bpf.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
To: bpf@vger.kernel.org, netdev@vger.kernel.org,
	daniel@iogearbox.net, ast@kernel.org, andrii@kernel.org
Cc: bjorn.topel@intel.com, magnus.karlsson@intel.com,
	ciara.loftus@intel.com, john.fastabend@gmail.com,
	toke@redhat.com,
	Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Subject: [PATCH v5 bpf-next 06/17] libbpf: xsk: use bpf_link
Date: Tue, 30 Mar 2021 00:43:05 +0200	[thread overview]
Message-ID: <20210329224316.17793-7-maciej.fijalkowski@intel.com> (raw)
In-Reply-To: <20210329224316.17793-1-maciej.fijalkowski@intel.com>

Currently, if there are multiple xdpsock instances running on a single
interface and in case one of the instances is terminated, the rest of
them are left in an inoperable state due to the fact of unloaded XDP
prog from interface.

Consider the scenario below:

// load xdp prog and xskmap and add entry to xskmap at idx 10
$ sudo ./xdpsock -i ens801f0 -t -q 10

// add entry to xskmap at idx 11
$ sudo ./xdpsock -i ens801f0 -t -q 11

terminate one of the processes and another one is unable to work due to
the fact that the XDP prog was unloaded from interface.

To address that, step away from setting bpf prog in favour of bpf_link.
This means that refcounting of BPF resources will be done automatically
by bpf_link itself.

Provide backward compatibility by checking if underlying system is
bpf_link capable. Do this by looking up/creating bpf_link on loopback
device. If it failed in any way, stick with netlink-based XDP prog.
therwise, use bpf_link-based logic.

When setting up BPF resources during xsk socket creation, check whether
bpf_link for a given ifindex already exists via set of calls to
bpf_link_get_next_id -> bpf_link_get_fd_by_id -> bpf_obj_get_info_by_fd
and comparing the ifindexes from bpf_link and xsk socket.

For case where resources exist but they are not AF_XDP related, bail out
and ask user to remove existing prog and then retry.

Lastly, do a bit of refactoring within __xsk_setup_xdp_prog and pull out
existing code branches based on prog_id value onto separate functions
that are responsible for resource initialization if prog_id was 0 and
for lookup existing resources for non-zero prog_id as that implies that
XDP program is present on the underlying net device. This in turn makes
it easier to follow, especially the teardown part of both branches.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
---
 tools/lib/bpf/xsk.c | 258 ++++++++++++++++++++++++++++++++++++--------
 1 file changed, 213 insertions(+), 45 deletions(-)

diff --git a/tools/lib/bpf/xsk.c b/tools/lib/bpf/xsk.c
index 526fc35c0b23..95da0e19f4a5 100644
--- a/tools/lib/bpf/xsk.c
+++ b/tools/lib/bpf/xsk.c
@@ -28,6 +28,7 @@
 #include <sys/mman.h>
 #include <sys/socket.h>
 #include <sys/types.h>
+#include <linux/if_link.h>
 
 #include "bpf.h"
 #include "libbpf.h"
@@ -70,8 +71,10 @@ struct xsk_ctx {
 	int ifindex;
 	struct list_head list;
 	int prog_fd;
+	int link_fd;
 	int xsks_map_fd;
 	char ifname[IFNAMSIZ];
+	bool has_bpf_link;
 };
 
 struct xsk_socket {
@@ -409,7 +412,7 @@ static int xsk_load_xdp_prog(struct xsk_socket *xsk)
 	static const int log_buf_size = 16 * 1024;
 	struct xsk_ctx *ctx = xsk->ctx;
 	char log_buf[log_buf_size];
-	int err, prog_fd;
+	int prog_fd;
 
 	/* This is the fallback C-program:
 	 * SEC("xdp_sock") int xdp_sock_prog(struct xdp_md *ctx)
@@ -499,14 +502,41 @@ static int xsk_load_xdp_prog(struct xsk_socket *xsk)
 		return prog_fd;
 	}
 
