[dpdk-dev] [PATCH v2 0/4] cryptodev vdev changes for -rc2

Jan Blunck jblunck at infradead.org
Thu Jul 27 21:31:54 CEST 2017


On Wed, Jul 19, 2017 at 9:31 AM, De Lara Guarch, Pablo
<pablo.de.lara.guarch at intel.com> wrote:
>
>
>
> > -----Original Message-----
> > From: Jan Blunck [mailto:jblunck at gmail.com] On Behalf Of Jan Blunck
> > Sent: Wednesday, July 12, 2017 8:59 PM
> > To: dev at dpdk.org
> > Cc: Doherty, Declan <declan.doherty at intel.com>; De Lara Guarch, Pablo
> > <pablo.de.lara.guarch at intel.com>
> > Subject: [PATCH v2 0/4] cryptodev vdev changes for -rc2
> >
> > This series is a preparation to move the vdev bus out of librte_eal. For that
> > the newly added cryptodev vdev functions need to change signature to not
> > require the rte_vdev.h header.
> >
> > Changes since v1:
> > - move params parsing into new header
> > - make rte_cryptodev_vdev_pmd_init() static inline
> >
> > Jan Blunck (4):
> >   cryptodev: remove obsolete include
> >   cryptodev: move initialization
> >   cryptodev: rework PMD init to not require rte_vdev.h
> >   cryptodev: move parameter parsing to its own header
> >
> >  lib/librte_cryptodev/Makefile                    |  1 +
> >  lib/librte_cryptodev/rte_cryptodev.c             |  3 +
> >  lib/librte_cryptodev/rte_cryptodev.h             |  1 -
> >  lib/librte_cryptodev/rte_cryptodev_pmd.c         | 37 +---------
> >  lib/librte_cryptodev/rte_cryptodev_vdev.h        | 71 ++++++++-----------
> >  lib/librte_cryptodev/rte_cryptodev_vdev_params.h | 89
> > ++++++++++++++++++++++++
> >  lib/librte_cryptodev/rte_cryptodev_version.map   |  1 -
> >  7 files changed, 122 insertions(+), 81 deletions(-)  create mode 100644
> > lib/librte_cryptodev/rte_cryptodev_vdev_params.h
> >
> > --
> > 2.13.2
>
> Hi Jan,
>
> Since there are some issues in this patchset and knowing that these changes are required
> for work to be done in 17.11 (vdev moving out of EAL), let's postpone this for the next release,
> as also there is no API breakage in this (correct me if I am wrong).

Unexporting a function does require a soname change. So this is affecting ABI.


More information about the dev mailing list