[dpdk-dev] [Bug 60] rte_event_port_unlink() causes subsequent events to end up in wrong port
https://dpdk.org/tracker/show_bug.cgi?id=60
Bug ID: 60
Summary: rte_event_port_unlink() causes subsequent events to
end up in wrong port
Product: DPDK
Version: 17.11
Hardware: x86
OS: Linux
Status: CONFIRMED
Severity: major
Priority: Normal
Component: eventdev
Assignee: dev@dpdk.org
Reporter: matias@nokia.com
Target Milestone: ---
Created attachment 8
--> https://dpdk.org/tracker/attachment.cgi?id=8&action=edit
Test application
I'm seeing some unexpected(?) behavior when calling rte_event_port_unlink()
with the SW eventdev driver (DPDK 17.11.2/18.02.1,
RTE_EVENT_MAX_QUEUES_PER_DEV=255). After calling rte_event_port_unlink(),
the enqueued events may end up either back to the unlinked port or to port
zero.
Scenario:
- Run SW evendev on a service core
- Start eventdev with e.g. 16 ports. Each core will have a dedicated port.
- Create 1 atomic queue and link all active ports to it (some ports may not
be linked).
- Allocate some events and enqueue them to the created queue
- Next, each worker core does a number of scheduling rounds concurrently.
E.g.
uint64_t rx_events = 0;
while(rx_events < SCHED_ROUNDS) {
num_deq = rte_event_dequeue_burst(dev_id, port_id, ev, 1, 0);
if (num_deq) {
rx_events++;
rte_event_enqueue_burst(dev_id, port_id, ev, 1);
}
}
- This works fine but problems occur when doing cleanup after the first
loop finishes on some core.
E.g.
rte_event_port_unlink(dev_id, port_id, NULL, 0);
while(1) {
num_deq = rte_event_dequeue_burst(dev_id, port_id, ev, 1, 0);
if (num_deq == 0)
break;
rte_event_enqueue_burst(dev_id, port_id, ev, 1);
}
- The events enqueued in the cleanup loop will ramdomly end up either back to
the same port (which has already been unlinked) or to port zero, which is not
used (mapping rte_lcore_id to port_id).
As far as I understand the eventdev API, an eventdev port shouldn't have to be
linked to the target queue for enqueue to work properly.
I've attached a simple test application for reproducing this issue.
# sudo ./eventdev --vdev event_sw0 -s 0x2
Below is an example rte_event_dev_dump() output when processing events with two
cores (ports 2 and 3). The rest of the ports are not linked at all but events
still end up to port zero stalling the system.
Regards,
Matias
EventDev todo-fix-name: ports 16, qids 1
rx 908342
drop 0
tx 908342
sched calls: 42577156
sched cq/qid call: 43120490
sched no IQ enq: 42122057
sched no CQ enq: 42122064
inflight 32, credits: 4064
Port 0
rx 0 drop 0 tx 2 inflight 2
Max New: 1024 Avg cycles PP: 0Credits: 0
Receive burst distribution:
0:-nan%
rx ring used:0 free: 4096
cq ring used:2 free: 14
Port 1
rx 0 drop 0 tx 0 inflight 0
Max New: 1024 Avg cycles PP: 0Credits: 0
Receive burst distribution:
0:-nan%
rx ring used:0 free: 4096
cq ring used:0 free: 16
Port 2
rx 524292 drop 0 tx 524290 inflight 0
Max New: 1024 Avg cycles PP: 190 Credits: 30
Receive burst distribution:
0:98% 1-4:1.82%
rx ring used:0 free: 4096
cq ring used:0 free: 16
Port 3
rx 384050 drop 0 tx 384050 inflight 0
Max New: 1024 Avg cycles PP: 191 Credits: 0
Receive burst distribution:
0:100% 1-4:0.04%
rx ring used:0 free: 4096
cq ring used:0 free: 16
...
Port 15
rx 0 drop 0 tx 0 inflight 0
Max New: 1024 Avg cycles PP: 0Credits: 0
Receive burst distribution:
0:-nan%
rx ring used:0 free: 4096
cq ring used:0 free: 16
Queue 0 (Atomic)
rx 908342 drop 0 tx 908342
Per Port Stats:
Port 0: Pkts: 2 Flows: 1
Port 1: Pkts: 0 Flows: 0
Port 2: Pkts: 524290 Flows: 0
Port 3: Pkts: 384050 Flows: 0
Port 4: Pkts: 0 Flows: 0
Port 5: Pkts: 0 Flows: 0
Port 6: Pkts: 0 Flows: 0
Port 7: Pkts: 0 Flows: 0
Port 8: Pkts: 0 Flows: 0
Port 9: Pkts: 0 Flows: 0
Port 10: Pkts: 0 Flows: 0
Port 11: Pkts: 0 Flows: 0
Port 12: Pkts: 0 Flows: 0
Port 13: Pkts: 0 Flows: 0
Port 14: Pkts: 0 Flows: 0
Port 15: Pkts: 0 Flows: 0
-- iqs empty --
--
You are receiving this mail because:
You are the assignee for the bug.
Re: [dpdk-dev] [RFC] net/mlx5: add support 32bit build
On Sun, Jun 03, 2018 at 06:29:13PM +0300, Moti Haimovsky wrote: > This RFC and patch aim to add support for building mlx5 PMD on 32bit > archs such as i686. > The attached patch was not tested on 32bit environments and is given > here for explanatory purposes. > > Prerequisites: > * RDMA-core with 32bit support libraries. > * compile and build tools for 32bit systems according to DPDK documentation. > > Assumptions: > * Kernel and drivers are 64bit. > * Only user programs and support libraries are 32Bit. > > Changes proposed in the PMD for adding 32bit support: > The attached patch provides the changes needed in the PMD in order to support > 32bit builds with the major changes being: > * Modifying the UAR access routine to support non-atomic 64bit writes > according to the mlx5 hardware reference manual. > * Handling 64bit address fields that are written to the device WQEs. > * Modify UAR address and size mapping to fit into the 32bit address space. > > Concerns: > * The major concern of this modification is the UAR mapping. > Current PMD assumes 64bit address and therefore feels free to map 4G size > UAR to virtual address just above the huge-page addresses which of course > cannot work for 32bit address space. > As a workaround I reduced the UAR size to half the original size but > this of course is not the solution. > * Will vectorized support work on 32bit builds ? > > I am interested in your inputs regarding the concerns listed above, > especially the UAR mapping issue mentioned there. >[...] Hi Moti, My biggest concern and why such support has been removed [1] is because Mellanox OFED is not supporting 32bits compilation. As Mellanox OFED is still supported, we need such guarantee. On Mellanox website there is no way to download such kind of ISO which means such support becomes only valid for upstream RDMA-Core library with almost newer kernels. If you can find a way to detect the PMD is compiled in front of Mellanox OFED to forbid such support, it would be great, otherwise I am afraid such support cannot be added back without creating a lot of issues for people wanting it on 32bits processors with Mellanox OFED. Thanks, [1] https://dpdk.org/browse/dpdk/commit/?id=ebbb81eb27daca0a89ee8f228fcf141d9eb6ef1c -- Nélio Laranjeiro 6WIND
Re: [dpdk-dev] [PATCH 0/4] net/virtio: Tx path selection and offload improvements
On 06/01/2018 02:47 PM, Maxime Coquelin wrote: Below are benchmarks results when offloads are enabled at "When *no* offload" should be read here... device level. Rx-mrg=off benchmarks: ++---+-+-+--+ |Run | PVP | Guest->Host | Host->Guest | Loopback | ++---+-+-+--+ | v18.05 | 14.47 | 17.02 | 17.57 |13.15 | | + series | 14.88 | 19.64 | 17.63 |13.11 | ++---+-+-+--+ Rx-mrg=on benchmarks: ++---+-+-+--+ |Run | PVP | Guest->Host | Host->Guest | Loopback | ++---+-+-+--+ | v18.05 | 9.53 | 13.80 | 16.70 |13.11 | | + series | 12.62 | 19.69 | 16.70 |13.11 | ++---+-+-+--+
Re: [dpdk-dev] [PATCH] ethdev: force RSS offload rules again
Sunday, June 3, 2018 5:15 PM, Ferruh Yigit: > Subject: Re: [PATCH] ethdev: force RSS offload rules again > > On 6/3/2018 11:41 AM, Shahaf Shuler wrote: > > Thursday, May 31, 2018 4:23 PM, Ferruh Yigit: > >> Subject: [PATCH] ethdev: force RSS offload rules again > >> > >> PMDs should provide supported RSS hash functions via > >> dev_info.flow_type_rss_offloads variable. > >> > >> There is a check in ethdev if requested RSS hash function is > >> supported by PMD or not. > >> This check has been relaxed in previous release to not return an > >> error when a non supported has function requested [1], this has been > >> done to not break the applications. > >> > >> Adding the error return back. > >> PMDs need to provide correct list of supported hash functions and > >> applications need to take care this information before configuring > >> the RSS otherwise they will get an error from APIs: > >> rte_eth_dev_rss_hash_update() > >> rte_eth_dev_configure() > > > > Are the current app/examples in DPDK tree behave accordingly? > > I tried a few which were good but I don't know about all. > > That is why we should merge this patch early so that we can detect and fix > ones fails. Acked-by: Shahaf Shuler
Re: [dpdk-dev] [Bug 60] rte_event_port_unlink() causes subsequent events to end up in wrong port
-Original Message-
> Date: Mon, 4 Jun 2018 07:21:18 +
> From: bugzi...@dpdk.org
> To: dev@dpdk.org
> Subject: [dpdk-dev] [Bug 60] rte_event_port_unlink() causes subsequent
> events to end up in wrong port
>
> https://dpdk.org/tracker/show_bug.cgi?id=60
>
> Bug ID: 60
>Summary: rte_event_port_unlink() causes subsequent events to
> end up in wrong port
>Product: DPDK
>Version: 17.11
> Hardware: x86
> OS: Linux
> Status: CONFIRMED
> Severity: major
> Priority: Normal
> Component: eventdev
> Assignee: dev@dpdk.org
> Reporter: matias@nokia.com
> Target Milestone: ---
>
> Created attachment 8
> --> https://dpdk.org/tracker/attachment.cgi?id=8&action=edit
> Test application
>
> I'm seeing some unexpected(?) behavior when calling rte_event_port_unlink()
> with the SW eventdev driver (DPDK 17.11.2/18.02.1,
> RTE_EVENT_MAX_QUEUES_PER_DEV=255). After calling rte_event_port_unlink(),
> the enqueued events may end up either back to the unlinked port or to port
> zero.
>
> Scenario:
>
> - Run SW evendev on a service core
> - Start eventdev with e.g. 16 ports. Each core will have a dedicated port.
> - Create 1 atomic queue and link all active ports to it (some ports may not
> be linked).
> - Allocate some events and enqueue them to the created queue
> - Next, each worker core does a number of scheduling rounds concurrently.
> E.g.
>
> uint64_t rx_events = 0;
> while(rx_events < SCHED_ROUNDS) {
> num_deq = rte_event_dequeue_burst(dev_id, port_id, ev, 1, 0);
>
> if (num_deq) {
> rx_events++;
> rte_event_enqueue_burst(dev_id, port_id, ev, 1);
> }
> }
>
> - This works fine but problems occur when doing cleanup after the first
> loop finishes on some core.
> E.g.
>
> rte_event_port_unlink(dev_id, port_id, NULL, 0);
>
> while(1) {
> num_deq = rte_event_dequeue_burst(dev_id, port_id, ev, 1, 0);
>
> if (num_deq == 0)
> break;
>
> rte_event_enqueue_burst(dev_id, port_id, ev, 1);
> }
>
> - The events enqueued in the cleanup loop will ramdomly end up either back to
> the same port (which has already been unlinked) or to port zero, which is not
> used (mapping rte_lcore_id to port_id).
>
> As far as I understand the eventdev API, an eventdev port shouldn't have to be
> linked to the target queue for enqueue to work properly.
That is a grey area in the spec. octeontx drivers works as the way you
described. I am not sure about SW driver(CC:
harry.van.haa...@intel.com), If there is no performance impact for none of
the drivers and it is do able for all HW and SW implementation then can
do that way(CC: all PMD maintainers)
No related to this question, Are you planning to use rte_event_port_unlink() in
fastpath?
Does rte_event_stop() works for you, if it is in slow path.
Re: [dpdk-dev] [dpdk-stable] Regression tests for stable releases from companies involved in DPDK
On Mon, 2018-06-04 at 05:24 +, Shreyansh Jain wrote: > Hello Luca > > > -Original Message- > > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Luca Boccassi > > Sent: Thursday, May 31, 2018 3:57 PM > > To: dev@dpdk.org > > Cc: sta...@dpdk.org; tho...@monjalon.net > > Subject: [dpdk-dev] Regression tests for stable releases from > > companies > > involved in DPDK > > > > Hello all, > > > > At this morning's release meeting (minutes coming soon from John), > > we > > briefly discussed the state of the regression testing for stable > > releases and agreed we need to formalise the process. > > > > At the moment we have a firm commitment from Intel and Mellanox to > > test > > all stable branches (and if I heard correctly from NXP as well? > > Please > > Yes, I confirmed that on call on behalf of NXP. But... > > > confirm!). AT&T committed to run regressions on the 16.11 branch. > > Not until 17.05 did NXP's first driver started appearing in the DPDK > upstream releases. Somehow, I misunderstood your request for 16.11 > with 17.11 stable. That's fine - the 17.11 branch is also getting ready for release (see relevant threads), and 18.02 will soon as well. -- Kind regards, Luca Boccassi
[dpdk-dev] [PATCH] net/mvpp2: convert to dynamic logging
Convert mvpp2 PMD to use dynamic logging.
Signed-off-by: Tomasz Duszynski
---
drivers/net/mvpp2/mrvl_ethdev.c | 147 +---
drivers/net/mvpp2/mrvl_ethdev.h | 8 +++
drivers/net/mvpp2/mrvl_flow.c | 32 -
drivers/net/mvpp2/mrvl_qos.c| 42 ++--
4 files changed, 124 insertions(+), 105 deletions(-)
diff --git a/drivers/net/mvpp2/mrvl_ethdev.c b/drivers/net/mvpp2/mrvl_ethdev.c
index ea6a786..ae8804a 100644
--- a/drivers/net/mvpp2/mrvl_ethdev.c
+++ b/drivers/net/mvpp2/mrvl_ethdev.c
@@ -94,6 +94,8 @@ struct pp2_bpool *mrvl_port_to_bpool_lookup[RTE_MAX_ETHPORTS];
int mrvl_port_bpool_size[PP2_NUM_PKT_PROC][PP2_BPOOL_NUM_POOLS][RTE_MAX_LCORE];
uint64_t cookie_addr_high = MRVL_COOKIE_ADDR_INVALID;
+int mrvl_logtype;
+
struct mrvl_ifnames {
const char *names[PP2_NUM_ETH_PPIO * PP2_NUM_PKT_PROC];
int idx;
@@ -206,7 +208,7 @@ mrvl_init_hif(int core_id)
ret = mrvl_reserve_bit(&used_hifs, MRVL_MUSDK_HIFS_MAX);
if (ret < 0) {
- RTE_LOG(ERR, PMD, "Failed to allocate hif %d\n", core_id);
+ MRVL_LOG(ERR, "Failed to allocate hif %d", core_id);
return ret;
}
@@ -216,7 +218,7 @@ mrvl_init_hif(int core_id)
params.out_size = MRVL_PP2_AGGR_TXQD_MAX;
ret = pp2_hif_init(¶ms, &hifs[core_id]);
if (ret) {
- RTE_LOG(ERR, PMD, "Failed to initialize hif %d\n", core_id);
+ MRVL_LOG(ERR, "Failed to initialize hif %d", core_id);
return ret;
}
@@ -235,7 +237,7 @@ mrvl_get_hif(struct mrvl_priv *priv, int core_id)
ret = mrvl_init_hif(core_id);
if (ret < 0) {
- RTE_LOG(ERR, PMD, "Failed to allocate hif %d\n", core_id);
+ MRVL_LOG(ERR, "Failed to allocate hif %d", core_id);
goto out;
}
@@ -265,7 +267,7 @@ static int
mrvl_configure_rss(struct mrvl_priv *priv, struct rte_eth_rss_conf *rss_conf)
{
if (rss_conf->rss_key)
- RTE_LOG(WARNING, PMD, "Changing hash key is not supported\n");
+ MRVL_LOG(WARNING, "Changing hash key is not supported");
if (rss_conf->rss_hf == 0) {
priv->ppio_params.inqs_params.hash_type = PP2_PPIO_HASH_T_NONE;
@@ -307,19 +309,19 @@ mrvl_dev_configure(struct rte_eth_dev *dev)
if (dev->data->dev_conf.rxmode.mq_mode != ETH_MQ_RX_NONE &&
dev->data->dev_conf.rxmode.mq_mode != ETH_MQ_RX_RSS) {
- RTE_LOG(INFO, PMD, "Unsupported rx multi queue mode %d\n",
+ MRVL_LOG(INFO, "Unsupported rx multi queue mode %d",
dev->data->dev_conf.rxmode.mq_mode);
return -EINVAL;
}
if (!(dev->data->dev_conf.rxmode.offloads & DEV_RX_OFFLOAD_CRC_STRIP)) {
- RTE_LOG(INFO, PMD,
- "L2 CRC stripping is always enabled in hw\n");
+ MRVL_LOG(INFO,
+ "L2 CRC stripping is always enabled in hw");
dev->data->dev_conf.rxmode.offloads |= DEV_RX_OFFLOAD_CRC_STRIP;
}
if (dev->data->dev_conf.rxmode.split_hdr_size) {
- RTE_LOG(INFO, PMD, "Split headers not supported\n");
+ MRVL_LOG(INFO, "Split headers not supported");
return -EINVAL;
}
@@ -343,7 +345,7 @@ mrvl_dev_configure(struct rte_eth_dev *dev)
if (dev->data->nb_rx_queues == 1 &&
dev->data->dev_conf.rxmode.mq_mode == ETH_MQ_RX_RSS) {
- RTE_LOG(WARNING, PMD, "Disabling hash for 1 rx queue\n");
+ MRVL_LOG(WARNING, "Disabling hash for 1 rx queue");
priv->ppio_params.inqs_params.hash_type = PP2_PPIO_HASH_T_NONE;
return 0;
@@ -467,7 +469,7 @@ mrvl_tx_queue_start(struct rte_eth_dev *dev, uint16_t
queue_id)
/* passing 1 enables given tx queue */
ret = pp2_ppio_set_outq_state(priv->ppio, queue_id, 1);
if (ret) {
- RTE_LOG(ERR, PMD, "Failed to start txq %d\n", queue_id);
+ MRVL_LOG(ERR, "Failed to start txq %d", queue_id);
return ret;
}
@@ -499,7 +501,7 @@ mrvl_tx_queue_stop(struct rte_eth_dev *dev, uint16_t
queue_id)
/* passing 0 disables given tx queue */
ret = pp2_ppio_set_outq_state(priv->ppio, queue_id, 0);
if (ret) {
- RTE_LOG(ERR, PMD, "Failed to stop txq %d\n", queue_id);
+ MRVL_LOG(ERR, "Failed to stop txq %d", queue_id);
return ret;
}
@@ -546,7 +548,7 @@ mrvl_dev_start(struct rte_eth_dev *dev)
priv->bpool_init_size += buffs_to_add;
ret = mrvl_fill_bpool(dev->data->rx_queues[0], buffs_to_add);
if (ret)
- RTE_LOG(ERR, PMD, "Failed to add buffers to bpool\n");
+ MRVL_LOG(ERR, "Failed to add buffers to bpool");
}
/*
@@ -561,7 +563,7 @@ mrvl_dev_start
[dpdk-dev] 18.02.2 patches review and test
Hi all, Here is a list of patches targeted for stable release 18.02.2. Please help review and test. The planned date for the final release is Thursday, the 14th of June. Before that, please shout if anyone has objections with these patches being applied. Also for the companies committed to running regression tests, please run the tests and report any issue before the release date. These patches are located at branch 18.02 of dpdk-stable repo: https://dpdk.org/browse/dpdk-stable/ Thanks. Luca Boccassi --- Aaron Conole (3): nfp: unlink the appropriate lock file nfp: allow for non-root user nfp: restore the unlink operation Adrien Mazarguil (20): net/mlx: fix rdma-core glue path with EAL plugins net/mlx4: fix RSS resource leak in case of error net/mlx4: fix ignored RSS hash types net/mlx5: fix RSS flow action bounds check net/bnxt: fix matching of flow API item masks app/testpmd: fix flow completion for RSS queues app/testpmd: fix lack of flow action configuration app/testpmd: fix RSS flow action configuration app/testpmd: fix missing RSS fields in flow action app/testpmd: fix missing boolean values in flow command ethdev: fix shallow copy of flow API RSS action ethdev: fix missing include in flow API net/mlx5: fix flow director rule deletion crash net/mlx4: fix Rx resource leak in case of error net/mlx4: fix default RSS hash fields net/mlx4: fix inner RSS support for broken kernels net/mlx: control netdevices through ioctl only net/mlx4: fix UDP flow rule limitation enforcement app/testpmd: fix empty list of RSS queues for flow net/mlx4: fix shifts of signed values in Tx Ajit Khaparde (20): net/bnxt: fix LRO disable net/bnxt: fix Rx drop setting net/bnxt: set padding flags in Rx descriptor net/bnxt: fix endianness of flag net/bnxt: fix incorrect ntuple flag setting net/bnxt: fix Rx checksum flags for tunnel frames net/bnxt: fix L2 filter cleanup net/bnxt: fix flow destroy net/bnxt: reset L2 filter id once filter is freed net/bnxt: free memory allocated for VF filters net/bnxt: avoid freeing memzone multiple times net/bnxt: avoid invalid vnic id in set L2 Rx mask net/bnxt: fix mbuf data offset initialization net/bnxt: fix MTU calculation net/bnxt: fix to reset status of initialization net/bnxt: fix usage of vnic id net/bnxt: fix Tx and Rx burst for secondary process net/bnxt: use first completion ring for fwd and async event net/bnxt: fix xstats for VF net/bnxt: fix Rx checksum flags Alejandro Lucero (5): net/nfp: fix assigning port id in mbuf net/nfp: fix barrier location net/nfp: fix link speed capabilities doc: fix NFP NIC guide grammar net/nfp: fix mbufs releasing when stop or close Allain Legacy (1): ip_frag: fix double free of chained mbufs Anatoly Burakov (17): eal: fix IPC timeout eal: fix IPC socket path eal: fix IPC request socket path eal: fix errno handling in IPC eal: fix race condition in IPC request vfio: fix headers for C++ support app/crypto-perf: fix IOVA translation mem: do not use physical addresses in IOVA as VA mode bus/fslmc: do not needlessly check for IOVA mode vfio: do not needlessly check for IOVA mode vfio: fix device hotplug when several devices per group vfio: export functions even when disabled ipc: fix use-after-free in synchronous requests ipc: fix missing mutex unlocks on failed send memzone: fix size on reserving biggest memzone eal: remove unused path pattern mempool: fix virtual address population Andrew Rybchenko (19): net/bonding: fix library version in meson build net/ixgbe: fix library version in meson build net/i40e: fix library version in meson build net/ring: fix library version in meson build net/null: fix library version in meson build bitratestats: fix library version in meson build cryptodev: fix library version in meson build eventdev: fix library version in meson build mempool: fix library version in meson build pdump: fix library version in meson build table: fix library version in meson build net/sfc: add missing defines for SAL annotation net/sfc/base: fix too long line net/sfc: fix mbuf data alignment calculation net/sfc/base: fix comparison always true warning mempool: fix leak when no objects are populated test/mempool: fix autotest retry net/sfc: ignore spec bits not covered by mask net/sfc: fix inner TCP/UDP checksum offload control Andy Green (40): eal: explicit cast of builtin for bsf32 eal: explicit cast of core id when getting index eal: declare trace buffer at top of own block spinlock/x86: move stack decla
[dpdk-dev] [PATCH] net/mvpp2: check pointer before using it
Avoid NULL dereference by checking pointer before using it.
