[dpdk-dev] [vmxnet3 PATCH] doc: describe kernel dependency and --vmware-tsc-map
On 07/25/2013 10:38 PM, Thomas Monjalon wrote: > Be more verbose on kernel dependency about version and parameters. > And add description of DPDK parameter for VMware. > > Signed-off-by: Thomas Monjalon > --- > README.rst | 17 - > 1 file changed, 12 insertions(+), 5 deletions(-) > > diff --git a/README.rst b/README.rst > index 55436a4..c718ec7 100644 > --- a/README.rst > +++ b/README.rst > @@ -49,6 +49,9 @@ HOW TO BUILD > > There are two build dependencies: Linux kernel headers and DPDK >= 1.2.3r3. > > +The kernel driver is based on vmxnet3-8.6.10 from VMware ESXi-5.0.0u2. > +So the latest supported Linux version is 3.2. > + > DPDK must be built first in order to have all its header files in the same > directory, that RTE_INCLUDE must point to. > > @@ -93,15 +96,19 @@ Other parameters have a global effect: > - ``num_rxds=int``: number of RX descriptors per queue > - ``num_txds=int``: number of TX descriptors per queue > > +Because the kernel module initializes interfaces, **it's very important that > +DPDK parameters match kernel ones**. It won't work otherwise. > + > For more information, use ``modinfo vmxnet3-usermap.ko``. > > -The values of the kernel module parameters must match the DPDK ones. > +In userland, librte_pmd_vmxnet3.so must be loaded by the DPDK using the > ``-d`` > +option. > > -Because the kernel module initializes interfaces, it's very important that > -DPDK parameters match kernel ones. It won't work otherwise. > +Another interesting option is ``--vmware-tsc-map``. > +It allows VMware TSC mapping which provides an access to physical TSC. > +When this option is used, take care of setting this ESXi parameter:: > > -In userland, librte_pmd_vmxnet3.so must be loaded by the DPDK using the "-d" > -option. > + monitor_control.pseudo_perfctr = TRUE > > ``testpmd`` example for two ports, matching the ``insmod`` command above:: > ack
[dpdk-dev] [vmxnet3 PATCH] pmd: show banner before initializing
On 07/25/2013 10:39 PM, Thomas Monjalon wrote: > Print title before running driver initialization. > > Signed-off-by: Thomas Monjalon ack
[dpdk-dev] [vmxnet3 PATCH] pmd: support dpdk-1.3
On 07/25/2013 10:39 PM, Thomas Monjalon wrote: > API changes in DPDK 1.3: > - queue arrays are already allocated by rte_ethdev > - queue could be released for reconfiguration (not supported) > - queue struct becomes an opaque pointer > - vlan_filter_set can return an error > - bit-fields vlan_macip have moved > > Signed-off-by: Thomas Monjalon ack
[dpdk-dev] [virtio PATCH] pmd: support dpdk-1.3
On Thu, Jul 25, 2013 at 10:40:56PM +0200, Thomas Monjalon wrote: > From: Damien Millescamps > > API changes in DPDK 1.3: > - queue arrays are already allocated by rte_ethdev > - queue release must be handled by the PMD > - queue struct becomes an opaque pointer > - vlan_filter_set can return an error > - bit-fields vlan_macip have moved > > Signed-off-by: Damien Millescamps > Signed-off-by: Thomas Monjalon [...] Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [vmxnet3 PATCH] doc: describe kernel dependency and --vmware-tsc-map
26/07/2013 11:01, Damien Millescamps : > On 07/25/2013 10:38 PM, Thomas Monjalon wrote: > > Be more verbose on kernel dependency about version and parameters. > > And add description of DPDK parameter for VMware. > > ack pushed -- Thomas
[dpdk-dev] [vmxnet3 PATCH] pmd: show banner before initializing
26/07/2013 11:01, Damien Millescamps : > On 07/25/2013 10:39 PM, Thomas Monjalon wrote: > > Print title before running driver initialization. > > ack pushed -- Thomas
[dpdk-dev] [vmxnet3 PATCH] pmd: support dpdk-1.3
26/07/2013 11:02, Damien Millescamps : > On 07/25/2013 10:39 PM, Thomas Monjalon wrote: > > API changes in DPDK 1.3: > > - queue arrays are already allocated by rte_ethdev > > - queue could be released for reconfiguration (not supported) > > - queue struct becomes an opaque pointer > > - vlan_filter_set can return an error > > - bit-fields vlan_macip have moved > > ack pushed -- Thomas
[dpdk-dev] [virtio PATCH] pmd: support dpdk-1.3
26/07/2013 11:05, Adrien Mazarguil : > On Thu, Jul 25, 2013 at 10:40:56PM +0200, Thomas Monjalon wrote: > > From: Damien Millescamps > > > > API changes in DPDK 1.3: > > - queue arrays are already allocated by rte_ethdev > > - queue release must be handled by the PMD > > - queue struct becomes an opaque pointer > > - vlan_filter_set can return an error > > - bit-fields vlan_macip have moved > > > > Signed-off-by: Damien Millescamps > > Signed-off-by: Thomas Monjalon > > [...] > > Acked-by: Adrien Mazarguil pushed -- Thomas
[dpdk-dev] [dpdk-announce] [vNICs] virtio-net-pmd-1.1 and vmxnet3-usermap-1.1 released
The new releases can be downloaded here: http://dpdk.org/browse/virtio-net-pmd/tag/?id=v1.1 http://dpdk.org/browse/vmxnet3-usermap/tag/?id=v1.1 These new versions are compatible with DPDK 1.3.1r1. -- Thomas
[dpdk-dev] [PATCH] config: disable KNI for 32-bit because cannot work
From: Jean-Mickael Guerin This is not supported, disable to avoid compilation error like: lib/librte_eal/linuxapp/kni/kni_misc.c:304:2: error: format '%llx' expects argument of type 'long long unsigned int', but argument 2 has type 'phys_addr_t' [-Werror=format] Signed-off-by: Jean-Mickael Guerin --- config/defconfig_i686-default-linuxapp-gcc |4 ++-- config/defconfig_i686-default-linuxapp-icc |4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/config/defconfig_i686-default-linuxapp-gcc b/config/defconfig_i686-default-linuxapp-gcc index 756f1cd..bcea6f9 100644 --- a/config/defconfig_i686-default-linuxapp-gcc +++ b/config/defconfig_i686-default-linuxapp-gcc @@ -222,9 +222,9 @@ CONFIG_RTE_LIBRTE_LPM_DEBUG=n CONFIG_RTE_LIBRTE_NET=y # -# Compile librte_kni +# KNI is not supported on 32-bit # -CONFIG_RTE_LIBRTE_KNI=y +CONFIG_RTE_LIBRTE_KNI=n # # Enable warning directives diff --git a/config/defconfig_i686-default-linuxapp-icc b/config/defconfig_i686-default-linuxapp-icc index e2d268b..125277e 100644 --- a/config/defconfig_i686-default-linuxapp-icc +++ b/config/defconfig_i686-default-linuxapp-icc @@ -222,9 +222,9 @@ CONFIG_RTE_LIBRTE_LPM_DEBUG=n CONFIG_RTE_LIBRTE_NET=y # -# Compile librte_kni +# KNI is not supported on 32-bit # -CONFIG_RTE_LIBRTE_KNI=y +CONFIG_RTE_LIBRTE_KNI=n # # Enable warning directives -- 1.7.10.4
[dpdk-dev] [PATCH] kni: fix build with 802.1p kernel support
C90 compilers forbid mixed declaration and code.
Signed-off-by: Thomas Monjalon
---
lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c |7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c
b/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c
index 85de817..947be44 100755
--- a/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c
+++ b/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c
@@ -314,6 +314,9 @@ void ixgbe_vlan_mode(struct net_device *netdev, u32
features)
#endif
{
struct ixgbe_adapter *adapter = netdev_priv(netdev);
+#ifdef HAVE_8021P_SUPPORT
+ bool enable;
+#endif
#ifdef HAVE_VLAN_RX_REGISTER
//if (!test_bit(__IXGBE_DOWN, &adapter->state))
@@ -326,9 +329,9 @@ void ixgbe_vlan_mode(struct net_device *netdev, u32
features)
#endif
#ifdef HAVE_8021P_SUPPORT
#ifdef HAVE_VLAN_RX_REGISTER
- bool enable = (grp || (adapter->flags & IXGBE_FLAG_DCB_ENABLED));
+ enable = (grp || (adapter->flags & IXGBE_FLAG_DCB_ENABLED));
#else
- bool enable = !!(features & NETIF_F_HW_VLAN_RX);
+ enable = !!(features & NETIF_F_HW_VLAN_RX);
#endif
if (enable)
/* enable VLAN tag insert/strip */
--
1.7.10.4
[dpdk-dev] [PATCH] kni: fix build with kernel < 3.3 with netdev_features_t backport
The netdev_features_t typedef appeared in Linux 3.3, but checking the kernel
version isn't enough with some distributions (such as Debian Wheezy) that
backported it into 3.2, causing a compilation failure due to redefinition.
Since the presence of a typedef can't be tested at compile time, this commit
adds type kni_netdev_features_t, which, depending on the kernel version,
translates either to u32 or netdev_features_t.
