[dpdk-dev] [PATCH v3 06/12] eal: pci: vfio: add rd/wr func for pci bar space
On Thu, Jan 7, 2016 at 11:49 PM, Stephen Hemminger <
stephen at networkplumber.org> wrote:
> On Thu, 7 Jan 2016 22:03:03 +0530
> Santosh Shukla wrote:
>
> >
> > +int rte_eal_pci_read_bar(const struct rte_pci_device *device
> __rte_unused,
> > + void *buf __rte_unused,
> > + size_t len __rte_unused,
> > + off_t offset __rte_unused,
> > + int bar_idx __rte_unused)
> > +{
> > +#ifdef VFIO_PRESENT
> > + const struct rte_intr_handle *intr_handle = &device->intr_handle;
> > + return pci_vfio_read_bar(intr_handle, buf, len, offset, bar_idx);
> > +#else
> > + return 0; /* UIO's not applicable */
> > +#endif
> > +}
>
> It seems wrong to declare all the parameters as unused but then use them.
> Maybe there is a way to have a macro for USED(x) in the #else case
>
Yes, I followed such practice in v2 series but missed on v3, Sorry for that
and we'll take care in v4.
[dpdk-dev] [PATCH v3 06/12] eal: pci: vfio: add rd/wr func for pci bar space
On Fri, Jan 8, 2016 at 12:09 AM, Stephen Hemminger <
stephen at networkplumber.org> wrote:
> On Thu, 7 Jan 2016 18:26:30 +
> "Wiles, Keith" wrote:
>
> > On 1/7/16, 12:19 PM, "dev on behalf of Stephen Hemminger" <
> dev-bounces at dpdk.org on behalf of stephen at networkplumber.org> wrote:
> >
> > >On Thu, 7 Jan 2016 22:03:03 +0530
> > >Santosh Shukla wrote:
> > >
> > >>
> > >> +int rte_eal_pci_read_bar(const struct rte_pci_device *device
> __rte_unused,
> > >> + void *buf __rte_unused,
> > >> + size_t len __rte_unused,
> > >> + off_t offset __rte_unused,
> > >> + int bar_idx __rte_unused)
> > >> +{
> > >> +#ifdef VFIO_PRESENT
> > >> + const struct rte_intr_handle *intr_handle = &device->intr_handle;
> > >> + return pci_vfio_read_bar(intr_handle, buf, len, offset, bar_idx);
> > >> +#else
> > >> + return 0; /* UIO's not applicable */
> > >> +#endif
> > >> +}
> > >
> > >It seems wrong to declare all the parameters as unused but then use
> them.
> > >Maybe there is a way to have a macro for USED(x) in the #else case
> >
> > I would suggest we create a macro '#define RTE_UNUSED(x) ((void)x)?,
> unless we have one and I missed it groping though the code.
> > >
> >
> >
> > Regards,
> > Keith
> >
> >
> >
> >
>
> Or just move the #ifdef outside the function and have two versions
>
> #ifdef VFIO_PRESENT
> int rte_eal_pci_read_bar(const struct rte_pci_device *device,
> void *buf, size_t len,
> off_t offset, int bar_idx)
> {
> const struct rte_intr_handle *intr_handle = &device->intr_handle;
>
> return pci_vfio_read_bar(intr_handle, buf, len, offset, bar_idx);
> }
> }
> #else
> int rte_eal_pci_read_bar(const struct rte_pci_device *device __rte_unused,
> void *buf __rte_unused,
> size_t len __rte_unused,
> off_t offset __rte_unused,
> int bar_idx __rte_unused)
> {
> return 0; /* UIO's not applicable */
> }
> #endif
>
I'll pick this one!, Thanks.
[dpdk-dev] [PATCH v3 04/12] linuxapp/vfio: ignore mapping for ioport region
On Thu, Jan 7, 2016 at 11:46 PM, Stephen Hemminger <
stephen at networkplumber.org> wrote:
> This looks like the right thing to do. Minor nits.
>
> > diff --git a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
> b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
> > index 74f91ba..4077eb6 100644
> > --- a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
> > +++ b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
> > @@ -760,6 +760,26 @@ pci_vfio_map_resource(struct rte_pci_device *dev)
> > return -1;
> > }
> >
> > + /* chk for io port region */
> > + uint32_t ioport_bar;
>
> In general DPDK has followed the kernel practice of putting declarations
> at the start of function/basic block. It is ok by me, but just noting that
> the rest of the code doesn't do it.
>
>
My bad, Thanks!
> > + ret = pread64(vfio_dev_fd, &ioport_bar, sizeof(ioport_bar),
> > +
> VFIO_GET_REGION_ADDR(VFIO_PCI_CONFIG_REGION_INDEX)
> > + + PCI_BASE_ADDRESS_0 + i*4);
> > +
> > + if (ret != sizeof(ioport_bar)) {
> > + RTE_LOG(ERR, EAL,
> > + "Cannot read command (%x) from PCI config"
> > + "space!\n", PCI_BASE_ADDRESS_0 + i*4);
>
> Please dont split the line of a log message string in mid sentence.
>
>
me to don't like splitting, This was deliberate to keep checkpatch happy,
If we are ok with debug message > 80 line warning I guess it will improve
code readability.
> + return -1;
> > + }
> > +
> > + if (ioport_bar & PCI_BASE_ADDRESS_SPACE_IO) {
> > + RTE_LOG(INFO, EAL, "\tIgnore mapping since Its a
> i/o"
> > +"port bar (%d) addr : %x\n", i,
> same here
>
> Agreed.
> > +ioport_bar);
> > + continue;
> > + }
> > +
>
[dpdk-dev] [PATCH 1/3] driver/net/mpipe: support native build on tilegx platform.
>-Original Message-
>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Liming Sun
>Sent: Tuesday, December 15, 2015 11:37 PM
>To: dev at dpdk.org
>Subject: [dpdk-dev] [PATCH 1/3] driver/net/mpipe: support native build on
tilegx
>platform.
>
>This submit updates the CROSS setting to support native build on TileGx
>platform. It also enable the combined library by default.
>
>Signed-off-by: Liming Sun
>---
> MAINTAINERS |3 ++-
> config/defconfig_tile-tilegx-linuxapp-gcc |4
> mk/arch/tile/rte.vars.mk |6 ++
> 3 files changed, 12 insertions(+), 1 deletions(-)
>
>diff --git a/MAINTAINERS b/MAINTAINERS
>index 3292e84..8f7e9ca 100644
>--- a/MAINTAINERS
>+++ b/MAINTAINERS
>@@ -138,8 +138,9 @@ M: Jianbo Liu
> F: lib/librte_eal/common/include/arch/arm/*_64.h
> F: lib/librte_acl/acl_run_neon.*
>
>-EZchip TILE-Gx
>+EZchip TILE-Gx/Mx
> M: Zhigang Lu
>+M: Liming Sun
> F: lib/librte_eal/common/include/arch/tile/
> F: drivers/net/mpipe/
>
>diff --git a/config/defconfig_tile-tilegx-linuxapp-gcc
>b/config/defconfig_tile-tilegx-linuxapp-gcc
>index 9df9d7f..fb61bcd 100644
>--- a/config/defconfig_tile-tilegx-linuxapp-gcc
>+++ b/config/defconfig_tile-tilegx-linuxapp-gcc
>@@ -70,3 +70,7 @@ CONFIG_RTE_LIBRTE_SCHED=n
>CONFIG_RTE_LIBRTE_PORT=n CONFIG_RTE_LIBRTE_TABLE=n
>CONFIG_RTE_LIBRTE_PIPELINE=n
>+CONFIG_RTE_LIBRTE_CXGBE_PMD=n
>+
>+# Compile combined lib by default.
