[dpdk-stable] [PATCH 19.11 2/2] net/netvsc: control use of external mbuf on Rx

[Long Li]

> Subject: Re: [dpdk-stable] [PATCH 19.11 2/2] net/netvsc: control use of
> external mbuf on Rx
> 
> On Tue, 15 Dec 2020 13:48:57 +0000
> Luca Boccassi <bluca at debian.org> wrote:
> 
> > On Thu, 2020-12-10 at 19:44 +0000, Long Li wrote:
> > > > Subject: Re: [dpdk-stable] [PATCH 19.11 2/2] net/netvsc: control
> > > > use of external mbuf on Rx
> > > >
> > > > On Wed, 2020-12-09 at 16:27 -0800, Long Li wrote:
> > > > > From: Long Li <longli at microsoft.com>
> > > > >
> > > > > [ upstream commit 096b31fc0d8c989cc455c35f4d1def24a4ed6dee ]
> > > > >
> > > > > When receiving packets, netvsp puts data in a buffer mapped through
> UIO.
> > > > > Depending on packet size, netvsc may attach the buffer as an
> > > > > external mbuf. This is not a problem if this mbuf is consumed in
> > > > > the application, and the application can correctly read data out
> > > > > of an external
> > > > mbuf.
> > > > > However, there are two problems with data in an external mbuf.
> > > > > 1. Due to the limitation of the kernel UIO implementation, physical
> > > > >    address of this external buffer is not exposed to the user-mode. If
> > > > >    this mbuf is passed to another driver, the other driver is unable to
> > > > >    map this buffer to iova.
> > > > > 2. Some DPDK applications are not aware of external mbuf, and may
> bug
> > > > >    when they receive an mbuf with external buffer attached.
> > > > >
> > > > > Introduce a driver parameter "rx_extmbuf_enable" to control if
> > > > > netvsc should use external mbuf for receiving packets. The default
> value is 0.
> > > > > (netvsc doesn't use external mbuf, it always allocates mbuf and
> > > > > copy data to mbuf) A non-zero value tells netvsc to attach
> > > > > external buffers to mbuf on receiving packets, thus avoid copying
> memory.
> > > > >
> > > > > Signed-off-by: Long Li <longli at microsoft.com>
> > > > > ---
> > > > >  doc/guides/nics/netvsc.rst     |  8 ++++++++
> > > > >  drivers/net/netvsc/hn_ethdev.c | 10 +++++++++-
> > > > >  drivers/net/netvsc/hn_rxtx.c   |  2 +-
> > > > >  drivers/net/netvsc/hn_var.h    |  3 +++
> > > > >  4 files changed, 21 insertions(+), 2 deletions(-)
> > > >
> > > > Correct me if I'm wrong, but these 2 patches look a bit more like
> > > > new features than bug fixes? It's new options for the PMD, right?
> > > >
> > > > In general, we do not take new features in LTS releases. Stable
> > > > means stable
> > > > - we make very few exceptions.
> > > >
> > > > Is the PMD broken/unusable without these options?
> > >
> > > This patch changes the default behavior of netvsc PMD to not use
> external mbufs on receiving data.
> > >
> > > Using external mbufs has shown problems in many applications. Changing
> the default behavior will fix those.
> >
> > Sorry but I don't think adding new options is appropriate for an LTS
> > release.
> > The expectations is that everything works as-is without any need to
> > even recompile, let alone changing options.

This patch effectively disables external mbuf without any changes to application.

> >
> 
> Can we just turn external mbuf usage in this driver off for the stable version
> instead?

That will require a separate patch not in upstream. If that's acceptable I can submit another patch. But this will create some confusion to someone doing bisecting in 19.11.