Signed-off-by: Adrien Mazarguil
---
lib/librte_eal/linuxapp/kni/ethtool/igb/igb_main.c | 6 +++---
lib/librte_eal/linuxapp/kni/ethtool/igb/kcompat.h | 3 ++-
lib/librte_eal/linuxapp/kni/ethtool/ixgbe/kcompat.h | 3 ++-
3 files changed, 7 insertions(+), 5 deletions(-)
diff --git a/lib/librte_eal/linuxapp/kni/ethtool/igb/igb_main.c
b/lib/librte_eal/linuxapp/kni/ethtool/igb/igb_main.c
index f0728bb..9d772fa 100644
--- a/lib/librte_eal/linuxapp/kni/ethtool/igb/igb_main.c
+++ b/lib/librte_eal/linuxapp/kni/ethtool/igb/igb_main.c
@@ -1831,8 +1831,8 @@ void igb_reset(struct igb_adapter *adapter)
}
#ifdef HAVE_NDO_SET_FEATURES
-static netdev_features_t igb_fix_features(struct net_device *netdev,
- netdev_features_t features)
+static kni_netdev_features_t igb_fix_features(struct net_device *netdev,
+ kni_netdev_features_t features)
{
/*
* Since there is no support for separate tx vlan accel
@@ -1849,7 +1849,7 @@ static netdev_features_t igb_fix_features(struct
net_device *netdev,
}
static int igb_set_features(struct net_device *netdev,
- netdev_features_t features)
+ kni_netdev_features_t features)
{
u32 changed = netdev->features ^ features;
diff --git a/lib/librte_eal/linuxapp/kni/ethtool/igb/kcompat.h
b/lib/librte_eal/linuxapp/kni/ethtool/igb/kcompat.h
index 553e1c0..a2aa361 100644
--- a/lib/librte_eal/linuxapp/kni/ethtool/igb/kcompat.h
+++ b/lib/librte_eal/linuxapp/kni/ethtool/igb/kcompat.h
@@ -3012,8 +3012,9 @@ static inline void __kc_skb_frag_unref(skb_frag_t *frag)
/*/
#if ( LINUX_VERSION_CODE < KERNEL_VERSION(3,3,0) )
-typedef u32 netdev_features_t;
+typedef u32 kni_netdev_features_t;
#else /* ! < 3.3.0 */
+typedef netdev_features_t kni_netdev_features_t;
#define HAVE_INT_NDO_VLAN_RX_ADD_VID
#ifdef ETHTOOL_SRXNTUPLE
#undef ETHTOOL_SRXNTUPLE
diff --git a/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/kcompat.h
b/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/kcompat.h
index a0e0698..6ac890a 100644
--- a/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/kcompat.h
+++ b/lib/librte_eal/linuxapp/kni/ethtool/ixgbe/kcompat.h
@@ -3078,8 +3078,9 @@ static inline void __kc_skb_frag_unref(skb_frag_t *frag)
/*/
#if ( LINUX_VERSION_CODE < KERNEL_VERSION(3,3,0) )
-typedef u32 netdev_features_t;
+typedef u32 kni_netdev_features_t;
#else /* ! < 3.3.0 */
+typedef netdev_features_t kni_netdev_features_t;
#define HAVE_INT_NDO_VLAN_RX_ADD_VID
#ifdef ETHTOOL_SRXNTUPLE
#undef ETHTOOL_SRXNTUPLE
--
1.7.12
[dpdk-dev] [PATCH] eal: fix type of pointer arithmetic result
Adding or subtracting a value to a pointer makes a new pointer
of unknown type.
So typeof() is replaced by (void*) in RTE_PTR_ADD() and RTE_PTR_SUB().
But RTE_PTR_ALIGN_* macros have in their explicit API to return a pointer
of the same type. Since RTE_PTR_ALIGN_CEIL is based on RTE_PTR_ADD, a
typeof() is added to keep the original behaviour.
Signed-off-by: Thomas Monjalon
---
lib/librte_eal/common/include/rte_common.h |6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/librte_eal/common/include/rte_common.h
b/lib/librte_eal/common/include/rte_common.h
index 96a..d2e096d 100644
--- a/lib/librte_eal/common/include/rte_common.h
+++ b/lib/librte_eal/common/include/rte_common.h
@@ -70,12 +70,12 @@ extern "C" {
/**
* add a byte-value offset from a pointer
*/
-#define RTE_PTR_ADD(ptr, x) ((typeof(ptr))((uintptr_t)ptr + (x)))
+#define RTE_PTR_ADD(ptr, x) ((void*)((uintptr_t)ptr + (x)))
/**
* subtract a byte-value offset from a pointer
*/
-#define RTE_PTR_SUB(ptr, x) ((typeof(ptr))((uintptr_t)ptr - (x)))
+#define RTE_PTR_SUB(ptr, x) ((void*)((uintptr_t)ptr - (x)))
/**
* get the difference between two pointer values, i.e. how far apart
@@ -131,7 +131,7 @@ rte_align_floor_int(uintptr_t ptr, uintptr_t align)
* must be a power-of-two value.
*/
#define RTE_PTR_ALIGN_CEIL(ptr, align) \
- RTE_PTR_ALIGN_FLOOR(RTE_PTR_ADD(ptr, align - 1), align)
+ RTE_PTR_ALIGN_FLOOR((typeof(ptr))RTE_PTR_ADD(ptr, align - 1), align)
/**
* Macro to align a value to a given power-of-two. The resultant value
--
1.7.10.4
[dpdk-dev] [PATCH] eal: fix recording of detected/enabled logical cores
From: Ivan Boule
1) In the EAL initialization phase, invoke the function rte_eal_cpu_init
to detect the set of running cores (and enable them by default) before
processing the [enabled] core mask option that is performed during the
parsing of EAL arguments.
2) In the function rte_eal_cpu_init():
- to parse the set of all running logical cores on the machine, do not
use the RTE_LCORE_FOREACH macro that considers the set of already
detected cores...
Instead, use a standard loop based on the RTE_MAX_LCORE constant.
- explicitely set to ROLE_RTE the role of each detected logical core
that is recorded in the EAL configuration, as all running cores are
enabled by default.
3) In the function eal_parse_coremask(), update the "lcore_count" field
of the EAL configuration with the effective number of logical cores
that are set in the mask of enabled logical cores.
Signed-off-by: Ivan Boule
---
lib/librte_eal/linuxapp/eal/eal.c |8 +---
lib/librte_eal/linuxapp/eal/eal_lcore.c | 12
2 files changed, 13 insertions(+), 7 deletions(-)
diff --git a/lib/librte_eal/linuxapp/eal/eal.c
b/lib/librte_eal/linuxapp/eal/eal.c
index d94c6bb..ed0e9b1 100644
--- a/lib/librte_eal/linuxapp/eal/eal.c
+++ b/lib/librte_eal/linuxapp/eal/eal.c
@@ -389,6 +389,8 @@ eal_parse_coremask(const char *coremask)
cfg->lcore_role[i] = ROLE_OFF;
}
}
+ /* Update the count of enabled logical cores of the EAL configuration */
+ cfg->lcore_count = count;
return 0;
}
@@ -817,6 +819,9 @@ rte_eal_init(int argc, char **argv)
if (rte_eal_log_early_init() < 0)
rte_panic("Cannot init early logs\n");
+ if (rte_eal_cpu_init() < 0)
+ rte_panic("Cannot detect lcores\n");
+
fctret = eal_parse_args(argc, argv);
if (fctret < 0)
exit(1);
@@ -849,9 +854,6 @@ rte_eal_init(int argc, char **argv)
rte_config_init();
- if (rte_eal_cpu_init() < 0)
- rte_panic("Cannot detect lcores\n");
-
if (rte_eal_memory_init() < 0)
rte_panic("Cannot init memory\n");
diff --git a/lib/librte_eal/linuxapp/eal/eal_lcore.c
b/lib/librte_eal/linuxapp/eal/eal_lcore.c
index 7487f34..768e333 100644
--- a/lib/librte_eal/linuxapp/eal/eal_lcore.c
+++ b/lib/librte_eal/linuxapp/eal/eal_lcore.c
@@ -152,15 +152,19 @@ rte_eal_cpu_init(void)
unsigned lcore_id;
unsigned count = 0;
- /* disable lcores that were not detected */
- RTE_LCORE_FOREACH(lcore_id) {
-
+ /*
+* Parse the maximum set of logical cores, detect the subset of running
+* ones and enable them by default.
