Message ID | 20190117233157.1467-1-edumazet@google.com |
---|---|
State | Changes Requested |
Delegated to: | BPF Maintainers |
Headers | show |
Series | [v2,bpf-next] bpf: allow BPF programs access skb_shared_info->gso_segs field | expand |
On Thu, Jan 17, 2019 at 03:31:57PM -0800, Eric Dumazet wrote: > This adds the ability to read gso_segs from a BPF program. > > v2: refined Eddie Hao patch to address Alexei feedback. > > Signed-off-by: Eric Dumazet <edumazet@google.com> > Cc: Eddie Hao <eddieh@google.com> > Cc: Vlad Dumitrescu <vladum@google.com> > Cc: Xiaotian Pei <xiaotian@google.com> > Cc: Yuchung Cheng <ycheng@google.com> > --- > include/uapi/linux/bpf.h | 1 + > net/core/filter.c | 21 ++++++++++++ > tools/include/uapi/linux/bpf.h | 1 + > tools/testing/selftests/bpf/test_verifier.c | 36 +++++++++++++++++++++ > 4 files changed, 59 insertions(+) > > diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h > index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 > --- a/include/uapi/linux/bpf.h > +++ b/include/uapi/linux/bpf.h > @@ -2540,6 +2540,7 @@ struct __sk_buff { > __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); > __u64 tstamp; > __u32 wire_len; > + __u32 gso_segs; > }; > > struct bpf_tunnel_key { > diff --git a/net/core/filter.c b/net/core/filter.c > index 2b3b436ef5457bf44c99780d6dec0b5f403f005c..a6ff5d9a04cf06926ee75cbc523456d12baf25ae 100644 > --- a/net/core/filter.c > +++ b/net/core/filter.c > @@ -6700,6 +6700,27 @@ static u32 bpf_convert_ctx_access(enum bpf_access_type type, > target_size)); > break; > > + case offsetof(struct __sk_buff, gso_segs): > + /* si->dst_reg = skb_shinfo(SKB); */ > +#ifdef NET_SKBUFF_DATA_USES_OFFSET > + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, head), > + si->dst_reg, si->src_reg, > + offsetof(struct sk_buff, head)); > + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), > + BPF_REG_TMP, si->src_reg, > + offsetof(struct sk_buff, end)); I am not sure BPF_REG_TMP can be used for non-classic BPF. The earlier insn could be using BPF_REG_TMP (which is BPF_REG_2) and R2 would become loss after this BPF_LDX_MEM. Daniel, can BPF_REG_AX be used here as a tmp? > + *insn++ = BPF_ALU64_REG(BPF_ADD, si->dst_reg, BPF_REG_TMP); > +#else > + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), > + si->dst_reg, si->src_reg, > + offsetof(struct sk_buff, end)); > +#endif > + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct skb_shared_info, gso_segs), > + si->dst_reg, si->dst_reg, > + bpf_target_off(struct skb_shared_info, > + gso_segs, 2, > + target_size)); > + break; > case offsetof(struct __sk_buff, wire_len): > BUILD_BUG_ON(FIELD_SIZEOF(struct qdisc_skb_cb, pkt_len) != 4); > > diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h > index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 > --- a/tools/include/uapi/linux/bpf.h > +++ b/tools/include/uapi/linux/bpf.h > @@ -2540,6 +2540,7 @@ struct __sk_buff { > __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); > __u64 tstamp; > __u32 wire_len; > + __u32 gso_segs; > }; > > struct bpf_tunnel_key { > diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c > index 2fd90d4568926d13542783c870507d43a6d6bb64..2c46531044bdf9ec1e4fa47e2c94c9edb0ac3d08 100644 > --- a/tools/testing/selftests/bpf/test_verifier.c > +++ b/tools/testing/selftests/bpf/test_verifier.c > @@ -5663,6 +5663,42 @@ static struct bpf_test tests[] = { > .result = ACCEPT, > .prog_type = BPF_PROG_TYPE_CGROUP_SKB, > }, > + { > + "read gso_segs from CGROUP_SKB", > + .insns = { > + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, > + offsetof(struct __sk_buff, gso_segs)), > + BPF_MOV64_IMM(BPF_REG_0, 0), > + BPF_EXIT_INSN(), > + }, > + .result = ACCEPT, > + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, > + }, > + { > + "write gso_segs from CGROUP_SKB", > + .insns = { > + BPF_MOV64_IMM(BPF_REG_0, 0), > + BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, > + offsetof(struct __sk_buff, gso_segs)), > + BPF_MOV64_IMM(BPF_REG_0, 0), > + BPF_EXIT_INSN(), > + }, > + .result = REJECT, > + .result_unpriv = REJECT, > + .errstr = "invalid bpf_context access off=164 size=4", > + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, > + }, > + { > + "read gso_segs from CLS", > + .insns = { > + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, > + offsetof(struct __sk_buff, gso_segs)), > + BPF_MOV64_IMM(BPF_REG_0, 0), > + BPF_EXIT_INSN(), > + }, > + .result = ACCEPT, > + .prog_type = BPF_PROG_TYPE_SCHED_CLS, > + }, > { > "multiple registers share map_lookup_elem result", > .insns = { > -- > 2.20.1.321.g9e740568ce-goog >
On 01/18/2019 07:42 PM, Martin Lau wrote: > On Thu, Jan 17, 2019 at 03:31:57PM -0800, Eric Dumazet wrote: >> This adds the ability to read gso_segs from a BPF program. >> >> v2: refined Eddie Hao patch to address Alexei feedback. >> >> Signed-off-by: Eric Dumazet <edumazet@google.com> >> Cc: Eddie Hao <eddieh@google.com> >> Cc: Vlad Dumitrescu <vladum@google.com> >> Cc: Xiaotian Pei <xiaotian@google.com> >> Cc: Yuchung Cheng <ycheng@google.com> >> --- >> include/uapi/linux/bpf.h | 1 + >> net/core/filter.c | 21 ++++++++++++ >> tools/include/uapi/linux/bpf.h | 1 + >> tools/testing/selftests/bpf/test_verifier.c | 36 +++++++++++++++++++++ >> 4 files changed, 59 insertions(+) >> >> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h >> index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 >> --- a/include/uapi/linux/bpf.h >> +++ b/include/uapi/linux/bpf.h >> @@ -2540,6 +2540,7 @@ struct __sk_buff { >> __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); >> __u64 tstamp; >> __u32 wire_len; >> + __u32 gso_segs; >> }; >> >> struct bpf_tunnel_key { >> diff --git a/net/core/filter.c b/net/core/filter.c >> index 2b3b436ef5457bf44c99780d6dec0b5f403f005c..a6ff5d9a04cf06926ee75cbc523456d12baf25ae 100644 >> --- a/net/core/filter.c >> +++ b/net/core/filter.c >> @@ -6700,6 +6700,27 @@ static u32 bpf_convert_ctx_access(enum bpf_access_type type, >> target_size)); >> break; >> >> + case offsetof(struct __sk_buff, gso_segs): >> + /* si->dst_reg = skb_shinfo(SKB); */ >> +#ifdef NET_SKBUFF_DATA_USES_OFFSET >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, head), >> + si->dst_reg, si->src_reg, >> + offsetof(struct sk_buff, head)); >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), >> + BPF_REG_TMP, si->src_reg, >> + offsetof(struct sk_buff, end)); > I am not sure BPF_REG_TMP can be used for non-classic BPF. > The earlier insn could be using BPF_REG_TMP (which is BPF_REG_2) and > R2 would become loss after this BPF_LDX_MEM. Yes, this will indeed corrupt R2 register. BPF_REG_TMP can only be used for reg mapping out of classic BPF. > Daniel, can BPF_REG_AX be used here as a tmp? BPF_REG_AX would work in this case, yes. Neither of the above insns are used in blinding nor would they collide with current verifier rewrites. >> + *insn++ = BPF_ALU64_REG(BPF_ADD, si->dst_reg, BPF_REG_TMP); >> +#else >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), >> + si->dst_reg, si->src_reg, >> + offsetof(struct sk_buff, end)); >> +#endif >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct skb_shared_info, gso_segs), >> + si->dst_reg, si->dst_reg, >> + bpf_target_off(struct skb_shared_info, >> + gso_segs, 2, >> + target_size)); >> + break; >> case offsetof(struct __sk_buff, wire_len): >> BUILD_BUG_ON(FIELD_SIZEOF(struct qdisc_skb_cb, pkt_len) != 4); >> >> diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h >> index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 >> --- a/tools/include/uapi/linux/bpf.h >> +++ b/tools/include/uapi/linux/bpf.h >> @@ -2540,6 +2540,7 @@ struct __sk_buff { >> __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); >> __u64 tstamp; >> __u32 wire_len; >> + __u32 gso_segs; >> }; >> >> struct bpf_tunnel_key { >> diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c >> index 2fd90d4568926d13542783c870507d43a6d6bb64..2c46531044bdf9ec1e4fa47e2c94c9edb0ac3d08 