Re: [PATCH bpf-next] selftests/bpf: add options and frags to xdp_hw_metadata

[Larysa Zaremba <larysa.zaremba@intel.com>]

On Mon, Oct 09, 2023 at 09:49:54AM -0700, Stanislav Fomichev wrote:
> On 10/09, Larysa Zaremba wrote:
> > This is a follow-up to the commit 9b2b86332a9b ("bpf: Allow to use kfunc
> > XDP hints and frags together").
> > 
> > The are some possible implementations problems that may arise when
> > providing metadata specifically for multi-buffer packets, therefore there
> > must be a possibility to test such option separately.
> > 
> > Add an option to use multi-buffer AF_XDP xdp_hw_metadata and mark used XDP
> > program as capable to use frags.
> > 
> > As for now, xdp_hw_metadata accepts no options, so add simple option
> > parsing logic and a help message.
> > 
> > For quick reference, also add an ingress packet generation command to the
> > help message. The command comes from [0].
> > 
> > Example of output for multi-buffer packet:
> > 
> > xsk_ring_cons__peek: 1
> > 0xead018: rx_desc[15]->addr=10000000000f000 addr=f100 comp_addr=f000
> > rx_hash: 0x5789FCBB with RSS type:0x29
> > rx_timestamp:  1696856851535324697 (sec:1696856851.5353)
> > XDP RX-time:   1696856843158256391 (sec:1696856843.1583)
> > 	delta sec:-8.3771 (-8377068.306 usec)
> > AF_XDP time:   1696856843158413078 (sec:1696856843.1584)
> > 	delta sec:0.0002 (156.687 usec)
> > 0xead018: complete idx=23 addr=f000
> > xsk_ring_cons__peek: 1
> > 0xead018: rx_desc[16]->addr=100000000008000 addr=8100 comp_addr=8000
> > 0xead018: complete idx=24 addr=8000
> > xsk_ring_cons__peek: 1
> > 0xead018: rx_desc[17]->addr=100000000009000 addr=9100 comp_addr=9000 EoP
> > 0xead018: complete idx=25 addr=9000
> > 
> > Metadata is printed for the first packet only.
> > 
> > [0] https://lore.kernel.org/all/20230119221536.3349901-18-sdf@google.com/
> > 
> > Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com>
> > ---
> >  .../selftests/bpf/progs/xdp_hw_metadata.c     |  2 +-
> >  tools/testing/selftests/bpf/xdp_hw_metadata.c | 92 ++++++++++++++++---
> >  2 files changed, 79 insertions(+), 15 deletions(-)
> > 
> > diff --git a/tools/testing/selftests/bpf/progs/xdp_hw_metadata.c b/tools/testing/selftests/bpf/progs/xdp_hw_metadata.c
> > index 63d7de6c6bbb..8767d919c881 100644
> > --- a/tools/testing/selftests/bpf/progs/xdp_hw_metadata.c
> > +++ b/tools/testing/selftests/bpf/progs/xdp_hw_metadata.c
> > @@ -21,7 +21,7 @@ extern int bpf_xdp_metadata_rx_timestamp(const struct xdp_md *ctx,
> >  extern int bpf_xdp_metadata_rx_hash(const struct xdp_md *ctx, __u32 *hash,
> >  				    enum xdp_rss_hash_type *rss_type) __ksym;
> >  
> > -SEC("xdp")
> > +SEC("xdp.frags")
> >  int rx(struct xdp_md *ctx)
> >  {
> >  	void *data, *data_meta, *data_end;
> > diff --git a/tools/testing/selftests/bpf/xdp_hw_metadata.c b/tools/testing/selftests/bpf/xdp_hw_metadata.c
> > index 17c980138796..25225720346b 100644
> > --- a/tools/testing/selftests/bpf/xdp_hw_metadata.c
> > +++ b/tools/testing/selftests/bpf/xdp_hw_metadata.c
> > @@ -26,6 +26,7 @@
> >  #include <linux/sockios.h>
> >  #include <sys/mman.h>
> >  #include <net/if.h>
> > +#include <ctype.h>
> >  #include <poll.h>
> >  #include <time.h>
> >  
> > @@ -49,19 +50,29 @@ struct xsk {
> >  struct xdp_hw_metadata *bpf_obj;
> >  struct xsk *rx_xsk;
> >  const char *ifname;
> > +bool use_frags;
> >  int ifindex;
> >  int rxq;
> >  
> >  void test__fail(void) { /* for network_helpers.c */ }
> >  
> > -static int open_xsk(int ifindex, struct xsk *xsk, __u32 queue_id)
> > +static struct xsk_socket_config gen_socket_config(void)
> >  {
> > -	int mmap_flags = MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE;
> > -	const struct xsk_socket_config socket_config = {
> > +	struct xsk_socket_config socket_config = {
> >  		.rx_size = XSK_RING_PROD__DEFAULT_NUM_DESCS,
> >  		.tx_size = XSK_RING_PROD__DEFAULT_NUM_DESCS,
> >  		.bind_flags = XDP_COPY,
> >  	};
> > +
> > +	if (use_frags)
> > +		socket_config.bind_flags |= XDP_USE_SG;
> > +	return socket_config;
> > +}
> 
> nit: why not drop const from socket_config and add this 'if (use_frags)'
> directly to open_xsk? Not sure separate function really buys us anything?
>

