From c9883ee9d110703ccb3dfe2ca13e0b7a01351077 Mon Sep 17 00:00:00 2001 From: Xin Liu Date: Sat, 10 Dec 2022 16:20:45 +0800 Subject: libbpf: Optimized return value in libbpf_strerror when errno is libbpf errno This is a small improvement in libbpf_strerror. When libbpf_strerror is used to obtain the system error description, if the length of the buf is insufficient, libbpf_sterror returns ERANGE and sets errno to ERANGE. However, this processing is not performed when the error code customized by libbpf is obtained. Make some minor improvements here, return -ERANGE and set errno to ERANGE when buf is not enough for custom description. Signed-off-by: Xin Liu Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20221210082045.233697-1-liuxin350@huawei.com --- tools/lib/bpf/libbpf_errno.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf_errno.c b/tools/lib/bpf/libbpf_errno.c index 96f67a772a1b..6b180172ec6b 100644 --- a/tools/lib/bpf/libbpf_errno.c +++ b/tools/lib/bpf/libbpf_errno.c @@ -39,14 +39,14 @@ static const char *libbpf_strerror_table[NR_ERRNO] = { int libbpf_strerror(int err, char *buf, size_t size) { + int ret; + if (!buf || !size) return libbpf_err(-EINVAL); err = err > 0 ? err : -err; if (err < __LIBBPF_ERRNO__START) { - int ret; - ret = strerror_r(err, buf, size); buf[size - 1] = '\0'; return libbpf_err_errno(ret); @@ -56,12 +56,20 @@ int libbpf_strerror(int err, char *buf, size_t size) const char *msg; msg = libbpf_strerror_table[ERRNO_OFFSET(err)]; - snprintf(buf, size, "%s", msg); + ret = snprintf(buf, size, "%s", msg); buf[size - 1] = '\0'; + /* The length of the buf and msg is positive. + * A negative number may be returned only when the + * size exceeds INT_MAX. Not likely to appear. + */ + if (ret >= size) + return libbpf_err(-ERANGE); return 0; } - snprintf(buf, size, "Unknown libbpf error %d", err); + ret = snprintf(buf, size, "Unknown libbpf error %d", err); buf[size - 1] = '\0'; + if (ret >= size) + return libbpf_err(-ERANGE); return libbpf_err(-ENOENT); } -- cgit From 872aec4b5f635d94111d48ec3c57fbe078d64e7d Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Mon, 12 Dec 2022 13:15:00 -0800 Subject: libbpf: Fix single-line struct definition output in btf_dump btf_dump APIs emit unnecessary tabs when emitting struct/union definition that fits on the single line. Before this patch we'd get: struct blah {}; This patch fixes this and makes sure that we get more natural: struct blah {}; Fixes: 44a726c3f23c ("bpftool: Print newline before '}' for struct with padding only fields") Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20221212211505.558851-2-andrii@kernel.org --- tools/lib/bpf/btf_dump.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c index deb2bc9a0a7b..69e80ee5f70e 100644 --- a/tools/lib/bpf/btf_dump.c +++ b/tools/lib/bpf/btf_dump.c @@ -959,9 +959,12 @@ static void btf_dump_emit_struct_def(struct btf_dump *d, * Keep `struct empty {}` on a single line, * only print newline when there are regular or padding fields. */ - if (vlen || t->size) + if (vlen || t->size) { btf_dump_printf(d, "\n"); - btf_dump_printf(d, "%s}", pfx(lvl)); + btf_dump_printf(d, "%s}", pfx(lvl)); + } else { + btf_dump_printf(d, "}"); + } if (packed) btf_dump_printf(d, " __attribute__((packed))"); } -- cgit From 21a9a1bcccaa4f0337a24d666fe55944abcb171e Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Mon, 12 Dec 2022 13:15:01 -0800 Subject: libbpf: Handle non-standardly sized enums better in BTF-to-C dumper Turns out C allows to force enum to be 1-byte or 8-byte explicitly using mode(byte) or mode(word), respecticely. Linux sources are using this in some cases. This is imporant to handle correctly, as enum size determines corresponding fields in a struct that use that enum type. And if enum size is incorrect, this will lead to invalid struct layout. So add mode(byte) and mode(word) attribute support to btf_dump APIs. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20221212211505.558851-3-andrii@kernel.org --- tools/lib/bpf/btf_dump.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c index 69e80ee5f70e..234e82334d56 100644 --- a/tools/lib/bpf/btf_dump.c +++ b/tools/lib/bpf/btf_dump.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include #include @@ -1076,6 +1077,43 @@ static void btf_dump_emit_enum_def(struct btf_dump *d, __u32 id, else btf_dump_emit_enum64_val(d, t, lvl, vlen); btf_dump_printf(d, "\n%s}", pfx(lvl)); + + /* special case enums with special sizes */ + if (t->size == 1) { + /* one-byte enums can be forced with mode(byte) attribute */ + btf_dump_printf(d, " __attribute__((mode(byte)))"); + } else if (t->size == 8 && d->ptr_sz == 8) { + /* enum can be 8-byte sized if one of the enumerator values + * doesn't fit in 32-bit integer, or by adding mode(word) + * attribute (but probably only on 64-bit architectures); do + * our best here to try to satisfy the contract without adding + * unnecessary attributes + */ + bool needs_word_mode; + + if (btf_is_enum(t)) { + /* enum can't represent 64-bit values, so we need word mode */ + needs_word_mode = true; + } else { + /* enum64 needs mode(word) if none of its values has + * non-zero upper 32-bits (which means that all values + * fit in 32-bit integers and won't cause compiler to + * bump enum to be 64-bit naturally + */ + int i; + + needs_word_mode = true; + for (i = 0; i < vlen; i++) { + if (btf_enum64(t)[i].val_hi32 != 0) { + needs_word_mode = false; + break; + } + } + } + if (needs_word_mode) + btf_dump_printf(d, " __attribute__((mode(word)))"); + } + } static void btf_dump_emit_fwd_def(struct btf_dump *d, __u32 id, -- cgit From 25a4481b4136af7794e1df2d6c90ed2f354d60ce Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Mon, 12 Dec 2022 13:15:03 -0800 Subject: libbpf: Fix btf__align_of() by taking into account field offsets btf__align_of() is supposed to be return alignment requirement of a requested BTF type. For STRUCT/UNION it doesn't always return correct value, because it calculates alignment only based on field types. But for packed structs this is not enough, we need to also check field offsets and struct size. If field offset isn't aligned according to field type's natural alignment, then struct must be packed. Similarly, if struct size is not a multiple of struct's natural alignment, then struct must be packed as well. This patch fixes this issue precisely by additionally checking these conditions. Fixes: 3d208f4ca111 ("libbpf: Expose btf__align_of() API") Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20221212211505.558851-5-andrii@kernel.org --- tools/lib/bpf/btf.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index 71e165b09ed5..8cbcef959456 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -688,8 +688,21 @@ int btf__align_of(const struct btf *btf, __u32 id) if (align <= 0) return libbpf_err(align); max_align = max(max_align, align); + + /* if field offset isn't aligned according to field + * type's alignment, then struct must be packed + */ + if (btf_member_bitfield_size(t, i) == 0 && + (m->offset % (8 * align)) != 0) + return 1; } + /* if struct/union size isn't a multiple of its alignment, + * then struct must be packed + */ + if ((t->size % max_align) != 0) + return 1; + return max_align; } default: -- cgit From ea2ce1ba99aa6a60c8d8a706e3abadf3de372163 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Mon, 12 Dec 2022 13:15:04 -0800 Subject: libbpf: Fix BTF-to-C converter's padding logic MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Turns out that btf_dump API doesn't handle a bunch of tricky corner cases, as reported by Per, and further discovered using his testing Python script ([0]). This patch revamps btf_dump's padding logic significantly, making it more correct and also avoiding unnecessary explicit padding, where compiler would pad naturally. This overall topic turned out to be very tricky and subtle, there are lots of subtle corner cases. The comments in the code tries to give some clues, but comments themselves are supposed to be paired with good understanding of C alignment and padding rules. Plus some experimentation to figure out subtle things like whether `long :0;` means that struct is now forced to be long-aligned (no, it's not, turns out). Anyways, Per's script, while not completely correct in some known situations, doesn't show any obvious cases where this logic breaks, so this is a nice improvement over the previous state of this logic. Some selftests had to be adjusted to accommodate better use of natural alignment rules, eliminating some unnecessary padding, or changing it to `type: 0;` alignment markers. Note also that for when we are in between bitfields, we emit explicit bit size, while otherwise we use `: 0`, this feels much more natural in practice. Next patch will add few more test cases, found through randomized Per's script. [0] https://lore.kernel.org/bpf/85f83c333f5355c8ac026f835b18d15060725fcb.camel@ericsson.com/ Reported-by: Per Sundström XP Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20221212211505.558851-6-andrii@kernel.org --- tools/lib/bpf/btf_dump.c | 169 ++++++++++++++++++++++++++++++++++------------- 1 file changed, 123 insertions(+), 46 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c index 234e82334d56..d6fd93a57f11 100644 --- a/tools/lib/bpf/btf_dump.c +++ b/tools/lib/bpf/btf_dump.c @@ -830,6 +830,25 @@ static void btf_dump_emit_type(struct btf_dump *d, __u32 id, __u32 cont_id) } } +static int btf_natural_align_of(const struct btf *btf, __u32 id) +{ + const struct btf_type *t = btf__type_by_id(btf, id); + int i, align, vlen; + const struct btf_member *m; + + if (!btf_is_composite(t)) + return btf__align_of(btf, id); + + align = 1; + m = btf_members(t); + vlen = btf_vlen(t); + for (i = 0; i < vlen; i++, m++) { + align = max(align, btf__align_of(btf, m->type)); + } + + return align; +} + static bool btf_is_struct_packed(const struct btf *btf, __u32 id, const struct btf_type *t) { @@ -837,16 +856,16 @@ static bool btf_is_struct_packed(const struct btf *btf, __u32 id, int align, i, bit_sz; __u16 vlen; - align = btf__align_of(btf, id); - /* size of a non-packed struct has to be a multiple of its alignment*/ - if (align && t->size % align) + align = btf_natural_align_of(btf, id); + /* size of a non-packed struct has to be a multiple of its alignment */ + if (align && (t->size % align) != 0) return true; m = btf_members(t); vlen = btf_vlen(t); /* all non-bitfield fields have to be naturally aligned */ for (i = 0; i < vlen; i++, m++) { - align = btf__align_of(btf, m->type); + align = btf_natural_align_of(btf, m->type); bit_sz = btf_member_bitfield_size(t, i); if (align && bit_sz == 0 && m->offset % (8 * align) != 0) return true; @@ -859,44 +878,97 @@ static bool btf_is_struct_packed(const struct btf *btf, __u32 id, return false; } -static int chip_away_bits(int total, int at_most) -{ - return total % at_most ? : at_most; -} - static void btf_dump_emit_bit_padding(const struct btf_dump *d, - int cur_off, int m_off, int m_bit_sz, - int align, int lvl) + int cur_off, int next_off, int next_align, + bool in_bitfield, int lvl) { - int off_diff = m_off - cur_off; - int ptr_bits = d->ptr_sz * 8; + const struct { + const char *name; + int bits; + } pads[] = { + {"long", d->ptr_sz * 8}, {"int", 32}, {"short", 16}, {"char", 8} + }; + int new_off, pad_bits, bits, i; + const char *pad_type; + + if (cur_off >= next_off) + return; /* no gap */ + + /* For filling out padding we want to take advantage of + * natural alignment rules to minimize unnecessary explicit + * padding. First, we find the largest type (among long, int, + * short, or char) that can be used to force naturally aligned + * boundary. Once determined, we'll use such type to fill in + * the remaining padding gap. In some cases we can rely on + * compiler filling some gaps, but sometimes we need to force + * alignment to close natural alignment with markers like + * `long: 0` (this is always the case for bitfields). Note + * that even if struct itself has, let's say 4-byte alignment + * (i.e., it only uses up to int-aligned types), using `long: + * X;` explicit padding doesn't actually change struct's + * overall alignment requirements, but compiler does take into + * account that type's (long, in this example) natural + * alignment requirements when adding implicit padding. We use + * this fact heavily and don't worry about ruining correct + * struct alignment requirement. + */ + for (i = 0; i < ARRAY_SIZE(pads); i++) { + pad_bits = pads[i].bits; + pad_type = pads[i].name; - if (off_diff <= 0) - /* no gap */ - return; - if (m_bit_sz == 0 && off_diff < align * 8) - /* natural padding will take care of a gap */ - return; + new_off = roundup(cur_off, pad_bits); + if (new_off <= next_off) + break; + } - while (off_diff > 0) { - const char *pad_type; - int pad_bits; - - if (ptr_bits > 32 && off_diff > 32) { - pad_type = "long"; - pad_bits = chip_away_bits(off_diff, ptr_bits); - } else if (off_diff > 16) { - pad_type = "int"; - pad_bits = chip_away_bits(off_diff, 32); - } else if (off_diff > 8) { - pad_type = "short"; - pad_bits = chip_away_bits(off_diff, 16); - } else { - pad_type = "char"; - pad_bits = chip_away_bits(off_diff, 8); + if (new_off > cur_off && new_off <= next_off) { + /* We need explicit `: 0` aligning mark if next + * field is right on alignment offset and its + * alignment requirement is less strict than 's + * alignment (so compiler won't naturally align to the + * offset we expect), or if subsequent `: X`, + * will actually completely fit in the remaining hole, + * making compiler basically ignore `: X` + * completely. + */ + if (in_bitfield || + (new_off == next_off && roundup(cur_off, next_align * 8) != new_off) || + (new_off != next_off && next_off - new_off <= new_off - cur_off)) + /* but for bitfields we'll emit explicit bit count */ + btf_dump_printf(d, "\n%s%s: %d;", pfx(lvl), pad_type, + in_bitfield ? new_off - cur_off : 0); + cur_off = new_off; + } + + /* Now we know we start at naturally aligned offset for a chosen + * padding type (long, int, short, or char), and so the rest is just + * a straightforward filling of remaining padding gap with full + * `: sizeof();` markers, except for the last one, which + * might need smaller than sizeof() padding. + */ + while (cur_off != next_off) { + bits = min(next_off - cur_off, pad_bits); + if (bits == pad_bits) { + btf_dump_printf(d, "\n%s%s: %d;", pfx(lvl), pad_type, pad_bits); + cur_off += bits; + continue; + } + /* For the remainder padding that doesn't cover entire + * pad_type bit length, we pick the smallest necessary type. + * This is pure aesthetics, we could have just used `long`, + * but having smallest necessary one communicates better the + * scale of the padding gap. + */ + for (i = ARRAY_SIZE(pads) - 1; i >= 0; i--) { + pad_type = pads[i].name; + pad_bits = pads[i].bits; + if (pad_bits < bits) + continue; + + btf_dump_printf(d, "\n%s%s: %d;", pfx(lvl), pad_type, bits); + cur_off += bits; + break; } - btf_dump_printf(d, "\n%s%s: %d;", pfx(lvl), pad_type, pad_bits); - off_diff -= pad_bits; } } @@ -916,9 +988,11 @@ static void btf_dump_emit_struct_def(struct btf_dump *d, { const struct btf_member *m = btf_members(t); bool is_struct = btf_is_struct(t); - int align, i, packed, off = 0; + bool packed, prev_bitfield = false; + int align, i, off = 0; __u16 vlen = btf_vlen(t); + align = btf__align_of(d->btf, id); packed = is_struct ? btf_is_struct_packed(d->btf, id, t) : 0; btf_dump_printf(d, "%s%s%s {", @@ -928,33 +1002,36 @@ static void btf_dump_emit_struct_def(struct btf_dump *d, for (i = 0; i < vlen; i++, m++) { const char *fname; - int m_off, m_sz; + int m_off, m_sz, m_align; + bool in_bitfield; fname = btf_name_of(d, m->name_off); m_sz = btf_member_bitfield_size(t, i); m_off = btf_member_bit_offset(t, i); - align = packed ? 