Re: [PATCHv7 bpf-next 1/4] bpf: run devmap xdp_prog on flush instead of bulk enqueue

[Martin KaFai Lau <kafai@fb.com>]

On Fri, Apr 16, 2021 at 12:03:41PM +0200, Toke Høiland-Jørgensen wrote:
> Martin KaFai Lau <kafai@fb.com> writes:
> 
> > On Thu, Apr 15, 2021 at 10:29:40PM +0200, Toke Høiland-Jørgensen wrote:
> >> Jesper Dangaard Brouer <brouer@redhat.com> writes:
> >> 
> >> > On Thu, 15 Apr 2021 10:35:51 -0700
> >> > Martin KaFai Lau <kafai@fb.com> wrote:
> >> >
> >> >> On Thu, Apr 15, 2021 at 11:22:19AM +0200, Toke Høiland-Jørgensen wrote:
> >> >> > Hangbin Liu <liuhangbin@gmail.com> writes:
> >> >> >   
> >> >> > > On Wed, Apr 14, 2021 at 05:17:11PM -0700, Martin KaFai Lau wrote:  
> >> >> > >> >  static void bq_xmit_all(struct xdp_dev_bulk_queue *bq, u32 flags)
> >> >> > >> >  {
> >> >> > >> >  	struct net_device *dev = bq->dev;
> >> >> > >> > -	int sent = 0, err = 0;
> >> >> > >> > +	int sent = 0, drops = 0, err = 0;
> >> >> > >> > +	unsigned int cnt = bq->count;
> >> >> > >> > +	int to_send = cnt;
> >> >> > >> >  	int i;
> >> >> > >> >  
> >> >> > >> > -	if (unlikely(!bq->count))
> >> >> > >> > +	if (unlikely(!cnt))
> >> >> > >> >  		return;
> >> >> > >> >  
> >> >> > >> > -	for (i = 0; i < bq->count; i++) {
> >> >> > >> > +	for (i = 0; i < cnt; i++) {
> >> >> > >> >  		struct xdp_frame *xdpf = bq->q[i];
> >> >> > >> >  
> >> >> > >> >  		prefetch(xdpf);
> >> >> > >> >  	}
> >> >> > >> >  
> >> >> > >> > -	sent = dev->netdev_ops->ndo_xdp_xmit(dev, bq->count, bq->q, flags);
> >> >> > >> > +	if (bq->xdp_prog) {  
> >> >> > >> bq->xdp_prog is used here
> >> >> > >>   
> >> >> > >> > +		to_send = dev_map_bpf_prog_run(bq->xdp_prog, bq->q, cnt, dev);
> >> >> > >> > +		if (!to_send)
> >> >> > >> > +			goto out;
> >> >> > >> > +
> >> >> > >> > +		drops = cnt - to_send;
> >> >> > >> > +	}
> >> >> > >> > +  
> >> >> > >> 
> >> >> > >> [ ... ]
> >> >> > >>   
> >> >> > >> >  static void bq_enqueue(struct net_device *dev, struct xdp_frame *xdpf,
> >> >> > >> > -		       struct net_device *dev_rx)
> >> >> > >> > +		       struct net_device *dev_rx, struct bpf_prog *xdp_prog)
> >> >> > >> >  {
> >> >> > >> >  	struct list_head *flush_list = this_cpu_ptr(&dev_flush_list);
> >> >> > >> >  	struct xdp_dev_bulk_queue *bq = this_cpu_ptr(dev->xdp_bulkq);
> >> >> > >> > @@ -412,18 +466,22 @@ static void bq_enqueue(struct net_device *dev, struct xdp_frame *xdpf,
> >> >> > >> >  	/* Ingress dev_rx will be the same for all xdp_frame's in
> >> >> > >> >  	 * bulk_queue, because bq stored per-CPU and must be flushed
> >> >> > >> >  	 * from net_device drivers NAPI func end.
> >> >> > >> > +	 *
> >> >> > >> > +	 * Do the same with xdp_prog and flush_list since these fields
> >> >> > >> > +	 * are only ever modified together.
> >> >> > >> >  	 */
> >> >> > >> > -	if (!bq->dev_rx)
> >> >> > >> > +	if (!bq->dev_rx) {
> >> >> > >> >  		bq->dev_rx = dev_rx;
> >> >> > >> > +		bq->xdp_prog = xdp_prog;  
> >> >> > >> bp->xdp_prog is assigned here and could be used later in bq_xmit_all().
> >> >> > >> How is bq->xdp_prog protected? Are they all under one rcu_read_lock()?
> >> >> > >> It is not very obvious after taking a quick look at xdp_do_flush[_map].
> >> >> > >> 
> >> >> > >> e.g. what if the devmap elem gets deleted.  
> >> >> > >
> >> >> > > Jesper knows better than me. From my veiw, based on the description of
> >> >> > > __dev_flush():
> >> >> > >
> >> >> > > On devmap tear down we ensure the flush list is empty before completing to
> >> >> > > ensure all flush operations have completed. When drivers update the bpf
> >> >> > > program they may need to ensure any flush ops are also complete.  
> >> >>
> >> >> AFAICT, the bq->xdp_prog is not from the dev. It is from a devmap's elem.
> >> >> 
> >> >> > 
> >> >> > Yeah, drivers call xdp_do_flush() before exiting their NAPI poll loop,
> >> >> > which also runs under one big rcu_read_lock(). So the storage in the
> >> >> > bulk queue is quite temporary, it's just used for bulking to increase
> >> >> > performance :)  
> >> >>
> >> >> I am missing the one big rcu_read_lock() part.  For example, in i40e_txrx.c,
> >> >> i40e_run_xdp() has its own rcu_read_lock/unlock().  dst->xdp_prog used to run
> >> >> in i40e_run_xdp() and it is fine.
> >> >> 
> >> >> In this patch, dst->xdp_prog is run outside of i40e_run_xdp() where the
> >> >> rcu_read_unlock() has already done.  It is now run in xdp_do_flush_map().
> >> >> or I missed the big rcu_read_lock() in i40e_napi_poll()?
> >> >>
> >> >> I do see the big rcu_read_lock() in mlx5e_napi_poll().
> >> >
> >> > I believed/assumed xdp_do_flush_map() was already protected under an
> >> > rcu_read_lock.  As the devmap and cpumap, which get called via
> >> > __dev_flush() and __cpu_map_flush(), have multiple RCU objects that we
> >> > are operating on.
> > What other rcu objects it is using during flush?
> 
> The bq_enqueue() function in cpumap.c puts the 'bq' pointer onto the
> flush_list, and 'bq' lives inside struct bpf_cpu_map_entry, so that's a
> reference to the map entry as well.
> 
> The devmap function used to work the same way, until we changed it in
> 75ccae62cb8d ("xdp: Move devmap bulk queue into struct net_device").
Got it. Thanks for the explanation in bq_enqueue() in cpumap.c.
I was under the impression that xdp_do_flush_map() should not
use any rcu object now since I don't see rcu_read_lock() there
and I use it as a hint in code reading.