100644 >> --- a/tools/testing/selftests/bpf/test_verifier.c >> +++ b/tools/testing/selftests/bpf/test_verifier.c >> @@ -5663,6 +5663,42 @@ static struct bpf_test tests[] = { >> .result = ACCEPT, >> .prog_type = BPF_PROG_TYPE_CGROUP_SKB, >> }, >> + { >> + "read gso_segs from CGROUP_SKB", >> + .insns = { >> + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, >> + offsetof(struct __sk_buff, gso_segs)), >> + BPF_MOV64_IMM(BPF_REG_0, 0), >> + BPF_EXIT_INSN(), >> + }, >> + .result = ACCEPT, >> + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, >> + }, >> + { >> + "write gso_segs from CGROUP_SKB", >> + .insns = { >> + BPF_MOV64_IMM(BPF_REG_0, 0), >> + BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, >> + offsetof(struct __sk_buff, gso_segs)), >> + BPF_MOV64_IMM(BPF_REG_0, 0), >> + BPF_EXIT_INSN(), >> + }, >> + .result = REJECT, >> + .result_unpriv = REJECT, >> + .errstr = "invalid bpf_context access off=164 size=4", >> + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, >> + }, >> + { >> + "read gso_segs from CLS", >> + .insns = { >> + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, >> + offsetof(struct __sk_buff, gso_segs)), >> + BPF_MOV64_IMM(BPF_REG_0, 0), >> + BPF_EXIT_INSN(), >> + }, >> + .result = ACCEPT, >> + .prog_type = BPF_PROG_TYPE_SCHED_CLS, >> + }, >> { >> "multiple registers share map_lookup_elem result", >> .insns = { >> -- >> 2.20.1.321.g9e740568ce-goog >>
On Wed, Jan 23, 2019 at 3:55 AM Daniel Borkmann <daniel@iogearbox.net> wrote: > > On 01/18/2019 07:42 PM, Martin Lau wrote: > > On Thu, Jan 17, 2019 at 03:31:57PM -0800, Eric Dumazet wrote: > >> This adds the ability to read gso_segs from a BPF program. > >> > >> v2: refined Eddie Hao patch to address Alexei feedback. > >> > >> Signed-off-by: Eric Dumazet <edumazet@google.com> > >> Cc: Eddie Hao <eddieh@google.com> > >> Cc: Vlad Dumitrescu <vladum@google.com> > >> Cc: Xiaotian Pei <xiaotian@google.com> > >> Cc: Yuchung Cheng <ycheng@google.com> > >> --- > >> include/uapi/linux/bpf.h | 1 + > >> net/core/filter.c | 21 ++++++++++++ > >> tools/include/uapi/linux/bpf.h | 1 + > >> tools/testing/selftests/bpf/test_verifier.c | 36 +++++++++++++++++++++ > >> 4 files changed, 59 insertions(+) > >> > >> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h > >> index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 > >> --- a/include/uapi/linux/bpf.h > >> +++ b/include/uapi/linux/bpf.h > >> @@ -2540,6 +2540,7 @@ struct __sk_buff { > >> __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); > >> __u64 tstamp; > >> __u32 wire_len; > >> + __u32 gso_segs; > >> }; > >> > >> struct bpf_tunnel_key { > >> diff --git a/net/core/filter.c b/net/core/filter.c > >> index 2b3b436ef5457bf44c99780d6dec0b5f403f005c..a6ff5d9a04cf06926ee75cbc523456d12baf25ae 100644 > >> --- a/net/core/filter.c > >> +++ b/net/core/filter.c > >> @@ -6700,6 +6700,27 @@ static u32 bpf_convert_ctx_access(enum bpf_access_type type, > >> target_size)); > >> break; > >> > >> + case offsetof(struct __sk_buff, gso_segs): > >> + /* si->dst_reg = skb_shinfo(SKB); */ > >> +#ifdef NET_SKBUFF_DATA_USES_OFFSET > >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, head), > >> + si->dst_reg, si->src_reg, > >> + offsetof(struct sk_buff, head)); > >> + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), > >> + BPF_REG_TMP, si->src_reg, > >> + offsetof(struct sk_buff, end)); > > I am not sure BPF_REG_TMP can be used for non-classic BPF. > > The earlier insn could be using BPF_REG_TMP (which is BPF_REG_2) and > > R2 would become loss after this BPF_LDX_MEM. > > Yes, this will indeed corrupt R2 register. BPF_REG_TMP can only be used for > reg mapping out of classic BPF. > > > Daniel, can BPF_REG_AX be used here as a tmp? > > BPF_REG_AX would work in this case, yes. Neither of the above insns are used > in blinding nor would they collide with current verifier rewrites. > OK, I'll send a v3 using REG_AX then, thanks Daniel & Martin !