1 : btf__align_of(d->btf, m->type); + m_align = packed ? 1 : btf__align_of(d->btf, m->type); + + in_bitfield = prev_bitfield && m_sz != 0; - btf_dump_emit_bit_padding(d, off, m_off, m_sz, align, lvl + 1); + btf_dump_emit_bit_padding(d, off, m_off, m_align, in_bitfield, lvl + 1); btf_dump_printf(d, "\n%s", pfx(lvl + 1)); btf_dump_emit_type_decl(d, m->type, fname, lvl + 1); if (m_sz) { btf_dump_printf(d, ": %d", m_sz); off = m_off + m_sz; + prev_bitfield = true; } else { m_sz = max((__s64)0, btf__resolve_size(d->btf, m->type)); off = m_off + m_sz * 8; + prev_bitfield = false; } + btf_dump_printf(d, ";"); } /* pad at the end, if necessary */ - if (is_struct) { - align = packed ? 1 : btf__align_of(d->btf, id); - btf_dump_emit_bit_padding(d, off, t->size * 8, 0, align, - lvl + 1); - } + if (is_struct) + btf_dump_emit_bit_padding(d, off, t->size * 8, align, false, lvl + 1); /* * Keep `struct empty {}` on a single line, -- cgit From 4fb877aaa179dcdb1676d55216482febaada457e Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Thu, 15 Dec 2022 10:36:05 -0800 Subject: libbpf: Fix btf_dump's packed struct determination Fix bug in btf_dump's logic of determining if a given struct type is packed or not. The notion of "natural alignment" is not needed and is even harmful in this case, so drop it altogether. The biggest difference in btf_is_struct_packed() compared to its original implementation is that we don't really use btf__align_of() to determine overall alignment of a struct type (because it could be 1 for both packed and non-packed struct, depending on specifci field definitions), and just use field's actual alignment to calculate whether any field is requiring packing or struct's size overall necessitates packing. Add two simple test cases that demonstrate the difference this change would make. Fixes: ea2ce1ba99aa ("libbpf: Fix BTF-to-C converter's padding logic") Reported-by: Eduard Zingerman Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Acked-by: Eduard Zingerman Link: https://lore.kernel.org/bpf/20221215183605.4149488-1-andrii@kernel.org --- tools/lib/bpf/btf_dump.c | 33 ++++++--------------------------- 1 file changed, 6 insertions(+), 27 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c index d6fd93a57f11..580985ee5545 100644 --- a/tools/lib/bpf/btf_dump.c +++ b/tools/lib/bpf/btf_dump.c @@ -830,47 +830,26 @@ static void btf_dump_emit_type(struct btf_dump *d, __u32 id, __u32 cont_id) } } -static int btf_natural_align_of(const struct btf *btf, __u32 id) -{ - const struct btf_type *t = btf__type_by_id(btf, id); - int i, align, vlen; - const struct btf_member *m; - - if (!btf_is_composite(t)) - return btf__align_of(btf, id); - - align = 1; - m = btf_members(t); - vlen = btf_vlen(t); - for (i = 0; i < vlen; i++, m++) { - align = max(align, btf__align_of(btf, m->type)); - } - - return align; -} - static bool btf_is_struct_packed(const struct btf *btf, __u32 id, const struct btf_type *t) { const struct btf_member *m; - int align, i, bit_sz; + int max_align = 1, align, i, bit_sz; __u16 vlen; - align = btf_natural_align_of(btf, id); - /* size of a non-packed struct has to be a multiple of its alignment */ - if (align && (t->size % align) != 0) - return true; - m = btf_members(t); vlen = btf_vlen(t); /* all non-bitfield fields have to be naturally aligned */ for (i = 0; i < vlen; i++, m++) { - align = btf_natural_align_of(btf, m->type); + align = btf__align_of(btf, m->type); bit_sz = btf_member_bitfield_size(t, i); if (align && bit_sz == 0 && m->offset % (8 * align) != 0) return true; + max_align = max(align, max_align); } - + /* size of a non-packed struct has to be a multiple of its alignment */ + if (t->size % max_align != 0) + return true; /* * if original struct was marked as packed, but its layout is * naturally aligned, we'll detect that it's not packed -- cgit From 1520e8466d683b6c5e1aa53aa65165ebd5da46cf Mon Sep 17 00:00:00 2001 From: Khem Raj Date: Mon, 19 Dec 2022 11:15:26 -0800 Subject: libbpf: Fix build warning on ref_ctr_off for 32-bit architectures Clang warns on 32-bit ARM on this comparision: libbpf.c:10497:18: error: result of comparison of constant 4294967296 with expression of type 'size_t' (aka 'unsigned int') is always false [-Werror,-Wtautological-constant-out-of-range-compare] if (ref_ctr_off >= (1ULL << PERF_UPROBE_REF_CTR_OFFSET_BITS)) ~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Typecast ref_ctr_off to __u64 in the check conditional, it is false on 32bit anyways. Signed-off-by: Khem Raj Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20221219191526.296264-1-raj.khem@gmail.com --- tools/lib/bpf/libbpf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 2a82f49ce16f..a5c67a3c93c5 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -9903,7 +9903,7 @@ static int perf_event_open_probe(bool uprobe, bool retprobe, const char *name, char errmsg[STRERR_BUFSIZE]; int type, pfd; - if (ref_ctr_off >= (1ULL << PERF_UPROBE_REF_CTR_OFFSET_BITS)) + if ((__u64)ref_ctr_off >= (1ULL << PERF_UPROBE_REF_CTR_OFFSET_BITS)) return -EINVAL; memset(&attr, 0, attr_sz); -- cgit From e6b4e1d759d3bfb7cb84c87cc8f1858da7db8dea Mon Sep 17 00:00:00 2001 From: Changbin Du Date: Sun, 18 Dec 2022 06:35:08 +0800 Subject: libbpf: Show error info about missing ".BTF" section Show the real problem instead of just saying "No such file or directory". Now will print below info: libbpf: failed to find '.BTF' ELF section in /home/changbin/work/linux/vmlinux Error: failed to load BTF from /home/changbin/work/linux/vmlinux: No such file or directory Signed-off-by: Changbin Du Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20221217223509.88254-2-changbin.du@gmail.com --- tools/lib/bpf/btf.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index 8cbcef959456..b03250007018 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -1003,6 +1003,7 @@ static struct btf *btf_parse_elf(const char *path, struct btf *base_btf, err = 0; if (!btf_data) { + pr_warn("failed to find '%s' ELF section in %s\n", BTF_ELF_SEC, path); err = -ENOENT; goto done; } -- cgit From 4ec38eda85b9d07942a1cc7a29bba8aa7c810780 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Wed, 21 Dec 2022 10:00:49 -0800 Subject: libbpf: start v1.2 development cycle Bump current version for new development cycle to v1.2. Signed-off-by: Andrii Nakryiko Acked-by: Stanislav Fomichev Link: https://lore.kernel.org/r/20221221180049.853365-1-andrii@kernel.org Signed-off-by: Martin KaFai Lau --- tools/lib/bpf/libbpf.map | 3 +++ tools/lib/bpf/libbpf_version.h | 2 +- 2 files changed, 4 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map index 71bf5691a689..11c36a3c1a9f 100644 --- a/tools/lib/bpf/libbpf.map +++ b/tools/lib/bpf/libbpf.map @@ -382,3 +382,6 @@ LIBBPF_1.1.0 { user_ring_buffer__reserve_blocking; user_ring_buffer__submit; } LIBBPF_1.0.0; + +LIBBPF_1.2.0 { +} LIBBPF_1.1.0; diff --git a/tools/lib/bpf/libbpf_version.h b/tools/lib/bpf/libbpf_version.h index e944f5bce728..1fd2eeac5cfc 100644 --- a/tools/lib/bpf/libbpf_version.h +++ b/tools/lib/bpf/libbpf_version.h @@ -4,6 +4,6 @@ #define __LIBBPF_VERSION_H #define LIBBPF_MAJOR_VERSION 1 -#define LIBBPF_MINOR_VERSION 1 +#define LIBBPF_MINOR_VERSION 2 #endif /* __LIBBPF_VERSION_H */ -- cgit From 07453245620c075779abefa2a9f469fa336e7510 Mon Sep 17 00:00:00 2001 From: Xin Liu Date: Fri, 23 Dec 2022 21:36:18 +0800 Subject: libbpf: fix errno is overwritten after being closed. In the ensure_good_fd function, if the fcntl function succeeds but the close function fails, ensure_good_fd returns a normal fd and sets errno, which may cause users to misunderstand. The close failure is not a serious problem, and the correct FD has been handed over to the upper-layer application. Let's restore errno here. Signed-off-by: Xin Liu Link: https://lore.kernel.org/r/20221223133618.10323-1-liuxin350@huawei.com Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/libbpf_internal.h | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf_internal.h b/tools/lib/bpf/libbpf_internal.h index 377642ff51fc..98333a6c38e9 100644 --- a/tools/lib/bpf/libbpf_internal.h +++ b/tools/lib/bpf/libbpf_internal.h @@ -543,10 +543,9 @@ static inline int ensure_good_fd(int fd) fd = fcntl(fd, F_DUPFD_CLOEXEC, 3); saved_errno = errno; close(old_fd); - if (fd < 0) { + errno = saved_errno; + if (fd < 0) pr_warn("failed to dup FD %d to FD > 2: %d\n", old_fd, -saved_errno); - errno = saved_errno; - } } return fd; } -- cgit From 7244eb669397f309c3d014264823cdc9cb3f8e6b Mon Sep 17 00:00:00 2001 From: "Daniel T. Lee" Date: Sat, 24 Dec 2022 16:15:27 +0900 Subject: libbpf: Fix invalid return address register in s390 There is currently an invalid register mapping in the s390 return address register. As the manual[1] states, the return address can be found at r14. In bpf_tracing.h, the s390 registers were named gprs(general purpose registers). This commit fixes the problem by correcting the mistyped mapping. [1]: https://uclibc.org/docs/psABI-s390x.pdf#page=14 Fixes: 3cc31d794097 ("libbpf: Normalize PT_REGS_xxx() macro definitions") Signed-off-by: Daniel T. Lee Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20221224071527.2292-7-danieltimlee@gmail.com --- tools/lib/bpf/bpf_tracing.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 2972dc25ff72..9c1b1689068d 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -137,7 +137,7 @@ struct pt_regs___s390 { #define __PT_PARM3_REG gprs[4] #define __PT_PARM4_REG gprs[5] #define __PT_PARM5_REG gprs[6] -#define __PT_RET_REG grps[14] +#define __PT_RET_REG gprs[14] #define __PT_FP_REG gprs[11] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG gprs[2] #define __PT_SP_REG gprs[15] -- cgit From 678a1c036199011e65b203367900f002a28da004 Mon Sep 17 00:00:00 2001 From: Xin Liu Date: Sat, 24 Dec 2022 19:20:58 +0800 Subject: libbpf: Added the description of some API functions Currently, many API functions are not described in the document. Add add API description of the following four API functions: - libbpf_set_print; - bpf_object__open; - bpf_object__load; - bpf_object__close. Signed-off-by: Xin Liu Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20221224112058.12038-1-liuxin350@huawei.com --- tools/lib/bpf/libbpf.h | 29 +++++++++++++++++++++++++++-- 1 file changed, 27 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index eee883f007f9..898db26e42e9 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -96,6 +96,12 @@ enum libbpf_print_level { typedef int (*libbpf_print_fn_t)(enum libbpf_print_level level, const char *, va_list ap); +/** + * @brief **libbpf_set_print()** sets user-provided log callback function to + * be used for libbpf warnings and informational messages. + * @param fn The log print function. If NULL, libbpf won't print anything. + * @return Pointer to old print function. + */ LIBBPF_API libbpf_print_fn_t libbpf_set_print(libbpf_print_fn_t fn); /* Hide internal to user */ @@ -174,6 +180,14 @@ struct bpf_object_open_opts { }; #define bpf_object_open_opts__last_field kernel_log_level +/** + * @brief **bpf_object__open()** creates a bpf_object by opening + * the BPF ELF object file pointed to by the passed path and loading it + * into memory. + * @param path BPF object file path. + * @return pointer to the new bpf_object; or NULL is returned on error, + * error code is stored in errno + */ LIBBPF_API struct bpf_object *bpf_object__open(const char *path); /** @@ -203,10 +217,21 @@ LIBBPF_API struct bpf_object * bpf_object__open_mem(const void *obj_buf, size_t obj_buf_sz, const struct bpf_object_open_opts *opts); -/* Load/unload object into/from kernel */ +/** + * @brief **bpf_object__load()** loads BPF object into kernel. + * @param obj Pointer to a valid BPF object instance returned by + * **bpf_object__open*()** APIs + * @return 0, on success; negative error code, otherwise, error code is + * stored in errno + */ LIBBPF_API int bpf_object__load(struct bpf_object *obj); -LIBBPF_API void bpf_object__close(struct bpf_object *object); +/** + * @brief **bpf_object__close()** closes a BPF object and releases all + * resources. + * @param obj Pointer to a valid BPF object + */ +LIBBPF_API void bpf_object__close(struct bpf_object *obj); /* pin_maps and unpin_maps can both be called with a NULL path, in which case * they will use the pin_path attribute of each map (and ignore all maps that -- cgit From bb5747cfbc4b7fe29621ca6cd4a695d2723bf2e8 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Thu, 29 Dec 2022 19:12:17 -0800 Subject: libbpf: Restore errno after pr_warn. pr_warn calls into user-provided callback, which can clobber errno, so `errno = saved_errno` should happen after pr_warn. Fixes: 07453245620c ("libbpf: fix errno is overwritten after being closed.") Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/libbpf_internal.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf_internal.h b/tools/lib/bpf/libbpf_internal.h index 98333a6c38e9..e4d05662a96c 100644 --- a/tools/lib/bpf/libbpf_internal.h +++ b/tools/lib/bpf/libbpf_internal.h @@ -544,8 +544,10 @@ static inline int ensure_good_fd(int fd) saved_errno = errno; close(old_fd); errno = saved_errno; - if (fd < 0) + if (fd < 0) { pr_warn("failed to dup FD %d to FD > 2: %d\n", old_fd, -saved_errno); + errno = saved_errno; + } } return fd; } -- cgit From 00883922ab404c0fc921709d8c2cec86f49c32f2 Mon Sep 17 00:00:00 2001 From: Hengqi Chen Date: Sat, 31 Dec 2022 18:07:57 +0800 Subject: libbpf: Add LoongArch support to bpf_tracing.h Add PT_REGS macros for LoongArch ([0]). [0]: https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html Signed-off-by: Hengqi Chen Signed-off-by: Daniel Borkmann Acked-by: Huacai Chen Link: https://lore.kernel.org/bpf/20221231100757.3177034-1-hengqi.chen@gmail.com --- tools/lib/bpf/bpf_tracing.h | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 9c1b1689068d..bdb0f6b5be84 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -32,6 +32,9 @@ #elif defined(__TARGET_ARCH_arc) #define bpf_target_arc #define bpf_target_defined +#elif defined(__TARGET_ARCH_loongarch) + #define bpf_target_loongarch + #define bpf_target_defined #else /* Fall back to what the compiler says */ @@ -62,6 +65,9 @@ #elif defined(__arc__) #define bpf_target_arc #define bpf_target_defined +#elif defined(__loongarch__) + #define bpf_target_loongarch + #define bpf_target_defined #endif /* no compiler target */ #endif @@ -258,6 +264,23 @@ struct pt_regs___arm64 { /* arc does not select ARCH_HAS_SYSCALL_WRAPPER. */ #define PT_REGS_SYSCALL_REGS(ctx) ctx +#elif defined(bpf_target_loongarch) + +/* https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html */ + +#define __PT_PARM1_REG regs[4] +#define __PT_PARM2_REG regs[5] +#define __PT_PARM3_REG regs[6] +#define __PT_PARM4_REG regs[7] +#define __PT_PARM5_REG regs[8] +#define __PT_RET_REG regs[1] +#define __PT_FP_REG regs[22] +#define __PT_RC_REG regs[4] +#define __PT_SP_REG regs[3] +#define __PT_IP_REG csr_era +/* loongarch does not select ARCH_HAS_SYSCALL_WRAPPER. */ +#define PT_REGS_SYSCALL_REGS(ctx) ctx + #endif #if defined(bpf_target_defined) -- cgit From acd3b7768048fe338248cdf43ccfbf8c084a6bc1 Mon Sep 17 00:00:00 2001 From: Changbin Du Date: Sat, 31 Dec 2022 23:14:36 +0800 Subject: libbpf: Return -ENODATA for missing btf section As discussed before, return -ENODATA (No data available) would be more meaningful than ENOENT (No such file or directory). Suggested-by: Leo Yan Signed-off-by: Changbin Du Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20221231151436.6541-1-changbin.du@gmail.com --- tools/lib/bpf/btf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index b03250007018..64841117fbb2 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -1004,7 +1004,7 @@ static struct btf *btf_parse_elf(const char *path, struct btf *base_btf, if (!btf_data) { pr_warn("failed to find '%s' ELF section in %s\n", BTF_ELF_SEC, path); - err = -ENOENT; + err = -ENODATA; goto done; } btf = btf_new(btf_data->d_buf, btf_data->d_size, base_btf); -- cgit From 6d0c4b11e743b84eb57898783e9adecb9083b269 Mon Sep 17 00:00:00 2001 From: Rong Tao Date: Thu, 5 Jan 2023 22:36:33 +0800 Subject: libbpf: Poison strlcpy() Since commit 9fc205b413b3("libbpf: Add sane strncpy alternative and use it internally") introduce libbpf_strlcpy(), thus add strlcpy() to a poison list to prevent accidental use of it. Signed-off-by: Rong Tao Signed-off-by: Daniel Borkmann Acked-by: Stanislav Fomichev Link: https://lore.kernel.org/bpf/tencent_5695A257C4D16B4413036BA1DAACDECB0B07@qq.com --- tools/lib/bpf/libbpf_internal.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf_internal.h b/tools/lib/bpf/libbpf_internal.h index e4d05662a96c..fbaf68335394 100644 --- a/tools/lib/bpf/libbpf_internal.h +++ b/tools/lib/bpf/libbpf_internal.h @@ -20,8 +20,8 @@ /* make sure libbpf doesn't use kernel-only integer typedefs */ #pragma GCC poison u8 u16 u32 u64 s8 s16 s32 s64 -/* prevent accidental re-addition of reallocarray() */ -#pragma GCC poison reallocarray +/* prevent accidental re-addition of reallocarray()/strlcpy() */ +#pragma GCC poison reallocarray strlcpy #include "libbpf.h" #include "btf.h" -- cgit From 6920b08661e3ad829206078b5c9879b24aea8dfc Mon Sep 17 00:00:00 2001 From: Ludovic L'Hours Date: Sun, 8 Jan 2023 19:20:18 +0100 Subject: libbpf: Fix map creation flags sanitization As BPF_F_MMAPABLE flag is now conditionnaly set (by map_is_mmapable), it should not be toggled but disabled if not supported by kernel. Fixes: 4fcac46c7e10 ("libbpf: only add BPF_F_MMAPABLE flag for data maps with global vars") Signed-off-by: Ludovic L'Hours Acked-by: Jiri Olsa Link: https://lore.kernel.org/r/20230108182018.24433-1-ludovic.lhours@gmail.com Signed-off-by: Martin KaFai Lau --- tools/lib/bpf/libbpf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index a5c67a3c93c5..f8dfee32c2bc 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -7355,7 +7355,7 @@ static int bpf_object__sanitize_maps(struct bpf_object *obj) if (!bpf_map__is_internal(m)) continue; if (!kernel_supports(obj, FEAT_ARRAY_MMAP)) - m->def.map_flags ^= BPF_F_MMAPABLE; + m->def.map_flags &= ~BPF_F_MMAPABLE; } return 0; -- cgit From 2fa074536590a82ee6c670d5f119f3dfb4ca6015 Mon Sep 17 00:00:00 2001 From: Menglong Dong Date: Fri, 13 Jan 2023 17:34:27 +0800 Subject: libbpf: Replace '.' with '_' in legacy kprobe event name '.' is not allowed in the event name of kprobe. Therefore, we will get a EINVAL if the kernel function name has a '.' in legacy kprobe attach case, such as 'icmp_reply.constprop.0'. In order to adapt this case, we need to replace the '.' with other char in gen_kprobe_legacy_event_name(). And I use '_' for this propose. Signed-off-by: Menglong Dong Signed-off-by: Andrii Nakryiko Reviewed-by: Alan Maguire Link: https://lore.kernel.org/bpf/20230113093427.1666466-1-imagedong@tencent.com --- tools/lib/bpf/libbpf.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index f8dfee32c2bc..27d9faa80471 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -9994,9 +9994,16 @@ static void gen_kprobe_legacy_event_name(char *buf, size_t buf_sz, const char *kfunc_name, size_t offset) { static int index = 0; + int i; snprintf(buf, buf_sz, "libbpf_%u_%s_0x%zx_%d", getpid(), kfunc_name, offset, __sync_fetch_and_add(&index, 1)); + + /* sanitize binary_path in the probe name */ + for (i = 0; buf[i]; i++) { + if (!isalnum(buf[i])) + buf[i] = '_'; + } } static int add_kprobe_event_legacy(const char *probe_name, bool retprobe, -- cgit From 3c59623d82947c8bfcc2fbf34c5eab02493b3b28 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:50 -0800 Subject: libbpf: Add support for fetching up to 8 arguments in kprobes Add BPF_KPROBE() and PT_REGS_PARMx() support for up to 8 arguments, if target architecture supports this. Currently all architectures are limited to only 5 register-placed arguments, which is limiting even on x86-64. This patch adds generic macro machinery to support up to 8 arguments both when explicitly fetching it from pt_regs through PT_REGS_PARMx() macros, as well as more ergonomic access in BPF_KPROBE(). Also, for i386 architecture we now don't have to define fake PARM4 and PARM5 definitions, they will be generically substituted, just like for PARM6 through PARM8. Subsequent patches will fill out architecture-specific definitions, where appropriate. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Alan Maguire # arm64 Tested-by: Ilya Leoshkevich # s390x Link: https://lore.kernel.org/bpf/20230120200914.3008030-2-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 41 +++++++++++++++++++++++++++++++++++++---- 1 file changed, 37 insertions(+), 4 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index bdb0f6b5be84..ef17d8fdd9d6 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -98,12 +98,10 @@ #ifdef __i386__ +/* i386 kernel is built with -mregparm=3 */ #define __PT_PARM1_REG eax #define __PT_PARM2_REG edx #define __PT_PARM3_REG ecx -/* i386 kernel is built with -mregparm=3 */ -#define __PT_PARM4_REG __unsupported__ -#define __PT_PARM5_REG __unsupported__ #define __PT_RET_REG esp #define __PT_FP_REG ebp #define __PT_RC_REG eax @@ -287,16 +285,39 @@ struct pt_regs___arm64 { struct pt_regs; -/* allow some architecutres to override `struct pt_regs` */ +/* allow some architectures to override `struct pt_regs` */ #ifndef __PT_REGS_CAST #define __PT_REGS_CAST(x) (x) #endif +/* + * Different architectures support different number of arguments passed + * through registers. i386 supports just 3, some arches support up to 8. + */ +#ifndef __PT_PARM4_REG +#define __PT_PARM4_REG __unsupported__ +#endif +#ifndef __PT_PARM5_REG +#define __PT_PARM5_REG __unsupported__ +#endif +#ifndef __PT_PARM6_REG +#define __PT_PARM6_REG __unsupported__ +#endif +#ifndef __PT_PARM7_REG +#define __PT_PARM7_REG __unsupported__ +#endif +#ifndef __PT_PARM8_REG +#define __PT_PARM8_REG __unsupported__ +#endif + #define PT_REGS_PARM1(x) (__PT_REGS_CAST(x)->__PT_PARM1_REG) #define PT_REGS_PARM2(x) (__PT_REGS_CAST(x)->__PT_PARM2_REG) #define PT_REGS_PARM3(x) (__PT_REGS_CAST(x)->__PT_PARM3_REG) #define PT_REGS_PARM4(x) (__PT_REGS_CAST(x)->__PT_PARM4_REG) #define PT_REGS_PARM5(x) (__PT_REGS_CAST(x)->__PT_PARM5_REG) +#define PT_REGS_PARM6(x) (__PT_REGS_CAST(x)->__PT_PARM6_REG) +#define PT_REGS_PARM7(x) (__PT_REGS_CAST(x)->__PT_PARM7_REG) +#define PT_REGS_PARM8(x) (__PT_REGS_CAST(x)->__PT_PARM8_REG) #define PT_REGS_RET(x) (__PT_REGS_CAST(x)->__PT_RET_REG) #define PT_REGS_FP(x) (__PT_REGS_CAST(x)->__PT_FP_REG) #define PT_REGS_RC(x) (__PT_REGS_CAST(x)->__PT_RC_REG) @@ -308,6 +329,9 @@ struct pt_regs; #define PT_REGS_PARM3_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM3_REG) #define PT_REGS_PARM4_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM4_REG) #define PT_REGS_PARM5_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM5_REG) +#define PT_REGS_PARM6_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM6_REG) +#define PT_REGS_PARM7_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM7_REG) +#define PT_REGS_PARM8_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM8_REG) #define PT_REGS_RET_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_RET_REG) #define PT_REGS_FP_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_FP_REG) #define PT_REGS_RC_CORE(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_RC_REG) @@ -360,6 +384,9 @@ struct pt_regs; #define PT_REGS_PARM3(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM4(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM5(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM6(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM7(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM8(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_RET(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_FP(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_RC(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) @@ -371,6 +398,9 @@ struct pt_regs; #define PT_REGS_PARM3_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM4_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM5_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM6_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM7_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM8_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_RET_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_FP_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_RC_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) @@ -576,6 +606,9 @@ struct pt_regs; #define ___bpf_kprobe_args3(x, args...) ___bpf_kprobe_args2(args), (void *)PT_REGS_PARM3(ctx) #define ___bpf_kprobe_args4(x, args...) ___bpf_kprobe_args3(args), (void *)PT_REGS_PARM4(ctx) #define ___bpf_kprobe_args5(x, args...) ___bpf_kprobe_args4(args), (void *)PT_REGS_PARM5(ctx) +#define ___bpf_kprobe_args6(x, args...) ___bpf_kprobe_args5(args), (void *)PT_REGS_PARM6(ctx) +#define ___bpf_kprobe_args7(x, args...) ___bpf_kprobe_args6(args), (void *)PT_REGS_PARM7(ctx) +#define ___bpf_kprobe_args8(x, args...) ___bpf_kprobe_args7(args), (void *)PT_REGS_PARM8(ctx) #define ___bpf_kprobe_args(args...) ___bpf_apply(___bpf_kprobe_args, ___bpf_narg(args))(args) /* -- cgit From 013290329a0707c9900f86a743dce70d6c345330 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:51 -0800 Subject: libbpf: Add 6th argument support for x86-64 in bpf_tracing.h Add r9 as register containing 6th argument on x86-64 architecture, as per its ABI. Add also a link to a page describing ABI for easier future reference. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-3-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index ef17d8fdd9d6..a47504c2f3cb 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -78,6 +78,10 @@ #if defined(bpf_target_x86) +/* + * https://en.wikipedia.org/wiki/X86_calling_conventions#System_V_AMD64_ABI + */ + #if defined(__KERNEL__) || defined(__VMLINUX_H__) #define __PT_PARM1_REG di @@ -85,6 +89,7 @@ #define __PT_PARM3_REG dx #define __PT_PARM4_REG cx #define __PT_PARM5_REG r8 +#define __PT_PARM6_REG r9 #define __PT_RET_REG sp #define __PT_FP_REG bp #define __PT_RC_REG ax @@ -115,6 +120,7 @@ #define __PT_PARM3_REG rdx #define __PT_PARM4_REG rcx #define __PT_PARM5_REG r8 +#define __PT_PARM6_REG r9 #define __PT_RET_REG rsp #define __PT_FP_REG rbp #define __PT_RC_REG rax -- cgit From 1dac40ac8742122a5f12b08f8073d03d6915f9ef Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:52 -0800 Subject: libbpf: Fix arm and arm64 specs in bpf_tracing.h Remove invalid support for PARM5 on 32-bit arm, as per ABI. Add three more argument registers for arm64. Also leave links to ABI specs for future reference. