Age | Commit message (Collapse) | Author | Files | Lines |
|
default RLIMIT_MEMLOCK is 64. if we use multi af_xdp interfaces or
load complex bpf program, libbpf will return permission error.
root cause is default 64 is not large enough. So we change it before
load bpf program.
Type: fix
Change-Id: Ia6aed19c9256c498cf1155586a54a32b3f444105
Signed-off-by: Chen Yahui <goodluckwillcomesoon@gmail.com>
|
|
Type: fix
Signed-off-by: Chen Yahui <goodluckwillcomesoon@gmail.com>
Change-Id: Ia447420f692f1487d343886845d648d766e43c27
Signed-off-by: Chen Yahui <goodluckwillcomesoon@gmail.com>
|
|
Type: improvement
Change-Id: If3a83848ae0741334887c654b65e424b99caa73c
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Originally XDP_FLAGS_REPLACE was passed to the function.
For kernels not defining this macro (for example 5.4) in if_link.h vpp
sets it to 0.
If kernel has this macro problem appears, replace flag requires
specifying the program to be replaced. bpf_set_link_xdp_fd wraps
around __bpf_set_link_xdp_fd_replace and passes 0 as old_fd, it leads to
an error while assessing for replacement (if 0 is passed
there is no assessing for replacement).
To address this issue no flag is passed to the function, only 0.
Type: improvement
Signed-off-by: Dastin Wilski <dastin.wilski@gmail.com>
Change-Id: I3689ce7eb8c71c699f0e589111929979c2bbe213
|
|
Type: improvement
Change-Id: I3659de6599f402c92e3855e3bf0e5e3388f2bea0
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
- per hw-interface-class handlers
- ethernet set_mtu callback
- driver can now refuse MTU change
Type: improvement
Change-Id: I3d37c9129930ebec7bb70caf4263025413873048
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Prep for supporting multiple callbacks, optional args, etc.
Type: improvement
Change-Id: I96244c098712e8213374678623f12527b0e7f387
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Make it shorter to type, easier to debug, make adding callbacks in
future simpler.
Type: improvement
Change-Id: I6cdd6375e36da23bd452a7c7273ff42789e94433
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Signed-off-by: arikachen <eaglesora@gmail.com>
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Change-Id: If8d57bcf033864935bd5e3a9912b2c1a7c712f44
|
|
In some situation, we support to deploy vpp as per host
and handler packet in container, so we use xdp to redirect
the flow.
Type: improvement
Signed-off-by: arikachen <eaglesora@gmail.com>
Change-Id: Iab42d6a0abb2b330a284d519018a90aff2fa4371
|
|
Attempting to create an af_xdp interface with zerocopy where the
underlying driver didn't support it would lead to a crash due to
queue creation silently failing.
Type: fix
Signed-off-by: Joshua Roys <roysjosh@gmail.com>
Change-Id: Ifd9070b8c2b3023d71120c5cf20f7e89d04e4cb3
|
|
Type: fix
Signed-off-by: arikachen <eaglesora@gmail.com>
Change-Id: Idb5e66d7a2a7ccb6fb5155341df54586186eb11f
|
|
xsk objects keep pointers to the rx and tx objects. If we re-allocate
the rx and tx vectors after initializing the associated xsk object, the
pointers in the xsk object will be staled.
To avoid this, we allocate the vectors to the max expected size instead
of growing them.
Type: fix
Change-Id: If30433a28c186787d66c12dbab34bf210c95b519
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
while xsk fd closed by clib_file_del_by_index fisrt,
xsk_get_mmap_offsets will get wrong off in xsk_socket__delete,
so munmap wrong place.
Signed-off-by: arikachen <eaglesora@gmail.com>
Change-Id: I95464e9b4eec99814bd32d7402c0d60a0605cef5
|
|
Instead of pre-programming the data offset on rx, use offset passed in
the descriptor. This is more robust and future-proof.
Type: fix
Change-Id: I2bd910d92b8b03d17be5be85a24108be711dc7b9
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
If interrupt mode is configured through startup.conf exec script, the
input function will not be polled and the rx ring will never be filled.
Always refill the ring when interface goes up so it is ready to receive
packets.
Type: fix
Change-Id: I4cf22c8ae00638679f2e8650303a6fe916c1319b
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Prior to Linux 5.6 there is a race condition between poll() and
sendmsg() in the kernel. This patch protects the syscalls with a lock
to prevent it, unless the NO_SYSCALL_LOCK flag is set at create time.
See
https://lore.kernel.org/bpf/BYAPR11MB365382C5DB1E5FCC53242609C1549@BYAPR11MB3653.namprd11.prod.outlook.com/
Type: fix
Change-Id: Ie7d4f5cb41f697b11a09b6046e54d190430d76df
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
trajectory trace has been broken for a while because we used to save the
buffer trajectory in a vector pointed to in opaque2. This does not work
well when opaque2 is copied (eg. because of a clone) as 2 buffers end up
sharing the same vector.
This dedicates a full cacheline in the buffer metadata instead when
trajectory is compiled in. No dynamic allocation, no sharing, no tears.
Type: refactor
Change-Id: I6a028ca1b48d38f393a36979e5e452c2dd48ad3f
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
This patch adds flags to represent the modern NICs capabilities.
Change-Id: I96d38d9ab7eac55974d72795cd100d8337168e1e
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Change-Id: I85a463b4ca15baf11e3eb70189f5190ba2585170
Type: refactor
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
Type: improvement
Change-Id: Icb23af5f5e458a555f416cb0a829e84646b25dd9
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: feature
Change-Id: I97176c2c90ea664a68078b3a7b7d44eb237a7f13
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Non-NUMA systems might report -1 as NUMA node.
Type: fix
Change-Id: I092c817ea670009d6f530cc70ad13d45e15fd363
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: feature
Change-Id: I85aa4ad6b68c1aa0e51938002dc691a4b11c545c
Signed-off-by: Damjan Marion <damarion@cisco.com>
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|