>+CONFIG_RTE_BUILD_COMBINE_LIBS=y
>diff --git a/mk/arch/tile/rte.vars.mk b/mk/arch/tile/rte.vars.mk index
>b518986..06dab18 100644
>--- a/mk/arch/tile/rte.vars.mk
>+++ b/mk/arch/tile/rte.vars.mk
>@@ -30,7 +30,13 @@
>
>
> ARCH ?= tile
>+
>+HOST_ARCH := ${shell uname -m}
>+ifneq ($(filter tile%,${HOST_ARCH}),)
>+CROSS =
>+else
> CROSS ?= tile-
>+endif
>
> CPU_CFLAGS ?=
> CPU_LDFLAGS ?=
>--
>1.7.1
Acked-by: Zhigang Lu
[dpdk-dev] [PATCH 2/3] driver/net/mpipe: optimize mpipe buffer return mechanism.
>-Original Message-
>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Liming Sun
>Sent: Tuesday, December 15, 2015 11:38 PM
>To: dev at dpdk.org
>Subject: [dpdk-dev] [PATCH 2/3] driver/net/mpipe: optimize mpipe buffer
return
>mechanism.
>
>This submit has changes to optimize the mpipe buffer return. When
>a packet is received, instead of allocating and refilling the
>buffer stack right away, it tracks the number of pending buffers,
>and use HW buffer return as an optimization when the pending
>number is below certain threshold, thus save two MMIO writes and
>improves performance especially for bidirectional traffic case.
>
>Signed-off-by: Liming Sun
>---
> drivers/net/mpipe/mpipe_tilegx.c | 50
>++---
> 1 files changed, 40 insertions(+), 10 deletions(-)
>
>diff --git a/drivers/net/mpipe/mpipe_tilegx.c
>b/drivers/net/mpipe/mpipe_tilegx.c
>index 35134ba..be7b6f2 100644
>--- a/drivers/net/mpipe/mpipe_tilegx.c
>+++ b/drivers/net/mpipe/mpipe_tilegx.c
>@@ -78,6 +78,13 @@ struct mpipe_context {
> struct mpipe_channel_config channels[MPIPE_MAX_CHANNELS];
> };
>
>+/* Per-core local data. */
>+struct mpipe_local {
>+ int mbuf_push_debt[RTE_MAX_ETHPORTS]; /* Buffer push debt. */
>+} __rte_cache_aligned;
>+
>+#define MPIPE_BUF_DEBT_THRESHOLD 32
>+static __thread struct mpipe_local mpipe_local;
> static struct mpipe_context mpipe_contexts[GXIO_MPIPE_INSTANCE_MAX];
> static int mpipe_instances;
> static const char *drivername = "MPIPE PMD";
>@@ -137,7 +144,7 @@ struct mpipe_dev_priv {
> int first_bucket; /* mPIPE bucket start index. */
> int first_ring; /* mPIPE notif ring start index. */
> int notif_group;/* mPIPE notif group. */
>- rte_atomic32_t dp_count;/* Active datapath thread count. */
>+ rte_atomic32_t dp_count __rte_cache_aligned;/* DP Entry count.
*/
> int tx_stat_mapping[RTE_ETHDEV_QUEUE_STAT_CNTRS];
> int rx_stat_mapping[RTE_ETHDEV_QUEUE_STAT_CNTRS];
> };
>@@ -461,6 +468,14 @@ mpipe_dp_wait(struct mpipe_dev_priv *priv)
> }
> }
>
>+static inline int
>+mpipe_mbuf_stack_index(struct mpipe_dev_priv *priv, struct rte_mbuf *mbuf)
>+{
>+ return (mbuf->port < RTE_MAX_ETHPORTS)?
>+ mpipe_priv(&rte_eth_devices[mbuf->port])->stack :
>+ priv->stack;
>+}
>+
> static inline struct rte_mbuf *
> mpipe_recv_mbuf(struct mpipe_dev_priv *priv, gxio_mpipe_idesc_t *idesc,
> int in_port)
>@@ -1267,6 +1282,7 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue,
>struct rte_mbuf **tx_pkts,
> unsigned nb_bytes = 0;
> unsigned nb_sent = 0;
> int nb_slots, i;
>+ uint8_t port_id;
>
> PMD_DEBUG_TX("Trying to transmit %d packets on %s:%d.\n",
>nb_pkts, mpipe_name(tx_queue->q.priv),
>@@ -1315,14 +1331,23 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue,
>struct rte_mbuf **tx_pkts,
> if (priv->tx_comps[idx])
> rte_pktmbuf_free_seg(priv->tx_comps[idx]);
>
>+ port_id = (mbuf->port < RTE_MAX_ETHPORTS)?
>+ mbuf->port : priv->port_id;
> desc = (gxio_mpipe_edesc_t) { {
> .va= rte_pktmbuf_mtod(mbuf,
uintptr_t),
> .xfer_size = rte_pktmbuf_data_len(mbuf),
> .bound = next ? 0 : 1,
>+ .stack_idx = mpipe_mbuf_stack_index(priv,
mbuf),
> } };
>+ if (mpipe_local.mbuf_push_debt[port_id] > 0) {
>+ mpipe_local.mbuf_push_debt[port_id]--;
>+ desc.hwb = 1;
>+ priv->tx_comps[idx] = NULL;
>+ }
>+ else
>+ priv->tx_comps[idx] = mbuf;
>
> nb_bytes += mbuf->data_len;
>- priv->tx_comps[idx] = mbuf;
> gxio_mpipe_equeue_put_at(equeue, desc, slot + i);
>
> PMD_DEBUG_TX("%s:%d: Sending packet %p, len %d\n",
>@@ -1443,17 +1468,22 @@ mpipe_do_recv(struct mpipe_rx_queue *rx_queue,
>struct rte_mbuf **rx_pkts,
> continue;
> }
>
>- mbuf = __rte_mbuf_raw_alloc(priv->rx_mpool);
>- if (unlikely(!mbuf)) {
>- nb_nomem++;
>- gxio_mpipe_iqueue_drop(iqueue, idesc);
>- PMD_DEBUG_RX("%s:%d: RX alloc failure\n",
>+ if (mpipe_local.mbuf_push_debt[in_port] <
>+ MPIPE_BUF_DEBT_THRESHOLD)
>+ mpipe_local.mbuf_push_debt[in_port]++;
>+ else {
>+ mbuf = __rte_mbuf_raw_alloc(priv->
[dpdk-dev] [PATCH 3/3] driver/net/mpipe: fix a mpipe link initialization ordering issue
>-Original Message-
>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Liming Sun
>Sent: Tuesday, December 15, 2015 11:38 PM
>To: dev at dpdk.org
>Subject: [dpdk-dev] [PATCH 3/3] driver/net/mpipe: fix a mpipe link
initialization
>ordering issue
>
>Mpipe link structure is initialized in function mpipe_link_init().
>Currently it's only called from the eth_dev_ops.dev_start, which
>caused crashes when link mgmt APIs (like promiscuous_enable)
>was called before eth_dev_ops.dev_start(). This submit fixed it
>by calling mpipe_link_init() in rte_pmd_mpipe_devinit().
