Fixes:
| In function 'memset',
| inlined from 'test_table_stub' at test_table_tables.c:151:4:
| /buildarea/tmp/work/intel_x86_64-wrs-linux/dpdk/19.11.5-r0/recipe-sysroot/usr/include/bits/string_fortified.h:59:10: error: '__builtin_memset' offset [0, 31] is out of the bounds [0, 0] [-Werror=array-bounds]
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
AF_XDP driver depend of libbpf. Enable it so that AF_XDP
driver can be compiled.
Signed-off-by: Mohamad Noor Alim Hussin <mohamad.noor.alim.hussin@intel.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Since commit 681a67288655 ("usertools: check if module is loaded
before binding") in dpdk, usertools/dpdk-devbind.py only checks
dynamic kernel drivers in /sys/module. So built-in kernel driver
cannot be bound.
Add "usertools-devbind-fix-binding-for-built-in-kernel-dr.patch" to
fix this issue.
Signed-off-by: Yongxin Liu <yongxin.liu@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
kernel/linux/kni/kni_dev.h:104:8: error: too many arguments to function 'get_user_pages_remote'
104 | ret = get_user_pages_remote(tsk, tsk->mm, iova, 1,
| ^~~~~~~~~~~~~~~~~~~~~
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Upstream renamed branch name 'master' to 'main',
so let's change it.
Signed-off-by: Yanfei Xu <yanfei.xu@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Since oe-core upgraded gcc to 10.1, there is a failure
...
|path_to_dpdk/19.11.1-r0/git/examples/ipsec-secgw/ipsec_process.c:131:34:
error: 'grp.m' may be used uninitialized in this function
[-Werror=maybe-uninitialized]
| 131 | grp[n].cnt = pkts + i - grp[n].m;
...
Add option `-Wno-error=maybe-uninitialized' to workaround
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
dpdk-devbind tool relies on lspci tool with specific options which are
not provided by lspci from busybox.
Signed-off-by: Eloi Bail <eloi.bail@savoirfairelinux.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
The commit 8c2fb51323 introduces an error
when installing test programs:
/buildarea//build/tmp/work/intel_corei7_64-poky-linux/dpdk/19.11.0-r0/git/x86_64-native-linuxapp-gcc/lib/librte_eal.a(eal_memory.o):
In function `eal_legacy_hugepage_init':
eal_memory.c:(.text+0x1f71): undefined reference to `memfd_create'
/buildarea/build/tmp/work/intel_corei7_64-poky-linux/dpdk/19.11.0-r0/git/x86_64-native-linuxapp-gcc/lib/librte_eal.a(eal_memalloc.o):
In function `get_seg_fd.constprop.0':
eal_memalloc.c:(.text+0x61f): undefined reference to `memfd_create'
eal_memalloc.c:(.text+0x696): undefined reference to `memfd_create'
/buildarea/build/tmp/work/intel_corei7_64-poky-linux/dpdk/19.11.0-r0/git/x86_64-native-linuxapp-gcc/lib/librte_eal.a(eal_memalloc.o):
In function `eal_memalloc_init':
eal_memalloc.c:(.text+0x21ad): undefined reference to `memfd_create'
/buildarea/build/tmp/work/intel_corei7_64-poky-linux/dpdk/19.11.0-r0/git/x86_64-native-linuxapp-gcc/lib/librte_pmd_memif.a(rte_eth_memif.o):rte_eth_memif.c:(.text+0xbc3): more undefined references to `memfd_create' follow
collect2: error: ld returned 1 exit status
The test programs have been complied and install to ${S}/test/app
directory. We just need to ship them to ${D} directory.
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
This reverts commit c29b7a480a.
Unfortunely, the reverted commit does not work in every cases. If
KERNEL_FEATURES was set in somewhere else other than a global configuration,
for example in kernel's bbappend, there would be no way for dpdk recipe to
detect if the wanted feature has been enabled.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
5eba511c5e ("dpdk: add recipe for 19.11") introduces CONFIG_NAME and
TEST_DIR for dpdk19 but not for dpdk18. Fix it.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
dpdk requires kernel option features/intel-dpdk/intel-dpdk.scc to build.
