When qemu close the unix socket fd of the vhostuser as a
server, and then immediately delete the vhostuser port on
openvswitch. There will be a deadlock.
A thread (fdset event thread): B thread:
1. fdset_event_dispatch rte_vhost_driver_unregister
2. set the fd busy to 1. lock vsocket->conn_mutex
3. vhost_user_read_cb fdset_del waits busy changed to 0.
4. vhost peer closed, remove the
conn from vsocket->conn_list:
lock vsocket->conn_mutex
5. set the fd busy to 0
Fixes: 65388b43f5 ("vhost: fix fd leaks for vhost-user server mode")
Cc: stable@dpdk.org
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
In VM2NIC case zero copy may need some tuning to get best performance.
This patch describes the zero copy starved case and provides a tuning
tip.
Signed-off-by: Junjie Chen <junjie.j.chen@intel.com>
Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Fix missing DEV_RX_OFFLOAD_CHECKSUM flag in RX offloads.
Remove depricated txq_flags field.
Fixes: a92870896b ("net/octeontx: use the new offload APIs")
Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
We are still using the txq_flags which is no longer needed with the
new offload API. Cleaning it up.
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Since we move to new offload APIs, txq_flags is no long needed.
This patch remove the dependence on that.
Fixes: 30f3ce999e ("net/fm10k: convert to new Tx offloads API")
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Since we move to new offload APIs, txq_flags is no long needed.
This patch remove the dependence on that.
Fixes: e5c05e6590 ("net/e1000: convert to new Tx offloads API")
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Since we move to new offload APIs, txq_flags is no long needed.
This patch remove the dependence on that.
Fixes: 51215925a3 ("net/ixgbe: convert to new Tx offloads API")
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
There is no necessity to fill in TxQ flags since ethdev maps
Tx offloads to TxQ flags on device info get for apps which are
not converted yet to Tx offloads API.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Tx offload will be converted to txq_flags automatically during
rte_eth_dev_info_get and rte_eth_tx_queue_info_get. So PMD can
clean the code to get rid of txq_flags at all while keep old APP
not be impacted.
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Tested-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
In ip_frag_process, some IP_FRAG_LOG content is wrong.
Fixes: 4f1a8f6338 ("ip_frag: add IPv6 reassembly")
Cc: stable@dpdk.org
Signed-off-by: Li Han <han.li1@zte.com.cn>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Patch adding configurable locking has reshuffled some init stages, but
did not put them back in correct order. Fix order of init by moving
everything that was before arguments parsing into correct places.
Fixes: e505d84c64 ("app/testpmd: make locking memory configurable")
Reported-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Ali Alnubani <alialnu@mellanox.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
- add EXPERIMENTAL tag for the section in MAINTAINERS.
- add EXPERIMENTAL tag to BPF public API files.
- add attribute __rte_experimental to BPF public API declarations.
Fixes: 94972f35a0 ("bpf: add BPF loading and execution framework")
Fixes: 5dba93ae5f ("bpf: add ability to load eBPF program from ELF object file")
Fixes: a93ff62a89 ("bpf: introduce basic Rx/Tx filters")
Reported-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Fixes the wrong argument in table action miss function.
Fixes: 4c387fcdf7 ("pipeline: add new functions for action handlers")
Cc: stable@dpdk.org
Signed-off-by: Jasvinder Singh <jasvinder.singh@intel.com>
In strlcpy function parameters there was no allowance for
null terminator, so ip address was copied without last character.
Fixes: ae943ebe1e ("examples/ipsec-secgw: replace strncpy with strlcpy")
Cc: stable@dpdk.org
Signed-off-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
The second parameter "name" in the function rte_eth_dev_detach
has been already redefined as "char *name __rte_unused",
"port_id" is printed instead of "name" in testpmd.
