John McNamara [Wed, 25 Nov 2015 14:26:00 +0000 (14:26 +0000)]
doc: increment minimum Linux kernel requirement
Update the documentation to reflect that the minimum Linux kernel
requirement for DPDK 2.2 has increased from 2.6.33 to 2.6.34.
Compatibility with kernel 2.6.33 was dropped, after discussion on
the mailing list, in the following commit: 2e6e9e215703 ("igb_uio: use existing PCI macros")
Signed-off-by: John McNamara <john.mcnamara@intel.com>
John McNamara [Fri, 11 Dec 2015 13:41:51 +0000 (13:41 +0000)]
doc: link examples to API
Enable Doxygen parsing of DPDK example files so that the example
code is linked to and from the API docs. This allows the reader
to see examples of API usage in context by following a link.
Signed-off-by: John McNamara <john.mcnamara@intel.com>
John McNamara [Sun, 1 Nov 2015 16:44:30 +0000 (16:44 +0000)]
doc: turn off permalink symbols in html guides
Turn off the permalink symbol in the Sphinx generated html docs.
This is a distracting mouseover when reading the docs and links
to sections can still be obtained from the doc index.
Signed-off-by: John McNamara <john.mcnamara@intel.com>
This project is missing a proper README which is used in
other projects and some git visualization services.
Only a starting point, please feel free to edit.
To keep the file short and current, I avoided putting any specific
information about features and versions.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Acked-by: John McNamara <john.mcnamara@intel.com>
Harish Patil [Fri, 11 Dec 2015 06:36:35 +0000 (22:36 -0800)]
bnx2x: fix build with clang
Fix for the following clang build error:
drivers/net/bnx2x/elink.c:10384:41: error: shifting a
negative signed value is undefined [-Werror,-Wshift-negative-value]
vars->eee_status &= ~SHMEM_EEE_1G_ADV <<
~~~~~~~~~~~~~~~~~ ^
Harish Patil [Fri, 11 Dec 2015 06:36:34 +0000 (22:36 -0800)]
bnx2x: update VF to support newer PF drivers
SR-IOV is supported using bnx2x poll mode driver running as VF driver and
native linux driver running as PF (in host/hypervisor). There is no issue
while running with the PF driver which is at the base version as that of
PMD. However, there is a compatibility issue between newer out-of-box PF
drivers with older VF driver. So the newer VFs would also need to send
BNX2X_VF_TLV_PHYS_PORT_ID (among other TLVs) to differentiate between
newer and older VFs.
Chas Williams [Wed, 9 Dec 2015 22:11:04 +0000 (17:11 -0500)]
bnx2x: fix restart
If you stop and start the driver, the rx queue will have the previous
index values when programming the adapter. Therefore, we should always
reset the queue indices when the rx ring is setup. Note: We need to
clear (write) the status block's completion queue index since it is
possibly in a read cache.
Tidy some init code to make it clearer what the defaults are.
Rahul Lakkireddy [Fri, 11 Dec 2015 08:25:30 +0000 (13:55 +0530)]
cxgbe: fix Rx queue setup error management
When refilling freelists for the first time and if it fails, the rxq
is freed and returns ENOMEM. There is a check while freeing hardware rxq
to pass freelist context id if the freelist exists or 0xffff otherwise.
The error path is only reached if freelist exists. So, fix is to remove
the useless check for freelist existence.
Coverity issue: 107108 Fixes: 92c8a63223e5 ("cxgbe: add device configuration and Rx support") Reported-by: John McNamara <john.mcnamara@intel.com> Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com> Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com> Acked-by: John McNamara <john.mcnamara@intel.com>
Cunming Liang [Fri, 11 Dec 2015 01:48:57 +0000 (09:48 +0800)]
eal/linux: fix Rx interrupt read error management
In eal_intr_proc_rxtx_intr, negative value may be used as argument to
a function expecting a positive value. If 'read' returns EAGAIN as
example, the bytes_read updates to a negative value which continue
be passed as argument for the next 'read'.
Coverity issue: 107115
Function read(fd, &buf, bytes_read) returns a negative number.
Assigning: signed variable bytes_read = read.
