Age | Commit message (Collapse) | Author | Files | Lines |
|
Type: improvement
Change-Id: I06130447b4e6f4726c4f5bffbe606385c45b8bd4
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
signs were changed here when calculating the offset:
d78ba5aa01ff1415bff0b06069ce21e0a78df89c
Type: fix
Signed-off-by: Artem Glazychev <artem.glazychev@xored.com>
Change-Id: I62b7a409caaf478e40efbdd6000922dcc7e92860
|
|
Type: refactor
Change-Id: Id10cbf52e8f2dd809080a228d8fa282308be84ac
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
The total_length_not_including_first_buffer field must be reset before
being updated otherwise it will quicly grows as stale values are reused.
Type: fix
Change-Id: Ic48c0822660998b0dfc0b5fdeadae6071b2d03f7
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Change-Id: I456cc0b0a6f2dc32b14791baf9d4a7f67279e8df
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Prefetch should be the next group of buffers instead of current buffers
Type: fix
Signed-off-by: Tianyu Li <tianyu.li@arm.com>
Change-Id: If13967e27f74c10856ae48892420d7ade6f0c92c
|
|
Type: fix
Change-Id: I6fce8c1f91fc5080271c1832bf40a9fb6bbcb7e3
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: refactor
Change-Id: I127b8c7131fd993a5fe605c7c36cc1d0b6672984
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I8169da230eb5f74651810a8e2490895620c38269
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: Idef30aee80b654ce424b6f1f1f730574ca68874c
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Signed-off-by: Ole Troan <ot@cisco.com>
Change-Id: I368b0410db2d633d3c52199c840e24d21952c1b4
|
|
Also rename counters in stat segment to make them more filesystem friendly.
Type: improvement
Signed-off-by: Ole Troan <ot@cisco.com>
Change-Id: I8a3f3ec318931f5475fcb181f8b4a079a1fa4b9c
Signed-off-by: Ole Troan <ot@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>
|
|
We hit a crash when the client sends us a bogus deescriptor which causes us
to access memory beyong the mapping. While the client clearly should not do
that, it is rather cheap for VPP to validate the descriptor instead of crash
and burn.
Type: fix
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: Id09035810939f5f98530f212f0b23e606132251d
|
|
Change-Id: Ifa8bccd8a34ec1b14e772ee53757e9083373e3de
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: feature
|
|
Type: style
Change-Id: I0d25d922312dfd60a665916cf47c3d2faa29ea49
Signed-off-by: Paul Vinciguerra <pvinci@vinciconsulting.com>
|
|
When using classifier to filter traces, not all packets will be traced.
In that case, we should only count traced packets.
Type: fix
Change-Id: I87d1e217b580ebff8c6ade7860eb43950420ae78
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
vlib_increment_combined_counter takes sw_if_index, not hw_if_index.
Type: fix
Signed-off-by: Steven Luong <sluong@cisco.com>
Change-Id: Iecde2697ed490940f0eff796d28d15381405b895
|
|
Type: fix
Signed-off-by: Mrityunjay Kumar <kumarnitp@gmail.com>
Change-Id: I31cc5e853b57e285064647503231b251e5152d3f
|
|
Type: feature
Change-Id: I913f08383ee1c24d610c3d2aac07cef402570e2c
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
When mode == ip, the variable next_index is not initialized.
Although insde the while loop, ni will be fixed to contain next[0],
it is easier to initialize it.
Type: fix
Change-Id: I6653a958bbc5105e7266bf89e3c8569ff00f0199
Signed-off-by: Steven Luong <sluong@cisco.com>
|
|
The fast path almost always has to deal with the real
pointers. Deriving the frame pointer from a frame_index requires a
load of the 32bit frame_index from memory, another 64bit load of the
heap base pointer and some calculations.
Lets store the full pointer instead and do a single 64bit load only.
This helps avoiding problems when the heap is grown and frames are
allocated below vm->heap_aligned_base.
Type: refactor
Change-Id: Ifa6e6e984aafe1e2755bff80f0a4dfcddee3623c
Signed-off-by: Andreas Schultz <andreas.schultz@travelping.com>
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Change-Id: I01c4f5755d579282773ac227b0bc24f8ddbb2bd1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I4e836244409c98739a13092ee252542a2c5fe259
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Example:
buffers {
default data-size 1536
}
Change-Id: I5b4436850ca18025c9fdcfc7ed648c2c2732d660
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: If514366be7385c64cafc03329f66b44380556daa
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I6aa030429c1740f7376e95daf82fce49efa6716b
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
This patch introduces following changes:
- deprecated free lists which are not used and not compatible
with external buffer managers (i.e. DPDK)
- introduces native support for per-numa buffer pools
- significantly improves performance of buffer alloc and free
Change-Id: I4a8e723ae47056717afd6cac0efe87cb731b5be7
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Storing buffer in local template seems to be better option....
Change-Id: I1a2fdd68cb956f99a5b36d2cd810fc623e089bcf
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ida2262238c0558d8340f05c678a84e76f990935d
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Id4f37f5d4a03160572954a416efa1ef9b3d79ad1
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Change-Id: Ied34720ca5a6e6e717eea4e86003e854031b6eab
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
This significantly reduces need for
...
in multiarch code. Simply constructor macros will jost create static unused
entry if CLIB_MARCH_VARIANT is defined and that will be optimized out by
compiler.