+*/
+ for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
lcore_config[lcore_id].detected = cpu_detected(lcore_id);
if (lcore_config[lcore_id].detected == 0) {
RTE_LOG(DEBUG, EAL, "Skip lcore %u (not detected)\n",
lcore_id);
config->lcore_role[lcore_id] = ROLE_OFF;
continue;
}
+ /* By default, each detected core is enabled */
+ config->lcore_role[lcore_id] = ROLE_RTE;
lcore_config[lcore_id].core_id = cpu_core_id(lcore_id);
lcore_config[lcore_id].socket_id = cpu_socket_id(lcore_id);
if (lcore_config[lcore_id].socket_id >= RTE_MAX_NUMA_NODES)
@@ -177,7 +181,7 @@ rte_eal_cpu_init(void)
lcore_config[lcore_id].socket_id);
count ++;
}
-
+ /* Set the count of enabled logical cores of the EAL configuration */
config->lcore_count = count;
return 0;
--
1.7.10.4
[dpdk-dev] [PATCH] config: disable KNI for 32-bit because cannot work
On Fri, Jul 26, 2013 at 11:49:59AM +0200, Thomas Monjalon wrote: > From: Jean-Mickael Guerin > > This is not supported, disable to avoid compilation error like: > lib/librte_eal/linuxapp/kni/kni_misc.c:304:2: error: > format '%llx' expects argument of type 'long long unsigned int', > but argument 2 has type 'phys_addr_t' [-Werror=format] > > Signed-off-by: Jean-Mickael Guerin > --- > config/defconfig_i686-default-linuxapp-gcc |4 ++-- > config/defconfig_i686-default-linuxapp-icc |4 ++-- > 2 files changed, 4 insertions(+), 4 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH] kni: fix build with 802.1p kernel support
On Fri, Jul 26, 2013 at 11:51:15AM +0200, Thomas Monjalon wrote: > C90 compilers forbid mixed declaration and code. > > Signed-off-by: Thomas Monjalon > --- > lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c |7 +-- > 1 file changed, 5 insertions(+), 2 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH] eal: fix type of pointer arithmetic result
On Fri, Jul 26, 2013 at 01:34:00PM +0200, Thomas Monjalon wrote: > Adding or subtracting a value to a pointer makes a new pointer > of unknown type. > So typeof() is replaced by (void*) in RTE_PTR_ADD() and RTE_PTR_SUB(). > > But RTE_PTR_ALIGN_* macros have in their explicit API to return a pointer > of the same type. Since RTE_PTR_ALIGN_CEIL is based on RTE_PTR_ADD, a > typeof() is added to keep the original behaviour. > > Signed-off-by: Thomas Monjalon > --- > lib/librte_eal/common/include/rte_common.h |6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH] eal: fix recording of detected/enabled logical cores
On Fri, Jul 26, 2013 at 01:55:49PM +0200, Thomas Monjalon wrote: > From: Ivan Boule > > 1) In the EAL initialization phase, invoke the function rte_eal_cpu_init >to detect the set of running cores (and enable them by default) before >processing the [enabled] core mask option that is performed during the >parsing of EAL arguments. > > 2) In the function rte_eal_cpu_init(): >- to parse the set of all running logical cores on the machine, do not > use the RTE_LCORE_FOREACH macro that considers the set of already > detected cores... > Instead, use a standard loop based on the RTE_MAX_LCORE constant. >- explicitely set to ROLE_RTE the role of each detected logical core > that is recorded in the EAL configuration, as all running cores are > enabled by default. > > 3) In the function eal_parse_coremask(), update the "lcore_count" field >of the EAL configuration with the effective number of logical cores >that are set in the mask of enabled logical cores. > > Signed-off-by: Ivan Boule > --- > lib/librte_eal/linuxapp/eal/eal.c |8 +--- > lib/librte_eal/linuxapp/eal/eal_lcore.c | 12 > 2 files changed, 13 insertions(+), 7 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH 0/3] real "make install" rule
The current "make install" rule don't install anything. It builds one or more targets with different configurations. These patches allow to install only files needed to build and run a DPDK appplication. The old behaviour is kept for compatibility and the new behaviour is a second stage triggered by setting the DESTDIR variable. So, the command "make install T=x86_64-default-linuxapp-gcc O=aaa DESTDIR=bbb" will build in aaa/x86_64-default-linuxapp-gcc/ and install in bbb/. The install directory can then be used as RTE_SDK path to build an application. -- Thomas --- Olivier Matz (3): mk: allow to specify O= in install rule mk: in install rule, don't overwrite .config if it already exists mk: allow to specify DESTDIR in build rule doc/build-sdk-quick.txt | 13 +++-- mk/rte.sdkbuild.mk | 16 mk/rte.sdkinstall.mk| 17 + 3 files changed, 32 insertions(+), 14 deletions(-) -- 1.7.10.4
[dpdk-dev] [PATCH 1/3] mk: allow to specify O= in install rule
From: Olivier Matz
This variable $(O) can be used to specify a build directory
when doing an "install" procedure. The default is ".", which
means that targets will be built in the source dpdk.
This option is useful to compile outside of the source tree that may be
read-only.
Signed-off-by: Olivier Matz
Acked-by: Thomas Monjalon
---
doc/build-sdk-quick.txt |4 ++--
mk/rte.sdkinstall.mk| 15 +++
2 files changed, 9 insertions(+), 10 deletions(-)
diff --git a/doc/build-sdk-quick.txt b/doc/build-sdk-quick.txt
index c839676..d66f0d5 100644
--- a/doc/build-sdk-quick.txt
+++ b/doc/build-sdk-quick.txt
@@ -5,7 +5,7 @@ Build commands
all same as build (default rule)
build build in a configured directory
clean remove files but keep configuration
- install build many targets (wildcard allowed) in fixed directories
+ install build many targets (wildcard allowed)
uninstall remove all installed targets
Build variables
CROSStoolchain prefix
@@ -14,7 +14,7 @@ Build variables
EXTRA_LDFLAGSlinker options
V verbose
D debug dependencies
- O output directory (default: build/) - cannot be used with install
+ O output directory (default: build/ - install default: ./)
T target template (install default: *) - used with config or install
format:
templates in config/defconfig_*
diff --git a/mk/rte.sdkinstall.mk b/mk/rte.sdkinstall.mk
index 022cd70..a280234 100644
--- a/mk/rte.sdkinstall.mk
+++ b/mk/rte.sdkinstall.mk
@@ -30,10 +30,11 @@
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
+# Build directory is given with O=
ifdef O
-ifeq ("$(origin O)", "command line")
-$(error "Cannot use O= with install target")
-endif
+BUILD_DIR=$(O)
+else
+BUILD_DIR=.
endif
# Targets to install can be specified in command line. It can be a
@@ -56,8 +57,8 @@ install: $(INSTALL_TARGETS)
%_install:
@echo == Installing $*
- $(Q)$(MAKE) config T=$* O=$*
- $(Q)$(MAKE) all O=$*
+ $(Q)$(MAKE) config T=$* O=$(BUILD_DIR)/$*
+ $(Q)$(MAKE) all O=$(BUILD_DIR)/$*
#
# uninstall: remove all built sdk
@@ -70,6 +71,4 @@ uninstall: $(UNINSTALL_TARGETS)
%_uninstall:
@echo == Uninstalling $*
- $(Q)rm -rf $*
-
-
+ $(Q)rm -rf $(BUILD_DIR)/$*
--
1.7.10.4
[dpdk-dev] [PATCH 2/3] mk: in install rule, don't overwrite .config if it already exists
From: Olivier Matz This allows the user to prepare a configuration with make config before using make install. Signed-off-by: Olivier Matz Acked-by: Thomas Monjalon --- mk/rte.sdkinstall.mk |4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/mk/rte.sdkinstall.mk b/mk/rte.sdkinstall.mk index a280234..ee98981 100644 --- a/mk/rte.sdkinstall.mk +++ b/mk/rte.sdkinstall.mk @@ -57,7 +57,9 @@ install: $(INSTALL_TARGETS) %_install: @echo == Installing $* - $(Q)$(MAKE) config T=$* O=$(BUILD_DIR)/$* + $(Q)if [ ! -f $(BUILD_DIR)/$*/.config ]; then \ + $(MAKE) config T=$* O=$(BUILD_DIR)/$*; \ + fi $(Q)$(MAKE) all O=$(BUILD_DIR)/$* # -- 1.7.10.4
[dpdk-dev] [PATCH 3/3] mk: allow to specify DESTDIR in build rule
From: Olivier Matz This will install the binary sdk (bin + modules + libs + headers + mk) in the specified directory. This directory can be used as RTE_SDK by external applications. Signed-off-by: Olivier Matz Acked-by: Thomas Monjalon --- doc/build-sdk-quick.txt | 13 +++-- mk/rte.sdkbuild.mk | 16 2 files changed, 23 insertions(+), 6 deletions(-) diff --git a/doc/build-sdk-quick.txt b/doc/build-sdk-quick.txt index d66f0d5..18b0ee6 100644 --- a/doc/build-sdk-quick.txt +++ b/doc/build-sdk-quick.txt @@ -5,16 +5,17 @@ Build commands all same as build (default rule) build build in a configured directory clean remove files but keep configuration - install build many targets (wildcard allowed) + install build many targets (wildcard allowed) and install in DESTDIR uninstall remove all installed targets Build variables - CROSStoolchain prefix EXTRA_CPPFLAGS preprocessor options EXTRA_CFLAGS compiler options EXTRA_LDFLAGSlinker options - V verbose - D debug dependencies - O output directory (default: build/ - install default: ./) - T target template (install default: *) - used with config or install + CROSS toolchain prefix + V verbose + D debug dependencies + O build directory (default: build/ - install default: ./) + DESTDIR second-stage install directory + T target template (install default: *) - used with config or install format: templates in config/defconfig_* diff --git a/mk/rte.sdkbuild.mk b/mk/rte.sdkbuild.mk index 5ea4202..087d3ab 100644 --- a/mk/rte.sdkbuild.mk +++ b/mk/rte.sdkbuild.mk @@ -30,6 +30,8 @@ # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # +# If DESTDIR variable is given, install binary dpdk + # # include rte.vars.mk if config file exists # @@ -60,6 +62,20 @@ CLEANDIRS = $(addsuffix _clean,$(ROOTDIRS-y) $(ROOTDIRS-n) $(ROOTDIRS-)) .PHONY: build build: $(ROOTDIRS-y) @echo Build complete +ifneq ($(DESTDIR),) + $(Q)mkdir -p $(DESTDIR) + $(Q)tar -C $(RTE_SDK) -cf - mk | tar -C $(DESTDIR) -x \ + --keep-newer-files --warning=no-ignore-newer -f - + $(Q)mkdir -p $(DESTDIR)/`basename $(RTE_OUTPUT)` + $(Q)tar -C $(RTE_OUTPUT) -chf - \ + --exclude app --exclude hostapp --exclude build \ + --exclude Makefile --exclude .depdirs . | \ + tar -C $(DESTDIR)/`basename $(RTE_OUTPUT)` -x --keep-newer-files \ + --warning=no-ignore-newer -f - + $(Q)install -D $(RTE_OUTPUT)/app/testpmd \ + $(DESTDIR)/`basename $(RTE_OUTPUT)`/app/testpmd + @echo Installation in $(DESTDIR) complete +endif .PHONY: clean clean: $(CLEANDIRS) -- 1.7.10.4