>
>Signed-off-by: Liming Sun
>---
> drivers/net/mpipe/mpipe_tilegx.c | 14 +++---
> 1 files changed, 7 insertions(+), 7 deletions(-)
>
>diff --git a/drivers/net/mpipe/mpipe_tilegx.c
>b/drivers/net/mpipe/mpipe_tilegx.c
>index be7b6f2..5845511 100644
>--- a/drivers/net/mpipe/mpipe_tilegx.c
>+++ b/drivers/net/mpipe/mpipe_tilegx.c
>@@ -752,13 +752,6 @@ mpipe_init(struct mpipe_dev_priv *priv)
> if (priv->initialized)
> return 0;
>
>- rc = mpipe_link_init(priv);
>- if (rc < 0) {
>- RTE_LOG(ERR, PMD, "%s: Failed to init link.\n",
>- mpipe_name(priv));
>- return rc;
>- }
>-
> rc = mpipe_recv_init(priv);
> if (rc < 0) {
> RTE_LOG(ERR, PMD, "%s: Failed to init rx.\n",
>@@ -1633,6 +1626,13 @@ rte_pmd_mpipe_devinit(const char *ifname,
> eth_dev->rx_pkt_burst = &mpipe_recv_pkts;
> eth_dev->tx_pkt_burst = &mpipe_xmit_pkts;
>
>+ rc = mpipe_link_init(priv);
>+ if (rc < 0) {
>+ RTE_LOG(ERR, PMD, "%s: Failed to init link.\n",
>+ mpipe_name(priv));
>+ return rc;
>+ }
>+
> return 0;
> }
>
>--
>1.7.1
Acked-by: Zhigang Lu
[dpdk-dev] DPDK on Xeon Phi Processors
Hi, I was reading up on the Xeon Phi processors - up to 61 cores?.. looks very interesting. Has anyone tried offloading packet processing to such a co-processor utilising the existing code base? Thanks, Padam
[dpdk-dev] [PATCH v3 04/12] linuxapp/vfio: ignore mapping for ioport region
On Fri, Jan 08, 2016 at 12:23:15AM +0530, Santosh Shukla wrote:
>
>
> On Thu, Jan 7, 2016 at 11:46 PM, Stephen Hemminger networkplumber.org>
> wrote:
> ?
>
> > +? ? ? ? ? ? ?ret = pread64(vfio_dev_fd, &ioport_bar,
> sizeof(ioport_bar),
> > +? ? ? ? ? ? ? ? ? ? ? ? ? ?VFIO_GET_REGION_ADDR
> (VFIO_PCI_CONFIG_REGION_INDEX)
> > +? ? ? ? ? ? ? ? ? ? ? ? ? ?+ PCI_BASE_ADDRESS_0 + i*4);
> > +
> > +? ? ? ? ? ? ?if (ret != sizeof(ioport_bar)) {
> > +? ? ? ? ? ? ? ? ? ? ?RTE_LOG(ERR, EAL,
> > +? ? ? ? ? ? ? ? ? ? ? ? ? ? ?"Cannot read command (%x) from PCI config"
> > +? ? ? ? ? ? ? ? ? ? ? ? ? ? ?"space!\n", PCI_BASE_ADDRESS_0 + i*4);
>
> Please dont split the line of a log message string in mid sentence.
>
First of all, it'd be good if you can fix your email client to reply
emails in a way open source world prefer, such as prefixing last email
with leading '> ', and do not auto fold lines.
> me to don't like splitting, This was deliberate to keep checkpatch happy, If
> we
> are ok with debug message > 80 line warning I guess ?it will improve code
> readability.
You may try to shorten your message to "Cannot read cmd %x from config
space!" :)
>
>
> > +? ? ? ? ? ? ? ? ? ? ?return -1;
> > +? ? ? ? ? ? ?}
> > +
> > +? ? ? ? ? ? ?if (ioport_bar & PCI_BASE_ADDRESS_SPACE_IO) {
> > +? ? ? ? ? ? ? ? ? ? ?RTE_LOG(INFO, EAL, "\tIgnore mapping since Its a
> i/
> o"
> > +? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? "port bar (%d) addr : %x\n", i,
> same here
>
>
> Agreed.
Ditto, maybe following is better
"Ignore mapping IO port bar %d, addr: %x\n".
--yliu
[dpdk-dev] DPDK on Xeon Phi Processors
AVX-512 in Xeon Phi Knights Landing and Skylake Xeon seems ideal for Hyperscan. -- Jim > On Jan 7, 2016, at 10:37 PM, Padam Jeet Singh > wrote: > > Hi, > > I was reading up on the Xeon Phi processors - up to 61 cores?.. looks very > interesting. Has anyone tried offloading packet processing to such a > co-processor utilising the existing code base? > > Thanks, > Padam
[dpdk-dev] [PATCH 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM
>-Original Message-
>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Liming Sun
>Sent: Wednesday, December 23, 2015 3:50 AM
>To: dev at dpdk.org
>Subject: [dpdk-dev] [PATCH 1/2] driver/net/mpipe: add rte_vect.h and enable
>CONFIG_RTE_LIBRTE_LPM
>
>rte_vect.h was missing earlier thus LPM was disabled and l3fwd is
>not able to compile. This commit implements the vector api and
>enable LPM in the tilegx configuration by default. It also includes
>a minor optimization to use __insn_fetchadd4() instead of
>rte_atomic32_xxx() in mpipe_dp_enter/mpipe_dp_exit to avoid the
>unnecessary memory fence.
>
>Signed-off-by: Liming Sun
>---
> config/defconfig_tile-tilegx-linuxapp-gcc |2 +-
> drivers/net/mpipe/mpipe_tilegx.c | 18 +++-
> lib/librte_eal/common/include/arch/tile/rte_vect.h | 93
>
> 3 files changed, 107 insertions(+), 6 deletions(-)
> create mode 100644 lib/librte_eal/common/include/arch/tile/rte_vect.h
>
>diff --git a/config/defconfig_tile-tilegx-linuxapp-gcc
>b/config/defconfig_tile-tilegx-linuxapp-gcc
>index fb61bcd..39794f6 100644
>--- a/config/defconfig_tile-tilegx-linuxapp-gcc
>+++ b/config/defconfig_tile-tilegx-linuxapp-gcc
>@@ -64,7 +64,7 @@ CONFIG_RTE_LIBRTE_ENIC_PMD=n
>
> # This following libraries are not available on the tile architecture.
> # So they're turned off.
>-CONFIG_RTE_LIBRTE_LPM=n
>+CONFIG_RTE_LIBRTE_LPM=y
> CONFIG_RTE_LIBRTE_ACL=n
> CONFIG_RTE_LIBRTE_SCHED=n
> CONFIG_RTE_LIBRTE_PORT=n
>diff --git a/drivers/net/mpipe/mpipe_tilegx.c
>b/drivers/net/mpipe/mpipe_tilegx.c
>index 5845511..8d006fa 100644
>--- a/drivers/net/mpipe/mpipe_tilegx.c
>+++ b/drivers/net/mpipe/mpipe_tilegx.c
>@@ -451,13 +451,13 @@ static inline void
> mpipe_dp_enter(struct mpipe_dev_priv *priv)
> {
> __insn_mtspr(SPR_DSTREAM_PF, 0);
>- rte_atomic32_inc(&priv->dp_count);
>+ __insn_fetchadd4(&priv->dp_count, 1);
> }
>
> static inline void
> mpipe_dp_exit(struct mpipe_dev_priv *priv)
> {
>- rte_atomic32_dec(&priv->dp_count);
>+ __insn_fetchadd4(&priv->dp_count, -1);
> }
>
> static inline void
>@@ -484,12 +484,20 @@ mpipe_recv_mbuf(struct mpipe_dev_priv *priv,
>gxio_mpipe_idesc_t *idesc,
> uint16_t size = gxio_mpipe_idesc_get_xfer_size(idesc);
> struct rte_mbuf *mbuf = RTE_PTR_SUB(va, priv->rx_offset);
>
>- rte_pktmbuf_reset(mbuf);
> mbuf->data_off = (uintptr_t)va - (uintptr_t)mbuf->buf_addr;
>- mbuf->port = in_port;
>- mbuf->data_len = size;
>+ mbuf->nb_segs = 1;
>+ mbuf->port = in_port;
>+ mbuf->ol_flags = 0;
>+ if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv4)
>+ mbuf->packet_type = RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L2_ETHER;
>+ else if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv6)
>+ mbuf->packet_type = RTE_PTYPE_L3_IPV6 | RTE_PTYPE_L2_ETHER;
>+ else
>+ mbuf->packet_type = RTE_PTYPE_UNKNOWN;
> mbuf->pkt_len = size;
>+ mbuf->data_len = size;
> mbuf->hash.rss = gxio_mpipe_idesc_get_flow_hash(idesc);
>+ mbuf->next = NULL;
>
> PMD_DEBUG_RX("%s: RX mbuf %p, buffer %p, buf_addr %p, size %d\n",
>mpipe_name(priv), mbuf, va, mbuf->buf_addr, size);
>diff --git a/lib/librte_eal/common/include/arch/tile/rte_vect.h
>b/lib/librte_eal/common/include/arch/tile/rte_vect.h
>new file mode 100644
>index 000..32d768a
>--- /dev/null
>+++ b/lib/librte_eal/common/include/arch/tile/rte_vect.h
>@@ -0,0 +1,93 @@
>+/*
>+ * BSD LICENSE
>+ *
>+ * Copyright (C) EZchip Semiconductor Ltd. 2015.