Note that the fixes tag is before the driver was renamed
to mvpp2, so the patch will not apply that far back.
Fixes: 3588aaa68eab ("net/mrvl: fix HIF objects allocation")
Signed-off-by: Tomasz Duszynski
---
drivers/net/mvpp2/mrvl_ethdev.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/net/mvpp2/mrvl_ethdev.c b/drivers/net/mvpp2/mrvl_ethdev.c
index ae8804a..d5eb1fe 100644
--- a/drivers/net/mvpp2/mrvl_ethdev.c
+++ b/drivers/net/mvpp2/mrvl_ethdev.c
@@ -1614,9 +1614,12 @@ mrvl_rx_queue_release(void *rxq)
if (core_id == LCORE_ID_ANY)
core_id = 0;
+ if (!q)
+ return;
+
hif = mrvl_get_hif(q->priv, core_id);
- if (!q || !hif)
+ if (!hif)
return;
tc = q->priv->rxq_map[q->queue_id].tc;
--
2.7.4
[dpdk-dev] [PATCH v2 1/2] librte_ip_frag: add function to delete expired entries
A fragmented packets is supposed to live no longer than max_cycles,
but the lib deletes an expired packet only occasionally when it scans
a bucket to find an empty slot while adding a new packet.
Therefore a fragment might sit in the table forever.
Signed-off-by: Alex Kiselev
---
lib/librte_ip_frag/ip_frag_common.h| 18 ++
lib/librte_ip_frag/ip_frag_internal.c | 18 --
lib/librte_ip_frag/rte_ip_frag.h | 19 ++-
lib/librte_ip_frag/rte_ip_frag_common.c| 21 +
lib/librte_ip_frag/rte_ip_frag_version.map | 6 ++
5 files changed, 63 insertions(+), 19 deletions(-)
diff --git a/lib/librte_ip_frag/ip_frag_common.h
b/lib/librte_ip_frag/ip_frag_common.h
index 197acf8d8..0fdcc7d0f 100644
--- a/lib/librte_ip_frag/ip_frag_common.h
+++ b/lib/librte_ip_frag/ip_frag_common.h
@@ -25,6 +25,12 @@
#define IPv6_KEY_BYTES_FMT \
"%08" PRIx64 "%08" PRIx64 "%08" PRIx64 "%08" PRIx64
+#ifdef RTE_LIBRTE_IP_FRAG_TBL_STAT
+#defineIP_FRAG_TBL_STAT_UPDATE(s, f, v)((s)->f += (v))
+#else
+#defineIP_FRAG_TBL_STAT_UPDATE(s, f, v)do {} while (0)
+#endif /* IP_FRAG_TBL_STAT */
+
/* internal functions declarations */
struct rte_mbuf * ip_frag_process(struct ip_frag_pkt *fp,
struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb,
@@ -149,4 +155,16 @@ ip_frag_reset(struct ip_frag_pkt *fp, uint64_t tms)
fp->frags[IP_FIRST_FRAG_IDX] = zero_frag;
}
+/* local frag table helper functions */
+static inline void
+ip_frag_tbl_del(struct rte_ip_frag_tbl *tbl, struct rte_ip_frag_death_row *dr,
+ struct ip_frag_pkt *fp)
+{
+ ip_frag_free(fp, dr);
+ ip_frag_key_invalidate(&fp->key);
+ TAILQ_REMOVE(&tbl->lru, fp, lru);
+ tbl->use_entries--;
+ IP_FRAG_TBL_STAT_UPDATE(&tbl->stat, del_num, 1);
+}
+
#endif /* _IP_FRAG_COMMON_H_ */
diff --git a/lib/librte_ip_frag/ip_frag_internal.c
b/lib/librte_ip_frag/ip_frag_internal.c
index 2560c7713..97470a872 100644
--- a/lib/librte_ip_frag/ip_frag_internal.c
+++ b/lib/librte_ip_frag/ip_frag_internal.c
@@ -14,24 +14,6 @@
#defineIP_FRAG_TBL_POS(tbl, sig) \
((tbl)->pkt + ((sig) & (tbl)->entry_mask))
-#ifdef RTE_LIBRTE_IP_FRAG_TBL_STAT
-#defineIP_FRAG_TBL_STAT_UPDATE(s, f, v)((s)->f += (v))
-#else
-#defineIP_FRAG_TBL_STAT_UPDATE(s, f, v)do {} while (0)
-#endif /* IP_FRAG_TBL_STAT */
-
-/* local frag table helper functions */
-static inline void
-ip_frag_tbl_del(struct rte_ip_frag_tbl *tbl, struct rte_ip_frag_death_row *dr,
- struct ip_frag_pkt *fp)
-{
- ip_frag_free(fp, dr);
- ip_frag_key_invalidate(&fp->key);
- TAILQ_REMOVE(&tbl->lru, fp, lru);
- tbl->use_entries--;
- IP_FRAG_TBL_STAT_UPDATE(&tbl->stat, del_num, 1);
-}
-
static inline void
ip_frag_tbl_add(struct rte_ip_frag_tbl *tbl, struct ip_frag_pkt *fp,
const struct ip_frag_key *key, uint64_t tms)
diff --git a/lib/librte_ip_frag/rte_ip_frag.h b/lib/librte_ip_frag/rte_ip_frag.h
index b3f3f78df..7f425f610 100644
--- a/lib/librte_ip_frag/rte_ip_frag.h
+++ b/lib/librte_ip_frag/rte_ip_frag.h
@@ -65,10 +65,13 @@ struct ip_frag_pkt {
#define IP_FRAG_DEATH_ROW_LEN 32 /**< death row size (in packets) */
+/* death row size in mbufs */
+#define IP_FRAG_DEATH_ROW_MBUF_LEN (IP_FRAG_DEATH_ROW_LEN * (IP_MAX_FRAG_NUM +
1))
+
/** mbuf death row (packets to be freed) */
struct rte_ip_frag_death_row {
uint32_t cnt; /**< number of mbufs currently on death row */
- struct rte_mbuf *row[IP_FRAG_DEATH_ROW_LEN * (IP_MAX_FRAG_NUM + 1)];
+ struct rte_mbuf *row[IP_FRAG_DEATH_ROW_MBUF_LEN];
/**< mbufs to be freed */
};
@@ -325,6 +328,20 @@ void rte_ip_frag_free_death_row(struct
rte_ip_frag_death_row *dr,
void
rte_ip_frag_table_statistics_dump(FILE * f, const struct rte_ip_frag_tbl *tbl);
+/**
+ * Delete expired fragments
+ *
+ * @param tbl
+ * Table to delete expired fragments from
+ * @param dr
+ * Death row to free buffers to
+ * @param tms
+ * Current timestamp
+ */
+void __rte_experimental
+rte_frag_table_del_expired_entries(struct rte_ip_frag_tbl *tbl,
+ struct rte_ip_frag_death_row *dr, uint64_t tms);
+
#ifdef __cplusplus
}
#endif
diff --git a/lib/librte_ip_frag/rte_ip_frag_common.c
b/lib/librte_ip_frag/rte_ip_frag_common.c
index 659a17951..a23f6f24f 100644
--- a/lib/librte_ip_frag/rte_ip_frag_common.c
+++ b/lib/librte_ip_frag/rte_ip_frag_common.c
@@ -121,3 +121,24 @@ rte_ip_frag_table_statistics_dump(FILE *f, const struct
rte_ip_frag_tbl *tbl)
fail_nospace,
fail_total - fail_nospace);
}
+
+/* Delete expired fragments */
+void __rte_experimental
+rte_frag_table_del_expired_entries(struct rte_ip_frag_tbl *tbl,
+ struct rte_ip_frag_death_row *dr, uint64_t tms)
+{
+ uint64_t max_cycles;
+ struct ip_frag_pkt *fp;
+
+ max_cycles = tbl->max_cycles;
+
+ TAILQ_F
[dpdk-dev] [PATCH v2 2/2] librte_ip_frag: add mbuf counter
There might be situations (kind of attack when a lot of
fragmented packets are sent to a dpdk application in order
to flood the fragmentation table) when no additional mbufs
must be added to the fragmentations table since it already
contains to many of them. Currently there is no way to
determine the number of mbufs holded int the fragmentation
table. This patch allows to keep track of the number of mbufs
holded in the fragmentation table.
Signed-off-by: Alex Kiselev
---
lib/librte_ip_frag/ip_frag_common.h| 16 +---
lib/librte_ip_frag/ip_frag_internal.c | 16 +---
lib/librte_ip_frag/rte_ip_frag.h | 18 +-
lib/librte_ip_frag/rte_ip_frag_common.c| 1 +
lib/librte_ip_frag/rte_ip_frag_version.map | 1 +
lib/librte_ip_frag/rte_ipv4_reassembly.c | 2 +-
lib/librte_ip_frag/rte_ipv6_reassembly.c | 2 +-
7 files changed, 39 insertions(+), 17 deletions(-)
diff --git a/lib/librte_ip_frag/ip_frag_common.h
b/lib/librte_ip_frag/ip_frag_common.h
index 0fdcc7d0f..9fe5c0559 100644
--- a/lib/librte_ip_frag/ip_frag_common.h
+++ b/lib/librte_ip_frag/ip_frag_common.h
@@ -32,15 +32,15 @@
#endif /* IP_FRAG_TBL_STAT */
/* internal functions declarations */
-struct rte_mbuf * ip_frag_process(struct ip_frag_pkt *fp,
- struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb,
- uint16_t ofs, uint16_t len, uint16_t more_frags);
+struct rte_mbuf *ip_frag_process(struct rte_ip_frag_tbl *tbl,
+ struct ip_frag_pkt *fp, struct rte_ip_frag_death_row *dr,
+ struct rte_mbuf *mb, uint16_t ofs, uint16_t len, uint16_t more_frags);
-struct ip_frag_pkt * ip_frag_find(struct rte_ip_frag_tbl *tbl,
+struct ip_frag_pkt *ip_frag_find(struct rte_ip_frag_tbl *tbl,
struct rte_ip_frag_death_row *dr,
const struct ip_frag_key *key, uint64_t tms);
-struct ip_frag_pkt * ip_frag_lookup(struct rte_ip_frag_tbl *tbl,
+struct ip_frag_pkt *ip_frag_lookup(struct rte_ip_frag_tbl *tbl,
const struct ip_frag_key *key, uint64_t tms,
struct ip_frag_pkt **free, struct ip_frag_pkt **stale);
@@ -91,7 +91,8 @@ ip_frag_key_cmp(const struct ip_frag_key * k1, const struct
ip_frag_key * k2)
/* put fragment on death row */
static inline void
-ip_frag_free(struct ip_frag_pkt *fp, struct rte_ip_frag_death_row *dr)
+ip_frag_free(struct rte_ip_frag_tbl *tbl, struct ip_frag_pkt *fp,
+ struct rte_ip_frag_death_row *dr)
{
uint32_t i, k;
@@ -100,6 +101,7 @@ ip_frag_free(struct ip_frag_pkt *fp, struct
rte_ip_frag_death_row *dr)
if (fp->frags[i].mb != NULL) {
dr->row[k++] = fp->frags[i].mb;
fp->frags[i].mb = NULL;
+ tbl->nb_mbufs--;
}
}
@@ -160,7 +162,7 @@ static inline void
ip_frag_tbl_del(struct rte_ip_frag_tbl *tbl, struct rte_ip_frag_death_row *dr,
struct ip_frag_pkt *fp)
{
- ip_frag_free(fp, dr);
+ ip_frag_free(tbl, fp, dr);
ip_frag_key_invalidate(&fp->key);
TAILQ_REMOVE(&tbl->lru, fp, lru);
tbl->use_entries--;
diff --git a/lib/librte_ip_frag/ip_frag_internal.c
b/lib/librte_ip_frag/ip_frag_internal.c
index 97470a872..4c47d3fb4 100644
--- a/lib/librte_ip_frag/ip_frag_internal.c
+++ b/lib/librte_ip_frag/ip_frag_internal.c
@@ -29,14 +29,13 @@ static inline void
ip_frag_tbl_reuse(struct rte_ip_frag_tbl *tbl, struct rte_ip_frag_death_row
*dr,
struct ip_frag_pkt *fp, uint64_t tms)
{
- ip_frag_free(fp, dr);
+ ip_frag_free(tbl, fp, dr);
ip_frag_reset(fp, tms);
TAILQ_REMOVE(&tbl->lru, fp, lru);
TAILQ_INSERT_TAIL(&tbl->lru, fp, lru);
IP_FRAG_TBL_STAT_UPDATE(&tbl->stat, reuse_num, 1);
}
-
static inline void
ipv4_frag_hash(const struct ip_frag_key *key, uint32_t *v1, uint32_t *v2)
{
@@ -88,8 +87,9 @@ ipv6_frag_hash(const struct ip_frag_key *key, uint32_t *v1,
uint32_t *v2)
}
struct rte_mbuf *
-ip_frag_process(struct ip_frag_pkt *fp, struct rte_ip_frag_death_row *dr,
- struct rte_mbuf *mb, uint16_t ofs, uint16_t len, uint16_t more_frags)
+ip_frag_process(struct rte_ip_frag_tbl *tbl, struct ip_frag_pkt *fp,
+ struct rte_ip_frag_death_row *dr, struct rte_mbuf *mb, uint16_t ofs,
+ uint16_t len, uint16_t more_frags)
{
uint32_t idx;
@@ -147,7 +147,7 @@ ip_frag_process(struct ip_frag_pkt *fp, struct
rte_ip_frag_death_row *dr,
fp->frags[IP_LAST_FRAG_IDX].len);
/* free all fragments, invalidate the entry. */
- ip_frag_free(fp, dr);
+ ip_frag_free(tbl, fp, dr);
ip_frag_key_invalidate(&fp->key);
IP_FRAG_MBUF2DR(dr, mb);
@@ -157,6 +157,7 @@ ip_frag_process(struct ip_frag_pkt *fp, struct
rte_ip_frag_death_row *dr,
fp->frags[idx].ofs = ofs;
fp->frags[idx].len = len;
fp->frags[idx].mb = mb;
+ tbl->nb_mbufs++;
mb = NUL
Re: [dpdk-dev] [PATCH 4/4] net/virtio: improve offload check performance
On Fri, Jun 01, 2018 at 02:47:58PM +0200, Maxime Coquelin wrote:
> Instead of checking the multiple Virtio features bits for
> every packet, let's do the check once at configure time and
> store it in virtio_hw struct.
>
> Signed-off-by: Maxime Coquelin
> ---
> drivers/net/virtio/virtio_ethdev.c | 19 +++
> drivers/net/virtio/virtio_pci.h| 2 ++
> drivers/net/virtio/virtio_rxtx.c | 29 ++---
> 3 files changed, 27 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_ethdev.c
> b/drivers/net/virtio/virtio_ethdev.c
> index d481b282e..981e0994a 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -1790,6 +1790,22 @@ rte_virtio_pmd_init(void)
> rte_pci_register(&rte_virtio_pmd);
> }
>
> +static inline int
Maybe it's better to return bool.
> +rx_offload_enabled(struct virtio_hw *hw)
> +{
> + return vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
> + vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
> + vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
> +}
> +
> +static inline int
Ditto.
> +tx_offload_enabled(struct virtio_hw *hw)
> +{
> + return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
> + vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
> + vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
> +}
> +
> /*
> * Configure virtio device
> * It returns 0 on success.
> @@ -1869,6 +1885,9 @@ virtio_dev_configure(struct rte_eth_dev *dev)
> return -ENOTSUP;
> }
>
> + hw->has_tx_offload = !!tx_offload_enabled(hw);
> + hw->has_rx_offload = !!rx_offload_enabled(hw);
> +
> if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
> /* Enable vector (0) for Link State Intrerrupt */
> if (VTPCI_OPS(hw)->set_config_irq(hw, 0) ==
> diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
> index a28ba8339..e0bb871f2 100644
> --- a/drivers/net/virtio/virtio_pci.h
> +++ b/drivers/net/virtio/virtio_pci.h
> @@ -233,6 +233,8 @@ struct virtio_hw {
> uint8_t modern;
> uint8_t use_simple_rx;
> uint8_t use_simple_tx;
> + uint8_t has_tx_offload;
> + uint8_t has_rx_offload;
I think it's better to use spaces, instead of
4 spaces width tabs after uint8_t.
> uint16_tport_id;
> uint8_t mac_addr[ETHER_ADDR_LEN];
> uint32_tnotify_off_multiplier;
> diff --git a/drivers/net/virtio/virtio_rxtx.c
> b/drivers/net/virtio/virtio_rxtx.c
> index 92fab2174..3f113a118 100644
> --- a/drivers/net/virtio/virtio_rxtx.c
> +++ b/drivers/net/virtio/virtio_rxtx.c
> @@ -225,13 +225,6 @@ virtio_tso_fix_cksum(struct rte_mbuf *m)
> }
> }
>
> -static inline int
> -tx_offload_enabled(struct virtio_hw *hw)
> -{
> - return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
> - vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
> - vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
> -}
>
> /* avoid write operation when necessary, to lessen cache issues */
> #define ASSIGN_UNLESS_EQUAL(var, val) do { \
> @@ -253,13 +246,15 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct
> rte_mbuf *cookie,
> struct virtio_net_hdr *hdr;
> int offload;
>
> - offload = tx_offload_enabled(vq->hw);
> head_idx = vq->vq_desc_head_idx;
> idx = head_idx;
> dxp = &vq->vq_descx[idx];
> dxp->cookie = (void *)cookie;
> dxp->ndescs = needed;
>
> + offload = vq->hw->has_tx_offload &&
> + (cookie->ol_flags & PKT_TX_OFFLOAD_MASK);
If features aren't negotiated, I think there is no need to
check ol_flags and update the net header.
> +
> start_dp = vq->vq_ring.desc;
>
> if (can_push) {
> @@ -270,7 +265,6 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct
> rte_mbuf *cookie,
>* which is wrong. Below subtract restores correct pkt size.
>*/
> cookie->pkt_len -= head_size;
> - /* if offload disabled, it is not zeroed below, do it now */
> if (offload == 0) {
> ASSIGN_UNLESS_EQUAL(hdr->csum_start, 0);
> ASSIGN_UNLESS_EQUAL(hdr->csum_offset, 0);
> @@ -686,14 +680,6 @@ virtio_rx_offload(struct rte_mbuf *m, struct
> virtio_net_hdr *hdr)
> return 0;
> }
[...]
Best regards,
Tiwei Bie
Re: [dpdk-dev] [PATCH 3/4] net/virtio: don't use simple Rx path if TCP LRO requested
On Fri, Jun 01, 2018 at 02:47:57PM +0200, Maxime Coquelin wrote: > Signed-off-by: Maxime Coquelin > --- > drivers/net/virtio/virtio_ethdev.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/virtio/virtio_ethdev.c > b/drivers/net/virtio/virtio_ethdev.c > index 5730620ed..d481b282e 100644 > --- a/drivers/net/virtio/virtio_ethdev.c > +++ b/drivers/net/virtio/virtio_ethdev.c > @@ -1893,7 +1893,8 @@ virtio_dev_configure(struct rte_eth_dev *dev) > } > > if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM | > -DEV_RX_OFFLOAD_TCP_CKSUM)) > +DEV_RX_OFFLOAD_TCP_CKSUM | > +DEV_RX_OFFLOAD_TCP_LRO)) Maybe we also need to consider not using simple Rx when DEV_RX_OFFLOAD_VLAN_STRIP is requested. Best regards, Tiwei Bie > hw->use_simple_rx = 0; > > if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM | > -- > 2.14.3 >
[dpdk-dev] [PATCH v2 02/26] net/ena: remove support of legacy LLQ
From: Rafal Kozik
The legacy LLQ should no longer be supported by the drivers, as this API
is deprecated. Because of that, it was removed from the driver.