[dpdk-dev] [PATCH] config: disable KNI for 32-bit because cannot work
26/07/2013 14:11, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 11:49:59AM +0200, Thomas Monjalon wrote: > > From: Jean-Mickael Guerin > > > > This is not supported, disable to avoid compilation error like: > > lib/librte_eal/linuxapp/kni/kni_misc.c:304:2: error: > > format '%llx' expects argument of type 'long long unsigned int', > > but argument 2 has type 'phys_addr_t' [-Werror=format] > > > > Signed-off-by: Jean-Mickael Guerin > > --- > > > > config/defconfig_i686-default-linuxapp-gcc |4 ++-- > > config/defconfig_i686-default-linuxapp-icc |4 ++-- > > 2 files changed, 4 insertions(+), 4 deletions(-) > > Acked-by: Adrien Mazarguil pushed -- Thomas
[dpdk-dev] [PATCH] kni: fix build with 802.1p kernel support
26/07/2013 14:12, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 11:51:15AM +0200, Thomas Monjalon wrote: > > C90 compilers forbid mixed declaration and code. > > > > Signed-off-by: Thomas Monjalon > > --- > > > > lib/librte_eal/linuxapp/kni/ethtool/ixgbe/ixgbe_main.c |7 +-- > > 1 file changed, 5 insertions(+), 2 deletions(-) > > Acked-by: Adrien Mazarguil pushed -- Thomas
[dpdk-dev] [PATCH] kni: fix build with kernel < 3.3 with netdev_features_t backport
26/07/2013 13:25, Adrien Mazarguil : > The netdev_features_t typedef appeared in Linux 3.3, but checking the > kernel version isn't enough with some distributions (such as Debian > Wheezy) that backported it into 3.2, causing a compilation failure due to > redefinition. > > Since the presence of a typedef can't be tested at compile time, this > commit adds type kni_netdev_features_t, which, depending on the kernel > version, translates either to u32 or netdev_features_t. > > Signed-off-by: Adrien Mazarguil acked and applied -- Thomas
[dpdk-dev] [PATCH] eal: fix type of pointer arithmetic result
26/07/2013 14:12, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 01:34:00PM +0200, Thomas Monjalon wrote: > > Adding or subtracting a value to a pointer makes a new pointer > > of unknown type. > > So typeof() is replaced by (void*) in RTE_PTR_ADD() and RTE_PTR_SUB(). > > > > But RTE_PTR_ALIGN_* macros have in their explicit API to return a pointer > > of the same type. Since RTE_PTR_ALIGN_CEIL is based on RTE_PTR_ADD, a > > typeof() is added to keep the original behaviour. > > > > Signed-off-by: Thomas Monjalon > > --- > > > > lib/librte_eal/common/include/rte_common.h |6 +++--- > > 1 file changed, 3 insertions(+), 3 deletions(-) > > Acked-by: Adrien Mazarguil pushed -- Thomas
[dpdk-dev] [PATCH] eal: fix recording of detected/enabled logical cores
26/07/2013 13:55, Thomas Monjalon : > From: Ivan Boule > > 1) In the EAL initialization phase, invoke the function rte_eal_cpu_init >to detect the set of running cores (and enable them by default) before >processing the [enabled] core mask option that is performed during the >parsing of EAL arguments. > > 2) In the function rte_eal_cpu_init(): >- to parse the set of all running logical cores on the machine, do not > use the RTE_LCORE_FOREACH macro that considers the set of already > detected cores... > Instead, use a standard loop based on the RTE_MAX_LCORE constant. >- explicitely set to ROLE_RTE the role of each detected logical core > that is recorded in the EAL configuration, as all running cores are > enabled by default. > > 3) In the function eal_parse_coremask(), update the "lcore_count" field >of the EAL configuration with the effective number of logical cores >that are set in the mask of enabled logical cores. > > Signed-off-by: Ivan Boule acked and pushed -- Thomas
[dpdk-dev] [PATCH 0/3] real "make install" rule
26/07/2013 15:13, Thomas Monjalon : > The current "make install" rule don't install anything. > It builds one or more targets with different configurations. > > These patches allow to install only files needed to build and run a DPDK > appplication. The old behaviour is kept for compatibility and the new > behaviour is a second stage triggered by setting the DESTDIR variable. > > So, the command "make install T=x86_64-default-linuxapp-gcc O=aaa > DESTDIR=bbb" will build in aaa/x86_64-default-linuxapp-gcc/ and install in > bbb/. > > The install directory can then be used as RTE_SDK path to build an > application. all pushed -- Thomas
[dpdk-dev] [PATCH] mem: more const qualifiers in malloc API
Some functions don't modify their parameter which should be marked as const.
Signed-off-by: Thomas Monjalon
---
lib/librte_malloc/malloc_elem.h |6 +++---
lib/librte_malloc/malloc_heap.c |2 +-
lib/librte_malloc/malloc_heap.h |2 +-
lib/librte_malloc/rte_malloc.c |4 ++--
lib/librte_malloc/rte_malloc.h |2 +-
5 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/lib/librte_malloc/malloc_elem.h b/lib/librte_malloc/malloc_elem.h
index 8a75e0c..fadf0a5 100644
--- a/lib/librte_malloc/malloc_elem.h
+++ b/lib/librte_malloc/malloc_elem.h
@@ -62,7 +62,7 @@ static const unsigned MALLOC_ELEM_TRAILER_LEN = 0;
/* dummy function - just check if pointer is non-null */
static inline int
-malloc_elem_cookies_ok(struct malloc_elem *elem){ return elem != NULL; }
+malloc_elem_cookies_ok(const struct malloc_elem *elem){ return elem != NULL; }
/* dummy function - no header if malloc_debug is not enabled */
static inline void
@@ -100,7 +100,7 @@ set_trailer(struct malloc_elem *elem)
/* check that the header and trailer cookies are set correctly */
static inline int
-malloc_elem_cookies_ok(struct malloc_elem *elem)
+malloc_elem_cookies_ok(const struct malloc_elem *elem)
{
return (elem != NULL &&
MALLOC_ELEM_HEADER(elem) == MALLOC_HEADER_COOKIE &&
@@ -117,7 +117,7 @@ static const unsigned MALLOC_ELEM_HEADER_LEN =
sizeof(struct malloc_elem);
* the actual malloc_elem header for that block.
*/
static inline struct malloc_elem *
-malloc_elem_from_data(void *data)
+malloc_elem_from_data(const void *data)
{
if (data == NULL)
return NULL;
diff --git a/lib/librte_malloc/malloc_heap.c b/lib/librte_malloc/malloc_heap.c
index b6d83a4..54b6b0c 100644
--- a/lib/librte_malloc/malloc_heap.c
+++ b/lib/librte_malloc/malloc_heap.c
@@ -201,7 +201,7 @@ malloc_heap_alloc(struct malloc_heap *heap,
* Function to retrieve data for heap on given socket
*/
int
-malloc_heap_get_stats(struct malloc_heap *heap,
+malloc_heap_get_stats(const struct malloc_heap *heap,
struct rte_malloc_socket_stats *socket_stats)
{
if (!heap->initialised)
diff --git a/lib/librte_malloc/malloc_heap.h b/lib/librte_malloc/malloc_heap.h
index e4a5063..e3eb906 100644
--- a/lib/librte_malloc/malloc_heap.h
+++ b/lib/librte_malloc/malloc_heap.h
@@ -53,7 +53,7 @@ malloc_heap_alloc(struct malloc_heap *heap, const char *type,
size_t size, unsigned align);
int
-malloc_heap_get_stats(struct malloc_heap *heap,
+malloc_heap_get_stats(const struct malloc_heap *heap,
struct rte_malloc_socket_stats *socket_stats);
int
diff --git a/lib/librte_malloc/rte_malloc.c b/lib/librte_malloc/rte_malloc.c
index 03db51f..e16ba9c 100644
--- a/lib/librte_malloc/rte_malloc.c
+++ b/lib/librte_malloc/rte_malloc.c
@@ -169,9 +169,9 @@ rte_realloc(void *ptr, size_t size, unsigned align)
}
int
-rte_malloc_validate(void *ptr, size_t *size)
+rte_malloc_validate(const void *ptr, size_t *size)
{
- struct malloc_elem *elem = malloc_elem_from_data(ptr);
+ const struct malloc_elem *elem = malloc_elem_from_data(ptr);
if (!malloc_elem_cookies_ok(elem))
return -1;
if (size != NULL)
diff --git a/lib/librte_malloc/rte_malloc.h b/lib/librte_malloc/rte_malloc.h
index 64ffaf0..acf2ae0 100644
--- a/lib/librte_malloc/rte_malloc.h
+++ b/lib/librte_malloc/rte_malloc.h
@@ -273,7 +273,7 @@ rte_free(void *ptr);
* 0 on success
*/
int
-rte_malloc_validate(void *ptr, size_t *size);
+rte_malloc_validate(const void *ptr, size_t *size);
/**
* Get heap statistics for the specified heap.
--
1.7.10.4
[dpdk-dev] Question regarding rte_eth_tx_burst()
Hi all,
Any ideas would be helpful, I have an application that uses a rte_ring
for queuing packets to be sent (a single consumer - multiple producer).
The problem I am facing is with rte_eth_tx_burst(), once I dequeue N
packets from tx ring, there is no guaranty that all packets will be
sent, if not all packets
are sent, rte_ring does not allow pushing packets at front (it is a
FIFO). So, if I re-queue them in the ring, packets will be network
reordered.
There is no mechanism to know the number of TX free descriptors in a
particular queue. I'm working with ixgbe and I found that nb_tx_free is
an approximation of that,
it is updated every time at ixgbe_xmit_cleanup() that is called during
ixgbe_xmit_pkts().
I was wondering to add a function that returns nb_tx_free value in order
to have an idea of how many packets can be safety sent.
