openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/tools/testing/selftests/bpf/progs/test_seg6_loop.c

261 lines
6.0 KiB
C
Raw Normal View History

selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
#include <stddef.h>
#include <inttypes.h>
#include <errno.h>
#include <linux/seg6_local.h>
#include <linux/bpf.h>
selftests: Use consistent include paths for libbpf Fix all selftests to include libbpf header files with the bpf/ prefix, to be consistent with external users of the library. Also ensure that all includes of exported libbpf header files (those that are exported on 'make install' of the library) use bracketed includes instead of quoted. To not break the build, keep the old include path until everything has been changed to the new one; a subsequent patch will remove that. Fixes: 6910d7d3867a ("selftests/bpf: Ensure bpf_helper_defs.h are taken from selftests dir") Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/bpf/157952560568.1683545.9649335788846513446.stgit@toke.dk
2020-01-20 21:06:45 +08:00
#include <bpf/bpf_helpers.h>
#include <bpf/bpf_endian.h>
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
/* Packet parsing state machine helpers. */
#define cursor_advance(_cursor, _len) \
({ void *_tmp = _cursor; _cursor += _len; _tmp; })
#define SR6_FLAG_ALERT (1 << 4)
#define BPF_PACKET_HEADER __attribute__((packed))
struct ip6_t {
unsigned int ver:4;
unsigned int priority:8;
unsigned int flow_label:20;
unsigned short payload_len;
unsigned char next_header;
unsigned char hop_limit;
unsigned long long src_hi;
unsigned long long src_lo;
unsigned long long dst_hi;
unsigned long long dst_lo;
} BPF_PACKET_HEADER;
struct ip6_addr_t {
unsigned long long hi;
unsigned long long lo;
} BPF_PACKET_HEADER;
struct ip6_srh_t {
unsigned char nexthdr;
unsigned char hdrlen;
unsigned char type;
unsigned char segments_left;
unsigned char first_segment;
unsigned char flags;
unsigned short tag;
struct ip6_addr_t segments[0];
} BPF_PACKET_HEADER;
struct sr6_tlv_t {
unsigned char type;
unsigned char len;
unsigned char value[0];
} BPF_PACKET_HEADER;
selftests: bpf: standardize to static __always_inline The progs for bpf selftests use several different notations to force function inlining. Standardize to what most of them use, static __always_inline. Suggested-by: Song Liu <liu.song.a23@gmail.com> Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-07-03 02:26:51 +08:00
static __always_inline struct ip6_srh_t *get_srh(struct __sk_buff *skb)
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
{
void *cursor, *data_end;
struct ip6_srh_t *srh;
struct ip6_t *ip;
uint8_t *ipver;
data_end = (void *)(long)skb->data_end;
cursor = (void *)(long)skb->data;
ipver = (uint8_t *)cursor;
if ((void *)ipver + sizeof(*ipver) > data_end)
return NULL;
if ((*ipver >> 4) != 6)
return NULL;
ip = cursor_advance(cursor, sizeof(*ip));
if ((void *)ip + sizeof(*ip) > data_end)
return NULL;
if (ip->next_header != 43)
return NULL;
srh = cursor_advance(cursor, sizeof(*srh));
if ((void *)srh + sizeof(*srh) > data_end)
return NULL;
if (srh->type != 4)
return NULL;
return srh;
}
selftests: bpf: standardize to static __always_inline The progs for bpf selftests use several different notations to force function inlining. Standardize to what most of them use, static __always_inline. Suggested-by: Song Liu <liu.song.a23@gmail.com> Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-07-03 02:26:51 +08:00
static __always_inline int update_tlv_pad(struct __sk_buff *skb,
uint32_t new_pad, uint32_t old_pad,
uint32_t pad_off)
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
{
int err;
if (new_pad != old_pad) {
err = bpf_lwt_seg6_adjust_srh(skb, pad_off,
(int) new_pad - (int) old_pad);
if (err)
return err;
}
if (new_pad > 0) {
char pad_tlv_buf[16] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0};
struct sr6_tlv_t *pad_tlv = (struct sr6_tlv_t *) pad_tlv_buf;
pad_tlv->type = SR6_TLV_PADDING;
pad_tlv->len = new_pad - 2;
err = bpf_lwt_seg6_store_bytes(skb, pad_off,
(void *)pad_tlv_buf, new_pad);
if (err)
return err;
}
return 0;
}
selftests: bpf: standardize to static __always_inline The progs for bpf selftests use several different notations to force function inlining. Standardize to what most of them use, static __always_inline. Suggested-by: Song Liu <liu.song.a23@gmail.com> Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-07-03 02:26:51 +08:00