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Alan Maguire # arm64 Link: https://lore.kernel.org/bpf/20230120200914.3008030-4-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index a47504c2f3cb..a263f600e309 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -157,11 +157,14 @@ struct pt_regs___s390 { #elif defined(bpf_target_arm) +/* + * https://github.com/ARM-software/abi-aa/blob/main/aapcs32/aapcs32.rst#machine-registers + */ + #define __PT_PARM1_REG uregs[0] #define __PT_PARM2_REG uregs[1] #define __PT_PARM3_REG uregs[2] #define __PT_PARM4_REG uregs[3] -#define __PT_PARM5_REG uregs[4] #define __PT_RET_REG uregs[14] #define __PT_FP_REG uregs[11] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG uregs[0] @@ -170,6 +173,10 @@ struct pt_regs___s390 { #elif defined(bpf_target_arm64) +/* + * https://github.com/ARM-software/abi-aa/blob/main/aapcs64/aapcs64.rst#machine-registers + */ + struct pt_regs___arm64 { unsigned long orig_x0; }; @@ -181,6 +188,9 @@ struct pt_regs___arm64 { #define __PT_PARM3_REG regs[2] #define __PT_PARM4_REG regs[3] #define __PT_PARM5_REG regs[4] +#define __PT_PARM6_REG regs[5] +#define __PT_PARM7_REG regs[6] +#define __PT_PARM8_REG regs[7] #define __PT_RET_REG regs[30] #define __PT_FP_REG regs[29] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG regs[0] -- cgit From 1222445a5bf622b37a9bb16c7ad0148e2a21eb0d Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:53 -0800 Subject: libbpf: Complete mips spec in bpf_tracing.h Add registers for PARM6 through PARM8. Add a link to an ABI. We don't distinguish between O32, N32, and N64, so document that we assume N64 right now. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-5-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index a263f600e309..f356955b059a 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -201,11 +201,19 @@ struct pt_regs___arm64 { #elif defined(bpf_target_mips) +/* + * N64 ABI is assumed right now. + * https://en.wikipedia.org/wiki/MIPS_architecture#Calling_conventions + */ + #define __PT_PARM1_REG regs[4] #define __PT_PARM2_REG regs[5] #define __PT_PARM3_REG regs[6] #define __PT_PARM4_REG regs[7] #define __PT_PARM5_REG regs[8] +#define __PT_PARM6_REG regs[9] +#define __PT_PARM7_REG regs[10] +#define __PT_PARM8_REG regs[11] #define __PT_RET_REG regs[31] #define __PT_FP_REG regs[30] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG regs[2] -- cgit From 2eb2be30b8dac137d9a21c6125a68993b9dc8a00 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:54 -0800 Subject: libbpf: Complete powerpc spec in bpf_tracing.h Add definitions of PARM6 through PARM8 for powerpc architecture. Add also a link to a functiona call sequence documentation for future reference. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-6-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index f356955b059a..1b9447f847b0 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -222,11 +222,19 @@ struct pt_regs___arm64 { #elif defined(bpf_target_powerpc) +/* + * http://refspecs.linux-foundation.org/elf/elfspec_ppc.pdf (page 3-14, + * section "Function Calling Sequence") + */ + #define __PT_PARM1_REG gpr[3] #define __PT_PARM2_REG gpr[4] #define __PT_PARM3_REG gpr[5] #define __PT_PARM4_REG gpr[6] #define __PT_PARM5_REG gpr[7] +#define __PT_PARM6_REG gpr[8] +#define __PT_PARM7_REG gpr[9] +#define __PT_PARM8_REG gpr[10] #define __PT_RET_REG regs[31] #define __PT_FP_REG __unsupported__ #define __PT_RC_REG gpr[3] -- cgit From 7f60f5d85e29441b297bf64f0236c70e624a9108 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:55 -0800 Subject: libbpf: Complete sparc spec in bpf_tracing.h Add PARM6 definition for sparc architecture. Leave a link to calling convention documentation. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-7-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 1b9447f847b0..0f9640ddbe1c 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -245,11 +245,16 @@ struct pt_regs___arm64 { #elif defined(bpf_target_sparc) +/* + * https://en.wikipedia.org/wiki/Calling_convention#SPARC + */ + #define __PT_PARM1_REG u_regs[UREG_I0] #define __PT_PARM2_REG u_regs[UREG_I1] #define __PT_PARM3_REG u_regs[UREG_I2] #define __PT_PARM4_REG u_regs[UREG_I3] #define __PT_PARM5_REG u_regs[UREG_I4] +#define __PT_PARM6_REG u_regs[UREG_I5] #define __PT_RET_REG u_regs[UREG_I7] #define __PT_FP_REG __unsupported__ #define __PT_RC_REG u_regs[UREG_I0] -- cgit From b13ed8ca7fba3d3e61621be021fdccd5c44b2634 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:56 -0800 Subject: libbpf: Complete riscv arch spec in bpf_tracing.h Add PARM6 through PARM8 definitions for RISC V (riscv) arch. Leave the link for ABI doc for future reference. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Pu Lehui # RISC-V Link: https://lore.kernel.org/bpf/20230120200914.3008030-8-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 0f9640ddbe1c..579e2f830532 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -268,12 +268,19 @@ struct pt_regs___arm64 { #elif defined(bpf_target_riscv) +/* + * https://github.com/riscv-non-isa/riscv-elf-psabi-doc/blob/master/riscv-cc.adoc#risc-v-calling-conventions + */ + #define __PT_REGS_CAST(x) ((const struct user_regs_struct *)(x)) #define __PT_PARM1_REG a0 #define __PT_PARM2_REG a1 #define __PT_PARM3_REG a2 #define __PT_PARM4_REG a3 #define __PT_PARM5_REG a4 +#define __PT_PARM6_REG a5 +#define __PT_PARM7_REG a6 +#define __PT_PARM8_REG a7 #define __PT_RET_REG ra #define __PT_FP_REG s0 #define __PT_RC_REG a0 -- cgit From 0ac086567916e189c7db71da7339e20b5747f8ac Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:57 -0800 Subject: libbpf: Fix and complete ARC spec in bpf_tracing.h Add PARM6 through PARM8 definitions. Also fix frame pointer (FP) register definition. Also leave a link to where to find ABI spec. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-9-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 579e2f830532..815c50ccedf4 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -291,6 +291,11 @@ struct pt_regs___arm64 { #elif defined(bpf_target_arc) +/* + * Section "Function Calling Sequence" (page 24): + * https://raw.githubusercontent.com/wiki/foss-for-synopsys-dwc-arc-processors/toolchain/files/ARCv2_ABI.pdf + */ + /* arc provides struct user_pt_regs instead of struct pt_regs to userspace */ #define __PT_REGS_CAST(x) ((const struct user_regs_struct *)(x)) #define __PT_PARM1_REG scratch.r0 @@ -298,8 +303,11 @@ struct pt_regs___arm64 { #define __PT_PARM3_REG scratch.r2 #define __PT_PARM4_REG scratch.r3 #define __PT_PARM5_REG scratch.r4 +#define __PT_PARM6_REG scratch.r5 +#define __PT_PARM7_REG scratch.r6 +#define __PT_PARM8_REG scratch.r7 #define __PT_RET_REG scratch.blink -#define __PT_FP_REG __unsupported__ +#define __PT_FP_REG scratch.fp #define __PT_RC_REG scratch.r0 #define __PT_SP_REG scratch.sp #define __PT_IP_REG scratch.ret -- cgit From 55ff00d5393b6d6f6cb90530aac546f5aa52456d Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:58 -0800 Subject: libbpf: Complete LoongArch (loongarch) spec in bpf_tracing.h Add PARM6 through PARM8 definitions. Add kernel docs link describing ABI for LoongArch. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-10-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 815c50ccedf4..510df96ff19d 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -316,13 +316,19 @@ struct pt_regs___arm64 { #elif defined(bpf_target_loongarch) -/* https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html */ +/* + * https://docs.kernel.org/loongarch/introduction.html + * https://loongson.github.io/LoongArch-Documentation/LoongArch-ELF-ABI-EN.html + */ #define __PT_PARM1_REG regs[4] #define __PT_PARM2_REG regs[5] #define __PT_PARM3_REG regs[6] #define __PT_PARM4_REG regs[7] #define __PT_PARM5_REG regs[8] +#define __PT_PARM6_REG regs[9] +#define __PT_PARM7_REG regs[10] +#define __PT_PARM8_REG regs[11] #define __PT_RET_REG regs[1] #define __PT_FP_REG regs[22] #define __PT_RC_REG regs[4] -- cgit From ac4afd6e6fa4d4f9a0ab6ca76f4953545e4da7e6 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:08:59 -0800 Subject: libbpf: Add BPF_UPROBE and BPF_URETPROBE macro aliases Add BPF_UPROBE and BPF_URETPROBE macros, aliased to BPF_KPROBE and BPF_KRETPROBE, respectively. This makes uprobe-based BPF program code much less confusing, especially to people new to tracing, at no cost in terms of maintainability. We'll use this macro in selftests in subsequent patch. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-11-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 510df96ff19d..9c4246fe8799 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -781,4 +781,11 @@ ____##name(struct pt_regs *ctx, ##args) #define BPF_KPROBE_SYSCALL BPF_KSYSCALL +/* BPF_UPROBE and BPF_URETPROBE are identical to BPF_KPROBE and BPF_KRETPROBE, + * but are named way less confusingly for SEC("uprobe") and SEC("uretprobe") + * use cases. + */ +#define BPF_UPROBE(name, args...) BPF_KPROBE(name, ##args) +#define BPF_URETPROBE(name, args...) BPF_KRETPROBE(name, ##args) + #endif -- cgit From 8ccabeef913311eb54599abc52a6bcfc3e7bf1d8 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:01 -0800 Subject: libbpf: Improve syscall tracing support in bpf_tracing.h Set up generic support in bpf_tracing.h for up to 7 syscall arguments tracing with BPF_KSYSCALL, which seems to be the limit according to syscall(2) manpage. Also change the way that syscall convention is specified to be more explicit. Subsequent patches will adjust and define proper per-architecture syscall conventions. __PT_PARM1_SYSCALL_REG through __PT_PARM6_SYSCALL_REG is added temporarily to keep everything working before each architecture has syscall reg tables defined. They will be removed afterwards. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Alan Maguire # arm64 Link: https://lore.kernel.org/bpf/20230120200914.3008030-13-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 71 ++++++++++++++++++++++++++++++++++++--------- 1 file changed, 58 insertions(+), 13 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 9c4246fe8799..e3bf510e1e55 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -367,6 +367,34 @@ struct pt_regs; #ifndef __PT_PARM8_REG #define __PT_PARM8_REG __unsupported__ #endif +/* + * Similarly, syscall-specific conventions might differ between function call + * conventions within each architecutre. All supported architectures pass + * either 6 or 7 syscall arguments in registers. + * + * See syscall(2) manpage for succinct table with information on each arch. + */ +#ifndef __PT_PARM1_SYSCALL_REG +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#endif +#ifndef __PT_PARM2_SYSCALL_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#endif +#ifndef __PT_PARM3_SYSCALL_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#endif +#ifndef __PT_PARM4_SYSCALL_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#endif +#ifndef __PT_PARM5_SYSCALL_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#endif +#ifndef __PT_PARM6_SYSCALL_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG +#endif +#ifndef __PT_PARM7_SYSCALL_REG +#define __PT_PARM7_SYSCALL_REG __unsupported__ +#endif #define PT_REGS_PARM1(x) (__PT_REGS_CAST(x)->__PT_PARM1_REG) #define PT_REGS_PARM2(x) (__PT_REGS_CAST(x)->__PT_PARM2_REG) @@ -416,24 +444,33 @@ struct pt_regs; #endif #ifndef PT_REGS_PARM1_SYSCALL -#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1(x) +#define PT_REGS_PARM1_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM1_SYSCALL_REG) +#define PT_REGS_PARM1_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM1_SYSCALL_REG) +#endif +#ifndef PT_REGS_PARM2_SYSCALL +#define PT_REGS_PARM2_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM2_SYSCALL_REG) +#define PT_REGS_PARM2_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM2_SYSCALL_REG) +#endif +#ifndef PT_REGS_PARM3_SYSCALL +#define PT_REGS_PARM3_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM3_SYSCALL_REG) +#define PT_REGS_PARM3_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM3_SYSCALL_REG) #endif -#define PT_REGS_PARM2_SYSCALL(x) PT_REGS_PARM2(x) -#define PT_REGS_PARM3_SYSCALL(x) PT_REGS_PARM3(x) #ifndef PT_REGS_PARM4_SYSCALL -#define PT_REGS_PARM4_SYSCALL(x) PT_REGS_PARM4(x) +#define PT_REGS_PARM4_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM4_SYSCALL_REG) +#define PT_REGS_PARM4_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM4_SYSCALL_REG) #endif -#define PT_REGS_PARM5_SYSCALL(x) PT_REGS_PARM5(x) - -#ifndef PT_REGS_PARM1_CORE_SYSCALL -#define PT_REGS_PARM1_CORE_SYSCALL(x) PT_REGS_PARM1_CORE(x) +#ifndef PT_REGS_PARM5_SYSCALL +#define PT_REGS_PARM5_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM5_SYSCALL_REG) +#define PT_REGS_PARM5_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM5_SYSCALL_REG) +#endif +#ifndef PT_REGS_PARM6_SYSCALL +#define PT_REGS_PARM6_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM6_SYSCALL_REG) +#define PT_REGS_PARM6_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM6_SYSCALL_REG) #endif -#define PT_REGS_PARM2_CORE_SYSCALL(x) PT_REGS_PARM2_CORE(x) -#define PT_REGS_PARM3_CORE_SYSCALL(x) PT_REGS_PARM3_CORE(x) -#ifndef PT_REGS_PARM4_CORE_SYSCALL -#define PT_REGS_PARM4_CORE_SYSCALL(x) PT_REGS_PARM4_CORE(x) +#ifndef PT_REGS_PARM7_SYSCALL +#define PT_REGS_PARM7_SYSCALL(x) (__PT_REGS_CAST(x)->__PT_PARM7_SYSCALL_REG) +#define PT_REGS_PARM7_CORE_SYSCALL(x) BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM7_SYSCALL_REG) #endif -#define PT_REGS_PARM5_CORE_SYSCALL(x) PT_REGS_PARM5_CORE(x) #else /* defined(bpf_target_defined) */ @@ -473,12 +510,16 @@ struct pt_regs; #define PT_REGS_PARM3_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM4_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM5_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM6_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM7_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM1_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM2_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM3_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM4_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #define PT_REGS_PARM5_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM6_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) +#define PT_REGS_PARM7_CORE_SYSCALL(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; }) #endif /* defined(bpf_target_defined) */ @@ -723,6 +764,8 @@ static __always_inline typeof(name(0)) ____##name(struct pt_regs *ctx, ##args) #define ___bpf_syscall_args3(x, args...) ___bpf_syscall_args2(args), (void *)PT_REGS_PARM3_SYSCALL(regs) #define ___bpf_syscall_args4(x, args...) ___bpf_syscall_args3(args), (void *)PT_REGS_PARM4_SYSCALL(regs) #define ___bpf_syscall_args5(x, args...) ___bpf_syscall_args4(args), (void *)PT_REGS_PARM5_SYSCALL(regs) +#define ___bpf_syscall_args6(x, args...) ___bpf_syscall_args5(args), (void *)PT_REGS_PARM6_SYSCALL(regs) +#define ___bpf_syscall_args7(x, args...) ___bpf_syscall_args6(args), (void *)PT_REGS_PARM7_SYSCALL(regs) #define ___bpf_syscall_args(args...) ___bpf_apply(___bpf_syscall_args, ___bpf_narg(args))(args) /* If kernel doesn't have CONFIG_ARCH_HAS_SYSCALL_WRAPPER, we have to BPF_CORE_READ from pt_regs */ @@ -732,6 +775,8 @@ static __always_inline typeof(name(0)) ____##name(struct pt_regs *ctx, ##args) #define ___bpf_syswrap_args3(x, args...) ___bpf_syswrap_args2(args), (void *)PT_REGS_PARM3_CORE_SYSCALL(regs) #define ___bpf_syswrap_args4(x, args...) ___bpf_syswrap_args3(args), (void *)PT_REGS_PARM4_CORE_SYSCALL(regs) #define ___bpf_syswrap_args5(x, args...) ___bpf_syswrap_args4(args), (void *)PT_REGS_PARM5_CORE_SYSCALL(regs) +#define ___bpf_syswrap_args6(x, args...) ___bpf_syswrap_args5(args), (void *)PT_REGS_PARM6_CORE_SYSCALL(regs) +#define ___bpf_syswrap_args7(x, args...) ___bpf_syswrap_args6(args), (void *)PT_REGS_PARM7_CORE_SYSCALL(regs) #define ___bpf_syswrap_args(args...) ___bpf_apply(___bpf_syswrap_args, ___bpf_narg(args))(args) /* -- cgit From d21fbceedd90ccc11775ca583889d21d3b66d88d Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:02 -0800 Subject: libbpf: Define x86-64 syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Remove now unnecessary overrides of PT_REGS_PARM5_[CORE_]SYSCALL macros. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-14-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 25 +++++++++++++++++++------ 1 file changed, 19 insertions(+), 6 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index e3bf510e1e55..5fd498821c29 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -90,14 +90,22 @@ #define __PT_PARM4_REG cx #define __PT_PARM5_REG r8 #define __PT_PARM6_REG r9 +/* + * Syscall uses r10 for PARM4. See arch/x86/entry/entry_64.S:entry_SYSCALL_64 + * comments in Linux sources. And refer to syscall(2) manpage. + */ +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG r10 +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG sp #define __PT_FP_REG bp #define __PT_RC_REG ax #define __PT_SP_REG sp #define __PT_IP_REG ip -/* syscall uses r10 for PARM4 */ -#define PT_REGS_PARM4_SYSCALL(x) ((x)->r10) -#define PT_REGS_PARM4_CORE_SYSCALL(x) BPF_CORE_READ(x, r10) #else @@ -121,14 +129,19 @@ #define __PT_PARM4_REG rcx #define __PT_PARM5_REG r8 #define __PT_PARM6_REG r9 + +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG r10 +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG rsp #define __PT_FP_REG rbp #define __PT_RC_REG rax #define __PT_SP_REG rsp #define __PT_IP_REG rip -/* syscall uses r10 for PARM4 */ -#define PT_REGS_PARM4_SYSCALL(x) ((x)->r10) -#define PT_REGS_PARM4_CORE_SYSCALL(x) BPF_CORE_READ(x, r10) #endif /* __i386__ */ -- cgit From ff00f9cbd2ddee60cd1542e58631638a5996b05f Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:03 -0800 Subject: libbpf: Define i386 syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-15-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 5fd498821c29..4d04c63fda27 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -115,6 +115,14 @@ #define __PT_PARM1_REG eax #define __PT_PARM2_REG edx #define __PT_PARM3_REG ecx +/* i386 syscall ABI is very different, refer to syscall(2) manpage */ +#define __PT_PARM1_SYSCALL_REG ebx +#define __PT_PARM2_SYSCALL_REG ecx +#define __PT_PARM3_SYSCALL_REG edx +#define __PT_PARM4_SYSCALL_REG esi +#define __PT_PARM5_SYSCALL_REG edi +#define __PT_PARM6_SYSCALL_REG ebp + #define __PT_RET_REG esp #define __PT_FP_REG ebp #define __PT_RC_REG eax -- cgit From e82b96a3a99f9a04566305490e88efef80a80247 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:04 -0800 Subject: libbpf: Define s390x syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Note that we need custom overrides for PT_REGS_PARM1_[CORE_]SYSCALL macros due to the need to use BPF CO-RE and custom local pt_regs definitions to fetch orig_gpr2, storing 1st argument. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Ilya Leoshkevich # s390x Acked-by: Ilya Leoshkevich Link: https://lore.kernel.org/bpf/20230120200914.3008030-16-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 4d04c63fda27..b63d538eb5f5 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -157,6 +157,10 @@ #elif defined(bpf_target_s390) +/* + * https://github.com/IBM/s390x-abi/releases/download/v1.6/lzsabi_s390x.pdf + */ + struct pt_regs___s390 { unsigned long orig_gpr2; }; @@ -168,13 +172,22 @@ struct pt_regs___s390 { #define __PT_PARM3_REG gprs[4] #define __PT_PARM4_REG gprs[5] #define __PT_PARM5_REG gprs[6] + +#define __PT_PARM1_SYSCALL_REG orig_gpr2 +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG gprs[7] +#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1_CORE_SYSCALL(x) +#define PT_REGS_PARM1_CORE_SYSCALL(x) \ + BPF_CORE_READ((const struct pt_regs___s390 *)(x), __PT_PARM1_SYSCALL_REG) + #define __PT_RET_REG gprs[14] #define __PT_FP_REG gprs[11] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG gprs[2] #define __PT_SP_REG gprs[15] #define __PT_IP_REG psw.addr -#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1_CORE_SYSCALL(x) -#define PT_REGS_PARM1_CORE_SYSCALL(x) BPF_CORE_READ((const struct pt_regs___s390 *)(x), orig_gpr2) #elif defined(bpf_target_arm) -- cgit From 3a95c42d65d5c8d836bc7f881b3ffd5229d1b73a Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:05 -0800 Subject: libbpf: Define arm syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-17-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index b63d538eb5f5..edd85a0ab612 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -199,6 +199,14 @@ struct pt_regs___s390 { #define __PT_PARM2_REG uregs[1] #define __PT_PARM3_REG uregs[2] #define __PT_PARM4_REG uregs[3] + +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM6_SYSCALL_REG uregs[5] +#define __PT_PARM7_SYSCALL_REG uregs[6] + #define __PT_RET_REG uregs[14] #define __PT_FP_REG uregs[11] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG uregs[0] -- cgit From 3488ea0584bb3888e2fcbe2d2ff738ff173d91ab Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:06 -0800 Subject: libbpf: Define arm64 syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. We need PT_REGS_PARM1_[CORE_]SYSCALL macros overrides, similarly to s390x, due to orig_x0 not being present in UAPI's pt_regs, so we need to utilize BPF CO-RE and custom pt_regs___arm64 definition. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Alan Maguire # arm64 Link: https://lore.kernel.org/bpf/20230120200914.3008030-18-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index edd85a0ab612..f753346e9be6 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -233,13 +233,22 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG regs[5] #define __PT_PARM7_REG regs[6] #define __PT_PARM8_REG regs[7] + +#define __PT_PARM1_SYSCALL_REG orig_x0 +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG +#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1_CORE_SYSCALL(x) +#define PT_REGS_PARM1_CORE_SYSCALL(x) \ + BPF_CORE_READ((const struct pt_regs___arm64 *)(x), __PT_PARM1_SYSCALL_REG) + #define __PT_RET_REG regs[30] #define __PT_FP_REG regs[29] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG regs[0] #define __PT_SP_REG sp #define __PT_IP_REG pc -#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1_CORE_SYSCALL(x) -#define PT_REGS_PARM1_CORE_SYSCALL(x) BPF_CORE_READ((const struct pt_regs___arm64 *)(x), orig_x0) #elif defined(bpf_target_mips) -- cgit From cfd0bbe91536095f5491e9ac71fb11304dd87dd2 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:07 -0800 Subject: libbpf: Define mips syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-19-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index f753346e9be6..a3da63b74b0b 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -265,6 +265,14 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG regs[9] #define __PT_PARM7_REG regs[10] #define __PT_PARM8_REG regs[11] + +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG /* only N32/N64 */ +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG /* only N32/N64 */ + #define __PT_RET_REG regs[31] #define __PT_FP_REG regs[30] /* Works only with CONFIG_FRAME_POINTER */ #define __PT_RC_REG regs[2] -- cgit From c1cc01a2d1d1d94079fa87ba44c34bd7bdf9175d Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:08 -0800 Subject: libbpf: Define powerpc syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Note that 7th arg is supported on 32-bit powerpc architecture, by not on powerpc64. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-20-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index a3da63b74b0b..b3647c4a6c0c 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -294,13 +294,24 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG gpr[8] #define __PT_PARM7_REG gpr[9] #define __PT_PARM8_REG gpr[10] + +/* powerpc does not select ARCH_HAS_SYSCALL_WRAPPER. */ +#define PT_REGS_SYSCALL_REGS(ctx) ctx +#define __PT_PARM1_SYSCALL_REG orig_gpr3 +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG +#if !defined(__arch64__) +#define __PT_PARM7_SYSCALL_REG __PT_PARM7_REG /* only powerpc (not powerpc64) */ +#endif + #define __PT_RET_REG regs[31] #define __PT_FP_REG __unsupported__ #define __PT_RC_REG gpr[3] #define __PT_SP_REG sp #define __PT_IP_REG nip -/* powerpc does not select ARCH_HAS_SYSCALL_WRAPPER. */ -#define PT_REGS_SYSCALL_REGS(ctx) ctx #elif defined(bpf_target_sparc) -- cgit From 377c15b1a2cd6e38db86be7a721c8311d8a73836 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:09 -0800 Subject: libbpf: Define sparc syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-21-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index b3647c4a6c0c..a2db267a6bce 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -325,6 +325,14 @@ struct pt_regs___arm64 { #define __PT_PARM4_REG u_regs[UREG_I3] #define __PT_PARM5_REG u_regs[UREG_I4] #define __PT_PARM6_REG u_regs[UREG_I5] + +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG u_regs[UREG_I7] #define __PT_FP_REG __unsupported__ #define __PT_RC_REG u_regs[UREG_I0] -- cgit From a0426216a32004bbb1c96dbffaeca7ac261960cf Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:10 -0800 Subject: libbpf: Define riscv syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Tested-by: Pu Lehui # RISC-V Link: https://lore.kernel.org/bpf/20230120200914.3008030-22-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index a2db267a6bce..047dc84c7cc9 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -359,13 +359,21 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG a5 #define __PT_PARM7_REG a6 #define __PT_PARM8_REG a7 + +/* riscv does not select ARCH_HAS_SYSCALL_WRAPPER. */ +#define PT_REGS_SYSCALL_REGS(ctx) ctx +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG ra #define __PT_FP_REG s0 #define __PT_RC_REG a0 #define __PT_SP_REG sp #define __PT_IP_REG pc -/* riscv does not select ARCH_HAS_SYSCALL_WRAPPER. */ -#define PT_REGS_SYSCALL_REGS(ctx) ctx #elif defined(bpf_target_arc) -- cgit From 2cf802737fb9cbc258ec65b15c06bd226881ae88 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:11 -0800 Subject: libbpf: Define arc syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-23-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 047dc84c7cc9..87fd2c774e73 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -392,13 +392,21 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG scratch.r5 #define __PT_PARM7_REG scratch.r6 #define __PT_PARM8_REG scratch.r7 + +/* arc does not select ARCH_HAS_SYSCALL_WRAPPER. */ +#define PT_REGS_SYSCALL_REGS(ctx) ctx +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG scratch.blink #define __PT_FP_REG scratch.fp #define __PT_RC_REG scratch.r0 #define __PT_SP_REG scratch.sp #define __PT_IP_REG scratch.ret -/* arc does not select ARCH_HAS_SYSCALL_WRAPPER. */ -#define PT_REGS_SYSCALL_REGS(ctx) ctx #elif defined(bpf_target_loongarch) -- cgit From 12a299f0b5c746c244bd06fe6a2b3672e9036e83 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:12 -0800 Subject: libbpf: Define loongarch syscall regs spec in bpf_tracing.h Define explicit table of registers used for syscall argument passing. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-24-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index 87fd2c774e73..b37d3c78b19a 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -423,13 +423,21 @@ struct pt_regs___arm64 { #define __PT_PARM6_REG regs[9] #define __PT_PARM7_REG regs[10] #define __PT_PARM8_REG regs[11] + +/* loongarch does not select ARCH_HAS_SYSCALL_WRAPPER. */ +#define PT_REGS_SYSCALL_REGS(ctx) ctx +#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG +#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG +#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG +#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG +#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG +#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG + #define __PT_RET_REG regs[1] #define __PT_FP_REG regs[22] #define __PT_RC_REG regs[4] #define __PT_SP_REG regs[3] #define __PT_IP_REG csr_era -/* loongarch does not select ARCH_HAS_SYSCALL_WRAPPER. */ -#define PT_REGS_SYSCALL_REGS(ctx) ctx #endif -- cgit From a4d325ae461c5d8e3c5e222082e0b5755b7b9b7f Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 20 Jan 2023 12:09:14 -0800 Subject: libbpf: Clean up now not needed __PT_PARM{1-6}_SYSCALL_REG defaults Each architecture supports at least 6 syscall argument registers, so now that specs for each architecture is defined in bpf_tracing.h, remove unnecessary macro overrides, which previously were required to keep existing BPF_KSYSCALL() uses compiling and working. Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230120200914.3008030-26-andrii@kernel.org --- tools/lib/bpf/bpf_tracing.h | 18 ------------------ 1 file changed, 18 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index b37d3c78b19a..6db88f41fa0d 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -476,24 +476,6 @@ struct pt_regs; * * See syscall(2) manpage for succinct table with information on each arch. */ -#ifndef __PT_PARM1_SYSCALL_REG -#define __PT_PARM1_SYSCALL_REG __PT_PARM1_REG -#endif -#ifndef __PT_PARM2_SYSCALL_REG -#define __PT_PARM2_SYSCALL_REG __PT_PARM2_REG -#endif -#ifndef __PT_PARM3_SYSCALL_REG -#define __PT_PARM3_SYSCALL_REG __PT_PARM3_REG -#endif -#ifndef __PT_PARM4_SYSCALL_REG -#define __PT_PARM4_SYSCALL_REG __PT_PARM4_REG -#endif -#ifndef __PT_PARM5_SYSCALL_REG -#define __PT_PARM5_SYSCALL_REG __PT_PARM5_REG -#endif -#ifndef __PT_PARM6_SYSCALL_REG -#define __PT_PARM6_SYSCALL_REG __PT_PARM6_REG -#endif #ifndef __PT_PARM7_SYSCALL_REG #define __PT_PARM7_SYSCALL_REG __unsupported__ #endif -- cgit From 913b2255c3d876254e968f1e8e2c817cca283a29 Mon Sep 17 00:00:00 2001 From: David Vernet Date: Wed, 25 Jan 2023 10:47:33 -0600 Subject: libbpf: Support sleepable struct_ops.s section In a prior change, the verifier was updated to support sleepable BPF_PROG_TYPE_STRUCT_OPS programs. A caller could set the program as sleepable with bpf_program__set_flags(), but it would be more ergonomic and more in-line with other sleepable program types if we supported suffixing a struct_ops section name with .s to indicate that it's sleepable. Signed-off-by: David Vernet Link: https://lore.kernel.org/r/20230125164735.785732-3-void@manifault.com Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/libbpf.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 27d9faa80471..eed5cec6f510 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -8605,6 +8605,7 @@ static const struct bpf_sec_def section_defs[] = { SEC_DEF("cgroup/setsockopt", CGROUP_SOCKOPT, BPF_CGROUP_SETSOCKOPT, SEC_ATTACHABLE), SEC_DEF("cgroup/dev", CGROUP_DEVICE, BPF_CGROUP_DEVICE, SEC_ATTACHABLE_OPT), SEC_DEF("struct_ops+", STRUCT_OPS, 0, SEC_NONE), + SEC_DEF("struct_ops.s+", STRUCT_OPS, 0, SEC_SLEEPABLE), SEC_DEF("sk_lookup", SK_LOOKUP, BPF_SK_LOOKUP, SEC_ATTACHABLE), }; -- cgit From 139df64d26fd5bb83cd04b330f7a849eefa1a889 Mon Sep 17 00:00:00 2001 From: Grant Seltzer Date: Wed, 25 Jan 2023 21:47:49 -0500 Subject: libbpf: Fix malformed documentation formatting This fixes the doxygen format documentation above the user_ring_buffer__* APIs. There has to be a newline before the @brief, otherwise doxygen won't render them for libbpf.readthedocs.org. Signed-off-by: Grant Seltzer Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230126024749.522278-1-grantseltzer@gmail.com --- tools/lib/bpf/libbpf.h | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index 898db26e42e9..830ad38e9f51 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -1064,7 +1064,8 @@ struct user_ring_buffer_opts { #define user_ring_buffer_opts__last_field sz -/* @brief **user_ring_buffer__new()** creates a new instance of a user ring +/** + * @brief **user_ring_buffer__new()** creates a new instance of a user ring * buffer. * * @param map_fd A file descriptor to a BPF_MAP_TYPE_USER_RINGBUF map. @@ -1075,7 +1076,8 @@ struct user_ring_buffer_opts { LIBBPF_API struct user_ring_buffer * user_ring_buffer__new(int map_fd, const struct user_ring_buffer_opts *opts); -/* @brief **user_ring_buffer__reserve()** reserves a pointer to a sample in the +/** + * @brief **user_ring_buffer__reserve()** reserves a pointer to a sample in the * user ring buffer. * @param rb A pointer to a user ring buffer. * @param size The size of the sample, in bytes. @@ -1095,7 +1097,8 @@ user_ring_buffer__new(int map_fd, const struct user_ring_buffer_opts *opts); */ LIBBPF_API void *user_ring_buffer__reserve(struct user_ring_buffer *rb, __u32 size); -/* @brief **user_ring_buffer__reserve_blocking()** reserves a record in the +/** + * @brief **user_ring_buffer__reserve_blocking()** reserves a record in the * ring buffer, possibly blocking for up to @timeout_ms until a sample becomes * available. * @param rb The user ring buffer. @@ -1139,7 +1142,8 @@ LIBBPF_API void *user_ring_buffer__reserve_blocking(struct user_ring_buffer *rb, __u32 size, int timeout_ms); -/* @brief **user_ring_buffer__submit()** submits a previously reserved sample +/** + * @brief **user_ring_buffer__submit()** submits a previously reserved sample * into the ring buffer. * @param rb The user ring buffer. * @param sample A reserved sample. @@ -1149,7 +1153,8 @@ LIBBPF_API void *user_ring_buffer__reserve_blocking(struct user_ring_buffer *rb, */ LIBBPF_API void user_ring_buffer__submit(struct user_ring_buffer *rb, void *sample); -/* @brief **user_ring_buffer__discard()** discards a previously reserved sample. +/** + * @brief **user_ring_buffer__discard()** discards a previously reserved sample. * @param rb The user ring buffer. * @param sample A reserved sample. * @@ -1158,7 +1163,8 @@ LIBBPF_API void user_ring_buffer__submit(struct user_ring_buffer *rb, void *samp */ LIBBPF_API void user_ring_buffer__discard(struct user_ring_buffer *rb, void *sample); -/* @brief **user_ring_buffer__free()** frees a ring buffer that was previously +/** + * @brief **user_ring_buffer__free()** frees a ring buffer that was previously * created with **user_ring_buffer__new()**. * @param rb The user ring buffer being freed. */ -- cgit From e4ce876f10a21fb4e550d1743468791021a430f5 Mon Sep 17 00:00:00 2001 From: Grant Seltzer Date: Wed, 25 Jan 2023 21:42:25 -0500 Subject: libbpf: Add documentation to map pinning API functions This adds documentation for the following API functions: - bpf_map__set_pin_path() - bpf_map__pin_path() - bpf_map__is_pinned() - bpf_map__pin() - bpf_map__unpin() - bpf_object__pin_maps() - bpf_object__unpin_maps() Signed-off-by: Grant Seltzer Signed-off-by: Andrii Nakryiko Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/20230126024225.520685-1-grantseltzer@gmail.com --- tools/lib/bpf/libbpf.h | 72 +++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 69 insertions(+), 3 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index 830ad38e9f51..8777ff21ea1d 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -233,11 +233,30 @@ LIBBPF_API int bpf_object__load(struct bpf_object *obj); */ LIBBPF_API void bpf_object__close(struct bpf_object *obj); -/* pin_maps and unpin_maps can both be called with a NULL path, in which case - * they will use the pin_path attribute of each map (and ignore all maps that - * don't have a pin_path set). +/** + * @brief **bpf_object__pin_maps()** pins each map contained within + * the BPF object at the passed directory. + * @param obj Pointer to a valid BPF object + * @param path A directory where maps should be pinned. + * @return 0, on success; negative error code, otherwise + * + * If `path` is NULL `bpf_map__pin` (which is being used on each map) + * will use the pin_path attribute of each map. In this case, maps that + * don't have a pin_path set will be ignored. */ LIBBPF_API int bpf_object__pin_maps(struct bpf_object *obj, const char *path); + +/** + * @brief **bpf_object__unpin_maps()** unpins each map contained within + * the BPF object found in the passed directory. + * @param obj Pointer to a valid BPF object + * @param path A directory where pinned maps should be searched for. + * @return 0, on success; negative error code, otherwise + * + * If `path` is NULL `bpf_map__unpin` (which is being used on each map) + * will use the pin_path attribute of each map. In this case, maps that + * don't have a pin_path set will be ignored. + */ LIBBPF_API int bpf_object__unpin_maps(struct bpf_object *obj, const char *path); LIBBPF_API int bpf_object__pin_programs(struct bpf_object *obj, @@ -848,10 +867,57 @@ LIBBPF_API const void *bpf_map__initial_value(struct bpf_map *map, size_t *psize * @return true, if the map is an internal map; false, otherwise */ LIBBPF_API bool bpf_map__is_internal(const struct bpf_map *map); + +/** + * @brief **bpf_map__set_pin_path()** sets the path attribute that tells where the + * BPF map should be pinned. This does not actually create the 'pin'. + * @param map The bpf_map + * @param path The path + * @return 0, on success; negative error, otherwise + */ LIBBPF_API int bpf_map__set_pin_path(struct bpf_map *map, const char *path); + +/** + * @brief **bpf_map__pin_path()** gets the path attribute that tells where the + * BPF map should be pinned. + * @param map The bpf_map + * @return The path string; which can be NULL + */ LIBBPF_API const char *bpf_map__pin_path(const struct bpf_map *map); + +/** + * @brief **bpf_map__is_pinned()** tells the caller whether or not the + * passed map has been pinned via a 'pin' file. + * @param map The bpf_map + * @return true, if the map is pinned; false, otherwise + */ LIBBPF_API bool bpf_map__is_pinned(const struct bpf_map *map); + +/** + * @brief **bpf_map__pin()** creates a file that serves as a 'pin' + * for the BPF map. This increments the reference count on the + * BPF map which will keep the BPF map loaded even after the + * userspace process which loaded it has exited. + * @param map The bpf_map to pin + * @param path A file path for the 'pin' + * @return 0, on success; negative error, otherwise + * + * If `path` is NULL the maps `pin_path` attribute will be used. If this is + * also NULL, an error will be returned and the map will not be pinned. + */ LIBBPF_API int bpf_map__pin(struct bpf_map *map, const char *path); + +/** + * @brief **bpf_map__unpin()** removes the file that serves as a + * 'pin' for the BPF map. + * @param map The bpf_map to unpin + * @param path A file path for the 'pin' + * @return 0, on success; negative error, otherwise + * + * The `path` parameter can be NULL, in which case the `pin_path` + * map attribute is unpinned. If both the `path` parameter and + * `pin_path` map attribute are set, they must be equal. + */ LIBBPF_API int bpf_map__unpin(struct bpf_map *map, const char *path); LIBBPF_API int bpf_map__set_inner_map_fd(struct bpf_map *map, int fd); -- cgit From e85465e420be1e408f9465f8b6fd9e2f7b17aea1 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Sat, 28 Jan 2023 01:06:40 +0100 Subject: libbpf: Simplify barrier_var() Use a single "+r" constraint instead of the separate "=r" and "0". Signed-off-by: Ilya Leoshkevich Link: https://lore.kernel.org/r/20230128000650.1516334-22-iii@linux.ibm.com Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/bpf_helpers.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_helpers.h b/tools/lib/bpf/bpf_helpers.h index d37c4fe2849d..5ec1871acb2f 100644 --- a/tools/lib/bpf/bpf_helpers.h +++ b/tools/lib/bpf/bpf_helpers.h @@ -109,7 +109,7 @@ * This is a variable-specific variant of more global barrier(). */ #ifndef barrier_var -#define barrier_var(var) asm volatile("" : "=r"(var) : "0"(var)) +#define barrier_var(var) asm volatile("" : "+r"(var)) #endif /* -- cgit From 25c76ed428219127aba385d40756f6d5814e96e9 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Sat, 28 Jan 2023 01:06:41 +0100 Subject: libbpf: Fix unbounded memory access in bpf_usdt_arg() Loading programs that use bpf_usdt_arg() on s390x fails with: ; if (arg_num >= BPF_USDT_MAX_ARG_CNT || arg_num >= spec->arg_cnt) 128: (79) r1 = *(u64 *)(r10 -24) ; frame1: R1_w=scalar(umax=4294967295,var_off=(0x0; 0xffffffff)) R10=fp0 129: (25) if r1 > 0xb goto pc+83 ; frame1: R1_w=scalar(umax=11,var_off=(0x0; 0xf)) ... ; arg_spec = &spec->args[arg_num]; 135: (79) r1 = *(u64 *)(r10 -24) ; frame1: R1_w=scalar(umax=4294967295,var_off=(0x0; 0xffffffff)) R10=fp0 ... ; switch (arg_spec->arg_type) { 139: (61) r1 = *(u32 *)(r2 +8) R2 unbounded memory access, make sure to bounds check any such access The reason is that, even though the C code enforces that arg_num < BPF_USDT_MAX_ARG_CNT, the verifier cannot propagate this constraint to the arg_spec assignment yet. Help it by forcing r1 back to stack after comparison. Signed-off-by: Ilya Leoshkevich Link: https://lore.kernel.org/r/20230128000650.1516334-23-iii@linux.ibm.com Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/usdt.bpf.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/usdt.bpf.h b/tools/lib/bpf/usdt.bpf.h index fdfd235e52c4..0bd4c135acc2 100644 --- a/tools/lib/bpf/usdt.bpf.h +++ b/tools/lib/bpf/usdt.bpf.h @@ -130,7 +130,10 @@ int bpf_usdt_arg(struct pt_regs *ctx, __u64 arg_num, long *res) if (!spec) return -ESRCH; - if (arg_num >= BPF_USDT_MAX_ARG_CNT || arg_num >= spec->arg_cnt) + if (arg_num >= BPF_USDT_MAX_ARG_CNT) + return -ENOENT; + barrier_var(arg_num); + if (arg_num >= spec->arg_cnt) return -ENOENT; arg_spec = &spec->args[arg_num]; -- cgit From 42fae973c2b168d527c8b4f6a02828ecd1e19626 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Sat, 28 Jan 2023 01:06:42 +0100 Subject: libbpf: Fix BPF_PROBE_READ{_STR}_INTO() on s390x BPF_PROBE_READ_INTO() and BPF_PROBE_READ_STR_INTO() should map to bpf_probe_read() and bpf_probe_read_str() respectively in order to work correctly on architectures with !ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE. Signed-off-by: Ilya Leoshkevich Link: https://lore.kernel.org/r/20230128000650.1516334-24-iii@linux.ibm.com Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/bpf_core_read.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf_core_read.h b/tools/lib/bpf/bpf_core_read.h index 496e6a8ee0dc..1ac57bb7ac55 100644 --- a/tools/lib/bpf/bpf_core_read.h +++ b/tools/lib/bpf/bpf_core_read.h @@ -364,7 +364,7 @@ enum bpf_enum_value_kind { /* Non-CO-RE variant of BPF_CORE_READ_INTO() */ #define BPF_PROBE_READ_INTO(dst, src, a, ...) ({ \ - ___core_read(bpf_probe_read, bpf_probe_read, \ + ___core_read(bpf_probe_read_kernel, bpf_probe_read_kernel, \ dst, (src), a, ##__VA_ARGS__) \ }) @@ -400,7 +400,7 @@ enum bpf_enum_value_kind { /* Non-CO-RE variant of BPF_CORE_READ_STR_INTO() */ #define BPF_PROBE_READ_STR_INTO(dst, src, a, ...) ({ \ - ___core_read(bpf_probe_read_str, bpf_probe_read, \ + ___core_read(bpf_probe_read_kernel_str, bpf_probe_read_kernel, \ dst, (src), a, ##__VA_ARGS__) \ }) -- cgit From 8f1669319c31dcdd07ae292822c9455aeb3bc7c7 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Wed, 1 Feb 2023 11:24:20 +0100 Subject: libbpf: add the capability to specify netlink proto in libbpf_netlink_send_recv This is a preliminary patch in order to introduce netlink_generic protocol support to libbpf. Signed-off-by: Lorenzo Bianconi Link: https://lore.kernel.org/r/7878a54667e74afeec3ee519999c044bd514b44c.1675245258.git.lorenzo@kernel.org Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/netlink.