> >> > Perhaps it is a bug in i40e?
> > A quick look into ixgbe falls into the same bucket.
> > didn't look at other drivers though.
> >
> >> >
> >> > We are running in softirq in NAPI context, when xdp_do_flush_map() is
> >> > call, which I think means that this CPU will not go-through a RCU grace
> >> > period before we exit softirq, so in-practice it should be safe.
> >> 
> >> Yup, this seems to be correct: rcu_softirq_qs() is only called between
> >> full invocations of the softirq handler, which for networking is
> >> net_rx_action(), and so translates into full NAPI poll cycles.
> >
> > I don't know enough to comment on the rcu/softirq part, may be someone
> > can chime in.  There is also a recent napi_threaded_poll().
> >
> > If it is the case, then some of the existing rcu_read_lock() is unnecessary?
> > At least, it sounds incorrect to only make an exception here while keeping
> > other rcu_read_lock() as-is.
> 
> I'd tend to agree that the correct thing to do is to fix any affected
> drivers so there's a wide rcu_read_lock() around the full xdp+flush. If
> nothing else, this serves as an annotation for the expected lifetime of
> the objects involved.
> 
> However, given that this is not a new issue, I don't think it should be
> holding up this patch series... We can start a new conversation on what
> the right way to fix this is - and maybe bring in Paul for advice on the
> RCU side? WDYT?
Yeah...it falls into the same issue as the current bq_enqueue() in cpumap.c.
I am fine to put them together into the solve later bucket.  I will delegate
this decision to the maintainers.

I would wait a bit on Paul's reply though.

Also, patch 2 does not necessary depend on patch 1?  Another option is to post
patch 1 separately later as an optimization when the rcu discussion concluded.