static __always_inline int is_valid_tlv_boundary(struct __sk_buff *skb,
struct ip6_srh_t *srh,
uint32_t *tlv_off,
uint32_t *pad_size,
uint32_t *pad_off)
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
{
uint32_t srh_off, cur_off;
int offset_valid = 0;
int err;
srh_off = (char *)srh - (char *)(long)skb->data;
// cur_off = end of segments, start of possible TLVs
cur_off = srh_off + sizeof(*srh) +
sizeof(struct ip6_addr_t) * (srh->first_segment + 1);
*pad_off = 0;
// we can only go as far as ~10 TLVs due to the BPF max stack size
selftests, bpf: Workaround an alu32 sub-register spilling issue Currently, with latest llvm trunk, selftest test_progs failed obj file test_seg6_loop.o with the following error in verifier: infinite loop detected at insn 76 The byte code sequence looks like below, and noted that alu32 has been turned off by default for better generated codes in general: 48: w3 = 100 49: *(u32 *)(r10 - 68) = r3 ... ; if (tlv.type == SR6_TLV_PADDING) { 76: if w3 == 5 goto -18 <LBB0_19> ... 85: r1 = *(u32 *)(r10 - 68) ; for (int i = 0; i < 100; i++) { 86: w1 += -1 87: if w1 == 0 goto +5 <LBB0_20> 88: *(u32 *)(r10 - 68) = r1 The main reason for verification failure is due to partial spills at r10 - 68 for induction variable "i". Current verifier only handles spills with 8-byte values. The above 4-byte value spill to stack is treated to STACK_MISC and its content is not saved. For the above example: w3 = 100 R3_w=inv100 fp-64_w=inv1086626730498 *(u32 *)(r10 - 68) = r3 R3_w=inv100 fp-64_w=inv1086626730498 ... r1 = *(u32 *)(r10 - 68) R1_w=inv(id=0,umax_value=4294967295,var_off=(0x0; 0xffffffff)) fp-64=inv1086626730498 To resolve this issue, verifier needs to be extended to track sub-registers in spilling, or llvm needs to enhanced to prevent sub-register spilling in register allocation phase. The former will increase verifier complexity and the latter will need some llvm "hacking". Let us workaround this issue by declaring the induction variable as "long" type so spilling will happen at non sub-register level. We can revisit this later if sub-register spilling causes similar or other verification issues. Signed-off-by: Yonghong Song <yhs@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/bpf/20191117214036.1309510-1-yhs@fb.com
2019-11-18 05:40:36 +08:00
// workaround: define induction variable "i" as "long" instead
// of "int" to prevent alu32 sub-register spilling.
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
#pragma clang loop unroll(disable)
selftests, bpf: Workaround an alu32 sub-register spilling issue Currently, with latest llvm trunk, selftest test_progs failed obj file test_seg6_loop.o with the following error in verifier: infinite loop detected at insn 76 The byte code sequence looks like below, and noted that alu32 has been turned off by default for better generated codes in general: 48: w3 = 100 49: *(u32 *)(r10 - 68) = r3 ... ; if (tlv.type == SR6_TLV_PADDING) { 76: if w3 == 5 goto -18 <LBB0_19> ... 85: r1 = *(u32 *)(r10 - 68) ; for (int i = 0; i < 100; i++) { 86: w1 += -1 87: if w1 == 0 goto +5 <LBB0_20> 88: *(u32 *)(r10 - 68) = r1 The main reason for verification failure is due to partial spills at r10 - 68 for induction variable "i". Current verifier only handles spills with 8-byte values. The above 4-byte value spill to stack is treated to STACK_MISC and its content is not saved. For the above example: w3 = 100 R3_w=inv100 fp-64_w=inv1086626730498 *(u32 *)(r10 - 68) = r3 R3_w=inv100 fp-64_w=inv1086626730498 ... r1 = *(u32 *)(r10 - 68) R1_w=inv(id=0,umax_value=4294967295,var_off=(0x0; 0xffffffff)) fp-64=inv1086626730498 To resolve this issue, verifier needs to be extended to track sub-registers in spilling, or llvm needs to enhanced to prevent sub-register spilling in register allocation phase. The former will increase verifier complexity and the latter will need some llvm "hacking". Let us workaround this issue by declaring the induction variable as "long" type so spilling will happen at non sub-register level. We can revisit this later if sub-register spilling causes similar or other verification issues. Signed-off-by: Yonghong Song <yhs@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Andrii Nakryiko <andriin@fb.com> Link: https://lore.kernel.org/bpf/20191117214036.1309510-1-yhs@fb.com
2019-11-18 05:40:36 +08:00
for (long i = 0; i < 100; i++) {
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