We'd better warn users in advance before they meet some confusing errors
at compile-time.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Introduce recipe to build 19.11 LTS.
This version changes the location of test directory to be under app and
also changes the name of config file to common_linux.
Introduce variables in shared inc to control these two differences.
These are set to values used by earlier versions by default.
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
When building examples and tests with GCC9, the following errors come up,
error: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Werror=cpp]
382 | # warning _FORTIFY_SOURCE requires compiling with optimization (-O)
| ^~~~~~~
cc1: all warnings being treated as errors
dpdk itself already appends -O3 the EXTRA_CFLAGS for some of those cases, but
dpdk.inc overwrites EXTRA_CFLAGS when evoking make. This patches adds it back.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
We have already had 18.11 and 18.05 is not a stable version.
This patch also removes the patches no longer needed.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Backport a patch from upstream to fix the following build failure.
x86_64-native-linuxapp-gcc/include/rte_ether.h:178:2: error:
converting a packed 'const struct ether_addr' pointer (alignment 1) to a 'unaligned_uint16_t' {aka 'const short
unsigned int'} pointer (alignment 2) may result in an unaligned pointer value [-Werror=address-of-packed-member]
| 178 | const unaligned_uint16_t *ea_words = (const unaligned_uint16_t *)ea;
| | ^~~~~
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
These two patches can't be applied to the new version DPDK, so move
them to the version specific bb.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Adjust the LICENSE according to the license files.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
This fixes dpdk build when using an external toolchain by adding
the HOST_CC_ARCH & TOOLCHAIN_OPTIONS to EXTRA_CFLAGS the way
standard Yocto does it to gather all the necessary flags for
compilation.
The TOOLCHAIN_OPTIONS variable also provides the sysroot flag, so
no need to explicitly provide the sysroot.
This commit also fixes the build when using a multilib toolchain
by adding the LDEMULATION flags to the LDFLAGS via TUNE_LDARGS
which are required while linking with a multilib toolchain.
Signed-off-by: Arsalan H. Awan <Arsalan_Awan@mentor.com>
Currently the examples&test applications would be installed into
directories like:
/usr/share/examples
/usr/share/test
These directories just seem too general, so change them to:
/usr/share/dpdk/examples
/usr/share/dpdk/test
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
This patch upgrades DPDK to the latest release of v18.05.
Signed-off-by: Tan, Raymond <raymond.tan@intel.com>
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
The TARGET_ARCH may be different for different x86 machines,
for example i586(qemux86) and i686(genericx86). So choose the "x86"
override to set the correct RTE_TARGET for these machines.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Fix the following parse error when building with unsupported arch, such
as arm:
bb.data_smart.ExpansionError: Failure expanding variable RTE_TARGET, expression was ${RTE_TARGET} which triggered exception Exception: variable RTE_TARGET references itself!
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Otherwise it always return "default" no matter what we set in
"DPDK_TARGET_MACHINE".
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
After changing to the git repo, these checksums become useless.
Just drop them.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
This cleans up the versioned recipes and moves the common
fragments to the include so to make the maintenance easy.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
This is more convenient for development.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Using the following commands to refresh the patches in order to
suppress the fuzz warnings.
devtool modify dpdk
devtool finish --force-patch-refresh dpdk meta-dpdk-dir
No code change.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
This is the latest stable version.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
The license files may be different between different versions. So move
them into the specific bb files.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Otherwise we get this:
| dpdk-18.02/lib/librte_eal/linuxapp/eal/eal_memory.c:27:10: fatal error: numa.h: No such file or directory
| #include <numa.h>
| ^~~~~~~~
| compilation terminated.
Use PACKAGECONFIG to optionally enable the numa option.
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
This upgrades the dpdk package to 18.02 which is
the latest release and drops a patch that has
already been merged upstream in this package.
Signed-off-by: Awais Belal <awais_belal@mentor.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Install and ship test programs into dpdk-test.
Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
We have multiple recipes in different layers providing the same libibverbs
library, so we need to use virtual/libibverbs to avoid potential errors.
Make dpdk depend on virtual/libibverbs instead of hardcoding
dpdk-dev-libibverbs.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