Right now my code does this:
nbulk = RTE_MIN(rte_ring_count(queue->tx_ring), (uint16_t)TX_DRAIN_PKTS);
ret = rte_ring_sc_dequeue_bulk(queue->tx_ring, (void **)ptrs, nbulk);
if(unlikely(ret < 0))
return 0; /* Not enough objects?, I asked a moment ago */
ret = rte_eth_tx_burst(queue->port, queue->id, ptrs, nbulk);
if(unlikely(ret > nbulk)) {
TRACE(RTE, ">>>Imposible sent more than existing\n");
return -1;
}
while(ret < nbulk) {
/* bad case: packet reordering */
ret2 = rte_ring_mp_enqueue(queue->tx_ring, (void *)&ptrs[ret]);
if(ret2 < 0) {
/* TODO: may be an emergency tx queue? */
/* wrost case: dropping */
rte_pktmbuf_free(ptrs[ret]);
}
ret++;
}
I would appreciate any ideas of how to overcome this situation.
Thanks and regards,
Jesus
[dpdk-dev] [PATCH 1/1] mem: get physical address of any rte_malloc buffer
Get physical address of any rte_malloc allocated buffer using
function rte_malloc_virt2phy(addr).
The rte_memzone pointer is now stored in each allocated memory block
header to allow simple computation of physical address of a block
using the memzone it comes from.
Declaration of memzone in malloc_elem structure adds a dependency between
rte_malloc.h and rte_memory.h; test source code are modified to
include both files in correct order.
Signed-off-by: Didier Pallard
---
app/test/test_hash.c|2 +-
app/test/test_hash_perf.c |2 +-
app/test/test_memcpy.c |1 +
app/test/test_memcpy_perf.c |1 +
lib/librte_malloc/malloc_elem.c |7 ---
lib/librte_malloc/malloc_elem.h |2 ++
lib/librte_malloc/malloc_heap.c |2 +-
lib/librte_malloc/rte_malloc.c | 12
lib/librte_malloc/rte_malloc.h | 13 +
9 files changed, 36 insertions(+), 6 deletions(-)
diff --git a/app/test/test_hash.c b/app/test/test_hash.c
index 5437e05..4de8cb1 100644
--- a/app/test/test_hash.c
+++ b/app/test/test_hash.c
@@ -41,10 +41,10 @@
#include
#include
-#include
#include
#include
#include
+#include
#include
#include
#include
diff --git a/app/test/test_hash_perf.c b/app/test/test_hash_perf.c
index 8a0feb3..311c2bd 100644
--- a/app/test/test_hash_perf.c
+++ b/app/test/test_hash_perf.c
@@ -42,10 +42,10 @@
#include
#include
-#include
#include
#include
#include
+#include
#include
#include
#include
diff --git a/app/test/test_memcpy.c b/app/test/test_memcpy.c
index da8..729a2ff 100644
--- a/app/test/test_memcpy.c
+++ b/app/test/test_memcpy.c
@@ -41,6 +41,7 @@
#include
#include
#include
+#include
#include
#include
diff --git a/app/test/test_memcpy_perf.c b/app/test/test_memcpy_perf.c
index 236b295..1e75f53 100644
--- a/app/test/test_memcpy_perf.c
+++ b/app/test/test_memcpy_perf.c
@@ -41,6 +41,7 @@
#include
#include
#include
+#include
#include
#include
diff --git a/lib/librte_malloc/malloc_elem.c b/lib/librte_malloc/malloc_elem.c
index 919d474..8da517f 100644
--- a/lib/librte_malloc/malloc_elem.c
+++ b/lib/librte_malloc/malloc_elem.c
@@ -57,9 +57,10 @@
*/
void
malloc_elem_init(struct malloc_elem *elem,
- struct malloc_heap *heap, size_t size)
+ struct malloc_heap *heap, const struct rte_memzone *mz, size_t
size)
{
elem->heap = heap;
+ elem->mz = mz;
elem->prev = elem->next_free = NULL;
elem->state = ELEM_FREE;
elem->size = size;
@@ -74,7 +75,7 @@ malloc_elem_init(struct malloc_elem *elem,
void
malloc_elem_mkend(struct malloc_elem *elem, struct malloc_elem *prev)
{
- malloc_elem_init(elem, prev->heap, 0);
+ malloc_elem_init(elem, prev->heap, prev->mz, 0);
elem->prev = prev;
elem->state = ELEM_BUSY; /* mark busy so its never merged */
}
@@ -117,7 +118,7 @@ split_elem(struct malloc_elem *elem, struct malloc_elem
*split_pt)
const unsigned old_elem_size = (uintptr_t)split_pt - (uintptr_t)elem;
const unsigned new_elem_size = elem->size - old_elem_size;
- malloc_elem_init(split_pt, elem->heap, new_elem_size);
+ malloc_elem_init(split_pt, elem->heap, elem->mz, new_elem_size);
split_pt->prev = elem;
next_elem->prev = split_pt;
elem->size = old_elem_size;
diff --git a/lib/librte_malloc/malloc_elem.h b/lib/librte_malloc/malloc_elem.h
index 8a75e0c..f1710cf 100644
--- a/lib/librte_malloc/malloc_elem.h
+++ b/lib/librte_malloc/malloc_elem.h
@@ -48,6 +48,7 @@ struct malloc_elem {
struct malloc_heap *heap;
struct malloc_elem *volatile prev; /* points to prev elem in
memzone */
struct malloc_elem *volatile next_free; /* to make list of free
elements */
+ const struct rte_memzone *mz;
volatile enum elem_state state;
uint32_t pad;
size_t size;
@@ -134,6 +135,7 @@ malloc_elem_from_data(void *data)
void
malloc_elem_init(struct malloc_elem *elem,
struct malloc_heap *heap,
+ const struct rte_memzone *mz,
size_t size);
/*
diff --git a/lib/librte_malloc/malloc_heap.c b/lib/librte_malloc/malloc_heap.c
index b6d83a4..bbf6bb8 100644
--- a/lib/librte_malloc/malloc_heap.c
+++ b/lib/librte_malloc/malloc_heap.c
@@ -100,7 +100,7 @@ malloc_heap_add_memzone(struct malloc_heap *heap, size_t
size, unsigned align)
end_elem = RTE_PTR_ALIGN_FLOOR(end_elem, CACHE_LINE_SIZE);
const unsigned elem_size = (uintptr_t)end_elem - (uintptr_t)start_elem;
- malloc_elem_init(start_elem, heap, elem_size);
+ malloc_elem_init(start_elem, heap, mz, elem_size);
malloc_elem_mkend(end_elem, start_elem);
start_elem->next_free = heap->free_head;
diff --git a/lib/librte_malloc/rte_malloc.c b/lib/librte_malloc/rte_malloc.c
index 03db51f..3c76c43 100644
--- a/lib/librte_malloc/rte_malloc.c
+++ b/lib/librte_malloc/rte_malloc.c
@@ -228,3 +228,15 @@ rte_mal
[dpdk-dev] [PATCH] mem: get memzone from any CPU socket when hugepages are disabled
When huge pages are disabled, memory is allocated for a single, undefined CPU socket using malloc(), causing rte_memzone_reserve_aligned() to fail most of the time. This patch causes that memory to use SOCKET_ID_ANY instead of 0, and allow it to be used in place of any socket ID specified by user. Signed-off-by: Adrien Mazarguil --- lib/librte_eal/common/eal_common_memzone.c | 1 + lib/librte_eal/linuxapp/eal/eal_memory.c | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/lib/librte_eal/common/eal_common_memzone.c b/lib/librte_eal/common/eal_common_memzone.c index 5532f0d..9872070 100644 --- a/lib/librte_eal/common/eal_common_memzone.c +++ b/lib/librte_eal/common/eal_common_memzone.c @@ -150,6 +150,7 @@ memzone_reserve_aligned_thread_unsafe(const char *name, uint64_t len, /* bad socket ID */ if (socket_id != SOCKET_ID_ANY && + free_memseg[i].socket_id != SOCKET_ID_ANY && socket_id != free_memseg[i].socket_id) continue; diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c b/lib/librte_eal/linuxapp/eal/eal_memory.c index ce0c2d8..d50726d 100644 --- a/lib/librte_eal/linuxapp/eal/eal_memory.c +++ b/lib/librte_eal/linuxapp/eal/eal_memory.c @@ -786,7 +786,7 @@ rte_eal_hugepage_init(void) mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr; mcfg->memseg[0].addr = addr; mcfg->memseg[0].len = internal_config.memory; - mcfg->memseg[0].socket_id = 0; + mcfg->memseg[0].socket_id = SOCKET_ID_ANY; return 0; } -- 1.7.12
[dpdk-dev] [PATCH] mem: more const qualifiers in malloc API
On Fri, Jul 26, 2013 at 04:06:40PM +0200, Thomas Monjalon wrote: > Some functions don't modify their parameter which should be marked as const. > > Signed-off-by: Thomas Monjalon > --- > lib/librte_malloc/malloc_elem.h |6 +++--- > lib/librte_malloc/malloc_heap.c |2 +- > lib/librte_malloc/malloc_heap.h |2 +- > lib/librte_malloc/rte_malloc.c |4 ++-- > lib/librte_malloc/rte_malloc.h |2 +- > 5 files changed, 8 insertions(+), 8 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH 0/2] fix mempool when --no-huge option is set
This set of patch permits to use the --no-huge option to create valid mempools. However, drivers necessitating a physically contiguous memory larger than a standard page won't work in this mode. Only vmxnet3-usermap will work in this mode among the available open source PMD. This can be useful for old kernels (< 2.6.33) and VM with limited amount of mempory. Damien Millescamps (2): mem: get hugepages config mem: fix mempool for --no-huge lib/librte_eal/common/include/rte_eal.h | 13 +++- lib/librte_eal/linuxapp/eal/eal.c|4 ++ lib/librte_eal/linuxapp/eal/eal_memory.c |2 +- lib/librte_mempool/rte_mempool.c | 54 +- lib/librte_mempool/rte_mempool.h | 20 +++ 5 files changed, 83 insertions(+), 10 deletions(-) -- 1.7.2.5
[dpdk-dev] [PATCH 1/2] mem: get hugepages config
Allow external libraries and applications to know if hugepages
are enabled.
