[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5e15faaac42e7_67ea2afd262665bc44@john-XPS-13-9370.notmuch>
Date: Wed, 08 Jan 2020 07:52:10 -0800
From: John Fastabend <john.fastabend@...il.com>
To: Björn Töpel <bjorn.topel@...il.com>,
John Fastabend <john.fastabend@...il.com>
Cc: Netdev <netdev@...r.kernel.org>,
Alexei Starovoitov <ast@...nel.org>,
Daniel Borkmann <daniel@...earbox.net>,
Björn Töpel <bjorn.topel@...el.com>,
bpf <bpf@...r.kernel.org>, David Miller <davem@...emloft.net>,
Jakub Kicinski <jakub.kicinski@...ronome.com>,
Jesper Dangaard Brouer <hawk@...nel.org>,
"Karlsson, Magnus" <magnus.karlsson@...el.com>,
Jonathan Lemon <jonathan.lemon@...il.com>,
Toke Høiland-Jørgensen <toke@...hat.com>
Subject: Re: [PATCH bpf-next v2 4/8] xsk: make xskmap flush_list common for
all map instances
Björn Töpel wrote:
> On Tue, 7 Jan 2020 at 18:54, John Fastabend <john.fastabend@...il.com> wrote:
> >
> > Björn Töpel wrote:
> > > From: Björn Töpel <bjorn.topel@...el.com>
> > >
> > > The xskmap flush list is used to track entries that need to flushed
> > > from via the xdp_do_flush_map() function. This list used to be
> > > per-map, but there is really no reason for that. Instead make the
> > > flush list global for all xskmaps, which simplifies __xsk_map_flush()
> > > and xsk_map_alloc().
> > >
> > > Acked-by: Toke Høiland-Jørgensen <toke@...hat.com>
> > > Signed-off-by: Björn Töpel <bjorn.topel@...el.com>
> > > ---
> >
> > Just to check. The reason this is OK is because xdp_do_flush_map()
> > is called from NAPI context and is per CPU so the only entries on
> > the list will be from the current cpu napi context?
>
> Correct!
>
> > Even in the case
> > where multiple xskmaps exist we can't have entries from more than
> > a single map on any list at the same time by my reading.
> >
>
> No, there can be entries from different (XSK) maps. Instead of
> focusing on maps to flush, focus on *entries* to flush. At the end of
> the poll function, all entries (regardless of map origin) will be
> flushed. Makes sense?
Ah OK. This would mean that a single program used multiple maps
though correct? Because we can only run a single BPF program per
NAPI context.
What I was after is checking that semantics haven't changed which
I believe is true, just checking.
>
>
> Björn
>
>
> > LGTM,
> > Acked-by: John Fastabend <john.fastabend@...il.com>
Powered by blists - more mailing lists