CID 107115 (#1 of 1): Argument cannot be negative
(NEGATIVE_RETURNS) bytes_read is passed to a parameter
that cannot be negative.
bytes_read = read(fd, &buf, bytes_read);
Fixes: c9f3ec1a0f3f ("eal/linux: add Rx interrupt control function") Signed-off-by: Cunming Liang <cunming.liang@intel.com> Acked-by: John McNamara <john.mcnamara@intel.com> Acked-by: David Marchand <david.marchand@6wind.com>
Ian Betts [Sat, 12 Dec 2015 08:12:08 +0000 (08:12 +0000)]
examples/performance-thread: fix return sign
The patch fixes an inverted return value in the
cond_destroy and cond_init APIs of the pthread shim
example.
These APIs are now demonstrated in the sample app by
having the mutexes and condition variables
explicitly destroyed before the appplication terminates.
Fixes: 433ba6228f9a77a9b5f4 ("add pthread_shim app") Signed-off-by: Ian Betts <ian.betts@intel.com>
Panu Matilainen [Fri, 11 Dec 2015 12:44:23 +0000 (14:44 +0200)]
scripts: fix relpath.sh output when $prefix is set in environment
When relpath.sh is called from install target with prefix set, eg
"make install DESTDIR=/tmp/dpdk-root prefix=/usr", the prefix
from the environment leaks to relpath.sh internal helper variable and
causes incorrect symlinks to be generated in sdk $(targetdir):
include -> /usr../../../include/dpdk
lib -> /usr../../../lib
Initialize the local variable to empty to avoid side-effects from
environment.
Fixes: 6b62a72a70d0 ("mk: install a standard cutomizable tree") Signed-off-by: Panu Matilainen <pmatilai@redhat.com>
Thomas Monjalon [Fri, 11 Dec 2015 23:26:24 +0000 (00:26 +0100)]
mbuf_offload: mark experimental state
Cryptodev was marked experimental and mbuf_offload depends on it.
The mbuf_offload library is one of the crypto area which requires
some discussions before having a stable API.
The experimental mark is also added to rte_cryptodev_configure()
to be sure one cannot miss it.
Fixes: 66874e55f550 ("cryptodev: mark experimental state") Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Ian Betts [Thu, 10 Dec 2015 12:24:58 +0000 (12:24 +0000)]
examples/performance-thread: mark as experimental
This commit removes the performance thread example from
examples/Makefile, and marks the example as "experimental"
in the release note, and it its API headers files.
Ian Betts [Tue, 8 Dec 2015 06:05:17 +0000 (06:05 +0000)]
examples/performance-thread: add pthread_shim app
This commit adds an example that illustrates how to implement
a pthread shim with the lthread subsystem included in the
performance thread example application.
Signed-off-by: Ian Betts <ian.betts@intel.com> Acked-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Add #ifndef MAX_QUEUES to change MAX_QUEUES at compile time if needed.
Change MAX_QUEUES from 512 to 128 to reduce the number of hugepages
required by the vhost-switch program.
To change MAX_QUEUES add '-D MAX_QUEUES=512' to the EXTRA_CFLAGS variable,
before building the application.
Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Pablo de Lara [Thu, 10 Dec 2015 09:28:27 +0000 (09:28 +0000)]
app/testpmd: swap MAC in IEEE1588 forwarding
IEEE1588 forwarding mode uses same port to receive and transmit packets.
In this mode, the MAC address in the packets are not touched,
and therefore, packets could be sent with the same destination
MAC address as the MAC address of the port that is transmitting,
which causes the packet to be dropped in some NICs.
In order to avoid this scenario, MAC addresses are swapped
before packet is sent out.
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com> Acked-by: John McNamara <john.mcnamara@intel.com>
The current implementation of VFIO will not with the new no-IOMMU mode
in 4.4 kernel. The original code assumed that IOMMU group zero would
never be used. Group numbers are assigned starting at zero, and up
until now the group numbers came from the hardware which is likely
to use group 0 for system devices that are not used with DPDK.
