On Wed, Oct 16, 2024 at 5:04 PM Stephen Hemminger <step...@networkplumber.org> wrote: > > On Wed, 16 Oct 2024 15:54:11 +0200 > Mattias Rönnblom <mattias.ronnb...@ericsson.com> wrote: > > > Extend checkpatches.sh to detect the use of __builtin_(ffs|ffsll). > > > > These intrinsics are not available in MSVC, and there are perfectly > > serviceable alternatives in <rte_bitops.h>. > > > > Signed-off-by: Mattias Rönnblom <mattias.ronnb...@ericsson.com> > > Suggested-by: David Marchand <david.march...@redhat.com> > > Shouldn't this apply to all _builtin_ functions. > > There are a lot of drivers still doing this.
- The updated check here is about use of __builtin_XXX instead of DPDK bit count ops. So when you refer to all builtin functions, there may be other builtins related to bit count we are missing, but at least the current patch is better than before. - On the larger topic of refusing *any* __builtin_, some use of them are "legit" under compiler checks. So the check would have to skip headers where wrappers are provided to the rest of DPDK. This seems possible, but a larger and probably more tricky change for now. Are you ok if we go with this patch as is, in a first step? -- David Marchand