>+ *
>+ * Redistribution and use in source and binary forms, with or without
>+ * modification, are permitted provided that the following conditions
>+ * are met:
>+ *
>+ * * Redistributions of source code must retain the above copyright
>+ * notice, this list of conditions and the following disclaimer.
>+ * * Redistributions in binary form must reproduce the above copyright
>+ * notice, this list of conditions and the following disclaimer in
>+ * the documentation and/or other materials provided with the
>+ * distribution.
>+ * * Neither the name of EZchip Semiconductor nor the names of its
>+ * contributors may be used to endorse or promote products derived
>+ * from this software without specific prior written permission.
>+ *
>+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
>CONTRIBUTORS
>+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT
>NOT
>+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
>FITNESS FOR
>+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
>COPYRIGHT
>+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
>INCIDENTAL,
>+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
>NOT
>+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
>OF USE,
>+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
>AND ON ANY
>+
[dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits
>-Original Message-
>From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Liming Sun
>Sent: Wednesday, December 23, 2015 3:50 AM
>To: dev at dpdk.org
>Subject: [dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue
when
>testpmd quits
>
>1. Fixed the compiling issue of the ethtool example on tilegx
> platform.
>2. Fixed the hung/crash issue when quitting testpmd under high
> traffic rate. The buffer error bit needs to be checked before
> processing the idesc and releasing the buffer. Code logic is
> also simplified.
>
>Signed-off-by: Liming Sun
>---
> drivers/net/mpipe/mpipe_tilegx.c | 46
>+++-
> .../common/include/arch/tile/rte_rwlock.h |1 +
> 2 files changed, 17 insertions(+), 30 deletions(-)
>
>diff --git a/drivers/net/mpipe/mpipe_tilegx.c
>b/drivers/net/mpipe/mpipe_tilegx.c
>index 8d006fa..4cb54c3 100644
>--- a/drivers/net/mpipe/mpipe_tilegx.c
>+++ b/drivers/net/mpipe/mpipe_tilegx.c
>@@ -134,7 +134,6 @@ struct mpipe_dev_priv {
> struct rte_mempool *rx_mpool; /* mpool used by the rx queues. */
> unsigned rx_offset; /* Receive head room. */
> unsigned rx_size_code; /* mPIPE rx buffer size code. */
>- unsigned rx_buffers;/* receive buffers on stack. */
> int is_xaui:1, /* Is this an xgbe or gbe? */
> initialized:1, /* Initialized port? */
> running:1; /* Running port? */
>@@ -529,7 +528,6 @@ mpipe_recv_fill_stack(struct mpipe_dev_priv *priv, int
>count)
> mpipe_recv_push(priv, mbuf);
> }
>
>- priv->rx_buffers += count;
> PMD_DEBUG_RX("%s: Filled %d/%d buffers\n", mpipe_name(priv), i,
>count); }
>
>@@ -539,10 +537,9 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
> const int offset = priv->rx_offset & ~RTE_MEMPOOL_ALIGN_MASK;
> uint8_t in_port = priv->port_id;
> struct rte_mbuf *mbuf;
>- unsigned count;
> void *va;
>
>- for (count = 0; count < priv->rx_buffers; count++) {
>+ while (1) {
> va = gxio_mpipe_pop_buffer(priv->context, priv->stack);
> if (!va)
> break;
>@@ -561,10 +558,6 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
>
> __rte_mbuf_raw_free(mbuf);
> }
>-
>- PMD_DEBUG_RX("%s: Returned %d/%d buffers\n",
>- mpipe_name(priv), count, priv->rx_buffers);
>- priv->rx_buffers -= count;
> }
>
> static void
>@@ -1246,31 +1239,23 @@ mpipe_recv_flush(struct mpipe_dev_priv *priv)
> gxio_mpipe_iqueue_t *iqueue;
> gxio_mpipe_idesc_t idesc;
> struct rte_mbuf *mbuf;
>- int retries = 0;
> unsigned queue;
>
>- do {
>- mpipe_recv_flush_stack(priv);
>-
>- /* Flush packets sitting in recv queues. */
>- for (queue = 0; queue < priv->nb_rx_queues; queue++) {
>- rx_queue = mpipe_rx_queue(priv, queue);
>- iqueue = &rx_queue->iqueue;
>- while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >=
0) {
>- mbuf = mpipe_recv_mbuf(priv, &idesc,
in_port);
>- rte_pktmbuf_free(mbuf);
>- priv->rx_buffers--;
>- }
>- rte_free(rx_queue->rx_ring_mem);
>- }
>- } while (retries++ < 10 && priv->rx_buffers);
>+ /* Release packets on the buffer stack. */
>+ mpipe_recv_flush_stack(priv);
>
>- if (priv->rx_buffers) {
>- RTE_LOG(ERR, PMD, "%s: Leaked %d receive buffers.\n",
>- mpipe_name(priv), priv->rx_buffers);
>- } else {
>- PMD_DEBUG_RX("%s: Returned all receive buffers.\n",
>- mpipe_name(priv));
>+ /* Flush packets sitting in recv queues. */
>+ for (queue = 0; queue < priv->nb_rx_queues; queue++) {
>+ rx_queue = mpipe_rx_queue(priv, queue);
>+ iqueue = &rx_queue->iqueue;
>+ while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >= 0) {
>+ /* Skip idesc with the 'buffer error' bit set. */
>+ if (idesc.be)
>+ continue;
>+ mbuf = mpipe_recv_mbuf(priv, &idesc, in_port);
>+ rte_pktmbuf_free(mbuf);
>+ }
>+ rte_free(rx_queue->rx_ring_mem);
> }
> }
>
>@@ -1339,6 +1324,7 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue,
>struct rte_mbuf **tx_pkts,
> .xfer_size = rte_pktmbuf_data_len(mbuf),
> .bound = next ? 0 : 1,
> .stack_idx = mpipe_mbuf_stack_index(priv,
mbuf),
>+ .size = priv->rx_size_code,
> } };
> if (mpipe_local.mbuf_push_debt[port_id] > 0
[dpdk-dev] [PATCH] mk: Fix examples install path
Hi Thomas, sorry being away so long due to the Christmas break. May bad assumption of being a typo was based on a coincidence combining a) former packaging in ubuntu had it in fact under /usr/share/doc/dpdk (I agree that just /usr/share/dpdk would be better) b) if you just and only ran a build and install-doc (no other install targets); the install failes just by missing the datadir That lead me to the wrong assumption it might be in /usr/share/doc/dpdk intentionally. But you are right, lets fix b) instead which is just a mkdir -p in the makefile. Alternative patch will follow. Christian Ehrhardt Software Engineer, Ubuntu Server Canonical Ltd On Wed, Dec 23, 2015 at 10:37 PM, Thomas Monjalon wrote: > Hi, > > 2015-12-22 14:13, Christian Ehrhardt: > > Depending on non-doc targets being built before and the setting of > DESTDIR > > the examples dir could in some cases not end up in the right target. > > Reason is just a typo variable reference in the copy target. > [...] > > - $(Q)cp -a $(RTE_SDK)/examples $(DESTDIR)$(datadir) > > + $(Q)cp -a $(RTE_SDK)/examples $(DESTDIR)$(docdir) > > No, it was not a typo. > Do you really think the examples code should be in the doc dir > (i.e. /usr/share/doc/dpdk) instead of datadir (i.e. /usr/share/dpdk)? >