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/netlink.c b/tools/lib/bpf/netlink.c index 35104580870c..d2468a04a6c3 100644 --- a/tools/lib/bpf/netlink.c +++ b/tools/lib/bpf/netlink.c @@ -41,7 +41,7 @@ struct xdp_id_md { struct xdp_link_info info; }; -static int libbpf_netlink_open(__u32 *nl_pid) +static int libbpf_netlink_open(__u32 *nl_pid, int proto) { struct sockaddr_nl sa; socklen_t addrlen; @@ -51,7 +51,7 @@ static int libbpf_netlink_open(__u32 *nl_pid) memset(&sa, 0, sizeof(sa)); sa.nl_family = AF_NETLINK; - sock = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC, NETLINK_ROUTE); + sock = socket(AF_NETLINK, SOCK_RAW | SOCK_CLOEXEC, proto); if (sock < 0) return -errno; @@ -212,14 +212,14 @@ done: } static int libbpf_netlink_send_recv(struct libbpf_nla_req *req, - __dump_nlmsg_t parse_msg, + int proto, __dump_nlmsg_t parse_msg, libbpf_dump_nlmsg_t parse_attr, void *cookie) { __u32 nl_pid = 0; int sock, ret; - sock = libbpf_netlink_open(&nl_pid); + sock = libbpf_netlink_open(&nl_pid, proto); if (sock < 0) return sock; @@ -271,7 +271,7 @@ static int __bpf_set_link_xdp_fd_replace(int ifindex, int fd, int old_fd, } nlattr_end_nested(&req, nla); - return libbpf_netlink_send_recv(&req, NULL, NULL, NULL); + return libbpf_netlink_send_recv(&req, NETLINK_ROUTE, NULL, NULL, NULL); } int bpf_xdp_attach(int ifindex, int prog_fd, __u32 flags, const struct bpf_xdp_attach_opts *opts) @@ -382,7 +382,7 @@ int bpf_xdp_query(int ifindex, int xdp_flags, struct bpf_xdp_query_opts *opts) xdp_id.ifindex = ifindex; xdp_id.flags = xdp_flags; - err = libbpf_netlink_send_recv(&req, __dump_link_nlmsg, + err = libbpf_netlink_send_recv(&req, NETLINK_ROUTE, __dump_link_nlmsg, get_xdp_info, &xdp_id); if (err) return libbpf_err(err); @@ -493,7 +493,7 @@ static int tc_qdisc_modify(struct bpf_tc_hook *hook, int cmd, int flags) if (ret < 0) return ret; - return libbpf_netlink_send_recv(&req, NULL, NULL, NULL); + return libbpf_netlink_send_recv(&req, NETLINK_ROUTE, NULL, NULL, NULL); } static int tc_qdisc_create_excl(struct bpf_tc_hook *hook) @@ -673,7 +673,8 @@ int bpf_tc_attach(const struct bpf_tc_hook *hook, struct bpf_tc_opts *opts) info.opts = opts; - ret = libbpf_netlink_send_recv(&req, get_tc_info, NULL, &info); + ret = libbpf_netlink_send_recv(&req, NETLINK_ROUTE, get_tc_info, NULL, + &info); if (ret < 0) return libbpf_err(ret); if (!info.processed) @@ -739,7 +740,7 @@ static int __bpf_tc_detach(const struct bpf_tc_hook *hook, return ret; } - return libbpf_netlink_send_recv(&req, NULL, NULL, NULL); + return libbpf_netlink_send_recv(&req, NETLINK_ROUTE, NULL, NULL, NULL); } int bpf_tc_detach(const struct bpf_tc_hook *hook, @@ -804,7 +805,8 @@ int bpf_tc_query(const struct bpf_tc_hook *hook, struct bpf_tc_opts *opts) info.opts = opts; - ret = libbpf_netlink_send_recv(&req, get_tc_info, NULL, &info); + ret = libbpf_netlink_send_recv(&req, NETLINK_ROUTE, get_tc_info, NULL, + &info); if (ret < 0) return libbpf_err(ret); if (!info.processed) -- cgit From 04d58f1b26a436c429581d286528ea3c01624462 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Wed, 1 Feb 2023 11:24:21 +0100 Subject: libbpf: add API to get XDP/XSK supported features Extend bpf_xdp_query routine in order to get XDP/XSK supported features of netdev over route netlink interface. Extend libbpf netlink implementation in order to support netlink_generic protocol. Co-developed-by: Kumar Kartikeya Dwivedi Signed-off-by: Kumar Kartikeya Dwivedi Co-developed-by: Marek Majtyka Signed-off-by: Marek Majtyka Signed-off-by: Lorenzo Bianconi Link: https://lore.kernel.org/r/a72609ef4f0de7fee5376c40dbf54ad7f13bfb8d.1675245258.git.lorenzo@kernel.org Signed-off-by: Alexei Starovoitov --- tools/lib/bpf/libbpf.h | 3 +- tools/lib/bpf/netlink.c | 96 +++++++++++++++++++++++++++++++++++++++++++++++++ tools/lib/bpf/nlattr.h | 12 +++++++ 3 files changed, 110 insertions(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index 8777ff21ea1d..b18581277eb2 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -1048,9 +1048,10 @@ struct bpf_xdp_query_opts { __u32 hw_prog_id; /* output */ __u32 skb_prog_id; /* output */ __u8 attach_mode; /* output */ + __u64 feature_flags; /* output */ size_t :0; }; -#define bpf_xdp_query_opts__last_field attach_mode +#define bpf_xdp_query_opts__last_field feature_flags LIBBPF_API int bpf_xdp_attach(int ifindex, int prog_fd, __u32 flags, const struct bpf_xdp_attach_opts *opts); diff --git a/tools/lib/bpf/netlink.c b/tools/lib/bpf/netlink.c index d2468a04a6c3..32b13b7a11b0 100644 --- a/tools/lib/bpf/netlink.c +++ b/tools/lib/bpf/netlink.c @@ -9,6 +9,7 @@ #include #include #include +#include #include #include #include @@ -39,6 +40,12 @@ struct xdp_id_md { int ifindex; __u32 flags; struct xdp_link_info info; + __u64 feature_flags; +}; + +struct xdp_features_md { + int ifindex; + __u64 flags; }; static int libbpf_netlink_open(__u32 *nl_pid, int proto) @@ -238,6 +245,43 @@ out: return ret; } +static int parse_genl_family_id(struct nlmsghdr *nh, libbpf_dump_nlmsg_t fn, + void *cookie) +{ + struct genlmsghdr *gnl = NLMSG_DATA(nh); + struct nlattr *na = (struct nlattr *)((void *)gnl + GENL_HDRLEN); + struct nlattr *tb[CTRL_ATTR_FAMILY_ID + 1]; + __u16 *id = cookie; + + libbpf_nla_parse(tb, CTRL_ATTR_FAMILY_ID, na, + NLMSG_PAYLOAD(nh, sizeof(*gnl)), NULL); + if (!tb[CTRL_ATTR_FAMILY_ID]) + return NL_CONT; + + *id = libbpf_nla_getattr_u16(tb[CTRL_ATTR_FAMILY_ID]); + return NL_DONE; +} + +static int libbpf_netlink_resolve_genl_family_id(const char *name, + __u16 len, __u16 *id) +{ + struct libbpf_nla_req req = { + .nh.nlmsg_len = NLMSG_LENGTH(GENL_HDRLEN), + .nh.nlmsg_type = GENL_ID_CTRL, + .nh.nlmsg_flags = NLM_F_REQUEST, + .gnl.cmd = CTRL_CMD_GETFAMILY, + .gnl.version = 2, + }; + int err; + + err = nlattr_add(&req, CTRL_ATTR_FAMILY_NAME, name, len); + if (err < 0) + return err; + + return libbpf_netlink_send_recv(&req, NETLINK_GENERIC, + parse_genl_family_id, NULL, id); +} + static int __bpf_set_link_xdp_fd_replace(int ifindex, int fd, int old_fd, __u32 flags) { @@ -357,6 +401,29 @@ static int get_xdp_info(void *cookie, void *msg, struct nlattr **tb) return 0; } +static int parse_xdp_features(struct nlmsghdr *nh, libbpf_dump_nlmsg_t fn, + void *cookie) +{ + struct genlmsghdr *gnl = NLMSG_DATA(nh); + struct nlattr *na = (struct nlattr *)((void *)gnl + GENL_HDRLEN); + struct nlattr *tb[NETDEV_CMD_MAX + 1]; + struct xdp_features_md *md = cookie; + __u32 ifindex; + + libbpf_nla_parse(tb, NETDEV_CMD_MAX, na, + NLMSG_PAYLOAD(nh, sizeof(*gnl)), NULL); + + if (!tb[NETDEV_A_DEV_IFINDEX] || !tb[NETDEV_A_DEV_XDP_FEATURES]) + return NL_CONT; + + ifindex = libbpf_nla_getattr_u32(tb[NETDEV_A_DEV_IFINDEX]); + if (ifindex != md->ifindex) + return NL_CONT; + + md->flags = libbpf_nla_getattr_u64(tb[NETDEV_A_DEV_XDP_FEATURES]); + return NL_DONE; +} + int bpf_xdp_query(int ifindex, int xdp_flags, struct bpf_xdp_query_opts *opts) { struct libbpf_nla_req req = { @@ -366,6 +433,10 @@ int bpf_xdp_query(int ifindex, int xdp_flags, struct bpf_xdp_query_opts *opts) .ifinfo.ifi_family = AF_PACKET, }; struct xdp_id_md xdp_id = {}; + struct xdp_features_md md = { + .ifindex = ifindex, + }; + __u16 id; int err; if (!OPTS_VALID(opts, bpf_xdp_query_opts)) @@ -393,6 +464,31 @@ int bpf_xdp_query(int ifindex, int xdp_flags, struct bpf_xdp_query_opts *opts) OPTS_SET(opts, skb_prog_id, xdp_id.info.skb_prog_id); OPTS_SET(opts, attach_mode, xdp_id.info.attach_mode); + if (!OPTS_HAS(opts, feature_flags)) + return 0; + + err = libbpf_netlink_resolve_genl_family_id("netdev", sizeof("netdev"), &id); + if (err < 0) + return libbpf_err(err); + + memset(&req, 0, sizeof(req)); + req.nh.nlmsg_len = NLMSG_LENGTH(GENL_HDRLEN); + req.nh.nlmsg_flags = NLM_F_REQUEST; + req.nh.nlmsg_type = id; + req.gnl.cmd = NETDEV_CMD_DEV_GET; + req.gnl.version = 2; + + err = nlattr_add(&req, NETDEV_A_DEV_IFINDEX, &ifindex, sizeof(ifindex)); + if (err < 0) + return err; + + err = libbpf_netlink_send_recv(&req, NETLINK_GENERIC, + parse_xdp_features, NULL, &md); + if (err) + return libbpf_err(err); + + opts->feature_flags = md.flags; + return 0; } diff --git a/tools/lib/bpf/nlattr.h b/tools/lib/bpf/nlattr.h index 4d15ae2ff812..d92d1c1de700 100644 --- a/tools/lib/bpf/nlattr.h +++ b/tools/lib/bpf/nlattr.h @@ -14,6 +14,7 @@ #include #include #include +#include /* avoid multiple definition of netlink features */ #define __LINUX_NETLINK_H @@ -58,6 +59,7 @@ struct libbpf_nla_req { union { struct ifinfomsg ifinfo; struct tcmsg tc; + struct genlmsghdr gnl; }; char buf[128]; }; @@ -89,11 +91,21 @@ static inline uint8_t libbpf_nla_getattr_u8(const struct nlattr *nla) return *(uint8_t *)libbpf_nla_data(nla); } +static inline uint16_t libbpf_nla_getattr_u16(const struct nlattr *nla) +{ + return *(uint16_t *)libbpf_nla_data(nla); +} + static inline uint32_t libbpf_nla_getattr_u32(const struct nlattr *nla) { return *(uint32_t *)libbpf_nla_data(nla); } +static inline uint64_t libbpf_nla_getattr_u64(const struct nlattr *nla) +{ + return *(uint64_t *)libbpf_nla_data(nla); +} + static inline const char *libbpf_nla_getattr_str(const struct nlattr *nla) { return (const char *)libbpf_nla_data(nla); -- cgit From d1d7730ff8756c6db20ff82096b577d8cfbaf547 Mon Sep 17 00:00:00 2001 From: Hao Xiang Date: Fri, 3 Feb 2023 23:48:42 +0000 Subject: libbpf: Correctly set the kernel code version in Debian kernel. In a previous commit, Ubuntu kernel code version is correctly set by retrieving the information from /proc/version_signature. commit<5b3d72987701d51bf31823b39db49d10970f5c2d> (libbpf: Improve LINUX_VERSION_CODE detection) The /proc/version_signature file doesn't present in at least the older versions of Debian distributions (eg, Debian 9, 10). The Debian kernel has a similar issue where the release information from uname() syscall doesn't give the kernel code version that matches what the kernel actually expects. Below is an example content from Debian 10. release: 4.19.0-23-amd64 version: #1 SMP Debian 4.19.269-1 (2022-12-20) x86_64 Debian reports incorrect kernel version in utsname::release returned by uname() syscall, which in older kernels (Debian 9, 10) leads to kprobe BPF programs failing to load due to the version check mismatch. Fortunately, the correct kernel code version presents in the utsname::version returned by uname() syscall in Debian kernels. This change adds another get kernel version function to handle Debian in addition to the previously added get kernel version function to handle Ubuntu. Some minor refactoring work is also done to make the code more readable. Signed-off-by: Hao Xiang Signed-off-by: Ho-Ren (Jack) Chuang Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20230203234842.2933903-1-hao.xiang@bytedance.com --- tools/lib/bpf/libbpf.c | 37 ------------------- tools/lib/bpf/libbpf_probes.c | 83 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 83 insertions(+), 37 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index eed5cec6f510..4191a78b2815 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -34,7 +34,6 @@ #include #include #include -#include #include #include #include @@ -870,42 +869,6 @@ bpf_object__add_programs(struct bpf_object *obj, Elf_Data *sec_data, return 0; } -__u32 get_kernel_version(void) -{ - /* On Ubuntu LINUX_VERSION_CODE doesn't correspond to info.release, - * but Ubuntu provides /proc/version_signature file, as described at - * https://ubuntu.com/kernel, with an example contents below, which we - * can use to get a proper LINUX_VERSION_CODE. - * - * Ubuntu 5.4.0-12.15-generic 5.4.8 - * - * In the above, 5.4.8 is what kernel is actually expecting, while - * uname() call will return 5.4.0 in info.release. - */ - const char *ubuntu_kver_file = "/proc/version_signature"; - __u32 major, minor, patch; - struct utsname info; - - if (faccessat(AT_FDCWD, ubuntu_kver_file, R_OK, AT_EACCESS) == 0) { - FILE *f; - - f = fopen(ubuntu_kver_file, "r"); - if (f) { - if (fscanf(f, "%*s %*s %d.%d.%d\n", &major, &minor, &patch) == 3) { - fclose(f); - return KERNEL_VERSION(major, minor, patch); - } - fclose(f); - } - /* something went wrong, fall back to uname() approach */ - } - - uname(&info); - if (sscanf(info.release, "%u.%u.%u", &major, &minor, &patch) != 3) - return 0; - return KERNEL_VERSION(major, minor, patch); -} - static const struct btf_member * find_member_by_offset(const struct btf_type *t, __u32 bit_offset) { diff --git a/tools/lib/bpf/libbpf_probes.c b/tools/lib/bpf/libbpf_probes.c index b44fcbb4b42e..4f3bc968ff8e 100644 --- a/tools/lib/bpf/libbpf_probes.c +++ b/tools/lib/bpf/libbpf_probes.c @@ -12,11 +12,94 @@ #include #include #include +#include #include "bpf.h" #include "libbpf.h" #include "libbpf_internal.h" +/* On Ubuntu LINUX_VERSION_CODE doesn't correspond to info.release, + * but Ubuntu provides /proc/version_signature file, as described at + * https://ubuntu.com/kernel, with an example contents below, which we + * can use to get a proper LINUX_VERSION_CODE. + * + * Ubuntu 5.4.0-12.15-generic 5.4.8 + * + * In the above, 5.4.8 is what kernel is actually expecting, while + * uname() call will return 5.4.0 in info.release. + */ +static __u32 get_ubuntu_kernel_version(void) +{ + const char *ubuntu_kver_file = "/proc/version_signature"; + __u32 major, minor, patch; + int ret; + FILE *f; + + if (faccessat(AT_FDCWD, ubuntu_kver_file, R_OK, AT_EACCESS) != 0) + return 0; + + f = fopen(ubuntu_kver_file, "r"); + if (!f) + return 0; + + ret = fscanf(f, "%*s %*s %u.%u.