Signed-off-by: Thomas Monjalon
Signed-off-by: Damien Millescamps
---
lib/librte_eal/common/include/rte_eal.h | 13 -
lib/librte_eal/linuxapp/eal/eal.c |4
2 files changed, 16 insertions(+), 1 deletions(-)
diff --git a/lib/librte_eal/common/include/rte_eal.h
b/lib/librte_eal/common/include/rte_eal.h
index 240530d..58a6f90 100644
--- a/lib/librte_eal/common/include/rte_eal.h
+++ b/lib/librte_eal/common/include/rte_eal.h
@@ -214,7 +214,18 @@ int rte_eal_init(int argc, char **argv);
} \
} while(0)
-
+/**
+ * Whether EAL is using huge pages (disabled by --no-huge option).
+ * The no-huge mode cannot be used with UIO poll-mode drivers like igb/ixgbe.
+ * It is useful for NIC drivers (e.g. librte_pmd_mlx4, librte_pmd_vmxnet3) or
+ * crypto drivers (e.g. librte_crypto_nitrox) provided by third-parties such
+ * as 6WIND.
+ *
+ * @return
+ * Nonzero if hugepages are enabled.
+ */
+int rte_eal_has_hugepages(void);
+
#ifdef __cplusplus
}
#endif
diff --git a/lib/librte_eal/linuxapp/eal/eal.c
b/lib/librte_eal/linuxapp/eal/eal.c
index ed0e9b1..52de2b6 100644
--- a/lib/librte_eal/linuxapp/eal/eal.c
+++ b/lib/librte_eal/linuxapp/eal/eal.c
@@ -945,3 +945,7 @@ rte_eal_process_type(void)
return (rte_config.process_type);
}
+int rte_eal_has_hugepages(void)
+{
+ return ! internal_config.no_hugetlbfs;
+}
--
1.7.2.5
[dpdk-dev] [PATCH 2/2] mem: fix mempool for --no-huge
In --no-huge mode, mempool provides objects with their associated
header/trailer fitting in a standard page (usually 4KB).
This means all non-UIO driver should work correctly in this mode,
since UIO drivers allocate ring sizes that cannot fit in a page.
Extend rte_mempool_virt2phy to obtain the correct physical address when
elements of the pool are not on the same physically contiguous memory region.
This is a first step for enhancement PR #29696.
Reason for this patch is to be able to run on a kernel < 2.6.37 without
the need to patch it, since all kernel below are either bugged or don't
have huge page support at all (< 2.6.28).
Signed-off-by: Damien Millescamps
---
lib/librte_eal/linuxapp/eal/eal_memory.c |2 +-
lib/librte_mempool/rte_mempool.c | 54 +-
lib/librte_mempool/rte_mempool.h | 20 +++
3 files changed, 67 insertions(+), 9 deletions(-)
diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c
b/lib/librte_eal/linuxapp/eal/eal_memory.c
index ce0c2d8..00b5952 100644
--- a/lib/librte_eal/linuxapp/eal/eal_memory.c
+++ b/lib/librte_eal/linuxapp/eal/eal_memory.c
@@ -780,7 +780,7 @@ rte_eal_hugepage_init(void)
/* get pointer to global configuration */
mcfg = rte_eal_get_configuration()->mem_config;
- /* for debug purposes, hugetlbfs can be disabled */
+ /* hugetlbfs can be disabled */
if (internal_config.no_hugetlbfs) {
addr = malloc(internal_config.memory);
mcfg->memseg[0].phys_addr = (phys_addr_t)(uintptr_t)addr;
diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
index aa8e76a..87183df 100644
--- a/lib/librte_mempool/rte_mempool.c
+++ b/lib/librte_mempool/rte_mempool.c
@@ -36,6 +36,7 @@
#include
#include
#include
+#include
#include
#include
#include
@@ -139,6 +140,8 @@ rte_mempool_create(const char *name, unsigned n, unsigned
elt_size,
uint32_t header_size, trailer_size;
unsigned i;
void *obj;
+ void *startaddr;
+ int page_size = getpagesize();
/* compilation-time checks */
RTE_BUILD_BUG_ON((sizeof(struct rte_mempool) &
@@ -227,6 +230,20 @@ rte_mempool_create(const char *name, unsigned n, unsigned
elt_size,
trailer_size);
trailer_size = new_size - header_size - elt_size;
}
+ if (! rte_eal_has_hugepages()) {
+ /*
+* compute trailer size so that pool elements fit exactly in
+* a standard page
+*/
+ int new_size = page_size - header_size - elt_size;
+ if (new_size < 0 || (unsigned int)new_size < trailer_size) {
+ printf("When hugepages are disabled, pool objects "
+ "can't exceed PAGE_SIZE: %d + %d + %d > %d\n",
+ header_size, elt_size, trailer_size, page_size);
+ return NULL;
+ }
+ trailer_size = new_size;
+ }
/* this is the size of an object, including header and trailer */
total_elt_size = header_size + elt_size + trailer_size;
@@ -235,8 +252,31 @@ rte_mempool_create(const char *name, unsigned n, unsigned
elt_size,
* cache-aligned */
private_data_size = (private_data_size +
CACHE_LINE_MASK) & (~CACHE_LINE_MASK);
+
+ if (! rte_eal_has_hugepages()) {
+ /*
+* expand private data size to a whole page, so that the
+* first pool element will start on a new standard page
+*/
+ int head = sizeof(struct rte_mempool);
+ int new_size = (private_data_size + head) % page_size;
+ if (new_size) {
+ private_data_size += page_size - new_size;
+ }
+ }
+
mempool_size = total_elt_size * n +
sizeof(struct rte_mempool) + private_data_size;
+
+ if (! rte_eal_has_hugepages()) {
+ /*
+* we want the memory pool to start on a page boundary,
+* because pool elements crossing page boundaries would
+* result in discontiguous physical addresses
+*/
+ mempool_size += page_size;
+ }
+
rte_snprintf(mz_name, sizeof(mz_name), "MP_%s", name);
mz = rte_memzone_reserve(mz_name, mempool_size, socket_id, mz_flags);
@@ -248,8 +288,20 @@ rte_mempool_create(const char *name, unsigned n, unsigned
elt_size,
if (mz == NULL)
goto exit;
+ if (rte_eal_has_hugepages()) {
+ startaddr = (void*)mz->addr;
+ } else {
+ /* align memory pool start address on a page boundary */
+ unsigned long addr = (unsigned long)mz->addr;
+ if (addr & (page_size - 1)) {
+ addr +=
[dpdk-dev] [PATCH] mem: more const qualifiers in malloc API
26/07/2013 16:37, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 04:06:40PM +0200, Thomas Monjalon wrote: > > Some functions don't modify their parameter which should be marked as > > const. > > > > Signed-off-by: Thomas Monjalon > > --- > > > > lib/librte_malloc/malloc_elem.h |6 +++--- > > lib/librte_malloc/malloc_heap.c |2 +- > > lib/librte_malloc/malloc_heap.h |2 +- > > lib/librte_malloc/rte_malloc.c |4 ++-- > > lib/librte_malloc/rte_malloc.h |2 +- > > 5 files changed, 8 insertions(+), 8 deletions(-) > > Acked-by: Adrien Mazarguil pushed -- Thomas
[dpdk-dev] [PATCH 1/1] mem: get physical address of any rte_malloc buffer
26/07/2013 16:25, Didier Pallard : > Get physical address of any rte_malloc allocated buffer using > function rte_malloc_virt2phy(addr). > The rte_memzone pointer is now stored in each allocated memory block > header to allow simple computation of physical address of a block > using the memzone it comes from. > Declaration of memzone in malloc_elem structure adds a dependency between > rte_malloc.h and rte_memory.h; test source code are modified to > include both files in correct order. > > Signed-off-by: Didier Pallard acked and applied -- Thomas
[dpdk-dev] [PATCH] mem: remove unneeded log
From: Olivier Matz Remove an error log in memzone_reserve_aligned_thread_unsafe(). It is up to the caller to log the error, and this is already done in DPDK code (especially in network drivers). Signed-off-by: Olivier Matz Acked-by: Thomas Monjalon --- lib/librte_eal/common/eal_common_memzone.c |3 --- 1 file changed, 3 deletions(-) diff --git a/lib/librte_eal/common/eal_common_memzone.c b/lib/librte_eal/common/eal_common_memzone.c index 5532f0d..cd60054 100644 --- a/lib/librte_eal/common/eal_common_memzone.c +++ b/lib/librte_eal/common/eal_common_memzone.c @@ -198,9 +198,6 @@ memzone_reserve_aligned_thread_unsafe(const char *name, uint64_t len, return memzone_reserve_aligned_thread_unsafe(name, len - align, socket_id, 0, align); - RTE_LOG(ERR, EAL, "%s(%s, %" PRIu64 ", %d): " - "No appropriate segment found\n", - __func__, name, requested_len, socket_id); rte_errno = ENOMEM; return NULL; } -- 1.7.10.4
[dpdk-dev] [PATCH] mem: fix rte_malloc(SOCKET_ID_ANY), try to allocate on other nodes
From: Olivier Matz
Before this patch, rte_malloc(SOCKET_ID_ANY) was equivalent to
rte_malloc(this_socket). If the user specifies SOCKET_ID_ANY, it means that
memory can be allocated on any socket. So fix the behavior of rte_malloc() in
order to do that. The current CPU socket is still the default, but if it fails,
other sockets are tested.