[dpdk-dev] [PATCH] mk: Fix examples install
Depending on non-doc targets being built before and the setting of DESTDIR the copy of the examples dir being part of install-doc could in some cases fail with a non existant "$(DESTDIR)$(datadir)" target directory. Add the conditional rte_mkdir for that to avoid the issue. Signed-off-by: Christian Ehrhardt --- [diffstat] rte.sdkinstall.mk |1 + 1 file changed, 1 insertion(+) [diff] diff --git a/mk/rte.sdkinstall.mk b/mk/rte.sdkinstall.mk index c159bf7..68e56b6 100644 --- a/mk/rte.sdkinstall.mk +++ b/mk/rte.sdkinstall.mk @@ -157,4 +157,5 @@ ifneq ($(wildcard $O/doc/*/*/*pdf),) $(Q)$(call rte_mkdir, $(DESTDIR)$(docdir)/guides) $(Q)cp -a $O/doc/*/*/*pdf $(DESTDIR)$(docdir)/guides endif + $(Q)$(call rte_mkdir, $(DESTDIR)$(datadir)) $(Q)cp -a $(RTE_SDK)/examples $(DESTDIR)$(datadir)
[dpdk-dev] [PATCH] mk: Fix examples install
Forgot to mention, an easy way to test-trigger this e.g. on the dpdk 2.2 tarball. make clean; make config T=x86_64-native-linuxapp-gcc && && make -j && make V=1 DESTDIR=Test install-doc In this example the install-doc fails with this: [...] Build complete [x86_64-native-linuxapp-gcc] make -f /home/ubuntu/dpdk-2.2.0/mk/rte.sdkinstall.mk install-doc cp -a /home/ubuntu/dpdk-2.2.0/examples Test/usr/local/share/dpdk cp: cannot create directory ?Test/usr/local/share/dpdk?: No such file or directory /home/ubuntu/dpdk-2.2.0/mk/rte.sdkinstall.mk:160: recipe for target 'install-doc' failed make[1]: *** [install-doc] Error 1 /home/ubuntu/dpdk-2.2.0/mk/rte.sdkroot.mk:101: recipe for target 'install-doc' failed make: *** [install-doc] Error 2 Christian Ehrhardt Software Engineer, Ubuntu Server Canonical Ltd On Fri, Jan 8, 2016 at 12:03 PM, Christian Ehrhardt < christian.ehrhardt at canonical.com> wrote: > Depending on non-doc targets being built before and the setting of DESTDIR > the copy of the examples dir being part of install-doc could in some cases > fail > with a non existant "$(DESTDIR)$(datadir)" target directory. > Add the conditional rte_mkdir for that to avoid the issue. > > Signed-off-by: Christian Ehrhardt > --- > > [diffstat] > rte.sdkinstall.mk |1 + > 1 file changed, 1 insertion(+) > > [diff] > diff --git a/mk/rte.sdkinstall.mk b/mk/rte.sdkinstall.mk > index c159bf7..68e56b6 100644 > --- a/mk/rte.sdkinstall.mk > +++ b/mk/rte.sdkinstall.mk > @@ -157,4 +157,5 @@ ifneq ($(wildcard $O/doc/*/*/*pdf),) > $(Q)$(call rte_mkdir, $(DESTDIR)$(docdir)/guides) > $(Q)cp -a $O/doc/*/*/*pdf $(DESTDIR)$(docdir)/guides > endif > + $(Q)$(call rte_mkdir, $(DESTDIR)$(datadir)) > $(Q)cp -a $(RTE_SDK)/examples $(DESTDIR)$(datadir) >
[dpdk-dev] time to kill rte_pci_dev_ids.h
Hello, On Wed, Jan 6, 2016 at 2:40 AM, Thomas Monjalon wrote: > 2016-01-05 11:37, Stephen Hemminger: > > Has anyone looked at getting rid of rte_pci_dev_ids.h? > > The current method with #ifdef's and putting all devices in one file > > really doesn't scale well. Something more like other OS's where > > the data is only in each device driver would be better. > > I agree. > The PCI ids should be embedded in drivers. > If I remember well, it's already the case for some of them. > I would love to see a tool able to list the supported devices > by reading a binary libraries. > I have done the first part of the job, moving all pci dev ids to the pmds, just rebased it. I can post this. In the last patch of the series, I put all those ids in a dedicated section, that can be dumped with objdump. Having a tool for this (like what I proposed some time ago) would be great but I did not find time to write it. The thing is, what do we want to achieve with such a tool ? My usecase at the moment is for "automatic" device binding to kernel drivers before starting dpdk or when hotplugging. But for this, I think that introducing a hook system in the pci probe functions is better. I am thinking of a hook system that would execute an external tool (shell scripts, C, ...) with all needed informations: pci device, pmd that wants it, driver flags etc... Return value would tell eal what to do next. This would make it possible to implement the blacklist / whitelist out of eal itself, implement auto bind and maybe more. If I go this way, I don't think that we need to expose the pci dev ids :-) -- David Marchand
[dpdk-dev] [PATCH v2 1/3] driver/net/mpipe: support native build on tilegx platform.
This submit updates the CROSS setting to support native build on
TileGx platform. It also enable the combined library by default.
Signed-off-by: Liming Sun
Acked-by: Zhigang Lu
---
MAINTAINERS | 3 ++-
config/defconfig_tile-tilegx-linuxapp-gcc | 4
mk/arch/tile/rte.vars.mk | 6 ++
3 files changed, 12 insertions(+), 1 deletion(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 3292e84..8f7e9ca 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -138,8 +138,9 @@ M: Jianbo Liu
F: lib/librte_eal/common/include/arch/arm/*_64.h
F: lib/librte_acl/acl_run_neon.*
-EZchip TILE-Gx
+EZchip TILE-Gx/Mx
M: Zhigang Lu
+M: Liming Sun
F: lib/librte_eal/common/include/arch/tile/
F: drivers/net/mpipe/
diff --git a/config/defconfig_tile-tilegx-linuxapp-gcc
b/config/defconfig_tile-tilegx-linuxapp-gcc
index 9df9d7f..fb61bcd 100644
--- a/config/defconfig_tile-tilegx-linuxapp-gcc
+++ b/config/defconfig_tile-tilegx-linuxapp-gcc
@@ -70,3 +70,7 @@ CONFIG_RTE_LIBRTE_SCHED=n
CONFIG_RTE_LIBRTE_PORT=n
CONFIG_RTE_LIBRTE_TABLE=n
CONFIG_RTE_LIBRTE_PIPELINE=n
+CONFIG_RTE_LIBRTE_CXGBE_PMD=n
+
+# Compile combined lib by default.