Signed-off-by: Rafal Kozik
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 29 +
1 file changed, 5 insertions(+), 24 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 40e669613..171655d33 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1327,16 +1327,11 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
adapter->regs = pci_dev->mem_resource[ENA_REGS_BAR].addr;
adapter->dev_mem_base = pci_dev->mem_resource[ENA_MEM_BAR].addr;
- /* Present ENA_MEM_BAR indicates available LLQ mode.
-* Use corresponding policy
-*/
- if (adapter->dev_mem_base)
- ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_DEV;
- else if (adapter->regs)
- ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
- else
+ if (!adapter->regs) {
PMD_INIT_LOG(CRIT, "Failed to access registers BAR(%d)",
ENA_REGS_BAR);
+ return -ENXIO;
+ }
ena_dev->reg_bar = adapter->regs;
ena_dev->dmadev = adapter->pdev;
@@ -1353,22 +1348,8 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
return -1;
}
- if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
- if (get_feat_ctx.max_queues.max_llq_num == 0) {
- PMD_INIT_LOG(ERR,
-"Trying to use LLQ but llq_num is 0.\n"
-"Fall back into regular queues.");
- ena_dev->tx_mem_queue_type =
- ENA_ADMIN_PLACEMENT_POLICY_HOST;
- adapter->num_queues =
- get_feat_ctx.max_queues.max_sq_num;
- } else {
- adapter->num_queues =
- get_feat_ctx.max_queues.max_llq_num;
- }
- } else {
- adapter->num_queues = get_feat_ctx.max_queues.max_sq_num;
- }
+ ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+ adapter->num_queues = get_feat_ctx.max_queues.max_sq_num;
queue_size = ena_calc_queue_size(ena_dev, &get_feat_ctx);
if ((queue_size <= 0) || (adapter->num_queues <= 0))
--
2.14.1
[dpdk-dev] [PATCH v2 00/26] net/ena: new features and fixes
The ENA driver was updated with the new features and few fixes and minor changes are introduced. First of all, the communication layer which is delivered by vendor was updated - the version in the HEAD is a bit outdated now. ENA is able to communicate with the driver through Admin queue by using admin interrupts instead of polling. Admin interrupts are also used for handling AENQ events, which are used for the following new features: - LSC handlers - watchdog and device rest - monitoring the admin queue - handling ENA notifications (getting hints from device) For the watchdog and admin queue monitoring, the timers had to be used, so the makefile was modified to do not cut out the librte_timer. >From other fixes and changes: - legacy LLQ was removed which is now deprecated API - Rx out of order completion was added to enable cleaning up packets out of order - Tx mbufs are now linearized if they exceed supported number of segments - pass information about maximum number of Tx and Rx descriptors - the IO queue number is now taking into consideration maximum number of sq and cq - Tx id requested for sending is now being validated and the reset is being triggered if it is invalid - branch predictioning was added for better performance - error checking and returned values were fixed - macros for allocating memory in communication layer were fixed - information about numa mode is now being passed to the NIC --- v2: * Rebased on top of dpdk-next-net * Added link speed patch * Added fix when allocating coherent memory in the PMD Michal Krawczyk (11): net/ena: update ena_com to the newer version net/ena: add interrupt handler for admin queue net/ena: add stop and uninit routines net/ena: add LSC intr support and AENQ handling net/ena: restart only initialized queues instead of all net/ena: add reset routine mk: link librte_timer with --whole-archive net/ena: add watchdog and keep alive AENQ handler net/ena: add checking for admin queue state net/ena: add RX out of order completion net/ena: store handle when allocating coherent memory node Rafal Kozik (15): net/ena: remove support of legacy LLQ net/ena: handle ENA notification net/ena: make watchdog configurable net/ena: linearize Tx mbuf net/ena: add info about max number of Tx/Rx descriptors net/ena: unimplemented handler error net/ena: rework configuration of IO queue numbers net/ena: validate Tx req id net/ena: add (un)likely statements net/ena: adjust error checking and cleaning net/ena: update numa node net/ena: check pointer before memset net/ena: change memory type net/ena: fix GENMASK_ULL macro net/ena: set link speed as none drivers/net/ena/base/ena_com.c | 717 +++--- drivers/net/ena/base/ena_com.h | 112 +-- drivers/net/ena/base/ena_defs/ena_admin_defs.h | 1164 +++ drivers/net/ena/base/ena_defs/ena_common_defs.h |8 +- drivers/net/ena/base/ena_defs/ena_eth_io_defs.h | 758 +-- drivers/net/ena/base/ena_defs/ena_gen_info.h|4 +- drivers/net/ena/base/ena_defs/ena_includes.h|2 - drivers/net/ena/base/ena_defs/ena_regs_defs.h | 36 + drivers/net/ena/base/ena_eth_com.c | 78 +- drivers/net/ena/base/ena_eth_com.h | 10 +- drivers/net/ena/base/ena_plat.h |2 - drivers/net/ena/base/ena_plat_dpdk.h| 74 +- drivers/net/ena/ena_ethdev.c| 718 +++--- drivers/net/ena/ena_ethdev.h| 32 +- mk/rte.app.mk |1 + 15 files changed, 1709 insertions(+), 2007 deletions(-) -- 2.14.1
[dpdk-dev] [PATCH v2 03/26] net/ena: add interrupt handler for admin queue
The ENA device is able to send MSI-X when it will complete an command
when polling mode is deactivated. Further, the same interrupt handler
will be used for the AENQ handling - services of the ENA device, that
allows to implement watchdog or lsc handler.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 23 +++
1 file changed, 23 insertions(+)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 171655d33..400f1eff8 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -246,6 +246,7 @@ static int ena_rss_reta_query(struct rte_eth_dev *dev,
struct rte_eth_rss_reta_entry64 *reta_conf,
uint16_t reta_size);
static int ena_get_sset_count(struct rte_eth_dev *dev, int sset);
+static void ena_interrupt_handler_rte(void *cb_arg);
static const struct eth_dev_ops ena_dev_ops = {
.dev_configure= ena_dev_configure,
@@ -459,9 +460,16 @@ static void ena_close(struct rte_eth_dev *dev)
{
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
+ struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+ struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
adapter->state = ENA_ADAPTER_STATE_STOPPED;
+ rte_intr_disable(intr_handle);
+ rte_intr_callback_unregister(intr_handle,
+ena_interrupt_handler_rte,
+adapter);
+
ena_rx_queue_release_all(dev);
ena_tx_queue_release_all(dev);
}
@@ -908,6 +916,8 @@ static int ena_start(struct rte_eth_dev *dev)
{
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
+ struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+ struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
int rc = 0;
if (!(adapter->state == ENA_ADAPTER_STATE_CONFIG ||
@@ -937,6 +947,11 @@ static int ena_start(struct rte_eth_dev *dev)
ena_stats_restart(dev);
+ rte_intr_callback_register(intr_handle,
+ ena_interrupt_handler_rte,
+ adapter);
+ rte_intr_enable(intr_handle);
+
adapter->state = ENA_ADAPTER_STATE_RUNNING;
return 0;
@@ -1291,6 +1306,14 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
return rc;
}
+static void ena_interrupt_handler_rte(__rte_unused void *cb_arg)
+{
+ struct ena_adapter *adapter = (struct ena_adapter *)cb_arg;
+ struct ena_com_dev *ena_dev = &adapter->ena_dev;
+
+ ena_com_admin_q_comp_intr_handler(ena_dev);
+}
+
static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
{
struct rte_pci_device *pci_dev;
--
2.14.1
[dpdk-dev] [PATCH v2 04/26] net/ena: add stop and uninit routines
Lack of uninit routine could lead to memory leak.
Stop was added to fulfill allowed PMD operations.
Checks for the PMD states in the start and configure routine were
removed, as the upper layer is already checking for them.
The interrupt setup was moved from start to init function.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 95 +---
drivers/net/ena/ena_ethdev.h | 3 +-
2 files changed, 56 insertions(+), 42 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 400f1eff8..6e228e050 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -223,6 +223,7 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count);
static void ena_init_rings(struct ena_adapter *adapter);
static int ena_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
static int ena_start(struct rte_eth_dev *dev);
+static void ena_stop(struct rte_eth_dev *dev);
static void ena_close(struct rte_eth_dev *dev);
static int ena_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats);
static void ena_rx_queue_release_all(struct rte_eth_dev *dev);
@@ -254,6 +255,7 @@ static const struct eth_dev_ops ena_dev_ops = {
.rx_queue_setup = ena_rx_queue_setup,
.tx_queue_setup = ena_tx_queue_setup,
.dev_start= ena_start,
+ .dev_stop = ena_stop,
.link_update = ena_link_update,
.stats_get= ena_stats_get,
.mtu_set = ena_mtu_set,
@@ -460,15 +462,9 @@ static void ena_close(struct rte_eth_dev *dev)
{
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
- struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
- struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-
- adapter->state = ENA_ADAPTER_STATE_STOPPED;
- rte_intr_disable(intr_handle);
- rte_intr_callback_unregister(intr_handle,
-ena_interrupt_handler_rte,
-adapter);
+ ena_stop(dev);
+ adapter->state = ENA_ADAPTER_STATE_CLOSED;
ena_rx_queue_release_all(dev);
ena_tx_queue_release_all(dev);
@@ -916,16 +912,8 @@ static int ena_start(struct rte_eth_dev *dev)
{
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
- struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
- struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
int rc = 0;
- if (!(adapter->state == ENA_ADAPTER_STATE_CONFIG ||
- adapter->state == ENA_ADAPTER_STATE_STOPPED)) {
- PMD_INIT_LOG(ERR, "API violation");
- return -1;
- }
-
rc = ena_check_valid_conf(adapter);
if (rc)
return rc;
@@ -947,16 +935,19 @@ static int ena_start(struct rte_eth_dev *dev)
ena_stats_restart(dev);
- rte_intr_callback_register(intr_handle,
- ena_interrupt_handler_rte,
- adapter);
- rte_intr_enable(intr_handle);
-
adapter->state = ENA_ADAPTER_STATE_RUNNING;
return 0;
}
+static void ena_stop(struct rte_eth_dev *dev)
+{
+ struct ena_adapter *adapter =
+ (struct ena_adapter *)(dev->data->dev_private);
+
+ adapter->state = ENA_ADAPTER_STATE_STOPPED;
+}
+
static int ena_queue_restart(struct ena_ring *ring)
{
int rc, bufs_num;
@@ -1317,6 +1308,7 @@ static void ena_interrupt_handler_rte(__rte_unused void
*cb_arg)
static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
{
struct rte_pci_device *pci_dev;
+ struct rte_intr_handle *intr_handle;
struct ena_adapter *adapter =
(struct ena_adapter *)(eth_dev->data->dev_private);
struct ena_com_dev *ena_dev = &adapter->ena_dev;
@@ -1347,6 +1339,8 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
pci_dev->addr.devid,
pci_dev->addr.function);
+ intr_handle = &pci_dev->intr_handle;
+
adapter->regs = pci_dev->mem_resource[ENA_REGS_BAR].addr;
adapter->dev_mem_base = pci_dev->mem_resource[ENA_MEM_BAR].addr;
@@ -1406,36 +1400,55 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
return -ENOMEM;
}
+ rte_intr_callback_register(intr_handle,
+ ena_interrupt_handler_rte,
+ adapter);
+ rte_intr_enable(intr_handle);
+ ena_com_set_admin_polling_mode(ena_dev, false);
+
adapters_found++;
adapter->state = ENA_ADAPTER_STATE_INIT;
return 0;
}
+static int eth_ena_dev_uninit(struct rte_eth_dev *eth_dev)
+{
+ struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
+ struct rte_intr_handle *intr_handle
[dpdk-dev] [PATCH v2 09/26] mk: link librte_timer with --whole-archive
ENA PMD is only PMD which is using librte_timer. In linking routine rte_timer is linked before PMDs, so if timers are not being used by the applications, they are just cut out. Signed-off-by: Michal Krawczyk --- mk/rte.app.mk | 1 + 1 file changed, 1 insertion(+) diff --git a/mk/rte.app.mk b/mk/rte.app.mk index 1e32c83e7..c70bc254e 100644 --- a/mk/rte.app.mk +++ b/mk/rte.app.mk @@ -60,6 +60,7 @@ endif _LDLIBS-y += --whole-archive +_LDLIBS-$(CONFIG_RTE_LIBRTE_TIMER) += -lrte_timer _LDLIBS-$(CONFIG_RTE_LIBRTE_CFGFILE)+= -lrte_cfgfile _LDLIBS-$(CONFIG_RTE_LIBRTE_HASH) += -lrte_hash _LDLIBS-$(CONFIG_RTE_LIBRTE_MEMBER) += -lrte_member -- 2.14.1
[dpdk-dev] [PATCH v2 08/26] net/ena: add reset routine
Reset routine can be used by the DPDK application to reset the device in
case of receiving RTE_ETH_EVENT_INTR_RESET from the PMD.
The reset event is not triggered by the driver, yet. It will be added in
next commits to enable error recovery in case of misfunctioning of the
device.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 64 +++-
drivers/net/ena/ena_ethdev.h | 2 ++
2 files changed, 65 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index fc4776327..ce05883d9 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -225,6 +225,7 @@ static int ena_mtu_set(struct rte_eth_dev *dev, uint16_t
mtu);
static int ena_start(struct rte_eth_dev *dev);
static void ena_stop(struct rte_eth_dev *dev);
static void ena_close(struct rte_eth_dev *dev);
+static int ena_dev_reset(struct rte_eth_dev *dev);
static int ena_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats);
static void ena_rx_queue_release_all(struct rte_eth_dev *dev);
static void ena_tx_queue_release_all(struct rte_eth_dev *dev);
@@ -262,6 +263,7 @@ static const struct eth_dev_ops ena_dev_ops = {
.rx_queue_release = ena_rx_queue_release,
.tx_queue_release = ena_tx_queue_release,
.dev_close= ena_close,
+ .dev_reset= ena_dev_reset,
.reta_update = ena_rss_reta_update,
.reta_query = ena_rss_reta_query,
};
@@ -470,6 +472,63 @@ static void ena_close(struct rte_eth_dev *dev)
ena_tx_queue_release_all(dev);
}
+static int
+ena_dev_reset(struct rte_eth_dev *dev)
+{
+ struct rte_mempool *mb_pool_rx[ENA_MAX_NUM_QUEUES];
+ struct rte_eth_dev *eth_dev;
+ struct rte_pci_device *pci_dev;
+ struct rte_intr_handle *intr_handle;
+ struct ena_com_dev *ena_dev;
+ struct ena_com_dev_get_features_ctx get_feat_ctx;
+ struct ena_adapter *adapter;
+ int nb_queues;
+ int rc, i;
+
+ adapter = (struct ena_adapter *)(dev->data->dev_private);
+ ena_dev = &adapter->ena_dev;
+ eth_dev = adapter->rte_dev;
+ pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
+ intr_handle = &pci_dev->intr_handle;
+ nb_queues = eth_dev->data->nb_rx_queues;
+
+ ena_com_set_admin_running_state(ena_dev, false);
+
+ ena_com_dev_reset(ena_dev, adapter->reset_reason);
+
+ for (i = 0; i < nb_queues; i++)
+ mb_pool_rx[i] = adapter->rx_ring[i].mb_pool;
+
+ ena_rx_queue_release_all(eth_dev);
+ ena_tx_queue_release_all(eth_dev);
+
+ rte_intr_disable(intr_handle);
+
+ ena_com_abort_admin_commands(ena_dev);
+ ena_com_wait_for_abort_completion(ena_dev);
+ ena_com_admin_destroy(ena_dev);
+ ena_com_mmio_reg_read_request_destroy(ena_dev);
+
+ rc = ena_device_init(ena_dev, &get_feat_ctx);
+ if (rc) {
+ PMD_INIT_LOG(CRIT, "Cannot initialize device\n");
+ return rc;
+ }
+
+ rte_intr_enable(intr_handle);
+ ena_com_set_admin_polling_mode(ena_dev, false);
+ ena_com_admin_aenq_enable(ena_dev);
+
+ for (i = 0; i < nb_queues; ++i)
+ ena_rx_queue_setup(eth_dev, i, adapter->rx_ring_size, 0, NULL,
+ mb_pool_rx[i]);
+
+ for (i = 0; i < nb_queues; ++i)
+ ena_tx_queue_setup(eth_dev, i, adapter->tx_ring_size, 0, NULL);
+
+ return 0;
+}
+
static int ena_rss_reta_update(struct rte_eth_dev *dev,
struct rte_eth_rss_reta_entry64 *reta_conf,
uint16_t reta_size)
@@ -1074,7 +1133,10 @@ static int ena_tx_queue_setup(struct rte_eth_dev *dev,
for (i = 0; i < txq->ring_size; i++)
txq->empty_tx_reqs[i] = i;
- txq->offloads = tx_conf->offloads | dev->data->dev_conf.txmode.offloads;
+ if (tx_conf != NULL) {
+ txq->offloads =
+ tx_conf->offloads | dev->data->dev_conf.txmode.offloads;
+ }
/* Store pointer to this queue in upper layer */
txq->configured = 1;
diff --git a/drivers/net/ena/ena_ethdev.h b/drivers/net/ena/ena_ethdev.h
index 16172a54a..79e9e655d 100644
--- a/drivers/net/ena/ena_ethdev.h
+++ b/drivers/net/ena/ena_ethdev.h
@@ -183,6 +183,8 @@ struct ena_adapter {
uint64_t rx_selected_offloads;
bool link_status;
+
+ enum ena_regs_reset_reason_types reset_reason;
};
#endif /* _ENA_ETHDEV_H_ */
--
2.14.1
[dpdk-dev] [PATCH v2 06/26] net/ena: handle ENA notification
From: Rafal Kozik
When ENA notifications are provided ena_notification handler is called.
It checks if received value is not corrupted and if necessary it
reports proper warnings.
Data received from NIC is parsed in ena_update_hints. Fields for
storing those information was added to ena_adapter structure.
