On x86-64, USDT arguments can be specified using Scale-Index-Base (SIB) addressing, e.g. "1@-96(%rbp,%rax,8)". The current USDT implementation in libbpf cannot parse this format, causing `bpf_program__attach_usdt()` to fail with -ENOENT (unrecognized register).

This patch fixes this by implementing the necessary changes: - add correct handling for SIB-addressed arguments in `bpf_usdt_arg`. - add adaptive support to `__bpf_usdt_arg_type` and `__bpf_usdt_arg_spec` to represent SIB addressing parameters.

Signed-off-by: Jiawei Zhao phoenix500526@163.com --- tools/lib/bpf/usdt.bpf.h | 44 +++++++++++++++++++++++++++++-- tools/lib/bpf/usdt.c | 57 ++++++++++++++++++++++++++++++++++++---- 2 files changed, 94 insertions(+), 7 deletions(-)

diff --git a/tools/lib/bpf/usdt.bpf.h b/tools/lib/bpf/usdt.bpf.h index 2a7865c8e3fe..2000b0aead75 100644 --- a/tools/lib/bpf/usdt.bpf.h +++ b/tools/lib/bpf/usdt.bpf.h @@ -34,13 +34,32 @@ enum __bpf_usdt_arg_type { BPF_USDT_ARG_CONST, BPF_USDT_ARG_REG, BPF_USDT_ARG_REG_DEREF, + BPF_USDT_ARG_SIB, };

+/* + * This struct layout is designed specifically to be backwards/forward + * compatible between libbpf versions for ARG_CONST, ARG_REG, and + * ARG_REG_DEREF modes. ARG_SIB requires libbpf v1.7+. + */ struct __bpf_usdt_arg_spec { /* u64 scalar interpreted depending on arg_type, see below */ __u64 val_off; +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ /* arg location case, see bpf_usdt_arg() for details */ - enum __bpf_usdt_arg_type arg_type; + enum __bpf_usdt_arg_type arg_type: 8; + /* index register offset within struct pt_regs */ + __u16 idx_reg_off: 12; + /* scale factor for index register (1, 2, 4, or 8) */ + __u16 scale: 4; + /* reserved for future use, keeps reg_off offset stable */ + __u8 __reserved: 8; +#else + __u8 __reserved: 8; + __u16 idx_reg_off: 12; + __u16 scale: 4; + enum __bpf_usdt_arg_type arg_type: 8; +#endif /* offset of referenced register within struct pt_regs */ short reg_off; /* whether arg should be interpreted as signed value */ @@ -149,7 +168,7 @@ int bpf_usdt_arg(struct pt_regs *ctx, __u64 arg_num, long *res) { struct __bpf_usdt_spec *spec; struct __bpf_usdt_arg_spec *arg_spec; - unsigned long val; + unsigned long val, idx; int err, spec_id;

*res = 0; @@ -202,6 +221,27 @@ int bpf_usdt_arg(struct pt_regs *ctx, __u64 arg_num, long *res) return err; #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ val >>= arg_spec->arg_bitshift; +#endif + break; + case BPF_USDT_ARG_SIB: + /* Arg is in memory addressed by SIB (Scale-Index-Base) mode + * (e.g., "-1@-96(%rbp,%rax,8)" in USDT arg spec). We first + * fetch the base register contents and the index register + * contents from pt_regs. Then we calculate the final address + * as base + (index * scale) + offset, and do a user-space + * probe read to fetch the argument value. + */ + err = bpf_probe_read_kernel(&val, sizeof(val), (void *)ctx + arg_spec->reg_off); + if (err) + return err; + err = bpf_probe_read_kernel(&idx, sizeof(idx), (void *)ctx + arg_spec->idx_reg_off); + if (err) + return err; + err = bpf_probe_read_user(&val, sizeof(val), (void *)(val + (idx * arg_spec->scale) + arg_spec->val_off)); + if (err) + return err; +#if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + val >>= arg_spec->arg_bitshift; #endif break; default: diff --git a/tools/lib/bpf/usdt.c b/tools/lib/bpf/usdt.c index 3373b9d45ac4..d90e0b694134 100644 --- a/tools/lib/bpf/usdt.c +++ b/tools/lib/bpf/usdt.c @@ -200,12 +200,23 @@ enum usdt_arg_type { USDT_ARG_CONST, USDT_ARG_REG, USDT_ARG_REG_DEREF, + USDT_ARG_SIB, };