-	err = bpf_set_link_xdp_fd(xsk->ctx->ifindex, prog_fd,
-				  xsk->config.xdp_flags);
+	ctx->prog_fd = prog_fd;
+	return 0;
+}
+
+static int xsk_create_bpf_link(struct xsk_socket *xsk)
+{
+	DECLARE_LIBBPF_OPTS(bpf_link_create_opts, opts);
+	struct xsk_ctx *ctx = xsk->ctx;
+	__u32 prog_id = 0;
+	int link_fd;
+	int err;
+
+	err = bpf_get_link_xdp_id(ctx->ifindex, &prog_id, xsk->config.xdp_flags);
 	if (err) {
-		close(prog_fd);
+		pr_warn("getting XDP prog id failed\n");
 		return err;
 	}
 
-	ctx->prog_fd = prog_fd;
+	/* if there's a netlink-based XDP prog loaded on interface, bail out
+	 * and ask user to do the removal by himself
+	 */
+	if (prog_id) {
+		pr_warn("Netlink-based XDP prog detected, please unload it in order to launch AF_XDP prog\n");
+		return -EINVAL;
+	}
+
+	opts.flags = xsk->config.xdp_flags & ~(XDP_FLAGS_UPDATE_IF_NOEXIST | XDP_FLAGS_REPLACE);
+
+	link_fd = bpf_link_create(ctx->prog_fd, ctx->ifindex, BPF_XDP, &opts);
+	if (link_fd < 0) {
+		pr_warn("bpf_link_create failed: %s\n", strerror(errno));
+		return link_fd;
+	}
+
+	ctx->link_fd = link_fd;
 	return 0;
 }
 
@@ -625,7 +655,6 @@ static int xsk_lookup_bpf_maps(struct xsk_socket *xsk)
 		close(fd);
 	}
 
-	err = 0;
 	if (ctx->xsks_map_fd == -1)
 		err = -ENOENT;
 
@@ -642,6 +671,98 @@ static int xsk_set_bpf_maps(struct xsk_socket *xsk)
 				   &xsk->fd, 0);
 }
 
+static int xsk_link_lookup(int ifindex, __u32 *prog_id, int *link_fd)
+{
+	struct bpf_link_info link_info;
+	__u32 link_len;
+	__u32 id = 0;
+	int err;
+	int fd;
+
+	while (true) {
+		err = bpf_link_get_next_id(id, &id);
+		if (err) {
+			if (errno == ENOENT) {
+				err = 0;
+				break;
+			}
+			pr_warn("can't get next link: %s\n", strerror(errno));
+			break;
+		}
+
+		fd = bpf_link_get_fd_by_id(id);
+		if (fd < 0) {
+			if (errno == ENOENT)
+				continue;
+			pr_warn("can't get link by id (%u): %s\n", id, strerror(errno));
+			err = -errno;
+			break;
+		}
+
+		link_len = sizeof(struct bpf_link_info);
+		memset(&link_info, 0, link_len);
+		err = bpf_obj_get_info_by_fd(fd, &link_info, &link_len);
+		if (err) {
+			pr_warn("can't get link info: %s\n", strerror(errno));
+			close(fd);
+			break;
+		}
+		if (link_info.type == BPF_LINK_TYPE_XDP) {
+			if (link_info.xdp.ifindex == ifindex) {
+				*link_fd = fd;
+				if (prog_id)
+					*prog_id = link_info.prog_id;
+				break;
+			}
+		}
+		close(fd);
+	}
+
+	return err;
+}
+
+static bool xsk_probe_bpf_link(void)
+{
+	DECLARE_LIBBPF_OPTS(bpf_link_create_opts, opts,
+			    .flags = XDP_FLAGS_SKB_MODE);
+	struct bpf_load_program_attr prog_attr;
+	struct bpf_insn insns[2] = {
+		BPF_MOV64_IMM(BPF_REG_0, XDP_PASS),
+		BPF_EXIT_INSN()
+	};
+	int prog_fd, link_fd = -1;
+	int ifindex_lo = 1;
+	bool ret = false;
+	int err;
+
+	err = xsk_link_lookup(ifindex_lo, NULL, &link_fd);
+	if (err)
+		return ret;
+
+	if (link_fd >= 0)
+		return true;
+
+	memset(&prog_attr, 0, sizeof(prog_attr));
+	prog_attr.prog_type = BPF_PROG_TYPE_XDP;
+	prog_attr.insns = insns;
+	prog_attr.insns_cnt = ARRAY_SIZE(insns);
+	prog_attr.license = "GPL";
+
+	prog_fd = bpf_load_program_xattr(&prog_attr, NULL, 0);
+	if (prog_fd < 0)
+		return ret;
+
+	link_fd = bpf_link_create(prog_fd, ifindex_lo, BPF_XDP, &opts);
+	close(prog_fd);
+
+	if (link_fd >= 0) {
+		ret = true;
+		close(link_fd);
+	}
+
+	return ret;
+}
+
 static int xsk_create_xsk_struct(int ifindex, struct xsk_socket *xsk)
 {
 	char ifname[IFNAMSIZ];
@@ -663,64 +784,108 @@ static int xsk_create_xsk_struct(int ifindex, struct xsk_socket *xsk)
 	ctx->ifname[IFNAMSIZ - 1] = 0;
 