+CONFIG_RTE_BUILD_COMBINE_LIBS=y
diff --git a/mk/arch/tile/rte.vars.mk b/mk/arch/tile/rte.vars.mk
index b518986..06dab18 100644
--- a/mk/arch/tile/rte.vars.mk
+++ b/mk/arch/tile/rte.vars.mk
@@ -30,7 +30,13 @@
ARCH ?= tile
+
+HOST_ARCH := ${shell uname -m}
+ifneq ($(filter tile%,${HOST_ARCH}),)
+CROSS =
+else
CROSS ?= tile-
+endif
CPU_CFLAGS ?=
CPU_LDFLAGS ?=
--
1.8.3.1
[dpdk-dev] [PATCH v2 2/3] driver/net/mpipe: optimize mpipe buffer return mechanism.
This submit has changes to optimize the mpipe buffer return. When
a packet is received, instead of allocating and refilling the
buffer stack right away, it tracks the number of pending buffers,
and use HW buffer return as an optimization when the pending
number is below certain threshold, thus save two MMIO writes and
improves performance especially for bidirectional traffic case.
Signed-off-by: Liming Sun
Acked-by: Zhigang Lu
---
drivers/net/mpipe/mpipe_tilegx.c | 50
1 file changed, 40 insertions(+), 10 deletions(-)
diff --git a/drivers/net/mpipe/mpipe_tilegx.c b/drivers/net/mpipe/mpipe_tilegx.c
index 35134ba..be7b6f2 100644
--- a/drivers/net/mpipe/mpipe_tilegx.c
+++ b/drivers/net/mpipe/mpipe_tilegx.c
@@ -78,6 +78,13 @@ struct mpipe_context {
struct mpipe_channel_config channels[MPIPE_MAX_CHANNELS];
};
+/* Per-core local data. */
+struct mpipe_local {
+ int mbuf_push_debt[RTE_MAX_ETHPORTS]; /* Buffer push debt. */
+} __rte_cache_aligned;
+
+#define MPIPE_BUF_DEBT_THRESHOLD 32
+static __thread struct mpipe_local mpipe_local;
static struct mpipe_context mpipe_contexts[GXIO_MPIPE_INSTANCE_MAX];
static int mpipe_instances;
static const char *drivername = "MPIPE PMD";
@@ -137,7 +144,7 @@ struct mpipe_dev_priv {
int first_bucket; /* mPIPE bucket start index. */
int first_ring; /* mPIPE notif ring start index. */
int notif_group;/* mPIPE notif group. */
- rte_atomic32_t dp_count;/* Active datapath thread count. */
+ rte_atomic32_t dp_count __rte_cache_aligned;/* DP Entry count. */
int tx_stat_mapping[RTE_ETHDEV_QUEUE_STAT_CNTRS];
int rx_stat_mapping[RTE_ETHDEV_QUEUE_STAT_CNTRS];
};
@@ -461,6 +468,14 @@ mpipe_dp_wait(struct mpipe_dev_priv *priv)
}
}
+static inline int
+mpipe_mbuf_stack_index(struct mpipe_dev_priv *priv, struct rte_mbuf *mbuf)
+{
+ return (mbuf->port < RTE_MAX_ETHPORTS)?
+ mpipe_priv(&rte_eth_devices[mbuf->port])->stack :
+ priv->stack;
+}
+
static inline struct rte_mbuf *
mpipe_recv_mbuf(struct mpipe_dev_priv *priv, gxio_mpipe_idesc_t *idesc,
int in_port)
@@ -1267,6 +1282,7 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue, struct
rte_mbuf **tx_pkts,
unsigned nb_bytes = 0;
unsigned nb_sent = 0;
int nb_slots, i;
+ uint8_t port_id;
PMD_DEBUG_TX("Trying to transmit %d packets on %s:%d.\n",
nb_pkts, mpipe_name(tx_queue->q.priv),
@@ -1315,14 +1331,23 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue, struct
rte_mbuf **tx_pkts,
if (priv->tx_comps[idx])
rte_pktmbuf_free_seg(priv->tx_comps[idx]);
+ port_id = (mbuf->port < RTE_MAX_ETHPORTS)?
+ mbuf->port : priv->port_id;
desc = (gxio_mpipe_edesc_t) { {
.va= rte_pktmbuf_mtod(mbuf, uintptr_t),
.xfer_size = rte_pktmbuf_data_len(mbuf),
.bound = next ? 0 : 1,
+ .stack_idx = mpipe_mbuf_stack_index(priv, mbuf),
} };
+ if (mpipe_local.mbuf_push_debt[port_id] > 0) {
+ mpipe_local.mbuf_push_debt[port_id]--;
+ desc.hwb = 1;
+ priv->tx_comps[idx] = NULL;
+ }
+ else
+ priv->tx_comps[idx] = mbuf;
nb_bytes += mbuf->data_len;
- priv->tx_comps[idx] = mbuf;
gxio_mpipe_equeue_put_at(equeue, desc, slot + i);
PMD_DEBUG_TX("%s:%d: Sending packet %p, len %d\n",
@@ -1443,17 +1468,22 @@ mpipe_do_recv(struct mpipe_rx_queue *rx_queue, struct
rte_mbuf **rx_pkts,
continue;
}
- mbuf = __rte_mbuf_raw_alloc(priv->rx_mpool);
- if (unlikely(!mbuf)) {
- nb_nomem++;
- gxio_mpipe_iqueue_drop(iqueue, idesc);
- PMD_DEBUG_RX("%s:%d: RX alloc failure\n",
+ if (mpipe_local.mbuf_push_debt[in_port] <
+ MPIPE_BUF_DEBT_THRESHOLD)
+ mpipe_local.mbuf_push_debt[in_port]++;
+ else {
+ mbuf = __rte_mbuf_raw_alloc(priv->rx_mpool);
+ if (unlikely(!mbuf)) {
+ nb_nomem++;
+ gxio_mpipe_iqueue_drop(iqueue, idesc);
+ PMD_DEBUG_RX("%s:%d: alloc failure\n",
[dpdk-dev] [PATCH v2 3/3] driver/net/mpipe: fix a mpipe link initialization ordering issue
Mpipe link structure is initialized in function mpipe_link_init().
Currently it's only called from the eth_dev_ops.dev_start, which
caused crashes when link mgmt APIs (like promiscuous_enable)
was called before eth_dev_ops.dev_start(). This submit fixed it
by calling mpipe_link_init() in rte_pmd_mpipe_devinit().