The fix is to allow 0 as a valid group and rearrange code
to split the return value from the group value.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Pablo de Lara [Wed, 9 Dec 2015 14:26:49 +0000 (14:26 +0000)]
examples/vhost: fix strict aliasing
Fixes following error on gcc 4.4.7:
examples/vhost/main.c: In function ‘new_device’:
rte_ring.h:740: error:
dereferencing pointer ‘mbuf.486’ does break strict-aliasing rules
examples/vhost/main.c:1503: note: initialized from here
...
rte_ring.h:740: error:
dereferencing pointer ‘({anonymous})’ does break strict-aliasing rules
examples/vhost/main.c:1804: note: initialized from here
Fixes: d19533e8 ("examples/vhost: copy old vhost example") Reported-by: Qian Xu <qian.q.xu@intel.com> Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com> Acked-by: Huawei Xie <huawei.xie@intel.com>
Michael Qiu [Wed, 9 Dec 2015 08:44:26 +0000 (16:44 +0800)]
examples/ip_pipeline: fix strict-aliasing
Compile ip_pipeline in CentOS 6.5 with kernel 2.6.32-431
GCC 4.4.7, will lead below error:
pipeline_routing_be.c: In function ‘pipeline_routing_msg_req_arp_add_handler’:
pipeline_routing_be.c:1817: error: dereferencing pointer ‘({anonymous})’
does break strict-aliasing rules
This because the code break strict-aliasing rule.
The patch solve this issue.
Fixes: 0ae7275810f1 (examples/ip_pipeline: add more functions to routing pipeline) Signed-off-by: Michael Qiu <michael.qiu@intel.com> Acked-by: Jasvinder Singh <jasvinder.singh@intel.com>
Piotr Azarewicz [Wed, 9 Dec 2015 09:33:50 +0000 (10:33 +0100)]
examples/ip_pipeline: fix coverity warnings
The source and destination both are the arrays of cmdline_parse_ctx_t.
So the goal is to copy elements size of cmdline_parse_ctx_t not
cmdline_parse_ctx_t*.
CID 120412: Code maintainability issues (SIZEOF_MISMATCH)
Passing argument "&app->cmds[app->n_cmds]" of type "cmdline_parse_ctx_t *"
and argument "n_cmds * 8UL /* sizeof (cmdline_parse_ctx_t *) */"
to function "memcpy" is suspicious.
In this case, "sizeof (cmdline_parse_ctx_t *)" is equal to
"sizeof (cmdline_parse_ctx_t)", but this is not a portable assumption.
When guest is booting (or any othertime guest is busy) it is possible
for the small receive ring (256) to get full. If this happens the
vhost library should just return normally. It's current behavior
of logging just creates massive log spew/overflow which could even
act as a DoS attack against host.
Reported-by: Nathan Law <nlaw@brocade.com> Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Yuanhan Liu [Tue, 8 Dec 2015 03:05:12 +0000 (11:05 +0800)]
virtio: fix wrong queue index
We should provide VIRTIO_PCI_QUEUE_SEL with vq->vq_queue_idx,
but not vq->queue_id.
vq->queue_id is the queue id from rte_eth_rx/tx_queue_setup(),
which always starts from 0 no matter which queue it is. However,
for virtio, even number is for RX queue, and odd number is for
TX queue.
Fixes: 5382b188fb8a ("virtio: add queue release") Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Chas Williams [Tue, 8 Dec 2015 14:48:24 +0000 (09:48 -0500)]
bnx2x: fix double init check
Check sc->state to prevent double init.
If the link is up, then the driver cannot be stopped and started
successfully. Instead of checking the link status, use the driver's
state.
Yuanhan Liu [Thu, 3 Dec 2015 02:27:02 +0000 (10:27 +0800)]
vhost: reserve some space in structures
So that we will not break ABI in future extension by adding few more
fields.
Struct vhost_virtqueue is reserved with 16 qwords (the later vhost-live
migration support would at least consume 3 of them), and struct virtio_net
is reserved with a bit more, 64 qwords, as there is only one instance for
a virtio nic instance.
Note that both reservation are not placed at the end of the struct, but
instead before the last field, since both the last field at the two struct
take a lot spaces. Putting the reservation after it would divide those
reserved fields to another cacheline. (we might need fix them in future, btw)
Suggested-by: Panu Matilainen <pmatilai@redhat.com> Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Lee Roberts [Mon, 7 Dec 2015 14:59:59 +0000 (07:59 -0700)]
kni: fix build on RHEL 7.2
RHEL 7.2 contains additional backports from newer upstream kernels.