Signed-off-by: Olivier Matz
Acked-by: Thomas Monjalon
---
lib/librte_malloc/rte_malloc.c | 28
1 file changed, 24 insertions(+), 4 deletions(-)
diff --git a/lib/librte_malloc/rte_malloc.c b/lib/librte_malloc/rte_malloc.c
index a70a6b0..3a9edb1 100644
--- a/lib/librte_malloc/rte_malloc.c
+++ b/lib/librte_malloc/rte_malloc.c
@@ -69,23 +69,43 @@ void rte_free(void *addr)
* Allocate memory on specified heap.
*/
void *
-rte_malloc_socket(const char *type, size_t size, unsigned align, int socket)
+rte_malloc_socket(const char *type, size_t size, unsigned align, int
socket_arg)
{
struct rte_mem_config *mcfg = rte_eal_get_configuration()->mem_config;
+ int socket, i;
+ void *ret;
/* return NULL if size is 0 or alignment is not power-of-2 */
if (size == 0 || !rte_is_power_of_2(align))
return NULL;
- if (socket == SOCKET_ID_ANY)
+ if (socket_arg == SOCKET_ID_ANY)
socket = malloc_get_numa_socket();
+ else
+ socket = socket_arg;
/* Check socket parameter */
if (socket >= RTE_MAX_NUMA_NODES)
return NULL;
- return malloc_heap_alloc(&mcfg->malloc_heaps[socket], type,
- size, align == 0 ? 1 : align);
+ ret = malloc_heap_alloc(&mcfg->malloc_heaps[socket], type,
+ size, align == 0 ? 1 : align);
+ if (ret != NULL || socket_arg != SOCKET_ID_ANY)
+ return ret;
+
+ /* try other heaps */
+ for (i = 0; i < RTE_MAX_NUMA_NODES; i++) {
+ /* we already tried this one */
+ if (i == socket)
+ continue;
+
+ ret = malloc_heap_alloc(&mcfg->malloc_heaps[i], type,
+ size, align == 0 ? 1 : align);
+ if (ret != NULL)
+ return ret;
+ }
+
+ return NULL;
}
/*
--
1.7.10.4
[dpdk-dev] [PATCH] mem: fix log for --no-huge
In some cases, it is possible to not use hugepages.
So a simple malloc is used to initialize DPDK memory.
Signed-off-by: Thomas Monjalon
---
lib/librte_eal/linuxapp/eal/eal_memory.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c
b/lib/librte_eal/linuxapp/eal/eal_memory.c
index 1a05d66..e429438 100644
--- a/lib/librte_eal/linuxapp/eal/eal_memory.c
+++ b/lib/librte_eal/linuxapp/eal/eal_memory.c
@@ -1142,7 +1142,7 @@ rte_eal_memdevice_init(void)
int
rte_eal_memory_init(void)
{
- RTE_LOG(INFO, EAL, "Setting up hugepage memory...\n");
+ RTE_LOG(INFO, EAL, "Setting up memory...\n");
const int retval = rte_eal_process_type() == RTE_PROC_PRIMARY ?
rte_eal_hugepage_init() :
rte_eal_hugepage_attach();
--
1.7.10.4
[dpdk-dev] [PATCH] mem: get memzone from any CPU socket when hugepages are disabled
On 07/26/2013 04:35 PM, Adrien Mazarguil wrote: > When huge pages are disabled, memory is allocated for a single, undefined > CPU socket using malloc(), causing rte_memzone_reserve_aligned() to fail > most of the time. > > This patch causes that memory to use SOCKET_ID_ANY instead of 0, and allow > it to be used in place of any socket ID specified by user. > > Signed-off-by: Adrien Mazarguil > --- > lib/librte_eal/common/eal_common_memzone.c | 1 + > lib/librte_eal/linuxapp/eal/eal_memory.c | 2 +- > 2 files changed, 2 insertions(+), 1 deletion(-) > ack
[dpdk-dev] [PATCH] mem: fix log for --no-huge
On 07/26/2013 04:56 PM, Thomas Monjalon wrote: > In some cases, it is possible to not use hugepages. > So a simple malloc is used to initialize DPDK memory. > > Signed-off-by: Thomas Monjalon > --- > lib/librte_eal/linuxapp/eal/eal_memory.c |2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > ack
[dpdk-dev] [PATCH 1/2] mem: get hugepages config
On Fri, Jul 26, 2013 at 04:39:12PM +0200, Damien Millescamps wrote: > Allow external libraries and applications to know if hugepages > are enabled. > > Signed-off-by: Thomas Monjalon > Signed-off-by: Damien Millescamps > --- > lib/librte_eal/common/include/rte_eal.h | 13 - > lib/librte_eal/linuxapp/eal/eal.c |4 > 2 files changed, 16 insertions(+), 1 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH 2/2] mem: fix mempool for --no-huge
On Fri, Jul 26, 2013 at 04:39:13PM +0200, Damien Millescamps wrote: > In --no-huge mode, mempool provides objects with their associated > header/trailer fitting in a standard page (usually 4KB). > This means all non-UIO driver should work correctly in this mode, > since UIO drivers allocate ring sizes that cannot fit in a page. > > Extend rte_mempool_virt2phy to obtain the correct physical address when > elements of the pool are not on the same physically contiguous memory region. > This is a first step for enhancement PR #29696. > > Reason for this patch is to be able to run on a kernel < 2.6.37 without > the need to patch it, since all kernel below are either bugged or don't > have huge page support at all (< 2.6.28). > > Signed-off-by: Damien Millescamps > --- > lib/librte_eal/linuxapp/eal/eal_memory.c |2 +- > lib/librte_mempool/rte_mempool.c | 54 > +- > lib/librte_mempool/rte_mempool.h | 20 +++ > 3 files changed, 67 insertions(+), 9 deletions(-) Acked-by: Adrien Mazarguil -- Adrien Mazarguil 6WIND
[dpdk-dev] [PATCH] mem: remove unneeded log
26/07/2013 16:54, Thomas Monjalon : > From: Olivier Matz > > Remove an error log in memzone_reserve_aligned_thread_unsafe(). > It is up to the caller to log the error, and this is already done > in DPDK code (especially in network drivers). > > Signed-off-by: Olivier Matz > Acked-by: Thomas Monjalon > --- > lib/librte_eal/common/eal_common_memzone.c |3 --- > 1 file changed, 3 deletions(-) > > diff --git a/lib/librte_eal/common/eal_common_memzone.c > b/lib/librte_eal/common/eal_common_memzone.c index 5532f0d..cd60054 100644 > --- a/lib/librte_eal/common/eal_common_memzone.c > +++ b/lib/librte_eal/common/eal_common_memzone.c > @@ -198,9 +198,6 @@ memzone_reserve_aligned_thread_unsafe(const char *name, > uint64_t len, return memzone_reserve_aligned_thread_unsafe(name, len - > align, socket_id, 0, align); > > - RTE_LOG(ERR, EAL, "%s(%s, %" PRIu64 ", %d): " > - "No appropriate segment found\n", > - __func__, name, requested_len, socket_id); > rte_errno = ENOMEM; > return NULL; > } pushed -- Thomas
[dpdk-dev] [PATCH] mem: fix rte_malloc(SOCKET_ID_ANY), try to allocate on other nodes
26/07/2013 16:55, Thomas Monjalon : > From: Olivier Matz > > Before this patch, rte_malloc(SOCKET_ID_ANY) was equivalent to > rte_malloc(this_socket). If the user specifies SOCKET_ID_ANY, it means that > memory can be allocated on any socket. So fix the behavior of rte_malloc() > in order to do that. The current CPU socket is still the default, but if > it fails, other sockets are tested. > > Signed-off-by: Olivier Matz > Acked-by: Thomas Monjalon pushed -- Thomas
[dpdk-dev] [PATCH] mem: get memzone from any CPU socket when hugepages are disabled
26/07/2013 16:56, Damien Millescamps : > On 07/26/2013 04:35 PM, Adrien Mazarguil wrote: > > When huge pages are disabled, memory is allocated for a single, undefined > > CPU socket using malloc(), causing rte_memzone_reserve_aligned() to fail > > most of the time. > > > > This patch causes that memory to use SOCKET_ID_ANY instead of 0, and > > allow it to be used in place of any socket ID specified by user. > > > > Signed-off-by: Adrien Mazarguil > > --- > > > > lib/librte_eal/common/eal_common_memzone.c | 1 + > > lib/librte_eal/linuxapp/eal/eal_memory.c | 2 +- > > 2 files changed, 2 insertions(+), 1 deletion(-) > > ack applied -- Thomas
[dpdk-dev] [PATCH 1/2] mem: get hugepages config
26/07/2013 16:59, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 04:39:12PM +0200, Damien Millescamps wrote: > > Allow external libraries and applications to know if hugepages > > are enabled. > > > > Signed-off-by: Thomas Monjalon > > Signed-off-by: Damien Millescamps > > --- > > > > lib/librte_eal/common/include/rte_eal.h | 13 - > > lib/librte_eal/linuxapp/eal/eal.c |4 > > 2 files changed, 16 insertions(+), 1 deletions(-) > > Acked-by: Adrien Mazarguil applied -- Thomas
[dpdk-dev] [PATCH 2/2] mem: fix mempool for --no-huge
26/07/2013 16:59, Adrien Mazarguil : > On Fri, Jul 26, 2013 at 04:39:13PM +0200, Damien Millescamps wrote: > > In --no-huge mode, mempool provides objects with their associated > > header/trailer fitting in a standard page (usually 4KB). > > This means all non-UIO driver should work correctly in this mode, > > since UIO drivers allocate ring sizes that cannot fit in a page. > > > > Extend rte_mempool_virt2phy to obtain the correct physical address when > > elements of the pool are not on the same physically contiguous memory > > region. This is a first step for enhancement PR #29696. > > > > Reason for this patch is to be able to run on a kernel < 2.6.37 without > > the need to patch it, since all kernel below are either bugged or don't > > have huge page support at all (< 2.6.28). > > > > Signed-off-by: Damien Millescamps > > --- > > > > lib/librte_eal/linuxapp/eal/eal_memory.c |2 +- > > lib/librte_mempool/rte_mempool.c | 54 > > +- lib/librte_mempool/rte_mempool.h > > | 20 +++ > > 3 files changed, 67 insertions(+), 9 deletions(-) > > Acked-by: Adrien Mazarguil applied -- Thomas