ENA notification are enabled by setting ENA_ADMIN_NOTIFICATION flag in
aenq_groups.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 41 +++--
1 file changed, 39 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 0d1b7c1ed..ae13916d1 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1290,7 +1290,8 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
goto err_admin_init;
}
- aenq_groups = BIT(ENA_ADMIN_LINK_CHANGE);
+ aenq_groups = BIT(ENA_ADMIN_LINK_CHANGE) |
+ BIT(ENA_ADMIN_NOTIFICATION);
aenq_groups &= get_feat_ctx->aenq.supported_groups;
rc = ena_com_set_aenq_config(ena_dev, aenq_groups);
@@ -1724,6 +1725,19 @@ eth_ena_prep_pkts(void *tx_queue, struct rte_mbuf
**tx_pkts,
return i;
}
+static void ena_update_hints(struct ena_adapter *adapter,
+struct ena_admin_ena_hw_hints *hints)
+{
+ if (hints->admin_completion_tx_timeout)
+ adapter->ena_dev.admin_queue.completion_timeout =
+ hints->admin_completion_tx_timeout * 1000;
+
+ if (hints->mmio_read_timeout)
+ /* convert to usec */
+ adapter->ena_dev.mmio_read.reg_read_to =
+ hints->mmio_read_timeout * 1000;
+}
+
static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
{
@@ -1918,6 +1932,29 @@ static void ena_update_on_link_change(void *adapter_data,
_rte_eth_dev_callback_process(eth_dev, RTE_ETH_EVENT_INTR_LSC, NULL);
}
+static void ena_notification(void *data,
+struct ena_admin_aenq_entry *aenq_e)
+{
+ struct ena_adapter *adapter = (struct ena_adapter *)data;
+ struct ena_admin_ena_hw_hints *hints;
+
+ if (aenq_e->aenq_common_desc.group != ENA_ADMIN_NOTIFICATION)
+ RTE_LOG(WARNING, PMD, "Invalid group(%x) expected %x\n",
+ aenq_e->aenq_common_desc.group,
+ ENA_ADMIN_NOTIFICATION);
+
+ switch (aenq_e->aenq_common_desc.syndrom) {
+ case ENA_ADMIN_UPDATE_HINTS:
+ hints = (struct ena_admin_ena_hw_hints *)
+ (&aenq_e->inline_data_w4);
+ ena_update_hints(adapter, hints);
+ break;
+ default:
+ RTE_LOG(ERR, PMD, "Invalid aenq notification link state %d\n",
+ aenq_e->aenq_common_desc.syndrom);
+ }
+}
+
/**
* This handler will called for unknown event group or unimplemented handlers
**/
@@ -1930,7 +1967,7 @@ static void unimplemented_aenq_handler(__rte_unused void
*data,
static struct ena_aenq_handlers aenq_handlers = {
.handlers = {
[ENA_ADMIN_LINK_CHANGE] = ena_update_on_link_change,
- [ENA_ADMIN_NOTIFICATION] = unimplemented_aenq_handler,
+ [ENA_ADMIN_NOTIFICATION] = ena_notification,
[ENA_ADMIN_KEEP_ALIVE] = unimplemented_aenq_handler
},
.unimplemented_handler = unimplemented_aenq_handler
--
2.14.1
[dpdk-dev] [PATCH v2 05/26] net/ena: add LSC intr support and AENQ handling
To make the LSC interrupt working, the AENQ must be configured properly
in the ENA device. The AENQ interrupt is common for all maintance
interrupts - the proper handler is then executed depending on the
received descriptor.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 52 ++--
drivers/net/ena/ena_ethdev.h | 2 ++
2 files changed, 47 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 6e228e050..0d1b7c1ed 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -201,7 +201,7 @@ static const struct rte_pci_id pci_id_ena_map[] = {
{ .device_id = 0 },
};
-static struct ena_aenq_handlers empty_aenq_handlers;
+static struct ena_aenq_handlers aenq_handlers;
static int ena_device_init(struct ena_com_dev *ena_dev,
struct ena_com_dev_get_features_ctx *get_feat_ctx);
@@ -732,8 +732,11 @@ static int ena_link_update(struct rte_eth_dev *dev,
__rte_unused int wait_to_complete)
{
struct rte_eth_link *link = &dev->data->dev_link;
+ struct ena_adapter *adapter;
+
+ adapter = (struct ena_adapter *)(dev->data->dev_private);
- link->link_status = ETH_LINK_UP;
+ link->link_status = adapter->link_status ? ETH_LINK_UP : ETH_LINK_DOWN;
link->link_speed = ETH_SPEED_NUM_10G;
link->link_duplex = ETH_LINK_FULL_DUPLEX;
@@ -1228,6 +1231,7 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count)
static int ena_device_init(struct ena_com_dev *ena_dev,
struct ena_com_dev_get_features_ctx *get_feat_ctx)
{
+ uint32_t aenq_groups;
int rc;
bool readless_supported;
@@ -1263,7 +1267,7 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
ena_dev->dma_addr_bits = ena_com_get_dma_width(ena_dev);
/* ENA device administration layer init */
- rc = ena_com_admin_init(ena_dev, &empty_aenq_handlers, true);
+ rc = ena_com_admin_init(ena_dev, &aenq_handlers, true);
if (rc) {
RTE_LOG(ERR, PMD,
"cannot initialize ena admin queue with device\n");
@@ -1286,6 +1290,15 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
goto err_admin_init;
}
+ aenq_groups = BIT(ENA_ADMIN_LINK_CHANGE);
+
+ aenq_groups &= get_feat_ctx->aenq.supported_groups;
+ rc = ena_com_set_aenq_config(ena_dev, aenq_groups);
+ if (rc) {
+ RTE_LOG(ERR, PMD, "Cannot configure aenq groups rc: %d\n", rc);
+ goto err_admin_init;
+ }
+
return 0;
err_admin_init:
@@ -1297,12 +1310,14 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
return rc;
}
-static void ena_interrupt_handler_rte(__rte_unused void *cb_arg)
+static void ena_interrupt_handler_rte(void *cb_arg)
{
struct ena_adapter *adapter = (struct ena_adapter *)cb_arg;
struct ena_com_dev *ena_dev = &adapter->ena_dev;
ena_com_admin_q_comp_intr_handler(ena_dev);
+ if (adapter->state != ENA_ADAPTER_STATE_CLOSED)
+ ena_com_aenq_intr_handler(ena_dev, adapter);
}
static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
@@ -1405,6 +1420,7 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
adapter);
rte_intr_enable(intr_handle);
ena_com_set_admin_polling_mode(ena_dev, false);
+ ena_com_admin_aenq_enable(ena_dev);
adapters_found++;
adapter->state = ENA_ADAPTER_STATE_INIT;
@@ -1842,6 +1858,9 @@ static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct
rte_mbuf **tx_pkts,
return sent_idx;
}
+/*
+ * PMD configuration
+ */
static int eth_ena_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
struct rte_pci_device *pci_dev)
{
@@ -1856,7 +1875,7 @@ static int eth_ena_pci_remove(struct rte_pci_device
*pci_dev)
static struct rte_pci_driver rte_ena_pmd = {
.id_table = pci_id_ena_map,
- .drv_flags = RTE_PCI_DRV_NEED_MAPPING,
+ .drv_flags = RTE_PCI_DRV_NEED_MAPPING | RTE_PCI_DRV_INTR_LSC,
.probe = eth_ena_pci_probe,
.remove = eth_ena_pci_remove,
};
@@ -1880,6 +1899,25 @@ ena_init_log(void)
/**
AENQ Handlers ***
*/
+static void ena_update_on_link_change(void *adapter_data,
+ struct ena_admin_aenq_entry *aenq_e)
+{
+ struct rte_eth_dev *eth_dev;
+ struct ena_adapter *adapter;
+ struct ena_admin_aenq_link_change_desc *aenq_link_desc;
[dpdk-dev] [PATCH v2 10/26] net/ena: add watchdog and keep alive AENQ handler
Keep alive is executing AENQ interrupt periodically. It allows to check
health of the device and trigger reset event if the device will stop
responding.
To check for the state of the device, the DPDK application must call
rte_timer_manage().
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 58 ++--
drivers/net/ena/ena_ethdev.h | 9 +++
2 files changed, 65 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index ce05883d9..98f1b6c0a 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -249,6 +249,7 @@ static int ena_rss_reta_query(struct rte_eth_dev *dev,
uint16_t reta_size);
static int ena_get_sset_count(struct rte_eth_dev *dev, int sset);
static void ena_interrupt_handler_rte(void *cb_arg);
+static void ena_timer_wd_callback(struct rte_timer *timer, void *arg);
static const struct eth_dev_ops ena_dev_ops = {
.dev_configure= ena_dev_configure,
@@ -979,6 +980,7 @@ static int ena_start(struct rte_eth_dev *dev)
{
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
+ uint64_t ticks;
int rc = 0;
rc = ena_check_valid_conf(adapter);
@@ -1002,6 +1004,13 @@ static int ena_start(struct rte_eth_dev *dev)
ena_stats_restart(dev);
+ adapter->timestamp_wd = rte_get_timer_cycles();
+ adapter->keep_alive_timeout = ENA_DEVICE_KALIVE_TIMEOUT;
+
+ ticks = rte_get_timer_hz();
+ rte_timer_reset(&adapter->timer_wd, ticks, PERIODICAL, rte_lcore_id(),
+ ena_timer_wd_callback, adapter);
+
adapter->state = ENA_ADAPTER_STATE_RUNNING;
return 0;
@@ -1012,6 +1021,8 @@ static void ena_stop(struct rte_eth_dev *dev)
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
+ rte_timer_stop_sync(&adapter->timer_wd);
+
adapter->state = ENA_ADAPTER_STATE_STOPPED;
}
@@ -1358,7 +1369,8 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
}
aenq_groups = BIT(ENA_ADMIN_LINK_CHANGE) |
- BIT(ENA_ADMIN_NOTIFICATION);
+ BIT(ENA_ADMIN_NOTIFICATION) |
+ BIT(ENA_ADMIN_KEEP_ALIVE);
aenq_groups &= get_feat_ctx->aenq.supported_groups;
rc = ena_com_set_aenq_config(ena_dev, aenq_groups);
@@ -1388,6 +1400,26 @@ static void ena_interrupt_handler_rte(void *cb_arg)
ena_com_aenq_intr_handler(ena_dev, adapter);
}
+static void ena_timer_wd_callback(__rte_unused struct rte_timer *timer,
+ void *arg)
+{
+ struct ena_adapter *adapter = (struct ena_adapter *)arg;
+ struct rte_eth_dev *dev = adapter->rte_dev;
+
+ if (adapter->keep_alive_timeout == ENA_HW_HINTS_NO_TIMEOUT)
+ return;
+
+ /* Within reasonable timing range no memory barriers are needed */
+ if ((rte_get_timer_cycles() - adapter->timestamp_wd) >=
+ adapter->keep_alive_timeout) {
+ RTE_LOG(ERR, PMD, "The ENA device is not responding - "
+ "performing device reset...");
+ adapter->reset_reason = ENA_REGS_RESET_KEEP_ALIVE_TO;
+ _rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_RESET,
+ NULL);
+ }
+}
+
static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
{
struct rte_pci_device *pci_dev;
@@ -1490,6 +1522,10 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
ena_com_set_admin_polling_mode(ena_dev, false);
ena_com_admin_aenq_enable(ena_dev);
+ if (adapters_found == 0)
+ rte_timer_subsystem_init();
+ rte_timer_init(&adapter->timer_wd);
+
adapters_found++;
adapter->state = ENA_ADAPTER_STATE_INIT;
@@ -1803,6 +1839,16 @@ static void ena_update_hints(struct ena_adapter *adapter,
/* convert to usec */
adapter->ena_dev.mmio_read.reg_read_to =
hints->mmio_read_timeout * 1000;
+
+ if (hints->driver_watchdog_timeout) {
+ if (hints->driver_watchdog_timeout == ENA_HW_HINTS_NO_TIMEOUT)
+ adapter->keep_alive_timeout = ENA_HW_HINTS_NO_TIMEOUT;
+ else
+ // Convert msecs to ticks
+ adapter->keep_alive_timeout =
+ (hints->driver_watchdog_timeout *
+ rte_get_timer_hz()) / 1000;
+ }
}
static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
@@ -2022,6 +2068,14 @@ static void ena_notification(void *data,
}
}
+static void ena_keep_alive(void *adapter_data,
+ __rte_unused struct ena_admin_aenq_entry *aenq_e)
+{
+ struct ena_adapter *adapter = (struct ena_adapter *)adapte
[dpdk-dev] [PATCH v2 07/26] net/ena: restart only initialized queues instead of all
There is no need to check for restart all queues. It is sufficient to
check only previously initialized queues.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 13 +
1 file changed, 9 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index ae13916d1..fc4776327 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -749,13 +749,18 @@ static int ena_queue_restart_all(struct rte_eth_dev *dev,
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
struct ena_ring *queues = NULL;
+ int nb_queues;
int i = 0;
int rc = 0;
- queues = (ring_type == ENA_RING_TYPE_RX) ?
- adapter->rx_ring : adapter->tx_ring;
-
- for (i = 0; i < adapter->num_queues; i++) {
+ if (ring_type == ENA_RING_TYPE_RX) {
+ queues = adapter->rx_ring;
+ nb_queues = dev->data->nb_rx_queues;
+ } else {
+ queues = adapter->tx_ring;
+ nb_queues = dev->data->nb_tx_queues;
+ }
+ for (i = 0; i < nb_queues; i++) {
if (queues[i].configured) {
if (ring_type == ENA_RING_TYPE_RX) {
ena_assert_msg(
--
2.14.1
[dpdk-dev] [PATCH v2 11/26] net/ena: add checking for admin queue state
The admin queue can stop responding or became inactive due to unexpected
behaviour of the device. In that case, the whole device should be
restarted.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 37 +
drivers/net/ena/ena_ethdev.h | 2 ++
2 files changed, 31 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 98f1b6c0a..d8447215a 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -527,6 +527,8 @@ ena_dev_reset(struct rte_eth_dev *dev)
for (i = 0; i < nb_queues; ++i)
ena_tx_queue_setup(eth_dev, i, adapter->tx_ring_size, 0, NULL);
+ adapter->trigger_reset = false;
+
return 0;
}
@@ -1400,21 +1402,40 @@ static void ena_interrupt_handler_rte(void *cb_arg)
ena_com_aenq_intr_handler(ena_dev, adapter);
}
+static void check_for_missing_keep_alive(struct ena_adapter *adapter)
+{
+ if (adapter->keep_alive_timeout == ENA_HW_HINTS_NO_TIMEOUT)
+ return;
+
+ if (unlikely((rte_get_timer_cycles() - adapter->timestamp_wd) >=
+ adapter->keep_alive_timeout)) {
+ RTE_LOG(ERR, PMD, "Keep alive timeout\n");
+ adapter->reset_reason = ENA_REGS_RESET_KEEP_ALIVE_TO;
+ adapter->trigger_reset = true;
+ }
+}
+
+/* Check if admin queue is enabled */
+static void check_for_admin_com_state(struct ena_adapter *adapter)
+{
+ if (unlikely(!ena_com_get_admin_running_state(&adapter->ena_dev))) {
+ RTE_LOG(ERR, PMD, "ENA admin queue is not in running state!\n");
+ adapter->reset_reason = ENA_REGS_RESET_ADMIN_TO;
+ adapter->trigger_reset = true;
+ }
+}
+
static void ena_timer_wd_callback(__rte_unused struct rte_timer *timer,
void *arg)
{
struct ena_adapter *adapter = (struct ena_adapter *)arg;
struct rte_eth_dev *dev = adapter->rte_dev;
- if (adapter->keep_alive_timeout == ENA_HW_HINTS_NO_TIMEOUT)
- return;
+ check_for_missing_keep_alive(adapter);
+ check_for_admin_com_state(adapter);
- /* Within reasonable timing range no memory barriers are needed */
- if ((rte_get_timer_cycles() - adapter->timestamp_wd) >=
- adapter->keep_alive_timeout) {
- RTE_LOG(ERR, PMD, "The ENA device is not responding - "
- "performing device reset...");
- adapter->reset_reason = ENA_REGS_RESET_KEEP_ALIVE_TO;
+ if (unlikely(adapter->trigger_reset)) {
+ RTE_LOG(ERR, PMD, "Trigger reset is on\n");
_rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_RESET,
NULL);
}
diff --git a/drivers/net/ena/ena_ethdev.h b/drivers/net/ena/ena_ethdev.h
index b44cca23e..1f6a7062f 100644
--- a/drivers/net/ena/ena_ethdev.h
+++ b/drivers/net/ena/ena_ethdev.h
@@ -194,6 +194,8 @@ struct ena_adapter {
struct rte_timer timer_wd;
uint64_t timestamp_wd;
uint64_t keep_alive_timeout;
+
+ bool trigger_reset;
};
#endif /* _ENA_ETHDEV_H_ */
--
2.14.1
[dpdk-dev] [PATCH v2 12/26] net/ena: make watchdog configurable
From: Rafal Kozik
Add variable wd_state to make driver functional without keep alive
AENQ handler.
The watchdog will be executed only if the aenq group has keep alive
enabled.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 19 +++
drivers/net/ena/ena_ethdev.h | 2 ++
2 files changed, 17 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index d8447215a..3f72272eb 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -204,7 +204,8 @@ static const struct rte_pci_id pci_id_ena_map[] = {
static struct ena_aenq_handlers aenq_handlers;
static int ena_device_init(struct ena_com_dev *ena_dev,
- struct ena_com_dev_get_features_ctx *get_feat_ctx);
+ struct ena_com_dev_get_features_ctx *get_feat_ctx,
+ bool *wd_state);
static int ena_dev_configure(struct rte_eth_dev *dev);
static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts);
@@ -485,6 +486,7 @@ ena_dev_reset(struct rte_eth_dev *dev)
struct ena_adapter *adapter;
int nb_queues;
int rc, i;
+ bool wd_state;
adapter = (struct ena_adapter *)(dev->data->dev_private);
ena_dev = &adapter->ena_dev;
@@ -510,11 +512,12 @@ ena_dev_reset(struct rte_eth_dev *dev)
ena_com_admin_destroy(ena_dev);
ena_com_mmio_reg_read_request_destroy(ena_dev);
- rc = ena_device_init(ena_dev, &get_feat_ctx);
+ rc = ena_device_init(ena_dev, &get_feat_ctx, &wd_state);
if (rc) {
PMD_INIT_LOG(CRIT, "Cannot initialize device\n");
return rc;
}
+ adapter->wd_state = wd_state;
rte_intr_enable(intr_handle);
ena_com_set_admin_polling_mode(ena_dev, false);
@@ -1309,7 +1312,8 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count)
}
static int ena_device_init(struct ena_com_dev *ena_dev,
- struct ena_com_dev_get_features_ctx *get_feat_ctx)
+ struct ena_com_dev_get_features_ctx *get_feat_ctx,
+ bool *wd_state)
{
uint32_t aenq_groups;
int rc;
@@ -1381,6 +1385,8 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
goto err_admin_init;
}
+ *wd_state = !!(aenq_groups & BIT(ENA_ADMIN_KEEP_ALIVE));
+
return 0;
err_admin_init:
@@ -1404,6 +1410,9 @@ static void ena_interrupt_handler_rte(void *cb_arg)
static void check_for_missing_keep_alive(struct ena_adapter *adapter)
{
+ if (!adapter->wd_state)
+ return;
+
if (adapter->keep_alive_timeout == ENA_HW_HINTS_NO_TIMEOUT)
return;
@@ -1452,6 +1461,7 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
int queue_size, rc;
static int adapters_found;
+ bool wd_state;
memset(adapter, 0, sizeof(struct ena_adapter));
ena_dev = &adapter->ena_dev;
@@ -1495,11 +1505,12 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
adapter->id_number);
/* device specific initialization routine */
- rc = ena_device_init(ena_dev, &get_feat_ctx);
+ rc = ena_device_init(ena_dev, &get_feat_ctx, &wd_state);
if (rc) {
PMD_INIT_LOG(CRIT, "Failed to init ENA device");
return -1;
}
+ adapter->wd_state = wd_state;
ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
adapter->num_queues = get_feat_ctx.max_queues.max_sq_num;
diff --git a/drivers/net/ena/ena_ethdev.h b/drivers/net/ena/ena_ethdev.h
index 1f6a7062f..594e643e2 100644
--- a/drivers/net/ena/ena_ethdev.h
+++ b/drivers/net/ena/ena_ethdev.h
@@ -196,6 +196,8 @@ struct ena_adapter {
uint64_t keep_alive_timeout;
bool trigger_reset;
+
+ bool wd_state;
};
#endif /* _ENA_ETHDEV_H_ */
--
2.14.1
[dpdk-dev] [PATCH v2 13/26] net/ena: add RX out of order completion
This feature allows RX packets to be cleaned up out of order.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 48
drivers/net/ena/ena_ethdev.h | 8 ++--
2 files changed, 50 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 3f72272eb..598c698a3 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -368,6 +368,19 @@ static inline void ena_tx_mbuf_prepare(struct rte_mbuf
*mbuf,
}
}
+static inline int validate_rx_req_id(struct ena_ring *rx_ring, uint16_t req_id)
+{
+ if (likely(req_id < rx_ring->ring_size))
+ return 0;
+
+ RTE_LOG(ERR, PMD, "Invalid rx req_id: %hu\n", req_id);
+
+ rx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_RX_REQ_ID;
+ rx_ring->adapter->trigger_reset = true;
+
+ return -EFAULT;
+}
+
static void ena_config_host_info(struct ena_com_dev *ena_dev)
{
struct ena_admin_host_info *host_info;
@@ -724,6 +737,10 @@ static void ena_rx_queue_release(void *queue)
rte_free(ring->rx_buffer_info);
ring->rx_buffer_info = NULL;
+ if (ring->empty_rx_reqs)
+ rte_free(ring->empty_rx_reqs);
+ ring->empty_rx_reqs = NULL;
+
ring->configured = 0;
RTE_LOG(NOTICE, PMD, "RX Queue %d:%d released\n",
@@ -1176,7 +1193,7 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,
(struct ena_adapter *)(dev->data->dev_private);
struct ena_ring *rxq = NULL;
uint16_t ena_qid = 0;
- int rc = 0;
+ int i, rc = 0;
struct ena_com_dev *ena_dev = &adapter->ena_dev;
rxq = &adapter->rx_ring[queue_idx];
@@ -1242,6 +1259,19 @@ static int ena_rx_queue_setup(struct rte_eth_dev *dev,
return -ENOMEM;
}
+ rxq->empty_rx_reqs = rte_zmalloc("rxq->empty_rx_reqs",
+sizeof(uint16_t) * nb_desc,
+RTE_CACHE_LINE_SIZE);
+ if (!rxq->empty_rx_reqs) {
+ RTE_LOG(ERR, PMD, "failed to alloc mem for empty rx reqs\n");
+ rte_free(rxq->rx_buffer_info);
+ rxq->rx_buffer_info = NULL;
+ return -ENOMEM;
+ }
+
+ for (i = 0; i < nb_desc; i++)
+ rxq->empty_tx_reqs[i] = i;
+
/* Store pointer to this queue in upper layer */
rxq->configured = 1;
dev->data->rx_queues[queue_idx] = rxq;
@@ -1256,7 +1286,7 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count)
uint16_t ring_size = rxq->ring_size;
uint16_t ring_mask = ring_size - 1;
uint16_t next_to_use = rxq->next_to_use;
- uint16_t in_use;
+ uint16_t in_use, req_id;
struct rte_mbuf **mbufs = &rxq->rx_buffer_info[0];
if (unlikely(!count))
@@ -1284,12 +1314,14 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count)
struct ena_com_buf ebuf;
rte_prefetch0(mbufs[((next_to_use + 4) & ring_mask)]);
+
+ req_id = rxq->empty_rx_reqs[next_to_use_masked];
/* prepare physical address for DMA transaction */
ebuf.paddr = mbuf->buf_iova + RTE_PKTMBUF_HEADROOM;
ebuf.len = mbuf->buf_len - RTE_PKTMBUF_HEADROOM;
/* pass resource to device */
rc = ena_com_add_single_rx_desc(rxq->ena_com_io_sq,
- &ebuf, next_to_use_masked);
+ &ebuf, req_id);
if (unlikely(rc)) {
rte_mempool_put_bulk(rxq->mb_pool, (void **)(&mbuf),
count - i);
@@ -1710,6 +1742,7 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct
rte_mbuf **rx_pkts,
unsigned int ring_mask = ring_size - 1;
uint16_t next_to_clean = rx_ring->next_to_clean;
uint16_t desc_in_use = 0;
+ uint16_t req_id;
unsigned int recv_idx = 0;
struct rte_mbuf *mbuf = NULL;
struct rte_mbuf *mbuf_head = NULL;
@@ -1750,7 +1783,12 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct
rte_mbuf **rx_pkts,
break;
while (segments < ena_rx_ctx.descs) {
- mbuf = rx_buff_info[next_to_clean & ring_mask];
+ req_id = ena_rx_ctx.ena_bufs[segments].req_id;
+ rc = validate_rx_req_id(rx_ring, req_id);
+ if (unlikely(rc))
+ break;
+
+ mbuf = rx_buff_info[req_id];
mbuf->data_len = ena_rx_ctx.ena_bufs[segments].len;
mbuf->data_off = RTE_PKTMBUF_HEADROOM;
mbuf->refcnt = 1;
@@ -1767,6 +1805,8 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct
[dpdk-dev] [PATCH v2 14/26] net/ena: linearize Tx mbuf
From: Rafal Kozik
Function ena_check_and_linearize_mbuf check Tx mbuf for number of
segments and linearize (defragment) it if necessary. It is called
before sending each packet.