Fixes: b65ecf1993 ("devargs: rename legacy API")
Cc: stable@dpdk.org
Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Acked-by: Matan Azrad <matan@mellanox.com>
With this change, DPAA and DPAA2 configuration compiled applications
are interoperable without any impact on the functional behavior.
Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>
The function pthread_create() expects void *(*func) (void *)
for function pointer, however, lthread_func_t was defined as
void (*func) (void *), so now gcc 8.1 warns that the cast is
incorrect, causing a compilation failure. This patch changes
the declaration of lthread_func_t from returning a void to
returning a void*, and then changes the sample app in the
relevant places that are affected by the typedef change.
Fixes: 116819b9ed ("examples/performance-thread: add lthread subsystem")
Cc: stable@dpdk.org
Signed-off-by: David Hunt <david.hunt@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
This patch provides fix for icc compilation issue with event
crypto adapter test application.
Currently, number of elements in the session mempool is defined
by max_nb_sessions & this is replaced with a macro.
Fixes: 3c2c535ecf ("test: add event crypto adapter auto-test")
Reported-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Signed-off-by: Abhinandan Gujjar <abhinandan.gujjar@intel.com>
Acked-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
When a removable device is plugged-out, a RMV interrupt is invoked and
the application can catch the event in order to stop the device
management.
The Testpmd wrong behavior in this case is to detach the removed device
using the EAL detach API.
The EAL API does not invalidate the ethdev port and the port keeps
appearing as valid from the ethdev point of view.
Thus, the next operations for the ethtev port X may trigger an invalid
rte_device access. For example, calling "show port info X" may cause
segfault.
Moreover, the removed port is not removed from the Testpmd data-path
structures. Therefore, the invalid device may still be used by the
Testpmd data-path.
Call the Testpmd detach_port() function which uses the ethdev detach
API, and prepare the Testpmd forward ports database for a new
forwarding session without the detached port.
Fixes: 284c908cc5 ("app/testpmd: request device removal interrupt")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
In the RMV device event callback, there is a call for the removed
device stop operation which triggers a link status operation for the
removed device.
It may casue an error from the removed device PMD.
Skip the link status operation in the above described case.
Fixes: 284c908cc5 ("app/testpmd: request device removal interrupt")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
When the user uses the synchronic hot-plug commands, attach\detach, in
order to insert\remove a port from the system, the forward ports list
update is missed in the current implementation.
Thus, an invalid port may be used for data-path in case of detach
because the detached port was not removed from the forward port list.
In addition, a new port is not used for data-path in case of attach, as the
default behavior of Testpmd, because the attached port was not inserted
to the forward port list.
Update the forward port list in the above cases to allow the correct
port usage for data-path in the next packet forwarding start.
Fixes: edab33b1c0 ("app/testpmd: support port hotplug")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
A port Rx queue flush is done when the packet forwarding starts in
order to clean the port statistics for a new traffic session.
The flush operation is wrongly called before the update of the new
forward ports, and may fail due to flush operation for an invalid port
configured by the old session.
Move the new forward port setup to be done before the Rx queue flush.
Fixes: 7741e4cf16 ("app/testpmd: VMDq and DCB updates")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
When the forward ports are changed either by new portlist\portmask
configurations or by a port detachment, all the old forward streams
are freed and new streams are allocated to be aligned with the new
forward ports.
If the number of the forward ports drops to 0, there is an attempt
to wrongly allocate 0 memory for the streams.
Skip the streams memory allocation if no forward ports are configured.
Fixes: ce8d561418 ("app/testpmd: add port configuration settings")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
There are several cases of an invalid port data access that causes the
printing of all the valid ports, for example, when the user asks to
receive a port information of an invalid port.
Wrongly, the port with id 0 is printed in all the above described
cases, regardless of its validity.
Print port 0 only if it is valid as done for the rest of the ports.
Fixes: af75078fec ("first public release")
Fixes: b6ea6408fb ("ethdev: store numa_node per device")
Fixes: edab33b1c0 ("app/testpmd: support port hotplug")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
Testpmd allows to create and control bonding devices by run time
command lines using the bonding PMD API.