Change-Id: I17d1c4ac0c903adcfadaa4a07de1b854c7ab14ac
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: If6970788396c85415634f12304f49eed0d812281
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
It is cheaper to get thread index from vlib_main_t if available...
Change-Id: I4582e160d06d9d7fccdc54271912f0635da79b50
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ibab5e27277f618ceb2d543b9d6a1a5f191e7d1db
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Id775efb2e85d850e510d00f1b48bb711a3342397
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I65306fb1f8e39221dd1d8c00737a7fb1c0129ba8
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
In version 1 of the protocol sender was always ring producer and
receiver was consumer. In version 2 slave is always producer,
and in case of master-to-slave rings, slave is responsible for
populating ring with empty buffers.
As this is major change, we need to bump version number.
In addition, descriptor size is reduced to 16 bytes.
This change allows zero-copy-slave operation (to be privided in the separate
patch).
Change-Id: I02115d232f455ffc05c0bd247f7d03f47252cfaf
Signed-off-by: Damjan Marion <damarion@cisco.com>
Signed-off-by: Jakub Grajciar <jgrajcia@cisco.com>
|
|
interface reconnect.
Change-Id: Ifc7eb2494a22c334d8899422545fca1a4bba4d05
Signed-off-by: Chun Li <chunl2@cisco.com>
|
|
Change-Id: I498ed1162eadf3eff2543f1ec02a9b1e5fdc05d8
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I881551e6c13503a71ae29a7a58bde4d193745d55
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ie849ab713ff086187c18a91ab32e58207fe94033
Signed-off-by: Damjan Marion <damarion@cisco.com>
Signed-off-by: Jakub Grajciar <Jakub.Grajciar@pantheon.tech>
|
|
- always use 'va_args' as pointer in all format_* functions
- u32 for all 'indent' params as it's declaration was inconsistent
Change-Id: Ic5799309a6b104c9b50fec309cba789c8da99e79
Signed-off-by: Christophe Fontaine <christophe.fontaine@enea.com>
|
|
In IP mode memif interface is L3 point-to-point interfaces and
we don't pass l2 header. There is no l2 header rewrite operation and
received packets are sent straight to ip4-input / ip6-input nodes.
Change-Id: I4177f3fce3004da7ecf14d235006ae053fcf3f09
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Current memif interface supports frame size up to 2048. This patch is to
enhance memif to support jumbo frames.
On tx (writing buffers to the ring), keep reading the next buffer in vlib
when the flag VLIB_BUFFER_NEXT_PRESENT and merge it to the same ring entry.
Use descriptor chaining if the buffer is not big enough.
On rx (reading buffers from the ring), if the packet is greater than 2048,
create multiple vlib buffers, chained with the VLIB_BUFFER_NEXT_PRESENT.
Testing:
Because the ping command provided by VPP does not support jumbo frames,
I have to use linux ping. Here is the set up that I use for testing.
VM1 --- vhost ---- VPP1 --- memif --- VPP2 --- vhost --- VM2
Create vhost-user interfaces between VM1 and VPP1 and between VPP2 and VM2
VM configuration:
Set the interface mtu on the VM, e.g 9216 to support jumbo frames.
create static route and static arp on VM1 to VM2 and vice versa.
Use iperf3 or ping -s 8000 from VM1 to VM2 or vice versa.
Sample run
sluong@ubuntu:~$ ping 131.1.1.1 -c1 -s 8000
ping 131.1.1.1 -c1 -s 8000
PING 131.1.1.1 (131.1.1.1) 8000(8028) bytes of data.
8008 bytes from 131.1.1.1: icmp_seq=1 ttl=62 time=0.835 ms
--- 131.1.1.1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.835/0.835/0.835/0.000 ms
sluong@ubuntu:~$
DBGvpp# sh interface memif0
Name Idx State Counter Count
memif0 1 up rx packets 1
rx bytes 8042
tx packets 1
tx bytes 8042
ip4 1
DBGvpp#
Change-Id: I469bece3d45a790dceaee1d6a8e976bd018feee2
Signed-off-by: Steven <sluong@cisco.com>
|
|
Change-Id: I4d41def83a23f13701f1ddcea722d481e4c85cbc
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
- Add rx-queues and tx-queues option to the create memif CLI
- Add vlib_worker_thread_barrier_sync () to memif_conn_fd_read_ready () as
the latter function may disconnect the ring and clean up the shared memory.
- On transmit, write the rid (queue number) to the socket.
- On receive, read the rid and trigger the interrupt for the corresponding
thread.
Change-Id: If1c7e26c7124174678f047909cbc33e931eaac8c
Signed-off-by: Steven <sluong@cisco.com>
|
|
Migrate memif to use vnet device infra APIs. No new function is added.
Change-Id: I70e440d2ae1e673876365041f31fe78997aceecf
Signed-off-by: Steven <sluong@cisco.com>
|
|
This patch deprecates stack-based thread identification,
Also removes requirement that thread stacks are adjacent.
Finally, possibly annoying for some folks, it renames
all occurences of cpu_index and cpu_number with thread
index. Using word "cpu" is misleading here as thread can
be migrated ti different CPU, and also it is not related
to linux cpu index.
Change-Id: I68cdaf661e701d2336fc953dcb9978d10a70f7c1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|