 	xsk->ctx = ctx;
+	xsk->ctx->has_bpf_link = xsk_probe_bpf_link();
 
 	return 0;
 }
 
-static int __xsk_setup_xdp_prog(struct xsk_socket *_xdp,
-				int *xsks_map_fd)
+static int xsk_init_xdp_res(struct xsk_socket *xsk,
+			    int *xsks_map_fd)
 {
-	struct xsk_socket *xsk = _xdp;
 	struct xsk_ctx *ctx = xsk->ctx;
-	__u32 prog_id = 0;
 	int err;
 
-	err = bpf_get_link_xdp_id(ctx->ifindex, &prog_id,
-				  xsk->config.xdp_flags);
+	err = xsk_create_bpf_maps(xsk);
 	if (err)
 		return err;
 
-	if (!prog_id) {
-		err = xsk_create_bpf_maps(xsk);
-		if (err)
-			return err;
+	err = xsk_load_xdp_prog(xsk);
+	if (err)
+		goto err_load_xdp_prog;
 
-		err = xsk_load_xdp_prog(xsk);
-		if (err) {
-			goto err_load_xdp_prog;
-		}
-	} else {
-		ctx->prog_fd = bpf_prog_get_fd_by_id(prog_id);
-		if (ctx->prog_fd < 0)
-			return -errno;
-		err = xsk_lookup_bpf_maps(xsk);
-		if (err) {
-			close(ctx->prog_fd);
-			return err;
-		}
-	}
+	if (ctx->has_bpf_link)
+		err = xsk_create_bpf_link(xsk);
+	else
+		err = bpf_set_link_xdp_fd(xsk->ctx->ifindex, ctx->prog_fd,
+					  xsk->config.xdp_flags);
 
-	if (xsk->rx) {
-		err = xsk_set_bpf_maps(xsk);
-		if (err) {
-			if (!prog_id) {
-				goto err_set_bpf_maps;
-			} else {
-				close(ctx->prog_fd);
-				return err;
-			}
-		}
-	}
-	if (xsks_map_fd)
-		*xsks_map_fd = ctx->xsks_map_fd;
+	if (err)
+		goto err_attach_xdp_prog;
 
-	return 0;
+	if (!xsk->rx)
+		return err;
+
+	err = xsk_set_bpf_maps(xsk);
+	if (err)
+		goto err_set_bpf_maps;
+
+	return err;
 
 err_set_bpf_maps:
+	if (ctx->has_bpf_link)
+		close(ctx->link_fd);
+	else
+		bpf_set_link_xdp_fd(ctx->ifindex, -1, 0);
+err_attach_xdp_prog:
 	close(ctx->prog_fd);
-	bpf_set_link_xdp_fd(ctx->ifindex, -1, 0);
 err_load_xdp_prog:
 	xsk_delete_bpf_maps(xsk);
+	return err;
+}
+
+static int xsk_lookup_xdp_res(struct xsk_socket *xsk, int *xsks_map_fd, int prog_id)
+{
+	struct xsk_ctx *ctx = xsk->ctx;
+	int err;
+
+	ctx->prog_fd = bpf_prog_get_fd_by_id(prog_id);
+	if (ctx->prog_fd < 0) {
+		err = -errno;
+		goto err_prog_fd;
+	}
+	err = xsk_lookup_bpf_maps(xsk);
+	if (err)
+		goto err_lookup_maps;
+
+	if (!xsk->rx)
+		return err;
+
+	err = xsk_set_bpf_maps(xsk);
+	if (err)
+		goto err_set_maps;
+
+	return err;
+
+err_set_maps:
+	close(ctx->xsks_map_fd);
+err_lookup_maps:
+	close(ctx->prog_fd);
+err_prog_fd:
+	if (ctx->has_bpf_link)
+		close(ctx->link_fd);
+	return err;
+}
+
+static int __xsk_setup_xdp_prog(struct xsk_socket *_xdp, int *xsks_map_fd)
+{
+	struct xsk_socket *xsk = _xdp;
+	struct xsk_ctx *ctx = xsk->ctx;
+	__u32 prog_id = 0;
+	int err;
+
+	if (ctx->has_bpf_link)
+		err = xsk_link_lookup(ctx->ifindex, &prog_id, &ctx->link_fd);
+	else
+		err = bpf_get_link_xdp_id(ctx->ifindex, &prog_id, xsk->config.xdp_flags);
+
+	if (err)
+		return err;
+
+	err = !prog_id ? xsk_init_xdp_res(xsk, xsks_map_fd) :
+			 xsk_lookup_xdp_res(xsk, xsks_map_fd, prog_id);
+
+	if (!err && xsks_map_fd)
+		*xsks_map_fd = ctx->xsks_map_fd;
 