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -2540,6 +2540,7 @@ struct __sk_buff { __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); __u64 tstamp; __u32 wire_len; + __u32 gso_segs; }; struct bpf_tunnel_key { diff --git a/net/core/filter.c b/net/core/filter.c index 2b3b436ef5457bf44c99780d6dec0b5f403f005c..a6ff5d9a04cf06926ee75cbc523456d12baf25ae 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -6700,6 +6700,27 @@ static u32 bpf_convert_ctx_access(enum bpf_access_type type, target_size)); break; + case offsetof(struct __sk_buff, gso_segs): + /* si->dst_reg = skb_shinfo(SKB); */ +#ifdef NET_SKBUFF_DATA_USES_OFFSET + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, head), + si->dst_reg, si->src_reg, + offsetof(struct sk_buff, head)); + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), + BPF_REG_TMP, si->src_reg, + offsetof(struct sk_buff, end)); + *insn++ = BPF_ALU64_REG(BPF_ADD, si->dst_reg, BPF_REG_TMP); +#else + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_buff, end), + si->dst_reg, si->src_reg, + offsetof(struct sk_buff, end)); +#endif + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct skb_shared_info, gso_segs), + si->dst_reg, si->dst_reg, + bpf_target_off(struct skb_shared_info, + gso_segs, 2, + target_size)); + break; case offsetof(struct __sk_buff, wire_len): BUILD_BUG_ON(FIELD_SIZEOF(struct qdisc_skb_cb, pkt_len) != 4); diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 91c43884f295f60a85268ddf0020bf8aa47f8329..2940a9854f6d8e493518ca894e0c9c630ae4ab7a 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -2540,6 +2540,7 @@ struct __sk_buff { __bpf_md_ptr(struct bpf_flow_keys *, flow_keys); __u64 tstamp; __u32 wire_len; + __u32 gso_segs; }; struct bpf_tunnel_key { diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c index 2fd90d4568926d13542783c870507d43a6d6bb64..2c46531044bdf9ec1e4fa47e2c94c9edb0ac3d08 100644 --- a/tools/testing/selftests/bpf/test_verifier.c +++ b/tools/testing/selftests/bpf/test_verifier.c @@ -5663,6 +5663,42 @@ static struct bpf_test tests[] = { .result = ACCEPT, .prog_type = BPF_PROG_TYPE_CGROUP_SKB, }, + { + "read gso_segs from CGROUP_SKB", + .insns = { + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, + offsetof(struct __sk_buff, gso_segs)), + BPF_MOV64_IMM(BPF_REG_0, 0), + BPF_EXIT_INSN(), + }, + .result = ACCEPT, + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, + }, + { + "write gso_segs from CGROUP_SKB", + .insns = { + BPF_MOV64_IMM(BPF_REG_0, 0), + BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, + offsetof(struct __sk_buff, gso_segs)), + BPF_MOV64_IMM(BPF_REG_0, 0), + BPF_EXIT_INSN(), + }, + .result = REJECT, + .result_unpriv = REJECT, + .errstr = "invalid bpf_context access off=164 size=4", + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, + }, + { + "read gso_segs from CLS", + .insns = { + BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1, + offsetof(struct __sk_buff, gso_segs)), + BPF_MOV64_IMM(BPF_REG_0, 0), + BPF_EXIT_INSN(), + }, + .result = ACCEPT, + .prog_type = BPF_PROG_TYPE_SCHED_CLS, + }, { "multiple registers share map_lookup_elem result", .insns = {
This adds the ability to read gso_segs from a BPF program. v2: refined Eddie Hao patch to address Alexei feedback. Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Eddie Hao <eddieh@google.com> Cc: Vlad Dumitrescu <vladum@google.com> Cc: Xiaotian Pei <xiaotian@google.com> Cc: Yuchung Cheng <ycheng@google.com> --- include/uapi/linux/bpf.h | 1 + net/core/filter.c | 21 ++++++++++++ tools/include/uapi/linux/bpf.h | 1 + tools/testing/selftests/bpf/test_verifier.c | 36 +++++++++++++++++++++ 4 files changed, 59 insertions(+)