struct sr6_tlv_t tlv;
if (cur_off == *tlv_off)
offset_valid = 1;
if (cur_off >= srh_off + ((srh->hdrlen + 1) << 3))
break;
err = bpf_skb_load_bytes(skb, cur_off, &tlv, sizeof(tlv));
if (err)
return err;
if (tlv.type == SR6_TLV_PADDING) {
*pad_size = tlv.len + sizeof(tlv);
*pad_off = cur_off;
if (*tlv_off == srh_off) {
*tlv_off = cur_off;
offset_valid = 1;
}
break;
} else if (tlv.type == SR6_TLV_HMAC) {
break;
}
cur_off += sizeof(tlv) + tlv.len;
} // we reached the padding or HMAC TLVs, or the end of the SRH
if (*pad_off == 0)
*pad_off = cur_off;
if (*tlv_off == -1)
*tlv_off = cur_off;
else if (!offset_valid)
return -EINVAL;
return 0;
}
selftests: bpf: standardize to static __always_inline The progs for bpf selftests use several different notations to force function inlining. Standardize to what most of them use, static __always_inline. Suggested-by: Song Liu <liu.song.a23@gmail.com> Signed-off-by: Jiri Benc <jbenc@redhat.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-07-03 02:26:51 +08:00
static __always_inline int add_tlv(struct __sk_buff *skb,
struct ip6_srh_t *srh, uint32_t tlv_off,
struct sr6_tlv_t *itlv, uint8_t tlv_size)
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
{
uint32_t srh_off = (char *)srh - (char *)(long)skb->data;
uint8_t len_remaining, new_pad;
uint32_t pad_off = 0;
uint32_t pad_size = 0;
uint32_t partial_srh_len;
int err;
if (tlv_off != -1)
tlv_off += srh_off;
if (itlv->type == SR6_TLV_PADDING || itlv->type == SR6_TLV_HMAC)
return -EINVAL;
err = is_valid_tlv_boundary(skb, srh, &tlv_off, &pad_size, &pad_off);
if (err)
return err;
err = bpf_lwt_seg6_adjust_srh(skb, tlv_off, sizeof(*itlv) + itlv->len);
if (err)
return err;
err = bpf_lwt_seg6_store_bytes(skb, tlv_off, (void *)itlv, tlv_size);
if (err)
return err;
// the following can't be moved inside update_tlv_pad because the
// bpf verifier has some issues with it
pad_off += sizeof(*itlv) + itlv->len;
partial_srh_len = pad_off - srh_off;
len_remaining = partial_srh_len % 8;
new_pad = 8 - len_remaining;
if (new_pad == 1) // cannot pad for 1 byte only
new_pad = 9;
else if (new_pad == 8)
new_pad = 0;
return update_tlv_pad(skb, new_pad, pad_size, pad_off);
}
// Add an Egress TLV fc00::4, add the flag A,
// and apply End.X action to fc42::1
SEC("lwt_seg6local")
int __add_egr_x(struct __sk_buff *skb)
{
unsigned long long hi = 0xfc42000000000000;
unsigned long long lo = 0x1;
struct ip6_srh_t *srh = get_srh(skb);
uint8_t new_flags = SR6_FLAG_ALERT;
struct ip6_addr_t addr;
int err, offset;
if (srh == NULL)
return BPF_DROP;
uint8_t tlv[20] = {2, 18, 0, 0, 0xfd, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0,
0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x4};
err = add_tlv(skb, srh, (srh->hdrlen+1) << 3,
(struct sr6_tlv_t *)&tlv, 20);
if (err)
return BPF_DROP;
offset = sizeof(struct ip6_t) + offsetof(struct ip6_srh_t, flags);
err = bpf_lwt_seg6_store_bytes(skb, offset,
(void *)&new_flags, sizeof(new_flags));
if (err)
return BPF_DROP;
selftests/bpf: introduce bpf_cpu_to_be64 and bpf_be64_to_cpu test_lwt_seg6local and test_seg6_loop use custom 64-bit endianness conversion macros. Centralize their definitions in bpf_endian.h in order to reduce code duplication. This will also be useful when bpf_endian.h is promoted to an offical libbpf header. Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-08-30 19:07:29 +08:00
addr.lo = bpf_cpu_to_be64(lo);
addr.hi = bpf_cpu_to_be64(hi);
selftests/bpf: add realistic loop tests Add a bunch of loop tests. Most of them are created by replacing '#pragma unroll' with '#pragma clang loop unroll(disable)' Several tests are artificially large: /* partial unroll. llvm will unroll loop ~150 times. * C loop count -> 600. * Asm loop count -> 4. * 16k insns in loop body. * Total of 5 such loops. Total program size ~82k insns. */ "./pyperf600.o", /* no unroll at all. * C loop count -> 600. * ASM loop count -> 600. * ~110 insns in loop body. * Total of 5 such loops. Total program size ~1500 insns. */ "./pyperf600_nounroll.o", /* partial unroll. 19k insn in a loop. * Total program size 20.8k insn. * ~350k processed_insns */ "./strobemeta.o", Signed-off-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
2019-06-16 03:12:24 +08:00
err = bpf_lwt_seg6_action(skb, SEG6_LOCAL_ACTION_END_X,
(void *)&addr, sizeof(addr));
if (err)
return BPF_DROP;
return BPF_REDIRECT;
}
char __license[] SEC("license") = "GPL";