Re: [dpdk-dev] [PATCH v4 4/6] net/ark: cleanup ark dynamic extension interface

[Ed Czeck <ed.czeck@atomicrules.com>]

On Tue, Mar 9, 2021 at 12:50 PM Ferruh Yigit <ferruh.yigit@intel.com> wrote:
>
> On 3/9/2021 4:08 PM, Ed Czeck wrote:
> > - Rename extension functions with rte_pmd_ark prefix
> > - Move extension prototype to rte_pmd_ark.h
> > - Update local function documentation
> >
> > Signed-off-by: Ed Czeck <ed.czeck@atomicrules.com>
> > ---
> > v3:
> > - split function rename from previous commit
> > v4:
> > - reorder patches renaming before adding
>
> <...>
>
> > diff --git a/drivers/net/ark/rte_pmd_ark.h b/drivers/net/ark/rte_pmd_ark.h
> > index 6f26d66b1..0f24a347d 100644
> > --- a/drivers/net/ark/rte_pmd_ark.h
> > +++ b/drivers/net/ark/rte_pmd_ark.h
> > @@ -1,18 +1,24 @@
> >   /* SPDX-License-Identifier: BSD-3-Clause
> > - * Copyright (c) 2020 Atomic Rules LLC
> > + * Copyright (c) 2020-2021 Atomic Rules LLC
> >    */
> >
> >   #ifndef RTE_PMD_ARK_H
> >   #define RTE_PMD_ARK_H
> >
> > +#include <rte_mbuf.h>
> > +#include <rte_mbuf_dyn.h>
> > +
> > +#include <stdint.h>
> > +struct rte_eth_dev;
> > +struct rte_mbuf;
> > +struct rte_ether_addr;
> > +struct rte_eth_stats;
> > +
>
> Is the declaring structures instead of including header preferred intentionally?
>
> I guess both works, and it may not differ for the PMD at all, but for the
> extension developer that is including this header, it may need to include
> required headers again, this header including all required headers and be a self
> contained may help extension code, up to you.

My experience/style has been to use a declaration rather than a full include to
avoid include bloat and speed up compiles.  If the extension developer needs
the declaration in the header, they can include it as needed.

>
> Plus just a detail but I guess mbuf struct is redundant for this patch, it can
> be added in next patch where mbufs header removed and hooks added.

Agreed.  Would you like another patch?

Thanks
Ed.