On 10/4/2018 10:42 AM, Igor Russkikh wrote: > Hi Ferruh, > > Thanks again for the review, we'll investigate and work on your comments! > >>> +CONFIG_RTE_LIBRTE_ATLANTIC_PMD=y >> Just to check, default enable in "common_base" means PMD supports all >> architectures (x86_64, i686, arm, ppc) and both Linux and FreeBSD. >> If this is not the case, please disable in common_base and enable in whatever >> supported. > > We actually don't have an ability to test and verify on PPC. > We are pretty confident on x86_64, i686 and arm. > > Thus, will it be ok to add `N` into defconfig_ppc_64-power8-linuxapp-gcc only?
If FreeBSD is also supported yes, above will be OK. > >> Thanks for the update. But a reminder, each patch should be build and >> functional, there are build errors when first patch applied. I think build is >> broken until 7/22. >> >> Can you please re-work the patchset to make each patch compiles? > > Thats strange because we explicitly tested that. > CI also states its OK: > http://patchwork.dpdk.org/project/dpdk/list/?series=1598&submitter=&state=*&q=&archive=&delegate= > > Could it be some custom build environment on your side, or may be you do > apply it > on different base, not master? I double checked, after fist patch applied getting following build error [1], and it seems valid because "atl_eth_dev_ops" not used at all. [1] .../drivers/net/atlantic/atl_ethdev.c:63:33: error: unused variable 'atl_eth_dev_ops' [-Werror,-Wunused-const-variable] static const struct eth_dev_ops atl_eth_dev_ops = { ^