/* should match exactly struct __bpf_usdt_arg_spec from usdt.bpf.h */ struct usdt_arg_spec { __u64 val_off; - enum usdt_arg_type arg_type; +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ + enum usdt_arg_type arg_type: 8; + __u16 idx_reg_off: 12; + __u16 scale: 4; + __u8 __reserved: 8; /* keep reg_off offset stable */ +#else + __u8 __reserved: 8; /* keep reg_off offset stable */ + __u16 idx_reg_off: 12; + __u16 scale: 4; + enum usdt_arg_type arg_type: 8; +#endif short reg_off; bool arg_signed; char arg_bitshift; @@ -1283,11 +1294,46 @@ static int calc_pt_regs_off(const char *reg_name)

static int parse_usdt_arg(const char *arg_str, int arg_num, struct usdt_arg_spec *arg, int *arg_sz) { - char reg_name[16]; - int len, reg_off; - long off; + char reg_name[16] = {0}, idx_reg_name[16] = {0}; + int len, reg_off, idx_reg_off, scale = 1; + long off = 0; + + if (sscanf(arg_str, " %d @ %ld ( %%%15[^,] , %%%15[^,] , %d ) %n", + arg_sz, &off, reg_name, idx_reg_name, &scale, &len) == 5 || + sscanf(arg_str, " %d @ ( %%%15[^,] , %%%15[^,] , %d ) %n", + arg_sz, reg_name, idx_reg_name, &scale, &len) == 4 || + sscanf(arg_str, " %d @ %ld ( %%%15[^,] , %%%15[^)] ) %n", + arg_sz, &off, reg_name, idx_reg_name, &len) == 4 || + sscanf(arg_str, " %d @ ( %%%15[^,] , %%%15[^)] ) %n", + arg_sz, reg_name, idx_reg_name, &len) == 3 + ) { + /* + * Scale Index Base case: + * 1@-96(%rbp,%rax,8) + * 1@(%rbp,%rax,8) + * 1@-96(%rbp,%rax) + * 1@(%rbp,%rax) + */ + arg->arg_type = USDT_ARG_SIB; + arg->val_off = off;

- if (sscanf(arg_str, " %d @ %ld ( %%%15[^)] ) %n", arg_sz, &off, reg_name, &len) == 3) { + reg_off = calc_pt_regs_off(reg_name); + if (reg_off < 0) + return reg_off; + arg->reg_off = reg_off; + + idx_reg_off = calc_pt_regs_off(idx_reg_name); + if (idx_reg_off < 0) + return idx_reg_off; + /* validate scale factor and set fields directly */ + if (scale != 1 && scale != 2 && scale != 4 && scale != 8) { + pr_warn("usdt: invalid SIB scale %d, expected 1, 2, 4, 8\n", scale); + return -EINVAL; + } + arg->idx_reg_off = idx_reg_off; + arg->scale = scale; + } else if (sscanf(arg_str, " %d @ %ld ( %%%15[^)] ) %n", + arg_sz, &off, reg_name, &len) == 3) { /* Memory dereference case, e.g., -4@-20(%rbp) */ arg->arg_type = USDT_ARG_REG_DEREF; arg->val_off = off; @@ -1306,6 +1352,7 @@ static int parse_usdt_arg(const char *arg_str, int arg_num, struct usdt_arg_spec } else if (sscanf(arg_str, " %d @ %%%15s %n", arg_sz, reg_name, &len) == 2) { /* Register read case, e.g., -4@%eax */ arg->arg_type = USDT_ARG_REG; + /* register read has no memory offset */ arg->val_off = 0;

reg_off = calc_pt_regs_off(reg_name);

When using GCC on x86-64 to compile an usdt prog with -O1 or higher optimization, the compiler will generate SIB addressing mode for global array and PC-relative addressing mode for global variable, e.g. "1@-96(%rbp,%rax,8)" and "-1@4+t1(%rip)".