[dpdk-dev] [PATCH] mem: fix log for --no-huge
26/07/2013 16:57, Damien Millescamps : > On 07/26/2013 04:56 PM, Thomas Monjalon wrote: > > In some cases, it is possible to not use hugepages. > > So a simple malloc is used to initialize DPDK memory. > > > > Signed-off-by: Thomas Monjalon > > --- > > > > lib/librte_eal/linuxapp/eal/eal_memory.c |2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > ack pushed -- Thomas
[dpdk-dev] [dpdk-announce] DPDK 1.3.1r2 released
The new release can be downloaded here: http://dpdk.org/browse/dpdk/tag/?id=v1.3.1r2 Changelog - fixes for: * KNI build * detection of CPU cores - enhancements: * real "make install" * improved malloc library * --no-huge mode is now usable with vmxnet3 -- Thomas
[dpdk-dev] NIC Stops Transmitting
Hi, I'm writing an application using DPDK that transmits a large number of packets (it doesn't receive any). When I transmit at 2 Gb/sec, everything will run fine for several seconds (receiver is receiving at correct rate), but then the NIC appears to get 'stuck' and doesn't transmit any more packets. In this state, rte_eth_tx_burst() is returning zero (suggesting that there are no available transmit descriptors), but even if I sleep() for a second and try again, rte_eth_tx_burst() still returns 0. It almost appears as if a packet gets stuck in the transmit ring and keeps everything from flowing. I'm using an Intel 82599EB NIC. Does anyone have any ideas of what might be going on? Thanks, Scott
[dpdk-dev] NIC Stops Transmitting
On Fri, 26 Jul 2013 15:39:18 -0400 (EDT) Scott Talbert wrote: > Hi, > > I'm writing an application using DPDK that transmits a large number of > packets (it doesn't receive any). When I transmit at 2 Gb/sec, everything > will run fine for several seconds (receiver is receiving at correct rate), > but then the NIC appears to get 'stuck' and doesn't transmit any more > packets. In this state, rte_eth_tx_burst() is returning zero (suggesting > that there are no available transmit descriptors), but even if I sleep() > for a second and try again, rte_eth_tx_burst() still returns 0. It almost > appears as if a packet gets stuck in the transmit ring and keeps > everything from flowing. I'm using an Intel 82599EB NIC. > > Does anyone have any ideas of what might be going on? > > Thanks, > Scott Make sure there is enough memory for mbufs. Also what is your ring size and transmit free threshold? It is easy to instrument the driver to see where it is saying "no space left" Also be careful with threshold values, many values of pthresh/hthresh/wthresh don't work. I would check the Intel reference manual for your hardware.
[dpdk-dev] NIC Stops Transmitting
On Fri, 26 Jul 2013, Stephen Hemminger wrote:
>> I'm writing an application using DPDK that transmits a large number of
>> packets (it doesn't receive any). When I transmit at 2 Gb/sec, everything
>> will run fine for several seconds (receiver is receiving at correct rate),
>> but then the NIC appears to get 'stuck' and doesn't transmit any more
>> packets. In this state, rte_eth_tx_burst() is returning zero (suggesting
>> that there are no available transmit descriptors), but even if I sleep()
>> for a second and try again, rte_eth_tx_burst() still returns 0. It almost
>> appears as if a packet gets stuck in the transmit ring and keeps
>> everything from flowing. I'm using an Intel 82599EB NIC.
>>
> Make sure there is enough memory for mbufs.
> Also what is your ring size and transmit free threshold?
> It is easy to instrument the driver to see where it is saying "no space left"
> Also be careful with threshold values, many values of pthresh/hthresh/wthresh
> don't work. I would check the Intel reference manual for your hardware.
Thanks for the tips. I don't think I'm running out of mbufs, but I'll
check that again. I am using these values from one of the examples -
which claim to be correct for the 82599EB.
/*
* These default values are optimized for use with the Intel(R) 82599 10
GbE
* Controller and the DPDK ixgbe PMD. Consider using other values for
other
* network controllers and/or network drivers.
*/
#define TX_PTHRESH 36 /**< Default values of TX prefetch threshold reg. */
#define TX_HTHRESH 0 /**< Default values of TX host threshold reg. */
#define TX_WTHRESH 0 /**< Default values of TX write-back threshold reg.
*/
static const struct rte_eth_txconf tx_conf = {
.tx_thresh = {
.pthresh = TX_PTHRESH,
.hthresh = TX_HTHRESH,
.wthresh = TX_WTHRESH,
},
.tx_free_thresh = 0, /* Use PMD default values */
.tx_rs_thresh = 0, /* Use PMD default values */
};
/*
* Configurable number of RX/TX ring descriptors
*/
#define RTE_TEST_TX_DESC_DEFAULT 512
static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT;
[dpdk-dev] NIC Stops Transmitting
On Fri, Jul 26, 2013 at 4:04 PM, Scott Talbert wrote:
> On Fri, 26 Jul 2013, Stephen Hemminger wrote:
>
>>> I'm writing an application using DPDK that transmits a large number of
>>> packets (it doesn't receive any). When I transmit at 2 Gb/sec,
>>> everything
>>> will run fine for several seconds (receiver is receiving at correct
>>> rate),
>>> but then the NIC appears to get 'stuck' and doesn't transmit any more
>>> packets. In this state, rte_eth_tx_burst() is returning zero (suggesting
>>> that there are no available transmit descriptors), but even if I sleep()
>>> for a second and try again, rte_eth_tx_burst() still returns 0. It
>>> almost
>>> appears as if a packet gets stuck in the transmit ring and keeps
>>> everything from flowing. I'm using an Intel 82599EB NIC.
>>>
>> Make sure there is enough memory for mbufs.
>> Also what is your ring size and transmit free threshold?
>> It is easy to instrument the driver to see where it is saying "no space
>> left"
>> Also be careful with threshold values, many values of
>> pthresh/hthresh/wthresh
>> don't work. I would check the Intel reference manual for your hardware.
>
>
> Thanks for the tips. I don't think I'm running out of mbufs, but I'll check
> that again. I am using these values from one of the examples - which claim
> to be correct for the 82599EB.
>
> /*
> * These default values are optimized for use with the Intel(R) 82599 10 GbE
> * Controller and the DPDK ixgbe PMD. Consider using other values for other
> * network controllers and/or network drivers.
> */
> #define TX_PTHRESH 36 /**< Default values of TX prefetch threshold reg. */
> #define TX_HTHRESH 0 /**< Default values of TX host threshold reg. */
> #define TX_WTHRESH 0 /**< Default values of TX write-back threshold reg. */
>
> static const struct rte_eth_txconf tx_conf = {
> .tx_thresh = {
> .pthresh = TX_PTHRESH,
> .hthresh = TX_HTHRESH,
> .wthresh = TX_WTHRESH,
> },
> .tx_free_thresh = 0, /* Use PMD default values */
> .tx_rs_thresh = 0, /* Use PMD default values */
> };
>
> /*
> * Configurable number of RX/TX ring descriptors
> */
> #define RTE_TEST_TX_DESC_DEFAULT 512
> static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT;
>
Scott,
I am wondering whether you use multiple cores accessing the same
receive queue. I had this problem before, but after I make the same
number of receiving queues as the number of receiving cores, the
problem disappeared. I did not dig more since I did not care how many
receive queues I have did not matter.
Jinho
[dpdk-dev] NIC Stops Transmitting
On Fri, 26 Jul 2013, jinho hwang wrote:
>> Thanks for the tips. I don't think I'm running out of mbufs, but I'll check
>> that again. I am using these values from one of the examples - which claim
>> to be correct for the 82599EB.
>>
>> /*
>> * These default values are optimized for use with the Intel(R) 82599 10 GbE
>> * Controller and the DPDK ixgbe PMD. Consider using other values for other
>> * network controllers and/or network drivers.
>> */
>> #define TX_PTHRESH 36 /**< Default values of TX prefetch threshold reg. */
>> #define TX_HTHRESH 0 /**< Default values of TX host threshold reg. */
>> #define TX_WTHRESH 0 /**< Default values of TX write-back threshold reg. */
>>
>> static const struct rte_eth_txconf tx_conf = {
>> .tx_thresh = {
>> .pthresh = TX_PTHRESH,
>> .hthresh = TX_HTHRESH,
>> .wthresh = TX_WTHRESH,
>> },
>> .tx_free_thresh = 0, /* Use PMD default values */
>> .tx_rs_thresh = 0, /* Use PMD default values */
>> };
>>
>> /*
>> * Configurable number of RX/TX ring descriptors
>> */
>> #define RTE_TEST_TX_DESC_DEFAULT 512
>> static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT;
>>
> I am wondering whether you use multiple cores accessing the same
> receive queue. I had this problem before, but after I make the same
> number of receiving queues as the number of receiving cores, the
> problem disappeared. I did not dig more since I did not care how many
> receive queues I have did not matter.
Jinho,
Thanks. I have only one queue (should I be using more?) but as far as I
know, I'm only using one core to transmit as well.
Scott
