Do not return HTTP errors to server on parse errors in client.
Type: fix
Change-Id: Id3e99d69626855848faa87af73002d559d948516
Signed-off-by: Matus Fabian <matfabia@cisco.com>
When the NIC does not support mac filter, we rely on ethernet-input
node to do the destination mac check, ie, when the interface is in L3,
the mac address for the packet must be the mac address of the
interface where the packet arrives. This works fine in ethernet-input
node when all packets in the frame might have different interfaces, ie,
ETH_INPUT_FRAME_F_SINGLE_SW_IF_ID is not set in the frame. However,
when all packets are having the same interface,
ETH_INPUT_FRAME_F_SINGLE_SW_IF_ID is set, ethernet-input node goes
through the optimized routine eth_input_single_int -> eth_input_process_frame.
That is where dmac check has a bug when all packets in the frame are
either, ip4, ip6, or mpls without vlan tags. Because without vlan tags,
the code handles all packets in fast path and ignores dmac check.
With vlan tags, the code goes to slow path where dmac check is handled
properly.
The fix is to check if we have a bad dmac in the fast path and force the
code to go to slow path which will handle dmac check properly.
Also do a wholesale correction on all the testcases which do not use
the proper dmac when sending L3 packets.
Type: fix
Change-Id: I73153a805cecdc24c4eefcc781676de04737ae2c
Signed-off-by: Steven Luong <sluong@cisco.com>
HTTP client should be able to send another request after large data
receiving.
Type: fix
Change-Id: I8352ea760a4b3de4b79159531c7e4c1fa4e03203
Signed-off-by: Matus Fabian <matfabia@cisco.com>
Notify client app that we errored out while trying to generate request.
Type: fix
Change-Id: I3e20fe22d8a927ae8e9bf8391297e10c974dc941
Signed-off-by: Matus Fabian <matfabia@cisco.com>
The VPP API auto-generated endian conversion functions are intended to
be symmetrical. They are used both by the API client and the API server.
Called on send to convert from host endian to network endian and on
receive to convert back.
For variable length arrays, we have to iterate over the array and call
a more specific handler for the array type. Unfortunately the length of
the array is part of the api definition, and if it's endian swapped
prior to the for loop, unexpected behaviour will ensue.
There was an earlier fix, for some specific messages, but unfortunately
that only fixed the problem from the VPP (server) side.
This adds a new parameters to the endian handler, so the boundary
argument to the loop can be treated differently depending on if this
message is to the network or from the network.
Type: fix
Change-Id: I43011aed384e3b847579a1dd2c390867ae17a9ad
Signed-off-by: Ole Troan <otroan@employees.org>
RTLD_DEEPBIND only exists in glibc. If it isn't defined we can just
ignore it.
Type: fix
Change-Id: I5b2d65b2462020808635b057d5f7665fb75a2444
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
Signed-off-by: Eric Sun <esun@meraki.com>
This change changes all instances of `uint` to `unsigned int` to avoid
use of the non-standard type. This fixes problems that some versions of
GCC may have.
Type: fix
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
Change-Id: I08cce367c2a8f31efb2fccabcb8f699d801b0f8a
Signed-off-by: Eric Sun <esun@meraki.com>
Some compilers were not clever enough to figure out these inits
weren't needed. Hence adding them.
Type: fix
Signed-off-by: Guillaume Solignac <gsoligna@cisco.com>
Change-Id: I50c4e978bfd2ef15ece7a29e517bb8ddba12f960
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
Signed-off-by: Dave Wallace <dwallacelf@gmail.com>
This reverts commit 71c32a898941e32b5d4f865b50fbe775560c582d.
Type: fix
Reason for revert: vnet pinning is not considered in this patch.
This causes keywords 'workers' and 'skip-cores' to be broken,
as well as keyword 'main-core auto' introduced in this patch.
If this patch is ever reconsidered, it must account for vnet
pinning fix in commit https://gerrit.fd.io/r/c/vpp/+/40711.
Change-Id: I1f3154a6c7e830b100f824375aa00e95b192f7f3
Signed-off-by: hsandid <halsandi@cisco.com>
--file-prefix is only supported by DPDK on Linux currently.