 	return err;
 }
@@ -898,6 +1063,7 @@ int xsk_socket__create_shared(struct xsk_socket **xsk_ptr,
 		}
 	}
 	xsk->ctx = ctx;
+	xsk->ctx->has_bpf_link = xsk_probe_bpf_link();
 
 	if (rx) {
 		err = setsockopt(xsk->fd, SOL_XDP, XDP_RX_RING,
@@ -1054,6 +1220,8 @@ void xsk_socket__delete(struct xsk_socket *xsk)
 	if (ctx->prog_fd != -1) {
 		xsk_delete_bpf_maps(xsk);
 		close(ctx->prog_fd);
+		if (ctx->has_bpf_link)
+			close(ctx->link_fd);
 	}
 
 	err = xsk_get_mmap_offsets(xsk->fd, &off);
-- 
2.20.1


  parent reply	other threads:[~2021-03-29 22:56 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-29 22:42 [PATCH v5 bpf-next 00/17] AF_XDP selftests improvements & bpf_link Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 01/17] selftests: xsk: don't call worker_pkt_dump() for stats test Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 02/17] selftests: xsk: remove struct ifaceconfigobj Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 03/17] selftests: xsk: remove unused function Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 04/17] selftests: xsk: remove inline keyword from source file Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 05/17] selftests: xsk: simplify frame traversal in dumping thread Maciej Fijalkowski
2021-03-29 22:43 ` Maciej Fijalkowski [this message]
2021-03-31 13:10   ` [PATCH v5 bpf-next 06/17] libbpf: xsk: use bpf_link Toke Høiland-Jørgensen
2021-03-29 22:43 ` [PATCH v5 bpf-next 07/17] samples: bpf: do not unload prog within xdpsock Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 08/17] selftests: xsk: remove thread for netns switch Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 09/17] selftests: xsk: split worker thread Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 10/17] selftests: xsk: remove Tx synchronization resources Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 11/17] selftests: xsk: refactor teardown/bidi test cases and testapp_validate Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 12/17] selftests: xsk: remove sync_mutex_tx and atomic var Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 13/17] veth: implement ethtool's get_channels() callback Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 14/17] selftests: xsk: implement bpf_link test Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 15/17] selftests: xsk: remove thread attribute Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 16/17] selftests: xsk: Remove mutex and condition variable Maciej Fijalkowski
2021-03-29 22:43 ` [PATCH v5 bpf-next 17/17] selftests: xsk: Remove unused defines Maciej Fijalkowski
2021-03-30 16:34 ` [PATCH v5 bpf-next 00/17] AF_XDP selftests improvements & bpf_link Alexei Starovoitov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210329224316.17793-7-maciej.fijalkowski@intel.com \
    --to=maciej.fijalkowski@intel.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bjorn.topel@intel.com \
    --cc=bpf@vger.kernel.org \
    --cc=ciara.loftus@intel.com \
    --cc=daniel@iogearbox.net \
    --cc=john.fastabend@gmail.com \
    --cc=magnus.karlsson@intel.com \
    --cc=netdev@vger.kernel.org \
    --cc=toke@redhat.com \
    --subject='Re: [PATCH v5 bpf-next 06/17] libbpf: xsk: use bpf_link' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).