pkt_len = frame_size;
+
/* Set the Max Rx frame length as 'mtu' +
* Maximum Ethernet header length
*/
ret = dpni_set_max_frame_length(dpni, CMD_PRI_LOW, priv->token,
- mtu + ETH_VLAN_HLEN);
+ frame_size);
if (ret) {
PMD_DRV_LOG(ERR, "setting the max frame length failed");
return -1;
--
2.7.4