Signed-off-by: Liming Sun
Acked-by: Zhigang Lu
---
drivers/net/mpipe/mpipe_tilegx.c | 14 +++---
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/net/mpipe/mpipe_tilegx.c b/drivers/net/mpipe/mpipe_tilegx.c
index be7b6f2..5845511 100644
--- a/drivers/net/mpipe/mpipe_tilegx.c
+++ b/drivers/net/mpipe/mpipe_tilegx.c
@@ -752,13 +752,6 @@ mpipe_init(struct mpipe_dev_priv *priv)
if (priv->initialized)
return 0;
- rc = mpipe_link_init(priv);
- if (rc < 0) {
- RTE_LOG(ERR, PMD, "%s: Failed to init link.\n",
- mpipe_name(priv));
- return rc;
- }
-
rc = mpipe_recv_init(priv);
if (rc < 0) {
RTE_LOG(ERR, PMD, "%s: Failed to init rx.\n",
@@ -1633,6 +1626,13 @@ rte_pmd_mpipe_devinit(const char *ifname,
eth_dev->rx_pkt_burst = &mpipe_recv_pkts;
eth_dev->tx_pkt_burst = &mpipe_xmit_pkts;
+ rc = mpipe_link_init(priv);
+ if (rc < 0) {
+ RTE_LOG(ERR, PMD, "%s: Failed to init link.\n",
+ mpipe_name(priv));
+ return rc;
+ }
+
return 0;
}
--
1.8.3.1
[dpdk-dev] [PATCH v2 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM
rte_vect.h was missing earlier thus LPM was disabled and l3fwd is
not able to compile. This commit implements the vector api and
enable LPM in the tilegx configuration by default. It also includes
a minor optimization to use __insn_fetchadd4() instead of
rte_atomic32_xxx() in mpipe_dp_enter/mpipe_dp_exit to avoid the
unnecessary memory fence.
Signed-off-by: Liming Sun
Acked-by: Zhigang Lu
---
config/defconfig_tile-tilegx-linuxapp-gcc | 2 +-
drivers/net/mpipe/mpipe_tilegx.c | 18 +++--
lib/librte_eal/common/include/arch/tile/rte_vect.h | 93 ++
3 files changed, 107 insertions(+), 6 deletions(-)
create mode 100644 lib/librte_eal/common/include/arch/tile/rte_vect.h
diff --git a/config/defconfig_tile-tilegx-linuxapp-gcc
b/config/defconfig_tile-tilegx-linuxapp-gcc
index fb61bcd..39794f6 100644
--- a/config/defconfig_tile-tilegx-linuxapp-gcc
+++ b/config/defconfig_tile-tilegx-linuxapp-gcc
@@ -64,7 +64,7 @@ CONFIG_RTE_LIBRTE_ENIC_PMD=n
# This following libraries are not available on the tile architecture.
# So they're turned off.
-CONFIG_RTE_LIBRTE_LPM=n
+CONFIG_RTE_LIBRTE_LPM=y
CONFIG_RTE_LIBRTE_ACL=n
CONFIG_RTE_LIBRTE_SCHED=n
CONFIG_RTE_LIBRTE_PORT=n
diff --git a/drivers/net/mpipe/mpipe_tilegx.c b/drivers/net/mpipe/mpipe_tilegx.c
index 5845511..8d006fa 100644
--- a/drivers/net/mpipe/mpipe_tilegx.c
+++ b/drivers/net/mpipe/mpipe_tilegx.c
@@ -451,13 +451,13 @@ static inline void
mpipe_dp_enter(struct mpipe_dev_priv *priv)
{
__insn_mtspr(SPR_DSTREAM_PF, 0);
- rte_atomic32_inc(&priv->dp_count);
+ __insn_fetchadd4(&priv->dp_count, 1);
}
static inline void
mpipe_dp_exit(struct mpipe_dev_priv *priv)
{
- rte_atomic32_dec(&priv->dp_count);
+ __insn_fetchadd4(&priv->dp_count, -1);
}
static inline void
@@ -484,12 +484,20 @@ mpipe_recv_mbuf(struct mpipe_dev_priv *priv,
gxio_mpipe_idesc_t *idesc,
uint16_t size = gxio_mpipe_idesc_get_xfer_size(idesc);
struct rte_mbuf *mbuf = RTE_PTR_SUB(va, priv->rx_offset);
- rte_pktmbuf_reset(mbuf);
mbuf->data_off = (uintptr_t)va - (uintptr_t)mbuf->buf_addr;
- mbuf->port = in_port;
- mbuf->data_len = size;
+ mbuf->nb_segs = 1;
+ mbuf->port = in_port;
+ mbuf->ol_flags = 0;
+ if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv4)
+ mbuf->packet_type = RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L2_ETHER;
+ else if (gxio_mpipe_idesc_get_ethertype(idesc) == ETHER_TYPE_IPv6)
+ mbuf->packet_type = RTE_PTYPE_L3_IPV6 | RTE_PTYPE_L2_ETHER;
+ else
+ mbuf->packet_type = RTE_PTYPE_UNKNOWN;
mbuf->pkt_len = size;
+ mbuf->data_len = size;
mbuf->hash.rss = gxio_mpipe_idesc_get_flow_hash(idesc);
+ mbuf->next = NULL;
PMD_DEBUG_RX("%s: RX mbuf %p, buffer %p, buf_addr %p, size %d\n",
mpipe_name(priv), mbuf, va, mbuf->buf_addr, size);
diff --git a/lib/librte_eal/common/include/arch/tile/rte_vect.h
b/lib/librte_eal/common/include/arch/tile/rte_vect.h
new file mode 100644
index 000..32d768a
--- /dev/null
+++ b/lib/librte_eal/common/include/arch/tile/rte_vect.h
@@ -0,0 +1,93 @@
+/*
+ * BSD LICENSE
+ *
+ * Copyright (C) EZchip Semiconductor Ltd. 2015.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of EZchip Semiconductor nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef _RTE_VECT_H_
+#define _RTE_VECT_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef __int1
[dpdk-dev] [PATCH v2 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits
1. Fixed the compiling issue of the ethtool example on tilegx
platform.
2. Fixed the hung/crash issue when quitting testpmd under high
traffic rate. The buffer error bit needs to be checked before
processing the idesc and releasing the buffer. Code logic is
also simplified.