Information about maximum number of segments is stored per each ring.
Maximum number of segments supported by NIC is taken from ENA COM in
ena_calc_queue_size function and stored in adapter structure.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 31 ++-
drivers/net/ena/ena_ethdev.h | 2 ++
2 files changed, 32 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 598c698a3..4178f9132 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -894,6 +894,7 @@ static int ena_check_valid_conf(struct ena_adapter *adapter)
static int
ena_calc_queue_size(struct ena_com_dev *ena_dev,
+ u16 *max_tx_sgl_size,
struct ena_com_dev_get_features_ctx *get_feat_ctx)
{
uint32_t queue_size = ENA_DEFAULT_RING_SIZE;
@@ -916,6 +917,9 @@ ena_calc_queue_size(struct ena_com_dev *ena_dev,
return -EFAULT;
}
+ *max_tx_sgl_size = RTE_MIN(ENA_PKT_MAX_BUFS,
+ get_feat_ctx->max_queues.max_packet_tx_descs);
+
return queue_size;
}
@@ -1491,6 +1495,7 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
struct ena_com_dev *ena_dev = &adapter->ena_dev;
struct ena_com_dev_get_features_ctx get_feat_ctx;
int queue_size, rc;
+ u16 tx_sgl_size = 0;
static int adapters_found;
bool wd_state;
@@ -1547,13 +1552,15 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
adapter->num_queues = get_feat_ctx.max_queues.max_sq_num;
- queue_size = ena_calc_queue_size(ena_dev, &get_feat_ctx);
+ queue_size = ena_calc_queue_size(ena_dev, &tx_sgl_size, &get_feat_ctx);
if ((queue_size <= 0) || (adapter->num_queues <= 0))
return -EFAULT;
adapter->tx_ring_size = queue_size;
adapter->rx_ring_size = queue_size;
+ adapter->max_tx_sgl_size = tx_sgl_size;
+
/* prepare ring structures */
ena_init_rings(adapter);
@@ -1652,6 +1659,7 @@ static void ena_init_rings(struct ena_adapter *adapter)
ring->id = i;
ring->tx_mem_queue_type = adapter->ena_dev.tx_mem_queue_type;
ring->tx_max_header_size = adapter->ena_dev.tx_max_header_size;
+ ring->sgl_size = adapter->max_tx_sgl_size;
}
for (i = 0; i < adapter->num_queues; i++) {
@@ -1923,6 +1931,23 @@ static void ena_update_hints(struct ena_adapter *adapter,
}
}
+static int ena_check_and_linearize_mbuf(struct ena_ring *tx_ring,
+ struct rte_mbuf *mbuf)
+{
+ int num_segments, rc;
+
+ num_segments = mbuf->nb_segs;
+
+ if (likely(num_segments < tx_ring->sgl_size))
+ return 0;
+
+ rc = rte_pktmbuf_linearize(mbuf);
+ if (unlikely(rc))
+ RTE_LOG(WARNING, PMD, "Mbuf linearize failed\n");
+
+ return rc;
+}
+
static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
{
@@ -1953,6 +1978,10 @@ static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct
rte_mbuf **tx_pkts,
for (sent_idx = 0; sent_idx < nb_pkts; sent_idx++) {
mbuf = tx_pkts[sent_idx];
+ rc = ena_check_and_linearize_mbuf(tx_ring, mbuf);
+ if (unlikely(rc))
+ break;
+
req_id = tx_ring->empty_tx_reqs[next_to_use & ring_mask];
tx_info = &tx_ring->tx_buffer_info[req_id];
tx_info->mbuf = mbuf;
diff --git a/drivers/net/ena/ena_ethdev.h b/drivers/net/ena/ena_ethdev.h
index bba5ad53a..73c110ab9 100644
--- a/drivers/net/ena/ena_ethdev.h
+++ b/drivers/net/ena/ena_ethdev.h
@@ -101,6 +101,7 @@ struct ena_ring {
int configured;
struct ena_adapter *adapter;
uint64_t offloads;
+ u16 sgl_size;
} __rte_cache_aligned;
enum ena_adapter_state {
@@ -167,6 +168,7 @@ struct ena_adapter {
/* TX */
struct ena_ring tx_ring[ENA_MAX_NUM_QUEUES] __rte_cache_aligned;
int tx_ring_size;
+ u16 max_tx_sgl_size;
/* RX */
struct ena_ring rx_ring[ENA_MAX_NUM_QUEUES] __rte_cache_aligned;
--
2.14.1
[dpdk-dev] [PATCH v2 15/26] net/ena: add info about max number of Tx/Rx descriptors
From: Rafal Kozik
In function ena_infos_get driver provides information about minimal
and maximal number of Rx and Tx descriptors.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 13 +
1 file changed, 13 insertions(+)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 4178f9132..3e3ae69bb 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -85,6 +85,9 @@
#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
+#define ENA_MAX_RING_DESC ENA_DEFAULT_RING_SIZE
+#define ENA_MIN_RING_DESC 128
+
enum ethtool_stringset {
ETH_SS_TEST = 0,
ETH_SS_STATS,
@@ -1740,6 +1743,16 @@ static void ena_infos_get(struct rte_eth_dev *dev,
adapter->tx_supported_offloads = tx_feat;
adapter->rx_supported_offloads = rx_feat;
+
+ dev_info->rx_desc_lim.nb_max = ENA_MAX_RING_DESC;
+ dev_info->rx_desc_lim.nb_min = ENA_MIN_RING_DESC;
+
+ dev_info->tx_desc_lim.nb_max = ENA_MAX_RING_DESC;
+ dev_info->tx_desc_lim.nb_min = ENA_MIN_RING_DESC;
+ dev_info->tx_desc_lim.nb_seg_max = RTE_MIN(ENA_PKT_MAX_BUFS,
+ feat.max_queues.max_packet_tx_descs);
+ dev_info->tx_desc_lim.nb_mtu_seg_max = RTE_MIN(ENA_PKT_MAX_BUFS,
+ feat.max_queues.max_packet_tx_descs);
}
static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
--
2.14.1
[dpdk-dev] [PATCH v2 16/26] net/ena: unimplemented handler error
From: Rafal Kozik
Enable AENQ FATAL_ERROR and WARNING callbacks by setting flags
in aenq_groups. They are handled by "unimplemented handler".
If unimplemented handler is called, error is logged.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 3e3ae69bb..2d9b104d6 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1415,7 +1415,9 @@ static int ena_device_init(struct ena_com_dev *ena_dev,
aenq_groups = BIT(ENA_ADMIN_LINK_CHANGE) |
BIT(ENA_ADMIN_NOTIFICATION) |
- BIT(ENA_ADMIN_KEEP_ALIVE);
+ BIT(ENA_ADMIN_KEEP_ALIVE) |
+ BIT(ENA_ADMIN_FATAL_ERROR) |
+ BIT(ENA_ADMIN_WARNING);
aenq_groups &= get_feat_ctx->aenq.supported_groups;
rc = ena_com_set_aenq_config(ena_dev, aenq_groups);
@@ -2196,7 +2198,8 @@ static void ena_keep_alive(void *adapter_data,
static void unimplemented_aenq_handler(__rte_unused void *data,
__rte_unused struct ena_admin_aenq_entry
*aenq_e)
{
- // Unimplemented handler
+ RTE_LOG(ERR, PMD, "Unknown event was received or event with "
+ "unimplemented handler\n");
}
static struct ena_aenq_handlers aenq_handlers = {
--
2.14.1
[dpdk-dev] [PATCH v2 17/26] net/ena: rework configuration of IO queue numbers
From: Rafal Kozik
Move configuration of IO queue numbers to separate function and take
into consideration max number of IO completion queues.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 21 -
1 file changed, 20 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 2d9b104d6..11fe40814 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1491,6 +1491,24 @@ static void ena_timer_wd_callback(__rte_unused struct
rte_timer *timer,
}
}
+static int ena_calc_io_queue_num(__rte_unused struct ena_com_dev *ena_dev,
+struct ena_com_dev_get_features_ctx
*get_feat_ctx)
+{
+ int io_sq_num, io_cq_num, io_queue_num;
+
+ io_sq_num = get_feat_ctx->max_queues.max_sq_num;
+ io_cq_num = get_feat_ctx->max_queues.max_cq_num;
+
+ io_queue_num = RTE_MIN(io_sq_num, io_cq_num);
+
+ if (unlikely(io_queue_num == 0)) {
+ RTE_LOG(ERR, PMD, "Number of IO queues should not be 0\n");
+ return -EFAULT;
+ }
+
+ return io_queue_num;
+}
+
static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
{
struct rte_pci_device *pci_dev;
@@ -1555,7 +1573,8 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
adapter->wd_state = wd_state;
ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
- adapter->num_queues = get_feat_ctx.max_queues.max_sq_num;
+ adapter->num_queues = ena_calc_io_queue_num(ena_dev,
+ &get_feat_ctx);
queue_size = ena_calc_queue_size(ena_dev, &tx_sgl_size, &get_feat_ctx);
if ((queue_size <= 0) || (adapter->num_queues <= 0))
--
2.14.1
[dpdk-dev] [PATCH v2 20/26] net/ena: adjust error checking and cleaning
From: Rafal Kozik
Adjust error checking and cleaning to Linux driver:
* add checking if MTU is to small,
* fix error messages (mismatched Rx and Tx),
* return error received from base driver or proper error
code instead of -1,
* in case of error release occupied resources,
* in case of Rx error trigger NIC reset.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 158 ---
drivers/net/ena/ena_ethdev.h | 2 +
2 files changed, 104 insertions(+), 56 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 4d4210040..a145c2264 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -436,9 +436,12 @@ static void ena_config_host_info(struct ena_com_dev
*ena_dev)
rc = ena_com_set_host_attributes(ena_dev);
if (rc) {
- RTE_LOG(ERR, PMD, "Cannot set host attributes\n");
- if (rc != -ENA_COM_UNSUPPORTED)
- goto err;
+ if (rc == -ENA_COM_UNSUPPORTED)
+ RTE_LOG(WARNING, PMD, "Cannot set host attributes\n");
+ else
+ RTE_LOG(ERR, PMD, "Cannot set host attributes\n");
+
+ goto err;
}
return;
@@ -489,9 +492,12 @@ static void ena_config_debug_area(struct ena_adapter
*adapter)
rc = ena_com_set_host_attributes(&adapter->ena_dev);
if (rc) {
- RTE_LOG(WARNING, PMD, "Cannot set host attributes\n");
- if (rc != -ENA_COM_UNSUPPORTED)
- goto err;
+ if (rc == -ENA_COM_UNSUPPORTED)
+ RTE_LOG(WARNING, PMD, "Cannot set host attributes\n");
+ else
+ RTE_LOG(ERR, PMD, "Cannot set host attributes\n");
+
+ goto err;
}
return;
@@ -534,7 +540,9 @@ ena_dev_reset(struct rte_eth_dev *dev)
ena_com_set_admin_running_state(ena_dev, false);
- ena_com_dev_reset(ena_dev, adapter->reset_reason);
+ rc = ena_com_dev_reset(ena_dev, adapter->reset_reason);
+ if (rc)
+ RTE_LOG(ERR, PMD, "Device reset failed\n");
for (i = 0; i < nb_queues; i++)
mb_pool_rx[i] = adapter->rx_ring[i].mb_pool;
@@ -579,7 +587,7 @@ static int ena_rss_reta_update(struct rte_eth_dev *dev,
struct ena_adapter *adapter =
(struct ena_adapter *)(dev->data->dev_private);
struct ena_com_dev *ena_dev = &adapter->ena_dev;
- int ret, i;
+ int rc, i;
u16 entry_value;
int conf_idx;
int idx;
@@ -591,8 +599,7 @@ static int ena_rss_reta_update(struct rte_eth_dev *dev,
RTE_LOG(WARNING, PMD,
"indirection table %d is bigger than supported (%d)\n",
reta_size, ENA_RX_RSS_TABLE_SIZE);
- ret = -EINVAL;
- goto err;
+ return -EINVAL;
}
for (i = 0 ; i < reta_size ; i++) {
@@ -604,29 +611,28 @@ static int ena_rss_reta_update(struct rte_eth_dev *dev,
if (TEST_BIT(reta_conf[conf_idx].mask, idx)) {
entry_value =
ENA_IO_RXQ_IDX(reta_conf[conf_idx].reta[idx]);
- ret = ena_com_indirect_table_fill_entry(ena_dev,
- i,
- entry_value);
- if (unlikely(ret && (ret != ENA_COM_UNSUPPORTED))) {
+
+ rc = ena_com_indirect_table_fill_entry(ena_dev,
+ i,
+ entry_value);
+ if (unlikely(rc && rc != ENA_COM_UNSUPPORTED)) {
RTE_LOG(ERR, PMD,
"Cannot fill indirect table\n");
- ret = -ENOTSUP;
- goto err;
+ return rc;
}
}
}
- ret = ena_com_indirect_table_set(ena_dev);
- if (unlikely(ret && (ret != ENA_COM_UNSUPPORTED))) {
+ rc = ena_com_indirect_table_set(ena_dev);
+ if (unlikely(rc && rc != ENA_COM_UNSUPPORTED)) {
RTE_LOG(ERR, PMD, "Cannot flush the indirect table\n");
- ret = -ENOTSUP;
- goto err;
+ return rc;
}
RTE_LOG(DEBUG, PMD, "%s(): RSS configured %d entries for port %d\n",
__func__, reta_size, adapter->rte_dev->data->port_id);
-err:
- return ret;
+
+ return 0;
}
/* Query redirection table. */
@@ -637,7 +643,7 @@ static int ena_rss_reta_query(struct rte_eth_dev *dev,
struct ena_adapter *adapter =
(struct ena_adapter *)
[dpdk-dev] [PATCH v2 19/26] net/ena: add (un)likely statements
From: Rafal Kozik
Add likely and unlikely statements to increase performance.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 4b853a964..4d4210040 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -936,7 +936,7 @@ ena_calc_queue_size(struct ena_com_dev *ena_dev,
if (!rte_is_power_of_2(queue_size))
queue_size = rte_align32pow2(queue_size >> 1);
- if (queue_size == 0) {
+ if (unlikely(queue_size == 0)) {
PMD_INIT_LOG(ERR, "Invalid queue size");
return -EFAULT;
}
@@ -1360,7 +1360,7 @@ static int ena_populate_rx_queue(struct ena_ring *rxq,
unsigned int count)
}
/* When we submitted free recources to device... */
- if (i > 0) {
+ if (likely(i > 0)) {
/* ...let HW know that it can fill buffers with data */
rte_wmb();
ena_com_write_sq_doorbell(rxq->ena_com_io_sq);
@@ -1466,7 +1466,7 @@ static void ena_interrupt_handler_rte(void *cb_arg)
struct ena_com_dev *ena_dev = &adapter->ena_dev;
ena_com_admin_q_comp_intr_handler(ena_dev);
- if (adapter->state != ENA_ADAPTER_STATE_CLOSED)
+ if (likely(adapter->state != ENA_ADAPTER_STATE_CLOSED))
ena_com_aenq_intr_handler(ena_dev, adapter);
}
@@ -1856,7 +1856,7 @@ static uint16_t eth_ena_recv_pkts(void *rx_queue, struct
rte_mbuf **rx_pkts,
mbuf->data_off = RTE_PKTMBUF_HEADROOM;
mbuf->refcnt = 1;
mbuf->next = NULL;
- if (segments == 0) {
+ if (unlikely(segments == 0)) {
mbuf->nb_segs = ena_rx_ctx.descs;
mbuf->port = rx_ring->port_id;
mbuf->pkt_len = 0;
--
2.14.1
[dpdk-dev] [PATCH v2 18/26] net/ena: validate Tx req id
From: Rafal Kozik
Validate Tx req id during clearing completed packets.
If id is wrong, trigger NIC reset.
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 25 +
1 file changed, 25 insertions(+)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 11fe40814..4b853a964 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -384,6 +384,27 @@ static inline int validate_rx_req_id(struct ena_ring
*rx_ring, uint16_t req_id)
return -EFAULT;
}
+static int validate_tx_req_id(struct ena_ring *tx_ring, u16 req_id)
+{
+ struct ena_tx_buffer *tx_info = NULL;
+
+ if (likely(req_id < tx_ring->ring_size)) {
+ tx_info = &tx_ring->tx_buffer_info[req_id];
+ if (likely(tx_info->mbuf))
+ return 0;
+ }
+
+ if (tx_info)
+ RTE_LOG(ERR, PMD, "tx_info doesn't have valid mbuf\n");
+ else
+ RTE_LOG(ERR, PMD, "Invalid req_id: %hu\n", req_id);
+
+ /* Trigger device reset */
+ tx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_TX_REQ_ID;
+ tx_ring->adapter->trigger_reset = true;
+ return -EFAULT;
+}
+
static void ena_config_host_info(struct ena_com_dev *ena_dev)
{
struct ena_admin_host_info *host_info;
@@ -2093,6 +2114,10 @@ static uint16_t eth_ena_xmit_pkts(void *tx_queue, struct
rte_mbuf **tx_pkts,
/* Clear complete packets */
while (ena_com_tx_comp_req_id_get(tx_ring->ena_com_io_cq, &req_id) >=
0) {
+ rc = validate_tx_req_id(tx_ring, req_id);
+ if (rc)
+ break;
+
/* Get Tx info & store how many descs were processed */
tx_info = &tx_ring->tx_buffer_info[req_id];
total_tx_descs += tx_info->tx_descs;
--
2.14.1
[dpdk-dev] [PATCH v2 22/26] net/ena: check pointer before memset
From: Rafal Kozik
Need to check if memory allocation succeed before using it.
Using memset on NULL pointer cause segfault.
Fixes: 9ba7981ec992 ("ena: add communication layer for DPDK")
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/base/ena_plat_dpdk.h | 22 --
1 file changed, 16 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ena/base/ena_plat_dpdk.h
b/drivers/net/ena/base/ena_plat_dpdk.h
index e2af4ee2c..d30153e92 100644
--- a/drivers/net/ena/base/ena_plat_dpdk.h
+++ b/drivers/net/ena/base/ena_plat_dpdk.h
@@ -210,10 +210,15 @@ extern uint32_t ena_alloc_cnt;
"ena_alloc_%d", ena_alloc_cnt++); \
mz = rte_memzone_reserve(z_name, size, SOCKET_ID_ANY, \
RTE_MEMZONE_IOVA_CONTIG); \
- memset(mz->addr, 0, size); \
- virt = mz->addr;\
- phys = mz->iova;\
handle = mz;\
+ if (mz == NULL) { \
+ virt = NULL;\
+ phys = 0; \
+ } else {\
+ memset(mz->addr, 0, size); \
+ virt = mz->addr;\
+ phys = mz->iova;\
+ } \
} while (0)
#define ENA_MEM_FREE_COHERENT(dmadev, size, virt, phys, handle)\
({ ENA_TOUCH(size); ENA_TOUCH(phys);\
@@ -230,9 +235,14 @@ extern uint32_t ena_alloc_cnt;
"ena_alloc_%d", ena_alloc_cnt++); \
mz = rte_memzone_reserve(z_name, size, node,\
RTE_MEMZONE_IOVA_CONTIG); \
- memset(mz->addr, 0, size); \
- virt = mz->addr;\
- phys = mz->iova;\
+ if (mz == NULL) { \
+ virt = NULL;\
+ phys = 0; \
+ } else {\
+ memset(mz->addr, 0, size); \
+ virt = mz->addr;\
+ phys = mz->iova;\
+ } \
(void)mem_handle; \
} while (0)
--
2.14.1
[dpdk-dev] [PATCH v2 21/26] net/ena: update numa node
From: Rafal Kozik During initializing tx queues update Non-Uniform Memory Access configuration in NIC firmware. Signed-off-by: Rafal Kozik Acked-by: Michal Krawczyk --- drivers/net/ena/ena_ethdev.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c index a145c2264..d2000f303 100644 --- a/drivers/net/ena/ena_ethdev.c +++ b/drivers/net/ena/ena_ethdev.c @@ -1175,6 +1175,8 @@ static int ena_tx_queue_setup(struct rte_eth_dev *dev, goto err_destroy_io_queue; } + ena_com_update_numa_node(txq->ena_com_io_cq, ctx.numa_node); + txq->port_id = dev->data->port_id; txq->next_to_clean = 0; txq->next_to_use = 0; -- 2.14.1
[dpdk-dev] [PATCH v2 25/26] net/ena: store handle after memory allocation
The pointer received from rte_memzone_reserve from macro
ENA_MEM_ALLOC_COHERENT_NODE was not stored anywhere, and as a result
memory allocated by this macro could not been released.
Signed-off-by: Michal Krawczyk
---
drivers/net/ena/base/ena_plat_dpdk.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ena/base/ena_plat_dpdk.h
b/drivers/net/ena/base/ena_plat_dpdk.h
index 8a04e84b9..900ba1a6b 100644
--- a/drivers/net/ena/base/ena_plat_dpdk.h
+++ b/drivers/net/ena/base/ena_plat_dpdk.h
@@ -237,6 +237,7 @@ extern uint32_t ena_alloc_cnt;
"ena_alloc_%d", ena_alloc_cnt++); \
mz = rte_memzone_reserve(z_name, size, node,\
RTE_MEMZONE_IOVA_CONTIG); \
+ mem_handle = mz;\
if (mz == NULL) { \
virt = NULL;\
phys = 0; \
@@ -245,7 +246,6 @@ extern uint32_t ena_alloc_cnt;
virt = mz->addr;\
phys = mz->iova;\
} \
- (void)mem_handle; \
} while (0)
#define ENA_MEM_ALLOC_NODE(dmadev, size, virt, node, dev_node) \
--
2.14.1
[dpdk-dev] [PATCH v2 26/26] net/ena: set link speed as none
From: Rafal Kozik
Link speed should is not limited to 10Gb/s and it shouldn't be hardcoded.
