> -----Original Message-----
> From: Leon Romanovsky <leon@xxxxxxxxxx>
> Sent: Friday, March 30, 2018 9:56 AM
> To: Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx>
> Cc: jgg@xxxxxxxxxxxx; dledford@xxxxxxxxxx; linux-rdma@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH v2 rdma-next 4/5] RDMA/NLDEV: helper functions to
> add provider attributes
>
> On Fri, Mar 30, 2018 at 09:51:09AM -0500, Steve Wise wrote:
> > > On Thu, Mar 29, 2018 at 09:09:54AM -0700, Steve Wise wrote:
> > > > These help rdma providers to fill out the provider entries.
> > > >
> > > > Signed-off-by: Steve Wise <swise@xxxxxxxxxxxxxxxxxxxxx>
> > > > ---
> > > > drivers/infiniband/core/nldev.c | 79
> > > +++++++++++++++++++++++++++++++++++++++++
> > > > include/rdma/restrack.h | 19 ++++++++++
> > > > 2 files changed, 98 insertions(+)
> > > >
> > > > diff --git a/drivers/infiniband/core/nldev.c
> > > b/drivers/infiniband/core/nldev.c
> > > > index 902517f..301646b 100644
> > > > --- a/drivers/infiniband/core/nldev.c
> > > > +++ b/drivers/infiniband/core/nldev.c
> > > > @@ -131,6 +131,85 @@ static int provider_fill_port_info(struct
> sk_buff
> > > *msg,
> > > > device->res.fill_port_info(msg, cb, device, port) :
0;
> > > > }
> > > >
> > > > +static int put_provider_name_print_type(struct sk_buff *msg, const
> char
> > > *name,
> > > > + enum rdma_nldev_print_type
> > > print_type)
> > > > +{
> > > > + if (nla_put_string(msg, RDMA_NLDEV_ATTR_PROVIDER_STRING,
> > > name))
> > > > + return -EMSGSIZE;
> > > > + if (print_type != RDMA_NLDEV_PRINT_TYPE_UNSPEC &&
> > > > + nla_put_u8(msg, RDMA_NLDEV_ATTR_PROVIDER_PRINT_TYPE,
> > > print_type))
> > > > + return -EMSGSIZE;
> > > > +
> > > > + return 0;
> > > > +}
> > > > +
> > > > +int rdma_nl_put_provider_u32(struct sk_buff *msg, const char
> *name,
> > > > + enum rdma_nldev_print_type print_type,
> > > > + u32 value)
> > > > +{
> > > > + if (put_provider_name_print_type(msg, name, print_type))
> > > > + return -EMSGSIZE;
> > > > + if (nla_put_u32(msg, RDMA_NLDEV_ATTR_PROVIDER_U32, value))
> > > > + return -EMSGSIZE;
> > > > +
> > > > + return 0;
> > > > +}
> > > > +EXPORT_SYMBOL(rdma_nl_put_provider_u32);
> > >
> > > Let's do two functions
> > > 1. rdma_nl_put_provider_u32(struct sk_buff *msg, const char
> *name,u32
> > > value)
> > > 2. rdma_nl_put_provider_u32_hex(struct sk_buff *msg, const char
> *name,
> > > u32 value)
> > >
> > > And save from drivers the need to specify type (and _UNSPEC)
> > >
> > > Please implement only functions which you are using, other functions
> > > will be implemented later.
> > >
> > > Thanks
> >
> > The last patch uses u64 as well. I need u32 and u64.
>
> I meant that if you don't need u32_hex, so don't implement it :)
> >
Let's nail this down so the next respin will stick:
- remove cb from provider fill API functions
- null out function pointers explicitly
- remove BIN print type
- have u32/u64 unspec and hex helpers, and only the ones I need for patch 4
(I'll add more if I need them in a subsequent series).
Is this list complete?
thanks
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