Some bonding device slaves operations (close, stop, etc) should not be
used by the application and must be managed by the bonding PMD.
Thus, Testpmd manages slave flags to prevent the special operations
calls and when a slave is added to bonding device by a run time command
line, the flag is set.
There is one more way to define the slaves for a bonding device using
EAL command line and Testpmd doesn't set the slave flag in this case
what causes to the special operations to be called by Testpmd.
Add one more check to detect bonding slave device.
Fixes: 41b05095c4 ("app/testpmd: fix bonding start")
Fixes: 0e545d3047 ("app/testpmd: check stopping port is not in bonding")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@mellanox.com>
Acked-by: Bernard Iremonger <bernard.iremonger@intel.com>
If mlockall() is called while allocated VA space is bigger than
amount of available RAM, FreeBSD kernel will deadlock and need
a hard reboot. We do allocate big amounts of memory because of
how new memory subsystem works, so calling mlockall() will cause
a deadlock. So, disable mlockall() by default on FreeBSD.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Add two new command-line parameters for either enabling or
disabling locking all memory at app startup.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Remy Horton <remy.horton@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Remove version tag from experimental block in linker version scripts
(.map files).
That label is not used by linker and information only. It is useful
for version blocks but not useful for experimental block but confusing.
Removing those labels.
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Currently, page deallocation might fail if allocator cannot get page
fd, which will leave VA space still mapped, and will also not mark
page as free.
Fix page deallocation function to always unmap space before trying
to get rid of the page itself, and always mark page as free even if
page deallocation failed.
Fixes: a5ff05d60f ("mem: support unmapping pages at runtime")
Fixes: 1a7dc2252f ("mem: revert to using flock and add per-segment lockfiles")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Return value should be zero for success, but if unlock and unlink
have succeeded, return value was 1, which triggered failure message
in calling code.
Fixes: a5ff05d60f ("mem: support unmapping pages at runtime")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Segment index was calculated incorrectly, causing free_seg to
attempt to free segments that do not exist.
Fixes: a5ff05d60f ("mem: support unmapping pages at runtime")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: Yong Liu <yong.liu@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
If total memory is already bigger than max memory, an underflow
will occur on subtraction. Fix it by simply stopping whenever
we already have amount of memory that is bigger than maximum.
Fixes: 66cc45e293 ("mem: replace memseg with memseg lists")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Currently, reserving a memzone with length set to 0 will not trigger
any memory allocations, and memzone will instead be looking through
already allocated memory only. Document this limitation.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Size of malloc heap elements include overhead, which should not
be counted as part of memzone.
Fixes: fafcc11985 ("mem: rework memzone to be allocated by malloc")
Cc: stable@dpdk.org
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Deallocation used the wrong function, which could have resulted in
race conditions because the function does not use locks internally.
Fixes: 1403f87d4f ("malloc: enable memory hotplug support")
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
When we ask to reserve virtual areas, we usually include
alignment in the mapping size, and that memory ends up
being wasted. Wasting a gigabyte of VA space while trying to
reserve one gigabyte is pretty expensive on 32-bit, so after
we're done mapping, unmap unneeded space.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Mapping size is a 64-bit integer, but mmap() will accept size_t for
size mappings. A user could request a mapping with an alignment, which
would have overflown size_t, so check if (size + alignment) will
overflow size_t.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
The code aimed to pick and remember the value of
mempool ops name from EAL command line arguments does not
copy the string and remembers the pointer provided
by getopt_long() directly. The latter could be clobbered
later and result in reading wrong mbuf pool ops name
by rte_mempool library.
Typically, this flaw could be avoided by using strdup()
to remember the string value of the option.
Fixes: a103a97e71 ("eal: allow user to override default mempool driver")
Cc: stable@dpdk.org
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