They link speed is set to none instead and the applications shouldn't
rely on this value when using ENA PMD.
Fixes: 1173fca ("ena: add polling-mode driver")
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/ena_ethdev.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index d2000f303..53b61592b 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -848,7 +848,7 @@ static int ena_link_update(struct rte_eth_dev *dev,
adapter = (struct ena_adapter *)(dev->data->dev_private);
link->link_status = adapter->link_status ? ETH_LINK_UP : ETH_LINK_DOWN;
- link->link_speed = ETH_SPEED_NUM_10G;
+ link->link_speed = ETH_SPEED_NUM_NONE;
link->link_duplex = ETH_LINK_FULL_DUPLEX;
return 0;
--
2.14.1
[dpdk-dev] [PATCH v2 23/26] net/ena: change memory type
From: Rafal Kozik
ENA_MEM_ALLOC_NODE not need to use contiguous physical memory.
Also using memset without checking if allocation succeed can cause
segmentation fault.
To avoid both issue use rte_zmalloc_socket.
Fixes: 3d3edc265fc8 ("net/ena: make coherent memory allocation NUMA-aware")
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/base/ena_plat_dpdk.h | 9 +
1 file changed, 1 insertion(+), 8 deletions(-)
diff --git a/drivers/net/ena/base/ena_plat_dpdk.h
b/drivers/net/ena/base/ena_plat_dpdk.h
index d30153e92..22d7a9cb1 100644
--- a/drivers/net/ena/base/ena_plat_dpdk.h
+++ b/drivers/net/ena/base/ena_plat_dpdk.h
@@ -248,15 +248,8 @@ extern uint32_t ena_alloc_cnt;
#define ENA_MEM_ALLOC_NODE(dmadev, size, virt, node, dev_node) \
do {\
- const struct rte_memzone *mz; \
- char z_name[RTE_MEMZONE_NAMESIZE]; \
ENA_TOUCH(dmadev); ENA_TOUCH(dev_node); \
- snprintf(z_name, sizeof(z_name),\
- "ena_alloc_%d", ena_alloc_cnt++); \
- mz = rte_memzone_reserve(z_name, size, node,\
- RTE_MEMZONE_IOVA_CONTIG); \
- memset(mz->addr, 0, size); \
- virt = mz->addr;\
+ virt = rte_zmalloc_socket(NULL, size, 0, node); \
} while (0)
#define ENA_MEM_ALLOC(dmadev, size) rte_zmalloc(NULL, size, 1)
--
2.14.1
[dpdk-dev] [PATCH v2 24/26] net/ena: fix GENMASK_ULL macro
From: Rafal Kozik
When use GENMASK_ULL(63,0) left shift by 64 bits is performed.
Shifting by number greater or equal then word length
is undefined operation and failed on some platforms.
Fixes: 9ba7981ec992 ("ena: add communication layer for DPDK")
Signed-off-by: Rafal Kozik
Acked-by: Michal Krawczyk
---
drivers/net/ena/base/ena_plat_dpdk.h | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ena/base/ena_plat_dpdk.h
b/drivers/net/ena/base/ena_plat_dpdk.h
index 22d7a9cb1..8a04e84b9 100644
--- a/drivers/net/ena/base/ena_plat_dpdk.h
+++ b/drivers/net/ena/base/ena_plat_dpdk.h
@@ -116,11 +116,13 @@ typedef uint64_t dma_addr_t;
#define ENA_MIN16(x, y) RTE_MIN((x), (y))
#define ENA_MIN8(x, y) RTE_MIN((x), (y))
+#define BITS_PER_LONG_LONG (__SIZEOF_LONG_LONG__ * 8)
#define U64_C(x) x ## ULL
#define BIT(nr) (1UL << (nr))
#define BITS_PER_LONG (__SIZEOF_LONG__ * 8)
#define GENMASK(h, l) (((~0UL) << (l)) & (~0UL >> (BITS_PER_LONG - 1 - (h
-#define GENMASK_ULL(h, l) (((U64_C(1) << ((h) - (l) + 1)) - 1) << (l))
+#define GENMASK_ULL(h, l) (((~0ULL) - (1ULL << (l)) + 1) & \
+ (~0ULL >> (BITS_PER_LONG_LONG - 1 - (h
#ifdef RTE_LIBRTE_ENA_COM_DEBUG
#define ena_trc_dbg(format, arg...)\
--
2.14.1
Re: [dpdk-dev] [PATCH 2/4] net/vhost: improve Tx path selection
On Fri, Jun 01, 2018 at 02:47:56PM +0200, Maxime Coquelin wrote:
> This patch improves the Tx path selection depending on
> whether the application request for offloads, and on whether
> offload features have been negotiated.
>
> When the application doesn't request for Tx offload features,
> the corresponding features bits aren't negotiated.
>
> When Tx offload virtio features have been negotiated, ensure
> the simple Tx path isn't selected.
>
> Signed-off-by: Maxime Coquelin
> ---
> drivers/net/virtio/virtio_ethdev.c | 20 ++--
> drivers/net/virtio/virtio_ethdev.h | 3 ---
> 2 files changed, 18 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/net/virtio/virtio_ethdev.c
> b/drivers/net/virtio/virtio_ethdev.c
> index e68e9d067..5730620ed 100644
> --- a/drivers/net/virtio/virtio_ethdev.c
> +++ b/drivers/net/virtio/virtio_ethdev.c
> @@ -1798,8 +1798,10 @@ static int
> virtio_dev_configure(struct rte_eth_dev *dev)
> {
> const struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
> + const struct rte_eth_txmode *txmode = &dev->data->dev_conf.txmode;
> struct virtio_hw *hw = dev->data->dev_private;
> uint64_t rx_offloads = rxmode->offloads;
> + uint64_t tx_offloads = txmode->offloads;
> uint64_t req_features;
> int ret;
>
> @@ -1821,6 +1823,15 @@ virtio_dev_configure(struct rte_eth_dev *dev)
> (1ULL << VIRTIO_NET_F_GUEST_TSO4) |
> (1ULL << VIRTIO_NET_F_GUEST_TSO6);
>
> + if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
> + DEV_TX_OFFLOAD_UDP_CKSUM))
I think it's better to keep DEV_TX_OFFLOAD_TCP/UDP_CKSUM
aligned, something like:
if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
DEV_TX_OFFLOAD_UDP_CKSUM))
> + req_features |= (1ULL << VIRTIO_NET_F_CSUM);
> +
> + if (tx_offloads & DEV_TX_OFFLOAD_TCP_TSO)
> + req_features |=
> + (1ULL << VIRTIO_NET_F_HOST_TSO4) |
> + (1ULL << VIRTIO_NET_F_HOST_TSO6);
> +
> /* if request features changed, reinit the device */
> if (req_features != hw->req_guest_features) {
> ret = virtio_init_device(dev, req_features);
> @@ -1885,6 +1896,11 @@ virtio_dev_configure(struct rte_eth_dev *dev)
> DEV_RX_OFFLOAD_TCP_CKSUM))
> hw->use_simple_rx = 0;
>
> + if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
> + DEV_TX_OFFLOAD_UDP_CKSUM |
> + DEV_TX_OFFLOAD_TCP_TSO))
Ditto. I think it's better to keep them aligned,
something like:
if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM |
DEV_TX_OFFLOAD_UDP_CKSUM |
DEV_TX_OFFLOAD_TCP_TSO))
Besides, we also need to consider not using simple Tx
when DEV_TX_OFFLOAD_VLAN_INSERT is requested.
Best regards,
Tiwei Bie
> + hw->use_simple_tx = 0;
> +
> return 0;
> }
>
> @@ -2138,14 +2154,14 @@ virtio_dev_info_get(struct rte_eth_dev *dev, struct
> rte_eth_dev_info *dev_info)
>
> dev_info->tx_offload_capa = DEV_TX_OFFLOAD_MULTI_SEGS |
> DEV_TX_OFFLOAD_VLAN_INSERT;
> - if (hw->guest_features & (1ULL << VIRTIO_NET_F_CSUM)) {
> + if (host_features & (1ULL << VIRTIO_NET_F_CSUM)) {
> dev_info->tx_offload_capa |=
> DEV_TX_OFFLOAD_UDP_CKSUM |
> DEV_TX_OFFLOAD_TCP_CKSUM;
> }
> tso_mask = (1ULL << VIRTIO_NET_F_HOST_TSO4) |
> (1ULL << VIRTIO_NET_F_HOST_TSO6);
> - if ((hw->guest_features & tso_mask) == tso_mask)
> + if ((host_features & tso_mask) == tso_mask)
> dev_info->tx_offload_capa |= DEV_TX_OFFLOAD_TCP_TSO;
> }
>
> diff --git a/drivers/net/virtio/virtio_ethdev.h
> b/drivers/net/virtio/virtio_ethdev.h
> index bb40064ea..b603665c7 100644
> --- a/drivers/net/virtio/virtio_ethdev.h
> +++ b/drivers/net/virtio/virtio_ethdev.h
> @@ -28,9 +28,6 @@
>1u << VIRTIO_NET_F_CTRL_VQ | \
>1u << VIRTIO_NET_F_CTRL_RX | \
>1u << VIRTIO_NET_F_CTRL_VLAN | \
> - 1u << VIRTIO_NET_F_CSUM | \
> - 1u << VIRTIO_NET_F_HOST_TSO4 | \
> - 1u << VIRTIO_NET_F_HOST_TSO6 | \
>1u << VIRTIO_NET_F_MRG_RXBUF | \
>1u << VIRTIO_NET_F_MTU | \
>1ULL << VIRTIO_NET_F_GUEST_ANNOUNCE | \
> --
> 2.14.3
>
Re: [dpdk-dev] [PATCH v5 2/2] doc: add a guide doc for cross compiling from x86
-Original Message- > Date: Mon, 4 Jun 2018 06:03:34 + > From: Gavin Hu > To: Jerin Jacob , Bruce Richardson > , Thomas Monjalon > CC: "dev@dpdk.org" > Subject: RE: [dpdk-dev] [PATCH v5 2/2] doc: add a guide doc for cross > compiling from x86 > > See my inline comments: > > > -Original Message- > > From: Jerin Jacob > > Sent: Thursday, May 31, 2018 3:36 AM > > To: Gavin Hu > > Cc: dev@dpdk.org > > Subject: Re: [dpdk-dev] [PATCH v5 2/2] doc: add a guide doc for cross > > compiling from x86 > > > > -Original Message- > > > Date: Tue, 29 May 2018 18:43:36 +0800 > > > From: Gavin Hu > > > To: dev@dpdk.org > > > CC: gavin...@arm.com > > > Subject: [dpdk-dev] [PATCH v5 2/2] doc: add a guide doc for cross > > > compiling from x86 > > > X-Mailer: git-send-email 2.1.4 > > > > > > + 1. EXTRA_CFLAGS and EXTRA_LDFLAGS should be added to include the > > NUMA headers and link the library respectively, > > > + if the step :ref:`argment_the_cross_toolcain_with_numa_support` was > > skipped therefore the toolchain was not > > > + argmented with NUMA support. > > > + > > > + 2. RTE_DEVEL_BUILD has to be disabled, otherwise the numa.h file > > > + gets > > > > If the warnings are from numa.h then please use -isystem > > instead of disabling RTE_DEVEL_BUILD. > > > [Gavin Hu] This is a good advice, I verified it okay and can upload a new > patch. > > > > + a lot of compiling errors of Werror=cast-qual, > > > Werror=strict-prototypes > > and Werror=old-style-definition. > > > + An example is given below: > > > + > > > + .. code-block:: console > > > + > > > + make -j CROSS=aarch64-linux-gnu- CONFIG_RTE_KNI_KMOD=n > > CONFIG_RTE_EAL_IGB_UIO=n > > > + RTE_DEVEL_BUILD=n EXTRA_CFLAGS="-I/include" > > EXTRA_LDFLAGS= > > > + "-L/lib -lnuma" > > > + > > > > As discussed earlier, meson cross build instruction is missing. > > > [Gavin Hu] I reproduced the meson build issue Bruce reported, as shown below. > It was not introduced by gcc, nor clang, it was actually introduced by > meson.build, see line #65 of > http://www.dpdk.org/browse/dpdk/tree/config/meson.build > Even worse, "has_argument" is not reliable(refer here: > http://mesonbuild.com/Compiler-properties.html#has-argument) for some > compilers. > This is the case of gcc and clang, which caused the 4 warning options were > included in the whole project, either the compiler is gcc or clang, cross or > native. > This finally caused the unrecognized warning options. > > I tried to disable the warning options, then the compiling got lots of noisy > warnings and errors. > > To fix this issue, we need to create a meson subproject for pmdinfogen, the > change is not little and I am not familiar with this. > > Any comments are welcome! If I am not wrong, This is specific to host compiler issue with specific version. Right? The build steps will remain same, so as far as this patch concerned, you can add the meson build step in this patch. > > [265/893] Compiling C object > 'buildtools/pmdinfogen/pmdinfogen@exe/pmdinfogen.c.o'. > warning: unknown warning option '-Wno-format-truncation' > [-Wunknown-warning-option] > 1 warning generated. > > > > > > diff --git a/doc/guides/linux_gsg/index.rst > > > b/doc/guides/linux_gsg/index.rst index 2a7bdfe..077f930 100644 > > > --- a/doc/guides/linux_gsg/index.rst > > > +++ b/doc/guides/linux_gsg/index.rst > > > @@ -13,6 +13,7 @@ Getting Started Guide for Linux > > > intro > > > sys_reqs > > > build_dpdk > > > +cross_build_dpdk_for_arm64 > > > linux_drivers > > > build_sample_apps > > > enable_func > > > -- > > > 2.1.4 > > > > IMPORTANT NOTICE: The contents of this email and any attachments are > confidential and may also be privileged. If you are not the intended > recipient, please notify the sender immediately and do not disclose the > contents to any other person, use it for any purpose, or store or copy the > information in any medium. Thank you.
[dpdk-dev] [PATCH] eventdev: fix Rx SW adapter stop
The Rx adapter stop call does not guarantee that the
SW service function will not execute after the
rte_event_eth_rx_adapter_stop() call.
Add a "started" flag to prevent the adapter from executing
if stop has been called.
Fixes: 9c38b704d280 ("eventdev: add eth Rx adapter implementation")
Signed-off-by: Nikhil Rao
CC: sta...@dpdk.org
---
lib/librte_eventdev/rte_event_eth_rx_adapter.c | 12 +++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/lib/librte_eventdev/rte_event_eth_rx_adapter.c
b/lib/librte_eventdev/rte_event_eth_rx_adapter.c
index d03f870..7b39e39 100644
--- a/lib/librte_eventdev/rte_event_eth_rx_adapter.c
+++ b/lib/librte_eventdev/rte_event_eth_rx_adapter.c
@@ -91,6 +91,8 @@ struct rte_event_eth_rx_adapter {
int socket_id;
/* Per adapter EAL service */
uint32_t service_id;
+ /* Adapter started flag */
+ uint8_t rxa_started;
} __rte_cache_aligned;
/* Per eth device */
@@ -556,6 +558,10 @@ static uint16_t gcd_u16(uint16_t a, uint16_t b)
if (rte_spinlock_trylock(&rx_adapter->rx_lock) == 0)
return 0;
+ if (!rx_adapter->rxa_started) {
+ return 0;
+ rte_spinlock_unlock(&rx_adapter->rx_lock);
+ }
eth_rx_poll(rx_adapter);
rte_spinlock_unlock(&rx_adapter->rx_lock);
return 0;
@@ -847,8 +853,12 @@ static int add_rx_queue(struct rte_event_eth_rx_adapter
*rx_adapter,
&rte_eth_devices[i]);
}
- if (use_service)
+ if (use_service) {
+ rte_spinlock_lock(&rx_adapter->rx_lock);
+ rx_adapter->rxa_started = start;
rte_service_runstate_set(rx_adapter->service_id, start);
+ rte_spinlock_unlock(&rx_adapter->rx_lock);
+ }
return 0;
}
--
1.8.3.1
Re: [dpdk-dev] [PATCH] net/bonding: add add/remove mac addrs
This is fine but you are missing one bit. If a slave is added after you
can configured the MAC
addresses, you will need to replay the configured MAC addresses into that
slave.
It's not clear to me what the contract is between bonding and DPDK as far
as slave
removal. Just looking at the behavior with the default MAC address, I
would guess
that bonding also needs to remove extra MAC addresses it added during slave
remove. Bonding is a PMD but also an application.
On Fri, May 25, 2018 at 12:21 PM, Alex Kiselev wrote:
> add functions to add/remove MAC addresses
>
> Signed-off-by: Alex Kiselev
> ---
> drivers/net/bonding/rte_eth_bond_pmd.c | 76
> +++---
> 1 file changed, 71 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/net/bonding/rte_eth_bond_pmd.c
> b/drivers/net/bonding/rte_eth_bond_pmd.c
> index 02d94b1..835b4e3 100644
> --- a/drivers/net/bonding/rte_eth_bond_pmd.c
> +++ b/drivers/net/bonding/rte_eth_bond_pmd.c
> @@ -25,6 +25,7 @@
>
> #define REORDER_PERIOD_MS 10
> #define DEFAULT_POLLING_INTERVAL_10_MS (10)
> +#define BOND_MAX_MAC_ADDRS 16
>
> #define HASH_L4_PORTS(h) ((h)->src_port ^ (h)->dst_port)
>
> @@ -2219,7 +2220,7 @@ bond_ethdev_info(struct rte_eth_dev *dev, struct
> rte_eth_dev_info *dev_info)
> uint16_t max_nb_rx_queues = UINT16_MAX;
> uint16_t max_nb_tx_queues = UINT16_MAX;
>
> - dev_info->max_mac_addrs = 1;
> + dev_info->max_mac_addrs = BOND_MAX_MAC_ADDRS;
>
> dev_info->max_rx_pktlen = internals->candidate_max_rx_pktlen ?
> internals->candidate_max_rx_pktlen :
> @@ -2905,6 +2906,65 @@ bond_filter_ctrl(struct rte_eth_dev *dev
> __rte_unused,
> return -ENOTSUP;
> }
>
> +static int
> +bond_ethdev_mac_addr_add(struct rte_eth_dev *dev, struct ether_addr
> *mac_addr,
> + __rte_unused uint32_t index, uint32_t vmdq)
> +{
> + struct rte_eth_dev *slave_eth_dev;
> + struct bond_dev_private *internals = dev->data->dev_private;
> + int ret, i;
> +
> + rte_spinlock_lock(&internals->lock);
> +
> + for (i = 0; i < internals->slave_count; i++) {
> + slave_eth_dev = &rte_eth_devices[internals->
> slaves[i].port_id];
> + if (*slave_eth_dev->dev_ops->mac_addr_add == NULL) {
> + ret = -ENOTSUP;
> + goto end;
> + }
> + }
> +
> + for (i = 0; i < internals->slave_count; i++) {
> + ret = rte_eth_dev_mac_addr_add(internals->slaves[i].port_id,
> mac_addr,
> + vmdq);
> + if (ret < 0)
> + goto end;
> + }
> +
> + ret = 0;
> +end:
> + rte_spinlock_unlock(&internals->lock);
> + return ret;
> +}
> +
> +static void
> +bond_ethdev_mac_addr_remove(struct rte_eth_dev *dev, uint32_t index)
> +{
> + struct rte_eth_dev *slave_eth_dev;
> + struct bond_dev_private *internals = dev->data->dev_private;
> + int ret, i;
> +
> + rte_spinlock_lock(&internals->lock);
> +
> + for (i = 0; i < internals->slave_count; i++) {
> + slave_eth_dev = &rte_eth_devices[internals->
> slaves[i].port_id];
> + if (*slave_eth_dev->dev_ops->mac_addr_remove == NULL)
> + goto end;
> + }
> +
> + struct ether_addr *mac_addr = &dev->data->mac_addrs[index];
> +
> + for (i = 0; i < internals->slave_count; i++) {
> + ret = rte_eth_dev_mac_addr_remove(
> internals->slaves[i].port_id,
> + mac_addr);
> + if (ret < 0)
> + goto end;
> + }
> +
> +end:
> + rte_spinlock_unlock(&internals->lock);
> +}
> +
> const struct eth_dev_ops default_dev_ops = {
> .dev_start= bond_ethdev_start,
> .dev_stop = bond_ethdev_stop,
> @@ -2927,7 +2987,9 @@ const struct eth_dev_ops default_dev_ops = {
> .rss_hash_conf_get= bond_ethdev_rss_hash_conf_get,
> .mtu_set = bond_ethdev_mtu_set,
> .mac_addr_set = bond_ethdev_mac_address_set,
> - .filter_ctrl = bond_filter_ctrl
> + .filter_ctrl = bond_filter_ctrl,
> + .mac_addr_add = bond_ethdev_mac_addr_add,
> + .mac_addr_remove = bond_ethdev_mac_addr_remove
> };
>
> static int
> @@ -2954,10 +3016,14 @@ bond_alloc(struct rte_vdev_device *dev, uint8_t
> mode)
> eth_dev->data->nb_rx_queues = (uint16_t)1;
> eth_dev->data->nb_tx_queues = (uint16_t)1;
>
> - eth_dev->data->mac_addrs = rte_zmalloc_socket(name,
> ETHER_ADDR_LEN, 0,
> - socket_id);
> + /* Allocate memory for storing MAC addresses */
> + eth_dev->data->mac_addrs = rte_zmalloc_socket(name, ETHER_ADDR_LEN
> *
> + BOND_MAX_MAC_ADDRS, 0,
> socket_id);
> if (eth_dev->data->mac_addrs == NULL
Re: [dpdk-dev] [PATCH 4/4] net/virtio: improve offload check performance
On 06/04/2018 01:55 PM, Tiwei Bie wrote:
On Fri, Jun 01, 2018 at 02:47:58PM +0200, Maxime Coquelin wrote:
Instead of checking the multiple Virtio features bits for
every packet, let's do the check once at configure time and
store it in virtio_hw struct.
