On Tue, Nov 03, 2020 at 10:45:37AM +0100, David Marchand wrote: > On Tue, Nov 3, 2020 at 10:19 AM David Marchand > <david.march...@redhat.com> wrote: > > > > On Tue, Nov 3, 2020 at 10:06 AM Thomas Monjalon <tho...@monjalon.net> wrote: > > > > > > 03/11/2020 09:19, David Marchand: > > > > On Tue, Nov 3, 2020 at 1:30 AM Thomas Monjalon <tho...@monjalon.net> > > > > wrote: > > > > > > > -name = 'octeontx2_regex' > > > > > > > > > > But it is not the same? > > > > > > > > > > The name will default to "octeontx2", which is fine. > > > > > But the fmt_name should not take this default. > > > > > I believe fmt_name should be "octeontx2_regex" as I did in my patch. > > > > > > > > fmt_name is only for maintaining config compat. > > > > This driver is new to 20.11. > > > > We can drop fmt_name too. > > > > > > If we don't set fmt_name, it defaults to "name", "octeontx2" here. > > > What is the consequence in compat definitions? > > > > Ok, got it, the problem is when we disable the net/octeontx2 driver. > Your patch correctly sets a RTE_LIBRTE_OCTEONTX2_REGEX_PMD compat > option that is unused but that does not overwrite the > RTE_LIBRTE_OCTEONTX2_PMD compat option (which indicates the presence > of the net/octeontx2 driver). > Yes, I forgot about compatibility macro settings. Given the context of this patchset, I was instead just looking at the library filesnames.
I'll drop this patch from v2, and you can take Thomas' instead. /Bruce