In this patch: - enrich subtest_basic_usdt test case to cover SIB addressing usdt argument spec handling logic

Signed-off-by: Jiawei Zhao phoenix500526@163.com --- tools/testing/selftests/bpf/prog_tests/usdt.c | 62 ++++++++++++++++++- tools/testing/selftests/bpf/progs/test_usdt.c | 32 ++++++++++ 2 files changed, 92 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/usdt.c b/tools/testing/selftests/bpf/prog_tests/usdt.c index 9057e983cc54..4b264f7d3324 100644 --- a/tools/testing/selftests/bpf/prog_tests/usdt.c +++ b/tools/testing/selftests/bpf/prog_tests/usdt.c @@ -10,6 +10,26 @@

int lets_test_this(int);

+#if defined(__x86_64__) || defined(__i386__) +/* + * SIB (Scale-Index-Base) addressing format: + * "size@(base_reg, index_reg, scale)" + * - 'size' is the size in bytes of the array element, and its sign indicates + * whether the type is signed (negative) or unsigned (positive). + * - 'base_reg' is the register holding the base address, normally rdx or edx + * - 'index_reg' is the register holding the index, normally rax or eax + * - 'scale' is the scaling factor (typically 1, 2, 4, or 8), which matches the + * size of the element type. + * + * For example, for an array of 'short' (signed 2-byte elements), the SIB spec would be: + * - size: -2 (negative because 'short' is signed) + * - scale: 2 (since sizeof(short) == 2) + * The resulting SIB format: "-2@(%%rdx,%%rax,2)" + */ +static volatile short array[] = {-1, -2, -3, -4}; +#define USDT_SIB_ARG_SPEC -2@(%%rdx,%%rax,2) +#endif + static volatile int idx = 2; static volatile __u64 bla = 0xFEDCBA9876543210ULL; static volatile short nums[] = {-1, -2, -3, -4}; @@ -25,6 +45,10 @@ unsigned short test_usdt0_semaphore SEC(".probes"); unsigned short test_usdt3_semaphore SEC(".probes"); unsigned short test_usdt12_semaphore SEC(".probes");

+#if defined(__x86_64__) || defined(__i386__) +unsigned short test_usdt_sib_semaphore SEC(".probes"); +#endif + static void __always_inline trigger_func(int x) { long y = 42;

@@ -40,12 +64,27 @@ static void __always_inline trigger_func(int x) { } }

+#if defined(__x86_64__) || defined(__i386__) +static void trigger_sib_spec(void) +{ + /* Base address + offset + (index * scale) */ + /* Force SIB addressing with inline assembly */ + asm volatile( + STAP_PROBE_ASM(test, usdt_sib, USDT_SIB_ARG_SPEC) + : + : "d"(array), "a"(0) + : "memory" + ); +} +#endif + static void subtest_basic_usdt(void) { LIBBPF_OPTS(bpf_usdt_opts, opts); struct test_usdt *skel; struct test_usdt__bss *bss; int err, i; + const __u64 expected_cookie = 0xcafedeadbeeffeed;

skel = test_usdt__open_and_load(); if (!ASSERT_OK_PTR(skel, "skel_open")) @@ -59,20 +98,29 @@ static void subtest_basic_usdt(void) goto cleanup;

/* usdt0 won't be auto-attached */ - opts.usdt_cookie = 0xcafedeadbeeffeed; + opts.usdt_cookie = expected_cookie; skel->links.usdt0 = bpf_program__attach_usdt(skel->progs.usdt0, 0 /*self*/, "/proc/self/exe", "test", "usdt0", &opts); if (!ASSERT_OK_PTR(skel->links.usdt0, "usdt0_link")) goto cleanup;