%u\n", &major, &minor, &patch); + fclose(f); + if (ret != 3) + return 0; + + return KERNEL_VERSION(major, minor, patch); +} + +/* On Debian LINUX_VERSION_CODE doesn't correspond to info.release. + * Instead, it is provided in info.version. An example content of + * Debian 10 looks like the below. + * + * utsname::release 4.19.0-22-amd64 + * utsname::version #1 SMP Debian 4.19.260-1 (2022-09-29) + * + * In the above, 4.19.260 is what kernel is actually expecting, while + * uname() call will return 4.19.0 in info.release. + */ +static __u32 get_debian_kernel_version(struct utsname *info) +{ + __u32 major, minor, patch; + char *p; + + p = strstr(info->version, "Debian "); + if (!p) { + /* This is not a Debian kernel. */ + return 0; + } + + if (sscanf(p, "Debian %u.%u.%u", &major, &minor, &patch) != 3) + return 0; + + return KERNEL_VERSION(major, minor, patch); +} + +__u32 get_kernel_version(void) +{ + __u32 major, minor, patch, version; + struct utsname info; + + /* Check if this is an Ubuntu kernel. */ + version = get_ubuntu_kernel_version(); + if (version != 0) + return version; + + uname(&info); + + /* Check if this is a Debian kernel. */ + version = get_debian_kernel_version(&info); + if (version != 0) + return version; + + if (sscanf(info.release, "%u.%u.%u", &major, &minor, &patch) != 3) + return 0; + + return KERNEL_VERSION(major, minor, patch); +} + static int probe_prog_load(enum bpf_prog_type prog_type, const struct bpf_insn *insns, size_t insns_cnt, char *log_buf, size_t log_buf_sz) -- cgit From 02fc0e73e852f78ec374004f924b9d84275d0006 Mon Sep 17 00:00:00 2001 From: Lorenzo Bianconi Date: Tue, 7 Feb 2023 12:11:03 +0100 Subject: libbpf: Always use libbpf_err to return an error in bpf_xdp_query() In order to properly set errno, rely on libbpf_err utility routine in bpf_xdp_query() to return an error to the caller. Fixes: 04d58f1b26a4 ("libbpf: add API to get XDP/XSK supported features") Signed-off-by: Lorenzo Bianconi Signed-off-by: Daniel Borkmann Link: https://lore.kernel.org/bpf/827d40181f9f90fb37702f44328e1614df7c0503.1675768112.git.lorenzo@kernel.org --- tools/lib/bpf/netlink.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/netlink.c b/tools/lib/bpf/netlink.c index 32b13b7a11b0..cb082a04ffa8 100644 --- a/tools/lib/bpf/netlink.c +++ b/tools/lib/bpf/netlink.c @@ -480,7 +480,7 @@ int bpf_xdp_query(int ifindex, int xdp_flags, struct bpf_xdp_query_opts *opts) err = nlattr_add(&req, NETDEV_A_DEV_IFINDEX, &ifindex, sizeof(ifindex)); if (err < 0) - return err; + return libbpf_err(err); err = libbpf_netlink_send_recv(&req, NETLINK_GENERIC, parse_xdp_features, NULL, &md); -- cgit From ab8684b8cecf711cc9e47c1cbb1a8f4b549f8893 Mon Sep 17 00:00:00 2001 From: Jon Doron Date: Tue, 7 Feb 2023 10:19:16 +0200 Subject: libbpf: Add sample_period to creation options Add option to set when the perf buffer should wake up, by default the perf buffer becomes signaled for every event that is being pushed to it. In case of a high throughput of events it will be more efficient to wake up only once you have X events ready to be read. So your application can wakeup once and drain the entire perf buffer. Signed-off-by: Jon Doron Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20230207081916.3398417-1-arilou@gmail.com --- tools/lib/bpf/libbpf.c | 9 +++++++-- tools/lib/bpf/libbpf.h | 4 +++- 2 files changed, 10 insertions(+), 3 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 4191a78b2815..35a698eb825d 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -11673,17 +11673,22 @@ struct perf_buffer *perf_buffer__new(int map_fd, size_t page_cnt, const size_t attr_sz = sizeof(struct perf_event_attr); struct perf_buffer_params p = {}; struct perf_event_attr attr; + __u32 sample_period; if (!OPTS_VALID(opts, perf_buffer_opts)) return libbpf_err_ptr(-EINVAL); + sample_period = OPTS_GET(opts, sample_period, 1); + if (!sample_period) + sample_period = 1; + memset(&attr, 0, attr_sz); attr.size = attr_sz; attr.config = PERF_COUNT_SW_BPF_OUTPUT; attr.type = PERF_TYPE_SOFTWARE; attr.sample_type = PERF_SAMPLE_RAW; - attr.sample_period = 1; - attr.wakeup_events = 1; + attr.sample_period = sample_period; + attr.wakeup_events = sample_period; p.attr = &attr; p.sample_cb = sample_cb; diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index b18581277eb2..2efd80f6f7b9 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -1247,8 +1247,10 @@ typedef void (*perf_buffer_lost_fn)(void *ctx, int cpu, __u64 cnt); /* common use perf buffer options */ struct perf_buffer_opts { size_t sz; + __u32 sample_period; + size_t :0; }; -#define perf_buffer_opts__last_field sz +#define perf_buffer_opts__last_field sample_period /** * @brief **perf_buffer__new()** creates BPF perfbuf manager for a specified -- cgit From 17bcd27a08a21397698edf143084d7c87ce17946 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Fri, 10 Feb 2023 01:12:01 +0100 Subject: libbpf: Fix alen calculation in libbpf_nla_dump_errormsg() The code assumes that everything that comes after nlmsgerr are nlattrs. When calculating their size, it does not account for the initial nlmsghdr. This may lead to accessing uninitialized memory. Fixes: bbf48c18ee0c ("libbpf: add error reporting in XDP") Signed-off-by: Ilya Leoshkevich Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20230210001210.395194-8-iii@linux.ibm.com --- tools/lib/bpf/nlattr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/nlattr.c b/tools/lib/bpf/nlattr.c index 3900d052ed19..975e265eab3b 100644 --- a/tools/lib/bpf/nlattr.c +++ b/tools/lib/bpf/nlattr.c @@ -178,7 +178,7 @@ int libbpf_nla_dump_errormsg(struct nlmsghdr *nlh) hlen += nlmsg_len(&err->msg); attr = (struct nlattr *) ((void *) err + hlen); - alen = nlh->nlmsg_len - hlen; + alen = (void *)nlh + nlh->nlmsg_len - (void *)attr; if (libbpf_nla_parse(tb, NLMSGERR_ATTR_MAX, attr, alen, extack_policy) != 0) { -- cgit From 55a9ed0e16baf4d025c160d46bc1e3fac0d4cdc4 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Wed, 15 Feb 2023 00:12:14 +0100 Subject: libbpf: Introduce bpf_{btf,link,map,prog}_get_info_by_fd() These are type-safe wrappers around bpf_obj_get_info_by_fd(). They found one problem in selftests, and are also useful for adding Memory Sanitizer annotations. Signed-off-by: Ilya Leoshkevich Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20230214231221.249277-2-iii@linux.ibm.com --- tools/lib/bpf/bpf.c | 20 ++++++++++++++++++++ tools/lib/bpf/bpf.h | 9 +++++++++ tools/lib/bpf/libbpf.map | 5 +++++ 3 files changed, 34 insertions(+) (limited to 'tools/lib') diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c index 9aff98f42a3d..e750b6f5fcc3 100644 --- a/tools/lib/bpf/bpf.c +++ b/tools/lib/bpf/bpf.c @@ -1044,6 +1044,26 @@ int bpf_obj_get_info_by_fd(int bpf_fd, void *info, __u32 *info_len) return libbpf_err_errno(err); } +int bpf_prog_get_info_by_fd(int prog_fd, struct bpf_prog_info *info, __u32 *info_len) +{ + return bpf_obj_get_info_by_fd(prog_fd, info, info_len); +} + +int bpf_map_get_info_by_fd(int map_fd, struct bpf_map_info *info, __u32 *info_len) +{ + return bpf_obj_get_info_by_fd(map_fd, info, info_len); +} + +int bpf_btf_get_info_by_fd(int btf_fd, struct bpf_btf_info *info, __u32 *info_len) +{ + return bpf_obj_get_info_by_fd(btf_fd, info, info_len); +} + +int bpf_link_get_info_by_fd(int link_fd, struct bpf_link_info *info, __u32 *info_len) +{ + return bpf_obj_get_info_by_fd(link_fd, info, info_len); +} + int bpf_raw_tracepoint_open(const char *name, int prog_fd) { const size_t attr_sz = offsetofend(union bpf_attr, raw_tracepoint); diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h index 7468978d3c27..9ed9bceb4111 100644 --- a/tools/lib/bpf/bpf.h +++ b/tools/lib/bpf/bpf.h @@ -386,6 +386,15 @@ LIBBPF_API int bpf_link_get_fd_by_id(__u32 id); LIBBPF_API int bpf_link_get_fd_by_id_opts(__u32 id, const struct bpf_get_fd_by_id_opts *opts); LIBBPF_API int bpf_obj_get_info_by_fd(int bpf_fd, void *info, __u32 *info_len); +/* Type-safe variants of bpf_obj_get_info_by_fd(). The callers still needs to + * pass info_len, which should normally be + * sizeof(struct bpf_{prog,map,btf,link}_info), in order to be compatible with + * different libbpf and kernel versions. + */ +LIBBPF_API int bpf_prog_get_info_by_fd(int prog_fd, struct bpf_prog_info *info, __u32 *info_len); +LIBBPF_API int bpf_map_get_info_by_fd(int map_fd, struct bpf_map_info *info, __u32 *info_len); +LIBBPF_API int bpf_btf_get_info_by_fd(int btf_fd, struct bpf_btf_info *info, __u32 *info_len); +LIBBPF_API int bpf_link_get_info_by_fd(int link_fd, struct bpf_link_info *info, __u32 *info_len); struct bpf_prog_query_opts { size_t sz; /* size of this struct for forward/backward compatibility */ diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map index 11c36a3c1a9f..50dde1f6521e 100644 --- a/tools/lib/bpf/libbpf.map +++ b/tools/lib/bpf/libbpf.map @@ -384,4 +384,9 @@ LIBBPF_1.1.0 { } LIBBPF_1.0.0; LIBBPF_1.2.0 { + global: + bpf_btf_get_info_by_fd; + bpf_link_get_info_by_fd; + bpf_map_get_info_by_fd; + bpf_prog_get_info_by_fd; } LIBBPF_1.1.0; -- cgit From 629dfc660cae86a6a48d19f5295226d03caae673 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Wed, 15 Feb 2023 00:12:15 +0100 Subject: libbpf: Use bpf_{btf,link,map,prog}_get_info_by_fd() Use the new type-safe wrappers around bpf_obj_get_info_by_fd(). Signed-off-by: Ilya Leoshkevich Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20230214231221.249277-3-iii@linux.ibm.com --- tools/lib/bpf/btf.c | 8 ++++---- tools/lib/bpf/libbpf.c | 14 +++++++------- tools/lib/bpf/netlink.c | 2 +- tools/lib/bpf/ringbuf.c | 4 ++-- 4 files changed, 14 insertions(+), 14 deletions(-) (limited to 'tools/lib') diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index 64841117fbb2..9181d36118d2 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -1350,9 +1350,9 @@ struct btf *btf_get_from_fd(int btf_fd, struct btf *base_btf) void *ptr; int err; - /* we won't know btf_size until we call bpf_obj_get_info_by_fd(). so + /* we won't know btf_size until we call bpf_btf_get_info_by_fd(). so * let's start with a sane default - 4KiB here - and resize it only if - * bpf_obj_get_info_by_fd() needs a bigger buffer. + * bpf_btf_get_info_by_fd() needs a bigger buffer. */ last_size = 4096; ptr = malloc(last_size); @@ -1362,7 +1362,7 @@ struct btf *btf_get_from_fd(int btf_fd, struct btf *base_btf) memset(&btf_info, 0, sizeof(btf_info)); btf_info.btf = ptr_to_u64(ptr); btf_info.btf_size = last_size; - err = bpf_obj_get_info_by_fd(btf_fd, &btf_info, &len); + err = bpf_btf_get_info_by_fd(btf_fd, &btf_info, &len); if (!err && btf_info.btf_size > last_size) { void *temp_ptr; @@ -1380,7 +1380,7 @@ struct btf *btf_get_from_fd(int btf_fd, struct btf *base_btf) btf_info.btf = ptr_to_u64(ptr); btf_info.btf_size = last_size; - err = bpf_obj_get_info_by_fd(btf_fd, &btf_info, &len); + err = bpf_btf_get_info_by_fd(btf_fd, &btf_info, &len); } if (err || btf_info.btf_size > last_size) { diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 35a698eb825d..05c4db355f28 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -4345,7 +4345,7 @@ int bpf_map__reuse_fd(struct bpf_map *map, int fd) char *new_name; memset(&info, 0, len); - err = bpf_obj_get_info_by_fd(fd, &info, &len); + err = bpf_map_get_info_by_fd(fd, &info, &len); if (err && errno == EINVAL) err = bpf_get_map_info_from_fdinfo(fd, &info); if (err) @@ -4729,7 +4729,7 @@ static int probe_module_btf(void) * kernel's module BTF support coincides with support for * name/name_len fields in struct bpf_btf_info. */ - err = bpf_obj_get_info_by_fd(fd, &info, &len); + err = bpf_btf_get_info_by_fd(fd, &info, &len); close(fd); return !err; } @@ -4892,7 +4892,7 @@ static bool map_is_reuse_compat(const struct bpf_map *map, int map_fd) int err; memset(&map_info, 0, map_info_len); - err = bpf_obj_get_info_by_fd(map_fd, &map_info, &map_info_len); + err = bpf_map_get_info_by_fd(map_fd, &map_info, &map_info_len); if (err && errno == EINVAL) err = bpf_get_map_info_from_fdinfo(map_fd, &map_info); if (err) { @@ -5437,7 +5437,7 @@ static int load_module_btfs(struct bpf_object *obj) info.name = ptr_to_u64(name); info.name_len = sizeof(name); - err = bpf_obj_get_info_by_fd(fd, &info, &len); + err = bpf_btf_get_info_by_fd(fd, &info, &len); if (err) { err = -errno; pr_warn("failed to get BTF object #%d info: %d\n", id, err); @@ -9030,9 +9030,9 @@ static int libbpf_find_prog_btf_id(const char *name, __u32 attach_prog_fd) int err; memset(&info, 0, info_len); - err = bpf_obj_get_info_by_fd(attach_prog_fd, &info, &info_len); + err = bpf_prog_get_info_by_fd(attach_prog_fd, &info, &info_len); if (err) { - pr_warn("failed bpf_obj_get_info_by_fd for FD %d: %d\n", + pr_warn("failed bpf_prog_get_info_by_fd for FD %d: %d\n", attach_prog_fd, err); return err; } @@ -11741,7 +11741,7 @@ static struct perf_buffer *__perf_buffer__new(int map_fd, size_t page_cnt, /* best-effort sanity checks */ memset(&map, 0, sizeof(map)); map_info_len = sizeof(map); - err = bpf_obj_get_info_by_fd(map_fd, &map, &map_info_len); + err = bpf_map_get_info_by_fd(map_fd, &map, &map_info_len); if (err) { err = -errno; /* if BPF_OBJ_GET_INFO_BY_FD is supported, will return diff --git a/tools/lib/bpf/netlink.c b/tools/lib/bpf/netlink.c index cb082a04ffa8..1653e7a8b0a1 100644 --- a/tools/lib/bpf/netlink.c +++ b/tools/lib/bpf/netlink.c @@ -689,7 +689,7 @@ static int tc_add_fd_and_name(struct libbpf_nla_req *req, int fd) int len, ret; memset(&info, 0, info_len); - ret = bpf_obj_get_info_by_fd(fd, &info, &info_len); + ret = bpf_prog_get_info_by_fd(fd, &info, &info_len); if (ret < 0) return ret; diff --git a/tools/lib/bpf/ringbuf.c b/tools/lib/bpf/ringbuf.c index 47855af25f3b..02199364db13 100644 --- a/tools/lib/bpf/ringbuf.c +++ b/tools/lib/bpf/ringbuf.c @@ -83,7 +83,7 @@ int ring_buffer__add(struct ring_buffer *rb, int map_fd, memset(&info, 0, sizeof(info)); - err = bpf_obj_get_info_by_fd(map_fd, &info, &len); + err = bpf_map_get_info_by_fd(map_fd, &info, &len); if (err) { err = -errno; pr_warn("ringbuf: failed to get map info for fd=%d: %d\n", @@ -359,7 +359,7 @@ static int user_ringbuf_map(struct user_ring_buffer *rb, int map_fd) memset(&info, 0, sizeof(info)); - err = bpf_obj_get_info_by_fd(map_fd, &info, &len); + err = bpf_map_get_info_by_fd(map_fd, &info, &len); if (err) { err = -errno; pr_warn("user ringbuf: failed to get map info for fd=%d: %d\n", map_fd, err); -- cgit