On 9/11/2017 1:44 PM, Ferruh Yigit wrote: > On 8/23/2017 4:44 PM, Stephen Hemminger wrote: >> There are many places that cast away the return value from >> rte_memcpy(). This was a common practice in the Unix BSD days because >> lint would complain about unused return values, but it is not done >> in modern Linux style because it clutters the code. >> >> Maybe it was done for some Coverity warnings. If so, that issue >> should be addressed with Coverity overrides not more code clutter. >> >> Stephen Hemminger (6): >> testpmd: remove unnecessary void casts >> bnx2x: remove unnecssary void cast of rte_memcpy >> sfc: remove unnecessary cast of rte_memcpy >> e1000: remove unnecessary cast of rte_memcpy >> i40e: remove unnecessary cast of rte_memcpy >> ixgbe: remove unnecessary cast of rte_memcpy > > My concern is if coverity complains about return value not checked and > someone starts fixing them adding void casts back. > > So I will evaluate the patchset for next-net, so that I can run coverity > on them before merged into main tree, and see the result.
Coverity result on latest next-net (including this patchset) https://scan.coverity.com/projects/dpdk-next-net rte_memcpy() return value not checked is not seen as an issue, so this set is OK.