Considering there will also be ZC/copy option, I thought it would be good to 
separate socket config creation. After giving this a sencond thought though, 
for now options would control bind_flags only. What do you this about removing 
gen_socket_config(), but introducing get_bind_flags()?
 
> > +static int open_xsk(int ifindex, struct xsk *xsk, __u32 queue_id)
> > +{
> > +	int mmap_flags = MAP_PRIVATE | MAP_ANONYMOUS | MAP_NORESERVE;
> > +	struct xsk_socket_config socket_config = gen_socket_config();
> >  	const struct xsk_umem_config umem_config = {
> >  		.fill_size = XSK_RING_PROD__DEFAULT_NUM_DESCS,
> >  		.comp_size = XSK_RING_CONS__DEFAULT_NUM_DESCS,
> > @@ -263,11 +274,14 @@ static int verify_metadata(struct xsk *rx_xsk, int rxq, int server_fd, clockid_t
> >  			verify_skb_metadata(server_fd);
> >  
> >  		for (i = 0; i < rxq; i++) {
> > +			bool first_seg = true;
> > +			bool is_eop = true;
> > +
> >  			if (fds[i].revents == 0)
> >  				continue;
> >  
> >  			struct xsk *xsk = &rx_xsk[i];
> > -
> > +peek:
> >  			ret = xsk_ring_cons__peek(&xsk->rx, 1, &idx);
> >  			printf("xsk_ring_cons__peek: %d\n", ret);
> >  			if (ret != 1)
> > @@ -276,12 +290,19 @@ static int verify_metadata(struct xsk *rx_xsk, int rxq, int server_fd, clockid_t
> >  			rx_desc = xsk_ring_cons__rx_desc(&xsk->rx, idx);
> >  			comp_addr = xsk_umem__extract_addr(rx_desc->addr);
> >  			addr = xsk_umem__add_offset_to_addr(rx_desc->addr);
> > -			printf("%p: rx_desc[%u]->addr=%llx addr=%llx comp_addr=%llx\n",
> > -			       xsk, idx, rx_desc->addr, addr, comp_addr);
> > -			verify_xdp_metadata(xsk_umem__get_data(xsk->umem_area, addr),
> > -					    clock_id);
> > +			is_eop = !(rx_desc->options & XDP_PKT_CONTD);
> > +			printf("%p: rx_desc[%u]->addr=%llx addr=%llx comp_addr=%llx%s\n",
> > +			       xsk, idx, rx_desc->addr, addr, comp_addr, is_eop ? " EoP" : "");
> > +			if (first_seg) {
> > +				verify_xdp_metadata(xsk_umem__get_data(xsk->umem_area, addr),
> > +						    clock_id);
> > +				first_seg = false;
> > +			}
> > +
> >  			xsk_ring_cons__release(&xsk->rx, 1);
> >  			refill_rx(xsk, comp_addr);
> > +			if (!is_eop)
> > +				goto peek;
> >  		}
> >  	}
> >  
> > @@ -404,6 +425,54 @@ static void timestamping_enable(int fd, int val)
> >  		error(1, errno, "setsockopt(SO_TIMESTAMPING)");
> >  }
> >  
> > +static void print_usage(void)
> > +{
> > +	const char *usage =
> > +		"  Usage: xdp_hw_metadata [OPTIONS] [IFNAME]\n"
> > +		"  Options:\n"
> > +		"  -m            Enable multi-buffer XDP for larger MTU\n"
> > +		"  -h            Display this help and exit\n\n"
> > +		"  Generate test packets on the other machine with:\n"
> > +		"    echo -n xdp | nc -u -q1 <dst_ip> 9091\n";
> 
> nit: any reason we have two spaces in the help description? I don't
> think it's a standard practice, so maybe drop them?

I have just copied usage from xskxceiver. As I see, this is not a standard 
practice indeed, will fix.