Signed-off-by: Maxime Coquelin
---
drivers/net/virtio/virtio_ethdev.c | 19 +++
drivers/net/virtio/virtio_pci.h| 2 ++
drivers/net/virtio/virtio_rxtx.c | 29 ++---
3 files changed, 27 insertions(+), 23 deletions(-)
diff --git a/drivers/net/virtio/virtio_ethdev.c
b/drivers/net/virtio/virtio_ethdev.c
index d481b282e..981e0994a 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c
@@ -1790,6 +1790,22 @@ rte_virtio_pmd_init(void)
rte_pci_register(&rte_virtio_pmd);
}
+static inline int
Maybe it's better to return bool.
Agree, I'll do that.
I just moved code from virtio_rxtx.c without change, but it is indeed
better to return bool.
+rx_offload_enabled(struct virtio_hw *hw)
+{
+ return vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
+ vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
+ vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
+}
+
+static inline int
Ditto.
+tx_offload_enabled(struct virtio_hw *hw)
+{
+ return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
+ vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
+ vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
+}
+
/*
* Configure virtio device
* It returns 0 on success.
@@ -1869,6 +1885,9 @@ virtio_dev_configure(struct rte_eth_dev *dev)
return -ENOTSUP;
}
+ hw->has_tx_offload = !!tx_offload_enabled(hw);
+ hw->has_rx_offload = !!rx_offload_enabled(hw);
+
if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
/* Enable vector (0) for Link State Intrerrupt */
if (VTPCI_OPS(hw)->set_config_irq(hw, 0) ==
diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h
index a28ba8339..e0bb871f2 100644
--- a/drivers/net/virtio/virtio_pci.h
+++ b/drivers/net/virtio/virtio_pci.h
@@ -233,6 +233,8 @@ struct virtio_hw {
uint8_t modern;
uint8_t use_simple_rx;
uint8_t use_simple_tx;
+ uint8_t has_tx_offload;
+ uint8_t has_rx_offload;
Acked.
I think it's better to use spaces, instead of
4 spaces width tabs after uint8_t.
uint16_tport_id;
uint8_t mac_addr[ETHER_ADDR_LEN];
uint32_tnotify_off_multiplier;
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index 92fab2174..3f113a118 100644
--- a/drivers/net/virtio/virtio_rxtx.c
+++ b/drivers/net/virtio/virtio_rxtx.c
@@ -225,13 +225,6 @@ virtio_tso_fix_cksum(struct rte_mbuf *m)
}
}
-static inline int
-tx_offload_enabled(struct virtio_hw *hw)
-{
- return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
- vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
- vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
-}
/* avoid write operation when necessary, to lessen cache issues */
#define ASSIGN_UNLESS_EQUAL(var, val) do {\
@@ -253,13 +246,15 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct
rte_mbuf *cookie,
struct virtio_net_hdr *hdr;
int offload;
- offload = tx_offload_enabled(vq->hw);
head_idx = vq->vq_desc_head_idx;
idx = head_idx;
dxp = &vq->vq_descx[idx];
dxp->cookie = (void *)cookie;
dxp->ndescs = needed;
+ offload = vq->hw->has_tx_offload &&
+ (cookie->ol_flags & PKT_TX_OFFLOAD_MASK);
If features aren't negotiated, I think there is no need to
check ol_flags and update the net header.
Isn't what the code is doing?
has_tx_offload will be false if none of the Tx offload features have
been negotiated, so ol_flags won't be checked in that case.
+
start_dp = vq->vq_ring.desc;
if (can_push) {
@@ -270,7 +265,6 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, struct
rte_mbuf *cookie,
* which is wrong. Below subtract restores correct pkt size.
*/
cookie->pkt_len -= head_size;
- /* if offload disabled, it is not zeroed below, do it now */
if (offload == 0) {
ASSIGN_UNLESS_EQUAL(hdr->csum_start, 0);
ASSIGN_UNLESS_EQUAL(hdr->csum_offset, 0);
@@ -686,14 +680,6 @@ virtio_rx_offload(struct rte_mbuf *m, struct
virtio_net_hdr *hdr)
return 0;
}
[...]
Best regards,
Tiwei Bie
Re: [dpdk-dev] [PATCH 3/4] net/virtio: don't use simple Rx path if TCP LRO requested
On 06/04/2018 01:59 PM, Tiwei Bie wrote: On Fri, Jun 01, 2018 at 02:47:57PM +0200, Maxime Coquelin wrote: Signed-off-by: Maxime Coquelin --- drivers/net/virtio/virtio_ethdev.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c index 5730620ed..d481b282e 100644 --- a/drivers/net/virtio/virtio_ethdev.c +++ b/drivers/net/virtio/virtio_ethdev.c @@ -1893,7 +1893,8 @@ virtio_dev_configure(struct rte_eth_dev *dev) } if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM | - DEV_RX_OFFLOAD_TCP_CKSUM)) + DEV_RX_OFFLOAD_TCP_CKSUM | + DEV_RX_OFFLOAD_TCP_LRO)) Maybe we also need to consider not using simple Rx when DEV_RX_OFFLOAD_VLAN_STRIP is requested. I think that makes sense. Thanks, Maxime Best regards, Tiwei Bie hw->use_simple_rx = 0; if (tx_offloads & (DEV_TX_OFFLOAD_TCP_CKSUM | -- 2.14.3
[dpdk-dev] [PATCH] net/mlx5: fix error number handling
rte_errno should be saved only if error has occurred because rte_errno
could have garbage value.
Fixes: a6d83b6a9209 ("net/mlx5: standardize on negative errno values")
Cc: sta...@dpdk.org
Signed-off-by: Yongseok Koh
---
drivers/net/mlx5/mlx5_flow.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
index 994be05be..eaffe7495 100644
--- a/drivers/net/mlx5/mlx5_flow.c
+++ b/drivers/net/mlx5/mlx5_flow.c
@@ -3561,7 +3561,8 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
/* The flow does not match. */
continue;
}
- ret = rte_errno; /* Save rte_errno before cleanup. */
+ if (ret)
+ ret = rte_errno; /* Save rte_errno before cleanup. */
if (flow)
mlx5_flow_list_destroy(dev, &priv->flows, flow);
exit:
--
2.11.0
[dpdk-dev] [Bug 61] build fails when shared libs enabled and optimization turned off
https://dpdk.org/tracker/show_bug.cgi?id=61 Bug ID: 61 Summary: build fails when shared libs enabled and optimization turned off Product: DPDK Version: 18.05 Hardware: All OS: All Status: CONFIRMED Severity: normal Priority: Normal Component: core Assignee: dev@dpdk.org Reporter: erik.g.carri...@intel.com Target Milestone: --- Overview: Build fails with CONFIG_RTE_BUILD_SHARED_LIB=y and EXTRA_CFLAGS="-O0" Steps to Reproduce: 1. Generate config file with: "$ make config T=x86_64-native-linuxapp-gcc O=x86_64-native-linuxapp-gcc" 2. Set CONFIG_RTE_BUILD_SHARED_LIB=y in .config 3. Run build with: "$ make V=1 EXTRA_CFLAGS='-O0'" Actual Results: Build fails with following error: <...snipped...> == Build drivers/bus/fslmc <...snipped...> gcc -Wp,-MD,./.fslmc_vfio.o.d.tmp -m64 -pthread -fPIC -march=native -DRTE_MACHINE_CPUFLAG_SSE -DRTE_MACHINE_CPUFLAG_SSE2 -DRTE_MACHINE_CPUFLAG_SSE3 -DRTE_MACHINE_CPUFLAG_SSSE3 -DRTE_MACHINE_CPUFLAG_SSE4_1 -DRTE_MACHINE_CPUFLAG_SS E4_2 -DRTE_MACHINE_CPUFLAG_AES -DRTE_MACHINE_CPUFLAG_PCLMULQDQ -DRTE_MACHINE_CPUFLAG_AVX -DRTE_MACHINE_CPUFLAG_RDRAND -DRTE_MACHINE_CPUFLAG_FSGSBASE -DRTE_MACHINE_CPUFLAG_F16C -DRTE_MACHINE_CPUFLAG_AVX2 -I/workspace/sw/egcarril/projects/upstream/dpdk/x86_64-native-linuxapp-gcc/include -include /workspace/sw/egcarril/projects/upstream/dpdk/x86_64-native-linuxapp-gcc/include/rte_config.h -DALLOW_EXPERIMENTAL_API -O3 -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations -Wold-style-definition -Wpointer-arith -Wcast-align -Wnested-externs -Wcast-qual -Wformat-nonliteral -Wformat-security -Wundef -Wwrite-strings -Wdeprecated -Werror -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/mc -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/qbman/include -I/workspace/sw/egcarril/projects/upstream/dpdk/lib/librte_eal/common -O0 -o fslmc_vfio.o -c /workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/fslmc_vfio.c gcc -Wp,-MD,./.fslmc_bus.o.d.tmp -m64 -pthread -fPIC -march=native -DRTE_MACHINE_CPUFLAG_SSE -DRTE_MACHINE_CPUFLAG_SSE2 -DRTE_MACHINE_CPUFLAG_SSE3 -DRTE_MACHINE_CPUFLAG_SSSE3 -DRTE_MACHINE_CPUFLAG_SSE4_1 -DRTE_MACHINE_CPUFLAG_SSE 4_2 -DRTE_MACHINE_CPUFLAG_AES -DRTE_MACHINE_CPUFLAG_PCLMULQDQ -DRTE_MACHINE_CPUFLAG_AVX -DRTE_MACHINE_CPUFLAG_RDRAND -DRTE_MACHINE_CPUFLAG_FSGSBASE -DRTE_MACHINE_CPUFLAG_F16C -DRTE_MACHINE_CPUFLAG_AVX2 -I/workspace/sw/egcarril/pro jects/upstream/dpdk/x86_64-native-linuxapp-gcc/include -include /workspace/sw/egcarril/projects/upstream/dpdk/x86_64-native-linuxapp-gcc/include/rte_config.h -DALLOW_EXPERIMENTAL_API -O3 -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wmissing-declarations -Wold-style-definition -Wpointer-arith -Wcast-align -Wnested-externs -Wcast-qual -Wformat-nonliteral -Wformat-security -Wundef -Wwrite-strings -Wdeprecated -Werror -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/mc -I/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/qbman/include -I/workspace/sw/egcarril/projects/upstream/dpdk/lib/librte_eal/common -O0 -o fslmc_bus.o -c /workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/fslmc_bus.c gcc -m64 -O0 -L/workspace/sw/egcarril/projects/upstream/dpdk/x86_64-native-linuxapp-gcc/lib -Wl,--version-script=/workspace/sw/egcarril/projects/upstream/dpdk/drivers/bus/fslmc/rte_bus_fslmc_version.map -shared qbman/qbman_portal. o qbman/qbman_debug.o mc/dpmng.o mc/dpbp.o mc/dpio.o mc/mc_sys.o mc/dpcon.o mc/dpci.o mc/dpdmai.o portal/dpaa2_hw_dpio.o portal/dpaa2_hw_dpbp.o portal/dpaa2_hw_dpci.o fslmc_vfio.o fslmc_bus.o -z defs -lrte_eal -lrte_mbuf -lrte_memp ool -lrte_ring -lrte_ethdev -Wl,-soname,librte_bus_fslmc.so.1.1 -o librte_bus_fslmc.so.1.1 portal/dpaa2_hw_dpio.o: In function `dpaa2_mem_ptov': dpaa2_hw_dpio.c:(.text+0xcc): undefined reference to `rte_dpaa2_memsegs' portal/dpaa2_hw_dpbp.o: In function `dpaa2_mem_ptov': dpaa2_hw_dpbp.c:(.text+0xa7): undefined reference to `rte_dpaa2_memsegs' portal/dpaa2_hw_dpci.o: In function `dpaa2_mem_ptov': dpaa2_hw_dpci.c:(.text+0xa7): undefined reference to `rte_dpaa2_memsegs' fslmc_vfio.o: In function `dpaa2_mem_ptov': fslmc_vfio.c:(.text+0xd9): undefined reference to `rte_dpaa2_memsegs' collect2: error: ld returned 1 exit status make[5]: *** [/workspace/sw/egcarril/projects/upstream/dpdk/mk/rte.lib.mk:100: librte_bus_fslmc.so.1.1] Error 1 make[4]: *** [/workspace/sw/egcarril/projects/upstream/dpdk/mk/rte.subdir.mk:37: fslmc] Error 2 make[3]: *** [/workspace/sw/egcarril/projects/upstream/dpdk/mk/rte.subdir.mk:37: bus] Error 2 make[2]: *** [/workspace/sw/egcarril/projects/upstream/dpdk/mk/rte.sdkbuild.mk:51: drivers] Error 2 make[1]: *** [/worksp
Re: [dpdk-dev] 18.02.2 patches review and test
On Mon, 2018-06-04 at 09:58 +0100, Luca Boccassi wrote: > Hi all, > > Here is a list of patches targeted for stable release 18.02.2. Please > help review and test. The planned date for the final release is > Thursday, > the 14th of June. Before that, please shout if anyone has objections > with these > patches being applied. > > Also for the companies committed to running regression tests, > please run the tests and report any issue before the release date. > > These patches are located at branch 18.02 of dpdk-stable repo: > https://dpdk.org/browse/dpdk-stable/ > > Thanks. > > Luca Boccassi > > --- The fix for a bad regression in the qat crypto pmd was missed, due to missing Fixes line in the commit message. It has now been queued on request of the author: Lee Roberts (1): crypto/qat: assign device to correct NUMA node -- Kind regards, Luca Boccassi
[dpdk-dev] [PATCH] net: rename u16 to fix shadowed declaration warning
This patch renames u16 to u16_buf. u16 as a variable name causes a shadowed
declaration warning if, for example, the application also typedefs u16
(e.g. by including a header containing "typedef unsigned short u16") and
the application is built with -Wshadow.
Signed-off-by: Gage Eads
---
lib/librte_net/rte_ip.h | 28 ++--
1 file changed, 14 insertions(+), 14 deletions(-)
diff --git a/lib/librte_net/rte_ip.h b/lib/librte_net/rte_ip.h
index 72dc2456a..f2a8904a2 100644
--- a/lib/librte_net/rte_ip.h
+++ b/lib/librte_net/rte_ip.h
@@ -108,25 +108,25 @@ __rte_raw_cksum(const void *buf, size_t len, uint32_t sum)
/* workaround gcc strict-aliasing warning */
uintptr_t ptr = (uintptr_t)buf;
typedef uint16_t __attribute__((__may_alias__)) u16_p;
- const u16_p *u16 = (const u16_p *)ptr;
-
- while (len >= (sizeof(*u16) * 4)) {
- sum += u16[0];
- sum += u16[1];
- sum += u16[2];
- sum += u16[3];
- len -= sizeof(*u16) * 4;
- u16 += 4;
+ const u16_p *u16_buf = (const u16_p *)ptr;
+
+ while (len >= (sizeof(*u16_buf) * 4)) {
+ sum += u16_buf[0];
+ sum += u16_buf[1];
+ sum += u16_buf[2];
+ sum += u16_buf[3];
+ len -= sizeof(*u16_buf) * 4;
+ u16_buf += 4;
}
- while (len >= sizeof(*u16)) {
- sum += *u16;
- len -= sizeof(*u16);
- u16 += 1;
+ while (len >= sizeof(*u16_buf)) {
+ sum += *u16_buf;
+ len -= sizeof(*u16_buf);
+ u16_buf += 1;
}
/* if length is in odd bytes */
if (len == 1)
- sum += *((const uint8_t *)u16);
+ sum += *((const uint8_t *)u16_buf);
return sum;
}
--
2.13.6
Re: [dpdk-dev] [PATCH] version: 18.08-rc0
01/06/2018 13:07, Thomas Monjalon: > Signed-off-by: Thomas Monjalon Applied > --- > > The new release cycle is open, and it will be a short one! > > Please consider reviewing your peers, and contribute to respect > the release schedule. > > If you think some patches are already ready for integration in master, > please notify me. > > Note that changes with big impact (renames or core features) > are preferred to be integrated early in the cycle. > > Thanks for your attention
Re: [dpdk-dev] cppcheck on dpdk
On 16/05/18 13:41, Ferruh Yigit wrote: > Today after listening Colin's Static Analysis talk, I run cppcheck on > v18.05-rc4 > code and it revealed some issues, sharing here for anyone to interested in > fixing them. At least I encourage to check maintainers to check their own > pieces. > > It is really easy to run cppcheck, in dpdk source folder: > cppcheck --force . FYI, it may be useful to run it with the following options: cppecheck --enable=all --force ..this may lead to more false positives, but it can catch some deeper issues. Colin > > With above command cppcheck verifies all #ifdef paths, some issues below seems > related to this and that is why these issues not seen in build tests. > > > [app/test-pmd/cmdline_mtr.c:115]: (error) Memory leak: dscp_table > [app/test-pmd/flowgen.c:160]: (error) Uninitialized variable: ol_flags > [app/test-pmd/tm.c:594]: (error) Memory leak: tnp.shared_shaper_id > [drivers/bus/dpaa/base/fman/fman.c:557]: (error) Uninitialized variable: __if > [drivers/bus/dpaa/base/qbman/qman.c:1220]: (error) Address of auto-variable > 'p->shadow_dqrr[DQRR_PTR2IDX(dq)]' returned > [drivers/bus/ifpga/ifpga_bus.c:433]: (warning) Possible null pointer > dereference: c2 > [drivers/crypto/ccp/ccp_pci.c:41]: (error) Resource leak: fp > [drivers/crypto/dpaa_sec/dpaa_sec.c:662]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:731]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:826]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:881]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:1020]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:1132]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:1258]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:1353]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/crypto/dpaa_sec/dpaa_sec.c:1392]: (error) Address of auto-variable > 'ctx->job' returned > [drivers/net/avf/base/avf_adminq.c:301]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/avf/base/avf_adminq.c:336]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/avf/base/avf_adminq.c:298]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/avf/base/avf_adminq.c:333]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/avf/base/avf_common.c:367]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/avf/base/avf_common.c:364]: (error) Shifting signed 32-bit value > by > 31 bits is undefined behaviour > [drivers/net/axgbe/axgbe_dev.c:808]: (error) Shifting signed 32-bit value by > 31 > bits is undefined behaviour > [drivers/net/axgbe/axgbe_dev.c] -> [drivers/net/axgbe/axgbe_dev.c]: (error) > Invalid value: 0x0204_BUSY_WIDTH > [drivers/net/axgbe/axgbe_ethdev.c] -> [drivers/net/axgbe/axgbe_ethdev.c]: > (error) Invalid value: 0x0008_PR_WIDTH > [drivers/net/axgbe/axgbe_i2c.c] -> [drivers/net/axgbe/axgbe_i2c.c]: (error) > Invalid value: 0x006c_EN_WIDTH > [drivers/net/axgbe/axgbe_phy_impl.c] -> [drivers/net/axgbe/axgbe_phy_impl.c]: > (error) Invalid value: 0x0080_ID_WIDTH > [drivers/net/axgbe/axgbe_rxtx.c:292]: (error) Shifting signed 32-bit value by > 31 > bits is undefined behaviour > [drivers/net/axgbe/axgbe_rxtx.c:592]: (error) Shifting signed 32-bit value by > 31 > bits is undefined behaviour > [drivers/net/axgbe/axgbe_rxtx.c] -> [drivers/net/axgbe/axgbe_rxtx.c]: (error) > Invalid value: 0x48_PRXQ_WIDTH > [drivers/net/bnx2x/bnx2x.c:3995]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/bnx2x.c:4000]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/bnx2x.c:8729]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/bnx2x.c:9765]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/elink.c:1042]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/elink.c:2711]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/elink.c:9662]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnx2x/elink.c:10295]: (error) Shifting signed 32-bit value by 31 > bits is undefined behaviour > [drivers/net/bnxt/bnxt_ethdev.c:594]: (error) Shifting signed 32-bit value by > 31 > bits is undefined behaviour > [drivers/net/bnxt/bnxt_ethdev.c:634]: (error) Shifting signed 32-bit value by > 31 > bits is undefined behaviour > [drivers/net/bnxt/bnxt_rxr.c:482]: (error) Uninitialized variable: ag_cons > [
[dpdk-dev] [PATCH] net/pcap: rx_iface_in stream type support
Support rx of in direction packets only
Useful for apps that also tx to eth_pcap ports in order not to see them
echoed back in as rx when out direction is also captured
Signed-off-by: ido goshen
---
drivers/net/pcap/rte_eth_pcap.c | 34 +++---
1 file changed, 31 insertions(+), 3 deletions(-)
diff --git a/drivers/net/pcap/rte_eth_pcap.c b/drivers/net/pcap/rte_eth_pcap.c
index 6bd4a7d..88bd550 100644
--- a/drivers/net/pcap/rte_eth_pcap.c
+++ b/drivers/net/pcap/rte_eth_pcap.c
@@ -26,6 +26,7 @@
#define ETH_PCAP_RX_PCAP_ARG "rx_pcap"
#define ETH_PCAP_TX_PCAP_ARG "tx_pcap"
#define ETH_PCAP_RX_IFACE_ARG "rx_iface"
+#define ETH_PCAP_RX_IFACE_IN_ARG "rx_iface_in"
#define ETH_PCAP_TX_IFACE_ARG "tx_iface"
#define ETH_PCAP_IFACE_ARG"iface"
@@ -83,6 +84,7 @@ struct pmd_devargs {
ETH_PCAP_RX_PCAP_ARG,
ETH_PCAP_TX_PCAP_ARG,
ETH_PCAP_RX_IFACE_ARG,
+ ETH_PCAP_RX_IFACE_IN_ARG,
ETH_PCAP_TX_IFACE_ARG,
ETH_PCAP_IFACE_ARG,
NULL
@@ -726,6 +728,22 @@ struct pmd_devargs {
return 0;
}
+static inline int
+set_iface_direction(const char *iface , const char *key, pcap_t *pcap)
+{
+ if (strcmp(key, ETH_PCAP_RX_IFACE_IN_ARG) == 0) {
+ if (pcap_setdirection(pcap, PCAP_D_IN) < 0) {
+ PMD_LOG(ERR,
+ "Setting %s pcap direction IN failed - %s\n",
+iface,
+pcap_geterr(pcap));
+ return -1;
+ }
+ PMD_LOG(INFO, "Setting %s pcap direction IN\n", iface);
+ }
+ return 0;
+}
+
/*
* Opens a NIC for reading packets from it
*/
@@ -740,11 +758,12 @@ struct pmd_devargs {
for (i = 0; i < rx->num_of_queue; i++) {
if (open_single_iface(iface, &pcap) < 0)
return -1;
+ if (set_iface_direction(iface, key, pcap) < 0)
+ return -1;
rx->queue[i].pcap = pcap;
rx->queue[i].name = iface;
rx->queue[i].type = key;
}
-
return 0;
}
@@ -963,7 +982,9 @@ struct pmd_devargs {
is_rx_pcap = 1;
else
pcaps.num_of_queue = rte_kvargs_count(kvlist,
- ETH_PCAP_RX_IFACE_ARG);
+ ETH_PCAP_RX_IFACE_ARG) +
+ rte_kvargs_count(kvlist,
+ ETH_PCAP_RX_IFACE_IN_ARG);
if (pcaps.num_of_queue > RTE_PMD_PCAP_MAX_QUEUES)
pcaps.num_of_queue = RTE_PMD_PCAP_MAX_QUEUES;
@@ -971,9 +992,15 @@ struct pmd_devargs {
if (is_rx_pcap)
ret = rte_kvargs_process(kvlist, ETH_PCAP_RX_PCAP_ARG,
&open_rx_pcap, &pcaps);
- else
+ else {
ret = rte_kvargs_process(kvlist, ETH_PCAP_RX_IFACE_ARG,
&open_rx_iface, &pcaps);
+ if (ret == 0)
+ ret = rte_kvargs_process(kvlist,
+ ETH_PCAP_RX_IFACE_IN_ARG,
+ &open_rx_iface,
+ &pcaps);
+ }
if (ret < 0)
goto free_kvlist;
@@ -1046,6 +1073,7 @@ struct pmd_devargs {
ETH_PCAP_RX_PCAP_ARG "= "
ETH_PCAP_TX_PCAP_ARG "= "
ETH_PCAP_RX_IFACE_ARG "= "
+ ETH_PCAP_RX_IFACE_IN_ARG "= "
ETH_PCAP_TX_IFACE_ARG "= "
ETH_PCAP_IFACE_ARG "=");
--
1.9.1
Re: [dpdk-dev] [PATCH] examples/performance-thread: convert license headers to SPDX
On Fri, May 25, 2018 at 12:10 PM, Thomas Monjalon wrote: > 20/04/2018 11:53, Pablo de Lara: >> Convert dual license headers with Intel and Dmitry Vyukov >> names to SPDX. >> >> Signed-off-by: Pablo de Lara >> --- >> examples/performance-thread/common/lthread_pool.h | 68 >> +- >> examples/performance-thread/common/lthread_queue.h | 68 >> +- >> 2 files changed, 6 insertions(+), 130 deletions(-) > > Waiting for confirmation from Dmitry Vyukov. I don't have any objections. I hereby grant my permission to change the the licence.