Signed-off-by: Liming Sun
Acked-by: Zhigang Lu
---
drivers/net/mpipe/mpipe_tilegx.c | 46 --
.../common/include/arch/tile/rte_rwlock.h | 1 +
2 files changed, 17 insertions(+), 30 deletions(-)
diff --git a/drivers/net/mpipe/mpipe_tilegx.c b/drivers/net/mpipe/mpipe_tilegx.c
index 8d006fa..4cb54c3 100644
--- a/drivers/net/mpipe/mpipe_tilegx.c
+++ b/drivers/net/mpipe/mpipe_tilegx.c
@@ -134,7 +134,6 @@ struct mpipe_dev_priv {
struct rte_mempool *rx_mpool; /* mpool used by the rx queues. */
unsigned rx_offset; /* Receive head room. */
unsigned rx_size_code; /* mPIPE rx buffer size code. */
- unsigned rx_buffers;/* receive buffers on stack. */
int is_xaui:1, /* Is this an xgbe or gbe? */
initialized:1, /* Initialized port? */
running:1; /* Running port? */
@@ -529,7 +528,6 @@ mpipe_recv_fill_stack(struct mpipe_dev_priv *priv, int
count)
mpipe_recv_push(priv, mbuf);
}
- priv->rx_buffers += count;
PMD_DEBUG_RX("%s: Filled %d/%d buffers\n", mpipe_name(priv), i, count);
}
@@ -539,10 +537,9 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
const int offset = priv->rx_offset & ~RTE_MEMPOOL_ALIGN_MASK;
uint8_t in_port = priv->port_id;
struct rte_mbuf *mbuf;
- unsigned count;
void *va;
- for (count = 0; count < priv->rx_buffers; count++) {
+ while (1) {
va = gxio_mpipe_pop_buffer(priv->context, priv->stack);
if (!va)
break;
@@ -561,10 +558,6 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
__rte_mbuf_raw_free(mbuf);
}
-
- PMD_DEBUG_RX("%s: Returned %d/%d buffers\n",
-mpipe_name(priv), count, priv->rx_buffers);
- priv->rx_buffers -= count;
}
static void
@@ -1246,31 +1239,23 @@ mpipe_recv_flush(struct mpipe_dev_priv *priv)
gxio_mpipe_iqueue_t *iqueue;
gxio_mpipe_idesc_t idesc;
struct rte_mbuf *mbuf;
- int retries = 0;
unsigned queue;
- do {
- mpipe_recv_flush_stack(priv);
-
- /* Flush packets sitting in recv queues. */
- for (queue = 0; queue < priv->nb_rx_queues; queue++) {
- rx_queue = mpipe_rx_queue(priv, queue);
- iqueue = &rx_queue->iqueue;
- while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >= 0) {
- mbuf = mpipe_recv_mbuf(priv, &idesc, in_port);
- rte_pktmbuf_free(mbuf);
- priv->rx_buffers--;
- }
- rte_free(rx_queue->rx_ring_mem);
- }
- } while (retries++ < 10 && priv->rx_buffers);
+ /* Release packets on the buffer stack. */
+ mpipe_recv_flush_stack(priv);
- if (priv->rx_buffers) {
- RTE_LOG(ERR, PMD, "%s: Leaked %d receive buffers.\n",
- mpipe_name(priv), priv->rx_buffers);
- } else {
- PMD_DEBUG_RX("%s: Returned all receive buffers.\n",
-mpipe_name(priv));
+ /* Flush packets sitting in recv queues. */
+ for (queue = 0; queue < priv->nb_rx_queues; queue++) {
+ rx_queue = mpipe_rx_queue(priv, queue);
+ iqueue = &rx_queue->iqueue;
+ while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >= 0) {
+ /* Skip idesc with the 'buffer error' bit set. */
+ if (idesc.be)
+ continue;
+ mbuf = mpipe_recv_mbuf(priv, &idesc, in_port);
+ rte_pktmbuf_free(mbuf);
+ }
+ rte_free(rx_queue->rx_ring_mem);
}
}
@@ -1339,6 +1324,7 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue, struct
rte_mbuf **tx_pkts,
.xfer_size = rte_pktmbuf_data_len(mbuf),
.bound = next ? 0 : 1,
.stack_idx = mpipe_mbuf_stack_index(priv, mbuf),
+ .size = priv->rx_size_code,
} };
if (mpipe_local.mbuf_push_debt[port_id] > 0) {
mpipe_local.mbuf_push_debt[port_id]--;
diff --git a/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
b/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
index 8f67a19..6d609e8 100644
--- a/lib/librte_eal/common/include/arch/tile/rte_r
[dpdk-dev] DPDK 2.1 on CentOS 7
Hi All,
I am facing an issue while compiling DPDK 2.1 on CentOS 7. Following is a
snapshot -
CC [M]
?/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_hwmon.o?
CC [M]
?/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.o/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_bridge_getlink?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:2284:2:
error: too few arguments to function ?ndo_dflt_bridge_getlink?? return
ndo_dflt_bridge_getlink(skb, pid, seq, dev, mode);? ^In file included from
/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/net/dst.h:13:0,? ? ? ? ? ? ?
? ?from /usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/net/sock.h:68,? ? ?
? ? ? ? ? ?from
/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/linux/tcp.h:23,? ? ? ? ? ? ?
? ?from
/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:34:/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/linux/rtnetlink.h:87:12:
note: declared here?extern int ndo_dflt_bridge_getlink(struct sk_buff *skb,
u32 pid, u32 seq,? ? ? ? ? ?
^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_xmit_frame_ring?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:5498:2:
error: implicit declaration of function ?vlan_tx_tag_present?
[-Werror=implicit-function-declaration]? if (vlan_tx_tag_present(skb)) {?
^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:5500:3:
error: implicit declaration of function ?vlan_tx_tag_ge ?
[-Werror=implicit-function-declaration]? ?tx_flags |= (vlan_tx_tag_get(skb) <<
IGB_TX_FLAGS_VLAN_SHIFT);?
?^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_get_vf_config?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:9467:5:
error: ?struct ifla_vf_info? has no member named ?tx_rate?? ivi->tx_rate =
adapter->vf_data[vf].tx_rate;? ?
?^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_bridge_getlink?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:2286:1:
error: control reaches end of non-void function [-Werror=return-type]?}?^cc1:
all warnings being treated as errorsmake[10]: ***
[/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.o]
Error 1make[9]: ***
[_module_/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni]
Error 2make[8]: *** [sub-make] Error 2make[7]: *** [rte_kni.ko] Error
2make[6]: *** [kni] Error 2make[5]: *** [linuxapp] Error 2make[4]: ***
[librte_eal] Error 2make[3]: *** [lib] Error 2make[2]: *** [all] Error
2make[1]: *** [x86_64-native-linuxapp-gcc_install] Error 2make: *** [install]
Error 2
>?
Can you please help me resolve this issue?
Thanks-sundar
[dpdk-dev] DPDK 2.1 on CentOS 7
The patch has been included in DPDK 2.2.0:
http://www.dpdk.org/browse/dpdk/commit/?id=261f0390841d1531d1d0fe88e2375e254cb63716
- Lee Roberts
-Original Message-
From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Sundar Ramakrishnan
Sent: Friday, January 08, 2016 2:17 PM
To: dev at dpdk.org
Subject: [dpdk-dev] DPDK 2.1 on CentOS 7
Hi All,
I am facing an issue while compiling DPDK 2.1 on CentOS 7. Following is a
snapshot -
CC [M]
?/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_hwmon.o?
CC [M]
?/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.o/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_bridge_getlink?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:2284:2:
error: too few arguments to function ?ndo_dflt_bridge_getlink?? return
ndo_dflt_bridge_getlink(skb, pid, seq, dev, mode);? ^In file included from
/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/net/dst.h:13:0,? ? ? ? ? ? ?
? ?from /usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/net/sock.h:68,? ? ?
? ? ? ? ? ?from
/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/linux/tcp.h:23,? ? ? ? ? ? ?
? ?from
/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:34:/usr/src/kernels/3.10.0-327.4.4.el7.x86_64/include/linux/rtnetlink.h:87:12:
note: declared here?extern int ndo_dflt_bridge_getlink(struct sk_buff *skb,
u32 pid, u32 seq,? ? ? ? ? ?
^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_xmit_frame_ring?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:5498:2:
error: implicit declaration of function ?vlan_tx_tag_present?
[-Werror=implicit-function-declaration]? if (vlan_tx_tag_present(skb)) {?
^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:5500:3:
error: implicit declaration of function ?vlan_tx_tag_ge ?
[-Werror=implicit-function-declaration]? ?tx_flags |= (vlan_tx_tag_get(skb) <<
IGB_TX_FLAGS_VLAN_SHIFT);?
?^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_get_vf_config?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:9467:5:
error: ?struct ifla_vf_info? has no member named ?tx_rate?? ivi->tx_rate =
adapter->vf_data[vf].tx_rate;? ?
?^/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:
In function
?igb_ndo_bridge_getlink?:/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.c:2286:1:
error: control reaches end of non-void function [-Werror=return-type]?}?^cc1:
all warnings being treated as errorsmake[10]: ***
[/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni/igb_main.o]
Error 1make[9]: ***
[_module_/home/dpdk-2.1.0/x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/kni]
Error 2make[8]: *** [sub-make] Error 2make[7]: *** [rte_kni.ko] Error
2make[6]: *** [kni] Error 2make[5]: *** [linuxapp] Error 2make[4]: ***
[librte_eal] Error 2make[3]: *** [lib] Error 2make[2]: *** [all] Error
2make[1]: *** [x86_64-native-linuxapp-gcc_install] Error 2make: *** [install]
Error 2
>?
Can you please help me resolve this issue?
Thanks-sundar