Add RHEL_RELEASE_CODE logic for RHEL_RELEASE_VERSION(7,2) to pick up
the changes to kernel functions.
Jianbo Liu [Thu, 3 Dec 2015 15:02:54 +0000 (23:02 +0800)]
eal/arm: fix timer read from PMU
CONFIG_* from config files can not be used in code.
Fixes: 12f45fa7e29b ("eal/arm: read timer from PMU if enabled") Signed-off-by: Jianbo Liu <jianbo.liu@linaro.org> Acked-by: Jan Viktorin <viktorin@rehivetech.com>
Further enhancements to the userspace ethtool implementation that was
submitted in 2.1 and packaged as a self-contained sample application.
Implements an rte_ethtool shim layer based on rte_ethdev API, along
with a command prompt driven demonstration application.
Panu Matilainen [Mon, 7 Dec 2015 12:56:43 +0000 (14:56 +0200)]
mk: fix warnings when adding extra warning flags
Starting with commit 9aa2053c6e81493b23346ff4e387903560de5c81
EXTRA_CFLAGS is sometimes being passed to the compiler without
WERROR_FLAGS which can cause spurious warnings by the dozen,
for example with when compiling with EXTRA_CFLAGS="-Wformat-security":
cc1: warning: -Wformat-security ignored without -Wformat [-Wformat-security]
Passing WERROR_FLAGS to AUTO_CPU helper makes the warning flag usage
consistent throughout the codebase, silencing the warnings.
Fixes: 9aa2053c6e81 ("mk: influence CPU flags with user input") Signed-off-by: Panu Matilainen <pmatilai@redhat.com> Acked-by: Simon Kagstrom <simon.kagstrom@netinsight.net>
Panu Matilainen [Mon, 7 Dec 2015 10:45:36 +0000 (12:45 +0200)]
mk: fix external shared library dependencies of drivers
Similar to commit 5f9115e58cc6f304ff4ade694cf5823d32887d1a, but
for qat and mpipe drivers. The former did not exist when the
previous patch was sent and latter I just missed.
Panu Matilainen [Thu, 3 Dec 2015 14:05:26 +0000 (16:05 +0200)]
scripts: support any git revisions as ABI validation range
In addition to git tags, support validating abi between any legal
gitrevisions(7) syntaxes, such as "validate-abi.sh -1 . <target>"
"validate-abi.sh master mybranch <target>" etc in addition to
validating between tags. Makes it easier to run the validator
for in-development work.
Signed-off-by: Panu Matilainen <pmatilai@redhat.com> Acked-by: Neil Horman <nhorman@tuxdriver.com>
Yong Liu [Mon, 7 Dec 2015 07:24:17 +0000 (15:24 +0800)]
examples/vm_power_manager: fix build with libvirt < 1.0
virNodeGetCPUMap introduced in libvirt 1.0. In some linux distributions
like Ubuntu12/14 and Fedora18, libvirt version is older than 1.0. So this
sample will not build pass.
Replace "virNodeGetCPUMap" with another libvirt API "virNodeGetInfo".
Fan Zhang [Mon, 7 Dec 2015 11:51:50 +0000 (12:51 +0100)]
examples/ip_pipeline: add missing files of flow actions pipeline
When first adding flow actions pipeline, some files were not pushed
to the repository by mistake. The original commit message is below.
Fixes: 9ef2593651f9 ("examples/ip_pipeline: add flow actions pipeline")
Flow actions pipeline is an extension of flow-classification pipeline.
Some of the operations of flow classification pipeline such as traffic
metering/marking(for e.g. Single Rate Three Color Marker (srTCM), Two
Rate Three Color Marker trTCM)), policer can be performed separately in
flow action pipeline to avoid excessive computational burden on the CPU
core running the flow-classification pipeline. The Flow action pipeline
implements various function such as traffic metering, policer, stats.
Traffic mettering can configured as per the required context, for
examples- per user, per traffic class or both. These contexts can be
applied by specifying parameters in configuration file as shown below;
The entries of flow and dscp tables of flow actions pipeline can be
modified through command-line interface. The commands to add or delete
entries to the flow table, DSCP(differentiated services code point)
table and for statistics collection, etc have been included. The key
functions such as Traffic Metering/marking and policer functions have
been implemented as flow-table action handler.