Re: [dpdk-dev] [PATCH 4/4] net/virtio: improve offload check performance
On Mon, Jun 04, 2018 at 04:29:56PM +0200, Maxime Coquelin wrote: > On 06/04/2018 01:55 PM, Tiwei Bie wrote: > > On Fri, Jun 01, 2018 at 02:47:58PM +0200, Maxime Coquelin wrote: [...] > > > @@ -253,13 +246,15 @@ virtqueue_enqueue_xmit(struct virtnet_tx *txvq, > > > struct rte_mbuf *cookie, > > > struct virtio_net_hdr *hdr; > > > int offload; > > > - offload = tx_offload_enabled(vq->hw); > > > head_idx = vq->vq_desc_head_idx; > > > idx = head_idx; > > > dxp = &vq->vq_descx[idx]; > > > dxp->cookie = (void *)cookie; > > > dxp->ndescs = needed; > > > + offload = vq->hw->has_tx_offload && > > > + (cookie->ol_flags & PKT_TX_OFFLOAD_MASK); > > > > If features aren't negotiated, I think there is no need to > > check ol_flags and update the net header. > > Isn't what the code is doing? > has_tx_offload will be false if none of the Tx offload features have > been negotiated, so ol_flags won't be checked in that case. Hmm.. Somehow I treated 'and' as 'or'.. I have another question. When 'can_push' is false and 'vq->hw->has_tx_offload' is true, there will be a chance that virtio net hdr won't be zeroed when ol_flags doesn't specify any Tx offloads. Best regards, Tiwei Bie
Re: [dpdk-dev] [PATCH] bus/dpaa: fix compilation issue with meson build
On 6/4/2018 8:22 AM, Jerin Jacob wrote:
ccache gcc -Idrivers/drivers@@tmp_rte_pmd_dpaa_sec@sta -Idrivers
In file included from ../drivers/bus/dpaa/include/fsl_usd.h:11,
from ../drivers/crypto/dpaa_sec/dpaa_sec.c:27:
../drivers/bus/dpaa/include/compat.h:53:
error: "__packed" redefined [-Werror]
#define __packed __rte_packed
In file included from /usr/include/bsd/string.h:39,
from ../lib/librte_eal/common/include/rte_string_fns.h:71,
from ../drivers/crypto/dpaa_sec/dpaa_sec.c:25:
/usr/include/bsd/sys/cdefs.h:120: note: this is the location
of the previous definition
# define __packed __attribute__((__packed__))
Cc: sta...@dpdk.org
Fixes: 39f373cf015a ("bus/dpaa: add compatibility and helper macros")
Signed-off-by: Jerin Jacob
---
Thanks Jerin.
Somehow, this didn't appear on my build environment (maybe because LIB
BSD is not enabled?). But, the change looks harmless.
Acked-by: Shreyansh Jain
[dpdk-dev] [PATCH] mbuf: remove experimental tag from pool ops functions
Signed-off-by: Hemant Agrawal
---
lib/librte_mbuf/rte_mbuf.c | 2 +-
lib/librte_mbuf/rte_mbuf.h | 2 +-
lib/librte_mbuf/rte_mbuf_pool_ops.c | 10 +-
lib/librte_mbuf/rte_mbuf_pool_ops.h | 13 +
lib/librte_mbuf/rte_mbuf_version.map | 4 ++--
5 files changed, 14 insertions(+), 17 deletions(-)
diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.c
index fca580e..e714c5a 100644
--- a/lib/librte_mbuf/rte_mbuf.c
+++ b/lib/librte_mbuf/rte_mbuf.c
@@ -107,7 +107,7 @@ rte_pktmbuf_init(struct rte_mempool *mp,
}
/* Helper to create a mbuf pool with given mempool ops name*/
-struct rte_mempool * __rte_experimental
+struct rte_mempool *
rte_pktmbuf_pool_create_by_ops(const char *name, unsigned int n,
unsigned int cache_size, uint16_t priv_size, uint16_t data_room_size,
int socket_id, const char *ops_name)
diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
index 8e6b4d2..5233d57 100644
--- a/lib/librte_mbuf/rte_mbuf.h
+++ b/lib/librte_mbuf/rte_mbuf.h
@@ -1146,7 +1146,7 @@ rte_pktmbuf_pool_create(const char *name, unsigned n,
*- EEXIST - a memzone with the same name already exists
*- ENOMEM - no appropriate memory area found in which to create memzone
*/
-struct rte_mempool * __rte_experimental
+struct rte_mempool *
rte_pktmbuf_pool_create_by_ops(const char *name, unsigned int n,
unsigned int cache_size, uint16_t priv_size, uint16_t data_room_size,
int socket_id, const char *ops_name);
diff --git a/lib/librte_mbuf/rte_mbuf_pool_ops.c
b/lib/librte_mbuf/rte_mbuf_pool_ops.c
index a1d4699..5722976 100644
--- a/lib/librte_mbuf/rte_mbuf_pool_ops.c
+++ b/lib/librte_mbuf/rte_mbuf_pool_ops.c
@@ -9,7 +9,7 @@
#include
#include
-int __rte_experimental
+int
rte_mbuf_set_platform_mempool_ops(const char *ops_name)
{
const struct rte_memzone *mz;
@@ -35,7 +35,7 @@ rte_mbuf_set_platform_mempool_ops(const char *ops_name)
return -EEXIST;
}
-const char * __rte_experimental
+const char *
rte_mbuf_platform_mempool_ops(void)
{
const struct rte_memzone *mz;
@@ -46,7 +46,7 @@ rte_mbuf_platform_mempool_ops(void)
return mz->addr;
}
-int __rte_experimental
+int
rte_mbuf_set_user_mempool_ops(const char *ops_name)
{
const struct rte_memzone *mz;
@@ -67,7 +67,7 @@ rte_mbuf_set_user_mempool_ops(const char *ops_name)
}
-const char * __rte_experimental
+const char *
rte_mbuf_user_mempool_ops(void)
{
const struct rte_memzone *mz;
@@ -79,7 +79,7 @@ rte_mbuf_user_mempool_ops(void)
}
/* Return mbuf pool ops name */
-const char * __rte_experimental
+const char *
rte_mbuf_best_mempool_ops(void)
{
/* User defined mempool ops takes the priority */
diff --git a/lib/librte_mbuf/rte_mbuf_pool_ops.h
b/lib/librte_mbuf/rte_mbuf_pool_ops.h
index ebf5bf0..7ed95a4 100644
--- a/lib/librte_mbuf/rte_mbuf_pool_ops.h
+++ b/lib/librte_mbuf/rte_mbuf_pool_ops.h
@@ -12,9 +12,6 @@
* These APIs are for configuring the mbuf pool ops names to be largely used by
* rte_pktmbuf_pool_create(). However, this can also be used to set and inquire
* the best mempool ops available.
- *
- * @warning
- * @b EXPERIMENTAL: this API may change without prior notice
*/
#include
@@ -34,7 +31,7 @@ extern "C" {
* - On success, zero.
* - On failure, a negative value.
*/
-int __rte_experimental
+int
rte_mbuf_set_platform_mempool_ops(const char *ops_name);
/**
@@ -46,7 +43,7 @@ rte_mbuf_set_platform_mempool_ops(const char *ops_name);
* - On success, platform pool ops name.
* - On failure, NULL.
*/
-const char * __rte_experimental
+const char *
rte_mbuf_platform_mempool_ops(void);
/**
@@ -60,7 +57,7 @@ rte_mbuf_platform_mempool_ops(void);
* - On success, zero.
* - On failure, a negative value.
*/
-int __rte_experimental
+int
rte_mbuf_set_user_mempool_ops(const char *ops_name);
/**
@@ -72,7 +69,7 @@ rte_mbuf_set_user_mempool_ops(const char *ops_name);
* - On success, user pool ops name..
* - On failure, NULL.
*/
-const char * __rte_experimental
+const char *
rte_mbuf_user_mempool_ops(void);
/**
@@ -87,7 +84,7 @@ rte_mbuf_user_mempool_ops(void);
* @return
* returns preferred mbuf pool ops name
*/
-const char * __rte_experimental
+const char *
rte_mbuf_best_mempool_ops(void);
diff --git a/lib/librte_mbuf/rte_mbuf_version.map
b/lib/librte_mbuf/rte_mbuf_version.map
index 1bb9538..cae68db 100644
--- a/lib/librte_mbuf/rte_mbuf_version.map
+++ b/lib/librte_mbuf/rte_mbuf_version.map
@@ -35,7 +35,7 @@ DPDK_16.11 {
} DPDK_2.1;
-EXPERIMENTAL {
+DPDK_18.08 {
global:
rte_mbuf_best_mempool_ops;
@@ -44,4 +44,4 @@ EXPERIMENTAL {
rte_mbuf_set_user_mempool_ops;
rte_mbuf_user_mempool_ops;
rte_pktmbuf_pool_create_by_ops;
-};
+} DPDK_16.11;
--
2.7.4
[dpdk-dev] [PATCH] vfio: remove experimental tag from vfio APIs
Signed-off-by: Hemant Agrawal
---
lib/librte_eal/bsdapp/eal/eal.c | 22 +-
lib/librte_eal/common/include/rte_vfio.h | 40 +---
lib/librte_eal/linuxapp/eal/eal_vfio.c | 36 ++--
lib/librte_eal/rte_eal_version.map | 17 ++
4 files changed, 57 insertions(+), 58 deletions(-)
diff --git a/lib/librte_eal/bsdapp/eal/eal.c b/lib/librte_eal/bsdapp/eal/eal.c
index dc27954..33edae5 100644
--- a/lib/librte_eal/bsdapp/eal/eal.c
+++ b/lib/librte_eal/bsdapp/eal/eal.c
@@ -866,21 +866,21 @@ int rte_vfio_clear_group(__rte_unused int vfio_group_fd)
return 0;
}
-int __rte_experimental
+int
rte_vfio_dma_map(uint64_t __rte_unused vaddr, __rte_unused uint64_t iova,
__rte_unused uint64_t len)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_dma_unmap(uint64_t __rte_unused vaddr, uint64_t __rte_unused iova,
__rte_unused uint64_t len)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_get_group_num(__rte_unused const char *sysfs_base,
__rte_unused const char *dev_addr,
__rte_unused int *iommu_group_num)
@@ -888,45 +888,45 @@ rte_vfio_get_group_num(__rte_unused const char
*sysfs_base,
return -1;
}
-int __rte_experimental
+int
rte_vfio_get_container_fd(void)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_get_group_fd(__rte_unused int iommu_group_num)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_create(void)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_destroy(__rte_unused int container_fd)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_group_bind(__rte_unused int container_fd,
__rte_unused int iommu_group_num)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_group_unbind(__rte_unused int container_fd,
__rte_unused int iommu_group_num)
{
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_dma_map(__rte_unused int container_fd,
__rte_unused uint64_t vaddr,
__rte_unused uint64_t iova,
@@ -935,7 +935,7 @@ rte_vfio_container_dma_map(__rte_unused int container_fd,
return -1;
}
-int __rte_experimental
+int
rte_vfio_container_dma_unmap(__rte_unused int container_fd,
__rte_unused uint64_t vaddr,
__rte_unused uint64_t iova,
diff --git a/lib/librte_eal/common/include/rte_vfio.h
b/lib/librte_eal/common/include/rte_vfio.h
index f90972f..5ca13fc 100644
--- a/lib/librte_eal/common/include/rte_vfio.h
+++ b/lib/librte_eal/common/include/rte_vfio.h
@@ -179,7 +179,7 @@ rte_vfio_clear_group(int vfio_group_fd);
* 0 if success.
* -1 on error.
*/
-int __rte_experimental
+int
rte_vfio_dma_map(uint64_t vaddr, uint64_t iova, uint64_t len);
@@ -200,7 +200,7 @@ rte_vfio_dma_map(uint64_t vaddr, uint64_t iova, uint64_t
len);
* -1 on error.
*/
-int __rte_experimental
+int
rte_vfio_dma_unmap(uint64_t vaddr, uint64_t iova, uint64_t len);
/**
* Parse IOMMU group number for a device
@@ -222,7 +222,7 @@ rte_vfio_dma_unmap(uint64_t vaddr, uint64_t iova, uint64_t
len);
* 0 for non-existent group or VFIO
* <0 for errors
*/
-int __rte_experimental
+int
rte_vfio_get_group_num(const char *sysfs_base,
const char *dev_addr, int *iommu_group_num);
@@ -236,7 +236,7 @@ rte_vfio_get_group_num(const char *sysfs_base,
* > 0 container fd
* < 0 for errors
*/
-int __rte_experimental
+int
rte_vfio_get_container_fd(void);
/**
@@ -252,13 +252,10 @@ rte_vfio_get_container_fd(void);
* > 0 group fd
* < 0 for errors
*/
-int __rte_experimental
+int
rte_vfio_get_group_fd(int iommu_group_num);
/**
- * @warning
- * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
- *
* Create a new container for device binding.
*
* @note Any newly allocated DPDK memory will not be mapped into these
@@ -269,13 +266,10 @@ rte_vfio_get_group_fd(int iommu_group_num);
* the container fd if successful
* <0 if failed
*/
-int __rte_experimental
+int
rte_vfio_container_create(void);
/**
- * @warning
- * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
- *
* Destroy the container, unbind all vfio groups within it.
*
* @param container_fd
@@ -285,13 +279,10 @@ rte_vfio_container_create(void);
*0 if successful
* <0 if failed
*/
-int __rte_experimental
+int
rte_vfio_container_destroy(int container_fd);
/**
- * @warning
- * @b EXPERIMENTAL: this API may change, or be removed, without prior notice
- *
* Bind a IOMMU group to a container.
*
* @param container_fd
@@ -304,13 +295,10 @@ rte_vfio_container_destroy(int container_fd);
* group fd if successful
* <0 if failed
*/
-int __rte_experimental
+int
rte
[dpdk-dev] [PATCH] mk: change TLS model for ARMv8 and DPAA machine
From: Sachin Saxena Random corruptions observed on ARM platfoms with using the dpdk library in shared mode with VPP software (plugin). sing traditional TLS scheme resolved the issue. Tested with VPP with DPDK as a plugin. Signed-off-by: Sachin Saxena --- mk/machine/armv8a/rte.vars.mk | 3 +++ mk/machine/dpaa/rte.vars.mk | 3 +++ mk/machine/dpaa2/rte.vars.mk | 3 +++ 3 files changed, 9 insertions(+) diff --git a/mk/machine/armv8a/rte.vars.mk b/mk/machine/armv8a/rte.vars.mk index 8252efb..6897cd6 100644 --- a/mk/machine/armv8a/rte.vars.mk +++ b/mk/machine/armv8a/rte.vars.mk @@ -29,3 +29,6 @@ # CPU_ASFLAGS = MACHINE_CFLAGS += -march=armv8-a+crc+crypto + +# To avoid TLS corruption issue. +MACHINE_CFLAGS += -mtls-dialect=trad diff --git a/mk/machine/dpaa/rte.vars.mk b/mk/machine/dpaa/rte.vars.mk index bddcb80..75df626 100644 --- a/mk/machine/dpaa/rte.vars.mk +++ b/mk/machine/dpaa/rte.vars.mk @@ -32,3 +32,6 @@ MACHINE_CFLAGS += -march=armv8-a+crc ifdef CONFIG_RTE_ARCH_ARM_TUNE MACHINE_CFLAGS += -mtune=$(CONFIG_RTE_ARCH_ARM_TUNE:"%"=%) endif + +# To avoid TLS corruption issue. +MACHINE_CFLAGS += -mtls-dialect=trad diff --git a/mk/machine/dpaa2/rte.vars.mk b/mk/machine/dpaa2/rte.vars.mk index 2fd2eac..aaa03c4 100644 --- a/mk/machine/dpaa2/rte.vars.mk +++ b/mk/machine/dpaa2/rte.vars.mk @@ -32,3 +32,6 @@ MACHINE_CFLAGS += -march=armv8-a+crc ifdef CONFIG_RTE_ARCH_ARM_TUNE MACHINE_CFLAGS += -mtune=$(CONFIG_RTE_ARCH_ARM_TUNE:"%"=%) endif + +# To avoid TLS corruption issue. +MACHINE_CFLAGS += -mtls-dialect=trad -- 2.7.4
[dpdk-dev] Out Of The Box Networks Developers Meetup in Taipei
Hi everyone, hope I can reach some DPDK developers in Taiwan via this mail. If you're one you should definitely join our newly funded Out of The Box Network Developers meetup: https://www.meetup.com/Out-Of-The-Box-Network-Developers-Taipei I will give a talk about "Fine-grained interaction control in userspace" (using DPDK) and we're also looking for more speakers. Please let me know if you're interested and feel free to forward this email to people how might be. Best, Marcus
Re: [dpdk-dev] [PATCH] net/mlx5: fix error number handling
On Mon, Jun 04, 2018 at 10:37:31AM -0700, Yongseok Koh wrote:
> rte_errno should be saved only if error has occurred because rte_errno
> could have garbage value.
>
> Fixes: a6d83b6a9209 ("net/mlx5: standardize on negative errno values")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Yongseok Koh
> ---
> drivers/net/mlx5/mlx5_flow.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c
> index 994be05be..eaffe7495 100644
> --- a/drivers/net/mlx5/mlx5_flow.c
> +++ b/drivers/net/mlx5/mlx5_flow.c
> @@ -3561,7 +3561,8 @@ mlx5_fdir_filter_delete(struct rte_eth_dev *dev,
> /* The flow does not match. */
> continue;
> }
> - ret = rte_errno; /* Save rte_errno before cleanup. */
> + if (ret)
> + ret = rte_errno; /* Save rte_errno before cleanup. */
> if (flow)
> mlx5_flow_list_destroy(dev, &priv->flows, flow);
> exit:
> --
> 2.11.0
This patch is not enough, the returned value being -rte_errno if no
error is detected by the function it cannot set rte_errno nor return it.
Thanks,
--
Nélio Laranjeiro
6WIND
