Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 0d830032 authored May 08, 2018 by Jakub Kicinski Committed by Daniel Borkmann May 09, 2018

bpf: xdp: allow offloads to store into rx_queue_index



It's fairly easy for offloaded XDP programs to select the RX queue
packets go to.  We need a way of expressing this in the software.
Allow write to the rx_queue_index field of struct xdp_md for
device-bound programs.

Skip convert_ctx_access callback entirely for offloads.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Quentin Monnet <quentin.monnet@netronome.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

parent a1d1f079

include/linux/bpf.h

+1 −1

Original line number	Diff line number	Diff line
		@@ -627,7 +627,7 @@ bool bpf_offload_dev_match(struct bpf_prog prog, struct bpf_map map);
		#if defined(CONFIG_NET) && defined(CONFIG_BPF_SYSCALL)
		int bpf_prog_offload_init(struct bpf_prog prog, union bpf_attr attr);

		static inline bool bpf_prog_is_dev_bound(struct bpf_prog_aux *aux)
		static inline bool bpf_prog_is_dev_bound(const struct bpf_prog_aux *aux)
		{
		return aux->offload_requested;
		}

kernel/bpf/verifier.c

+1 −1

Original line number	Diff line number	Diff line
		@@ -5215,7 +5215,7 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env)
		}
		}

		if (!ops->convert_ctx_access)
		if (!ops->convert_ctx_access \|\| bpf_prog_is_dev_bound(env->prog->aux))
		return 0;

		insn = env->prog->insnsi + delta;

net/core/filter.c

+8 −1

Original line number	Diff line number	Diff line
		@@ -4645,8 +4645,15 @@ static bool xdp_is_valid_access(int off, int size,
		const struct bpf_prog *prog,
		struct bpf_insn_access_aux *info)
		{
		if (type == BPF_WRITE)
		if (type == BPF_WRITE) {
		if (bpf_prog_is_dev_bound(prog->aux)) {
		switch (off) {
		case offsetof(struct xdp_md, rx_queue_index):
		return __is_valid_xdp_access(off, size);
		}
		}
		return false;
		}

		switch (off) {
		case offsetof(struct xdp_md, data):