John McNamara [Fri, 13 Nov 2015 11:45:12 +0000 (11:45 +0000)]
doc: fix examples in netmap compatibility guide
Fix the examples in the netmap compatibility sample application
docs which referred to the packet_reordering application.
Also fix some minor rst formatting issues.
Reported-by: Qian Xu <qian.q.xu@intel.com> Signed-off-by: John McNamara <john.mcnamara@intel.com> Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Aaron Conole [Fri, 23 Oct 2015 20:03:57 +0000 (16:03 -0400)]
power: remove duplicate definition
The CHANNEL_CMDS_MAX_VM_CHANNELS is duplicated in the channel_commands
header file. This commit removes that duplication.
Fixes: 210c383e247b ("power: packet format for vm power management") Signed-off-by: Aaron Conole <aconole@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
app/testpmd: add description of queue info command
Fixes: ab3257e13dc4 ("app/testpmd: add command to display queue info") Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com> Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Correct the sample code in the pcap_ring.rst file to match the latest
rte_eth_ring.c code.
The parameters to the rte_eth_from_rings() function have changed since
the documentation was written.
The API change occurred before DPDK 1.8 when the rst files were added.
The original documentation on which the pcap_ring.rst file was based was
not correct.
Fixes: fc1f2750a3ec ("doc: programmers guide") Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com> Acked-by: John McNamara <john.mcnamara@intel.com>
If eth_ring vdevs are created from the command line with the --vdev option,
they create their own rings which are not shared by other vdevs.
Some of tests in this suite require that the vdevs share rings, so some
of the tests fail.
For vdevs to share rings they must be created in the test code with the
rte_eth_from_rings() function using rings created with the rte_ring_create()
function.
Use the command line option --vdev=eth_ring0 to create port 0.
This option is not mandatory for the tests to pass. It allows some additional
functional tests on cmd-line created rings to be executed.
Create two rings and five ethdevs in test_pmd_ring.c for ports 1 to 5,
and then use these to run the unit tests on the pmd ring functionality.
Improve test output by adding the port number to printf statements,
and adding a printf describing each test.
Fixes: 572eb3cd834c ("ring: simplify unit tests") Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
David Hunt [Mon, 23 Nov 2015 11:45:48 +0000 (11:45 +0000)]
app/test: fix crash in performance test
Fix crash in pmd_perf_test autotest (div by 0) when no packets received
Also fixes the fact that the test passes even if exec_burst fails
To repeat the issue:
The system must be incorrectly set up so that all packets will be lost,
i.e. no loopback cable, etc. This is an edge case, but still the test
should not crash or pass when failing.
run the test app
RTE>> set_rxtx_sc poll_before_xmit
RTE>> pmd_perf_autotest
--snip--
> Generate 4096 packets @socket 1
> start to receive total expect 4096
> 4096 packets lost, IDLE 10000 times
> Floating point exception (core dumped)
Harish Patil [Sun, 8 Nov 2015 19:39:56 +0000 (11:39 -0800)]
examples/l3fwd: fix crash with unaligned L2 header
l3fwd app expects PMDs to return packets whose L2 header is
16-byte aligned due to usage of _mm_load_si128()/_mm_store_si128()
intrinsics in the app. However, most of the protocol stacks expects
packets such that its IP/L3 header be aligned on a 16-byte boundary.
Based on the recommendations received on dpdk-dev, we are changing
the l3fwd app to use _mm_loadu_si128()/_mm_loadu_si128() so that the
address need not be 16-byte aligned and thereby preventing crash.
We have tested that there is no performance impact due to this
change.
Signed-off-by: Harish Patil <harish.patil@qlogic.com> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Pablo de Lara [Tue, 3 Nov 2015 20:13:48 +0000 (20:13 +0000)]
examples/l3fwd: fix crash with IPv6
Lookup burst size was changed for exact match
from 4 to 8, for both ipv4 and ipv6, but actually only
4 keys were being looked up for ipv6, instead of 8,
causing random segmentation faults.
Fixes: 80fcb4d4 ("examples/l3fwd: increase lookup burst size to 8") Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com> Acked-by: Reshma Pattan <reshma.pattan@intel.com>
git.droids-corp.org / dpdk.git / log