Re: [PATCH for-next v3 4/6] RDMA-rxe: Isolate mr code from atomic_write_reply()

[Jason Gunthorpe <jgg@nvidia.com>]

On Tue, Jan 17, 2023 at 09:36:02AM +0800, Zhu Yanjun wrote:
> On Sat, Jan 14, 2023 at 7:28 AM Bob Pearson <rpearsonhpe@gmail.com> wrote:
> >
> > Isolate mr specific code from atomic_write_reply() in rxe_resp.c into
> > a subroutine rxe_mr_do_atomic_write() in rxe_mr.c.
> > Check length for atomic write operation.
> > Make iova_to_vaddr() static.
> >
> > Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
> > ---
> > v3:
> >   Fixed bug reported by kernel test robot. Ifdef'ed out atomic 8 byte
> >   write if CONFIG_64BIT is not defined as orignally intended by the
> >   developers of the atomic write implementation.
> > link: https://lore.kernel.org/linux-rdma/202301131143.CmoyVcul-lkp@intel.com/
> >
> >  drivers/infiniband/sw/rxe/rxe_loc.h  |  1 +
> >  drivers/infiniband/sw/rxe/rxe_mr.c   | 50 ++++++++++++++++++++++++
> >  drivers/infiniband/sw/rxe/rxe_resp.c | 58 +++++++++++-----------------
> >  3 files changed, 73 insertions(+), 36 deletions(-)
> >
> > diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h
> > index bcb1bbcf50df..fd70c71a9e4e 100644
> > --- a/drivers/infiniband/sw/rxe/rxe_loc.h
> > +++ b/drivers/infiniband/sw/rxe/rxe_loc.h
> > @@ -74,6 +74,7 @@ int rxe_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg,
> >  void *iova_to_vaddr(struct rxe_mr *mr, u64 iova, int length);
> >  int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
> >                         u64 compare, u64 swap_add, u64 *orig_val);
> > +int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, void *addr);
> >  struct rxe_mr *lookup_mr(struct rxe_pd *pd, int access, u32 key,
> >                          enum rxe_mr_lookup_type type);
> >  int mr_check_range(struct rxe_mr *mr, u64 iova, size_t length);
> > diff --git a/drivers/infiniband/sw/rxe/rxe_mr.c b/drivers/infiniband/sw/rxe/rxe_mr.c
> > index 791731be6067..1e74f5e8e10b 100644
> > --- a/drivers/infiniband/sw/rxe/rxe_mr.c
> > +++ b/drivers/infiniband/sw/rxe/rxe_mr.c
> > @@ -568,6 +568,56 @@ int rxe_mr_do_atomic_op(struct rxe_mr *mr, u64 iova, int opcode,
> >         return 0;
> >  }
> >
> > +/**
> > + * rxe_mr_do_atomic_write() - write 64bit value to iova from addr
> > + * @mr: memory region
> > + * @iova: iova in mr
> > + * @addr: source of data to write
> > + *
> > + * Returns:
> > + *      0 on success
> > + *     -1 for misaligned address
> > + *     -2 for access errors
> > + *     -3 for cpu without native 64 bit support
> > + */
> > +int rxe_mr_do_atomic_write(struct rxe_mr *mr, u64 iova, void *addr)
> > +{
> > +#if defined CONFIG_64BIT
> 
> IS_ENABLED is better?

is_enabled won't work here because the code doesn't compile.

Jason