Re: [PATCH for-next 0/7] Refactor control path of bnxt_re driver

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sat, Jan 25, 2020 at 11:34 PM Jason Gunthorpe <jgg@xxxxxxxxxxxx> wrote:
>
> On Fri, Jan 24, 2020 at 12:52:38AM -0500, Devesh Sharma wrote:
> > This is the first series out of few more forthcoming series to refactor
> > Broadcom's RoCE driver. This series contains patches to refactor control
> > path. Since this is first series, there may be few code section which may
> > look redundant or overkill but those will be taken care in future patche
> > series.
> >
> > These patches apply clean on tip of for-next branch.
> > Each patch in this series is tested against user and kernel functionality.
> >
> > Devesh Sharma (7):
> >   RDMA/bnxt_re: Refactor queue pair creation code
> >   RDMA/bnxt_re: Replace chip context structure with pointer
> >   RDMA/bnxt_re: Refactor hardware queue memory allocation
> >   RDMA/bnxt_re: Refactor net ring allocation function
> >   RDMA/bnxt_re: Refactor command queue management code
> >   RDMA/bnxt_re: Refactor notification queue management code
> >   RDMA/bnxt_re: Refactor doorbell management functions
> >
> >  drivers/infiniband/hw/bnxt_re/bnxt_re.h    |  24 +-
> >  drivers/infiniband/hw/bnxt_re/ib_verbs.c   | 670 +++++++++++++++++++----------
> >  drivers/infiniband/hw/bnxt_re/main.c       | 134 +++---
> >  drivers/infiniband/hw/bnxt_re/qplib_fp.c   | 423 +++++++++---------
> >  drivers/infiniband/hw/bnxt_re/qplib_fp.h   |  94 ++--
> >  drivers/infiniband/hw/bnxt_re/qplib_rcfw.c | 474 ++++++++++++--------
> >  drivers/infiniband/hw/bnxt_re/qplib_rcfw.h |  85 ++--
> >  drivers/infiniband/hw/bnxt_re/qplib_res.c  | 475 ++++++++++++--------
> >  drivers/infiniband/hw/bnxt_re/qplib_res.h  | 145 ++++++-
> >  drivers/infiniband/hw/bnxt_re/qplib_sp.c   |  52 ++-
> >  10 files changed, 1579 insertions(+), 997 deletions(-)
>
> Usually when you 'refactor' something the code gets smaller, not
> larger. What is going on here?
I agree with this fact however this series is adding object attribute
structures at various places and the attribute initialization before
object creation is adding up code lines.
>
> Jason



[Index of Archives]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Photo]     [Yosemite News]     [Yosemite Photos]     [Linux Kernel]     [Linux SCSI]     [XFree86]

  Powered by Linux