Type: improvement
Change-Id: I47765ea666b7d14011804e7cc90d9035bce1bc93
Signed-off-by: Tom Jones <thj@freebsd.org>
FreeBSD dynmically grows pipes based on usage and available system
memory. Don't try to resize pipes on FreeBSD for now.
Type: improvement
Change-Id: I78b06dead5d42a3a7bdf634a67b43ef854e510f8
Signed-off-by: Tom Jones <thj@freebsd.org>
FreeBSD sed isn't compatible with gnu seds in place option. On FreeBSD
explicitly use gsed, on Linux and other platforms this is available just
as 'sed'.
Type: improvement
Change-Id: I6e4fcaf11d4d22cd3e4cd0c3327b16393542845f
Signed-off-by: Tom Jones <thj@freebsd.org>
This patch compares packet length with maximum supported
packet length and drops the packet accordingly.
Type: fix
Change-Id: I80ef453d43149818936649e1e58ae90c84a34ab9
Signed-off-by: Monendra Singh Kushwaha <kmonendra@marvell.com>
If peek touches all data in the fifo, unset fifo event to request new
events when more data is enqueued.
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I878fa017f18ada5cef557600673bba920f230542
set_interface_name may crash if wrong sw_if_index is passed e.g. ~0
Type: fix
Signed-off-by: Stanislav Zaikin <stanislav.zaikin@46labs.com>
Change-Id: Ic7e400c914fb33c2f9eac4f2985bb5b163a18d57
App name is used in HTTP response Server header.
Type: fix
Change-Id: Ie3b2d985dd7d554a0352f7e602891f878bebd031
Signed-off-by: Matus Fabian <matfabia@cisco.com>
On glibc this builds fine; on musl the byteswap header doesn't seem to be
included properly. This change adds it, fixing the build.
Type: fix
Change-Id: I402812409395d471357b6317084774afba39548f
Signed-off-by: Eric Sun <esun@meraki.com>
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
HTTP client was relying on synchronous rx notifications to the client
app when moving lage data from underlying transport proto.
Recent change in session layer made such notifications asynchronous
making http client not working. This patch fixes the issue.
Type: fix
Change-Id: I4b24c6185a594a0fe8d5d87c149c53d3b40d7110
Signed-off-by: Filip Tehlar <ftehlar@cisco.com>
Signed-off-by: Matus Fabian <matfabia@cisco.com>
Effective CPUs file is different between cgroups v1 and v2.
Type: test
Change-Id: Ic1b9059435b3f0944624120f57e72da45adca223
Signed-off-by: Matus Fabian <matfabia@cisco.com>
When building with musl, PATH_MAX is not available if we don't add its header
to pci.c.
Type: fix
Signed-off-by: Guillaume Solignac <gsoligna@cisco.com>
Change-Id: I3d4e7cf5754d7c1d480834d7cd038d5bda46fd2f
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
Previously all of the vcl library was excluded from the build on
FreeBSD, this was too agressive as only the ldpreload component requeres
platform specific implementation for FreeBSD
Type: improvement
Change-Id: I94e70d7730878cc98060561eedf07cb7e4d1868a
Signed-off-by: Tom Jones <thj@freebsd.org>
FreeBSD has a different interface to request memory mapping information.
Add a FreeBSD specific method for reading physical addresses and make it
available at build time.
Type: improvement
Change-Id: I3588dde8e0a6f6d53333040245341ed09cebef9d
Signed-off-by: Tom Jones <thj@freebsd.org>
Implement a subset of the pci interface with unimplemented methods
marked with a comment. This is enough functionality to allow dpdk to
attach to pci devices and forward traffic on FreeBSD.
Type: improvement
Change-Id: I0d87778cd92ae332a6802c11083d3d99988678a1
Signed-off-by: Tom Jones <thj@freebsd.org>
FreeBSD doesn't support SCM_CREDENTIALS, but has an equivalent mechanism
called SCM_CRED. This is documented in unix(4), use SCM_CREDS on
FreeBSD.
Type: improvement
Change-Id: Iaf4ec0d68445239c06fa0eef6c25db7e3328e989
Signed-off-by: Tom Jones <thj@freebsd.org>
The PCI class values used to identify devices are different between
platforms. Add a selector for the values used on FreeBSD.
Type: improvement
Change-Id: I981f0dc24954457068fc94bcc4681b120ee253db
Signed-off-by: Tom Jones <thj@freebsd.org>