+#if defined(__x86_64__) || defined(__i386__) + opts.usdt_cookie = expected_cookie; + skel->links.usdt_sib = bpf_program__attach_usdt(skel->progs.usdt_sib, + 0 /*self*/, "/proc/self/exe", + "test", "usdt_sib", &opts); + if (!ASSERT_OK_PTR(skel->links.usdt_sib, "usdt_sib_link")) + goto cleanup; +#endif + trigger_func(1);

ASSERT_EQ(bss->usdt0_called, 1, "usdt0_called"); ASSERT_EQ(bss->usdt3_called, 1, "usdt3_called"); ASSERT_EQ(bss->usdt12_called, 1, "usdt12_called");

- ASSERT_EQ(bss->usdt0_cookie, 0xcafedeadbeeffeed, "usdt0_cookie"); + ASSERT_EQ(bss->usdt0_cookie, expected_cookie, "usdt0_cookie"); ASSERT_EQ(bss->usdt0_arg_cnt, 0, "usdt0_arg_cnt"); ASSERT_EQ(bss->usdt0_arg_ret, -ENOENT, "usdt0_arg_ret"); ASSERT_EQ(bss->usdt0_arg_size, -ENOENT, "usdt0_arg_size"); @@ -156,6 +204,16 @@ static void subtest_basic_usdt(void) ASSERT_EQ(bss->usdt3_args[1], 42, "usdt3_arg2"); ASSERT_EQ(bss->usdt3_args[2], (uintptr_t)&bla, "usdt3_arg3");

+#if defined(__x86_64__) || defined(__i386__) + trigger_sib_spec(); + ASSERT_EQ(bss->usdt_sib_called, 1, "usdt_sib_called"); + ASSERT_EQ(bss->usdt_sib_cookie, expected_cookie, "usdt_sib_cookie"); + ASSERT_EQ(bss->usdt_sib_arg_cnt, 1, "usdt_sib_arg_cnt"); + ASSERT_EQ(bss->usdt_sib_arg, nums[0], "usdt_sib_arg"); + ASSERT_EQ(bss->usdt_sib_arg_ret, 0, "usdt_sib_arg_ret"); + ASSERT_EQ(bss->usdt_sib_arg_size, sizeof(nums[0]), "usdt_sib_arg_size"); +#endif + cleanup: test_usdt__destroy(skel); } diff --git a/tools/testing/selftests/bpf/progs/test_usdt.c b/tools/testing/selftests/bpf/progs/test_usdt.c index 096488f47fbc..63db72253316 100644 --- a/tools/testing/selftests/bpf/progs/test_usdt.c +++ b/tools/testing/selftests/bpf/progs/test_usdt.c @@ -107,4 +107,36 @@ int BPF_USDT(usdt12, int a1, int a2, long a3, long a4, unsigned a5, return 0; }

+ +int usdt_sib_called; +u64 usdt_sib_cookie; +int usdt_sib_arg_cnt; +int usdt_sib_arg_ret; +u64 usdt_sib_arg; +int usdt_sib_arg_size; + +/* + * usdt_sib is only tested on x86-related architectures, so it requires + * manual attach since auto-attach will panic tests under other architectures + */ +SEC("usdt") +int usdt_sib(struct pt_regs *ctx) +{ + long tmp; + + if (my_pid != (bpf_get_current_pid_tgid() >> 32)) + return 0; + + __sync_fetch_and_add(&usdt_sib_called, 1); + + usdt_sib_cookie = bpf_usdt_cookie(ctx); + usdt_sib_arg_cnt = bpf_usdt_arg_cnt(ctx); + + usdt_sib_arg_ret = bpf_usdt_arg(ctx, 0, &tmp); + usdt_sib_arg = (short)tmp; + usdt_sib_arg_size = bpf_usdt_arg_size(ctx, 0); + + return 0; +} + char _license[] SEC("license") = "GPL";