[dpdk-dev] [PATCH v2 1/1] gpu/cuda: introduce CUDA driver

[Elena Agostini]

> From: Stephen Hemminger <stephen at networkplumber.org>
> Date: Monday, 8 November 2021 at 19:59
> To: Elena Agostini <eagostini at nvidia.com>
> Cc: dev at dpdk.org <dev at dpdk.org>
> Subject: Re: [dpdk-dev] [PATCH v2 1/1] gpu/cuda: introduce CUDA driver>
> On Mon, 8 Nov 2021 18:39:36 +0000
> Elena Agostini <eagostini at nvidia.com> wrote:>
> > > From: Stephen Hemminger <stephen at networkplumber.org>
> > > Date: Monday, 8 November 2021 at 19:35
> > > To: Elena Agostini <eagostini at nvidia.com>
> > > Cc: dev at dpdk.org <dev at dpdk.org>
> > > Subject: Re: [dpdk-dev] [PATCH v2 1/1] gpu/cuda: introduce CUDA driver
> > > External email: Use caution opening links or attachments>
> > >
> > > On Thu, 4 Nov 2021 02:01:28 +0000
> > > <eagostini at nvidia.com> wrote:>
> > > > +/* Single entry of the memory list */
> > > > +struct mem_entry {
> > > > +     CUdeviceptr ptr_d;
> > > > +     void *ptr_h;
> > > > +     size_t size;
> > > > +     struct rte_gpu *dev;
> > > > +     CUcontext ctx;>
> > > Not sure where these types CUdeviceptr and CUcontext are coming
> > > from, but the code looks like Windows style not DPDK or Linux.>
> > > Please don't introduce CamelCase typedef's and never typedefs
> > > for pointers.
> >
> > These are CUDA Driver API specific types, I can’t change them.>
> Could you at least avoid using the pointer typedefs.
> When a pointer is typedef'd it leads to confusion to readers and
> some dumb static analyzers.

There is not pointer typedef here.
The typedef is about uintptr_t