wireguard-linux/tools/lib/bpf, branch stable

libbpf: Fix powerpc's stack register definition in bpf_tracing.h

2025-10-23T18:25:16Z

retsnoop's build on powerpc (ppc64le) architecture ([0]) failed due to wrong definition of PT_REGS_SP() macro. Looking at powerpc's implementation of stack unwinding in perf_callchain_user_64() clearly shows that stack pointer register is gpr[1]. Fix libbpf's definition of __PT_SP_REG for powerpc to fix all this. [0] https://kojipkgs.fedoraproject.org/work/tasks/1544/137921544/build.log Fixes: 138d6153a139 ("samples/bpf: Enable powerpc support") Signed-off-by: Andrii Nakryiko Reviewed-by: Naveen N Rao (AMD) Link: https://lore.kernel.org/r/20251020203643.989467-1-andrii@kernel.org Signed-off-by: Alexei Starovoitov

libbpf: Fix undefined behavior in {get,put}_unaligned_be32()

2025-10-06T16:16:29Z

These violate aliasing rules and may be miscompiled unless -fno-strict-aliasing is used. Replace them with the standard memcpy() solution. Note that compilers know how to optimize this properly. Fixes: 4a1c9e544b8d ("libbpf: remove linux/unaligned.h dependency for libbpf_sha256()") Signed-off-by: Eric Biggers Signed-off-by: Andrii Nakryiko Acked-by: Yonghong Song Link: https://lore.kernel.org/bpf/20251006012037.159295-1-ebiggers@kernel.org

libbpf: Fix missing #pragma in libbpf_utils.c

2025-10-02T21:33:57Z

The recent sha256 patch uses a GCC pragma to suppress compile errors for a packed struct, but omits a needed pragma (see related link) and thus still raises errors: (e.g. on GCC 12.3 armhf) libbpf_utils.c:153:29: error: packed attribute causes inefficient alignment for ‘__val’ [-Werror=attributes] 153 | struct __packed_u32 { __u32 __val; } __attribute__((packed)); | ^~~~~ Resolve by adding the GCC diagnostic pragma to ignore "-Wattributes". Link: https://lore.kernel.org/bpf/CAP-5=fXURWoZu2j6Y8xQy23i7=DfgThq3WC1RkGFBx-4moQKYQ@mail.gmail.com/ Fixes: 4a1c9e544b8d ("libbpf: remove linux/unaligned.h dependency for libbpf_sha256()") Signed-off-by: Tony Ambardar Acked-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov

libbpf: remove linux/unaligned.h dependency for libbpf_sha256()

2025-10-01T22:27:25Z

linux/unaligned.h include dependency is causing issues for libbpf's Github mirror due to {get,put}_unaligned_be32() usage. So get rid of it by implementing custom variants of those macros that will work both in kernel and Github mirror repos. Also switch round_up() to roundup(), as the former is not available in Github mirror (and is just a subtly more specific variant of roundup() anyways). Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20251001171326.3883055-6-andrii@kernel.org Signed-off-by: Alexei Starovoitov Acked-by: Eduard Zingerman

libbpf: move libbpf_sha256() implementation into libbpf_utils.c

2025-10-01T22:27:25Z

Move sha256 implementation out of already large and unwieldy libbpf.c into libbpf_utils.c where we'll keep reusable helpers. Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20251001171326.3883055-5-andrii@kernel.org Signed-off-by: Alexei Starovoitov Acked-by: Eduard Zingerman

libbpf: move libbpf_errstr() into libbpf_utils.c

2025-10-01T22:27:25Z

Get rid of str_err.{c,h} by moving implementation of libbpf_errstr() into libbpf_utils.c and declarations into libbpf_internal.h. Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20251001171326.3883055-4-andrii@kernel.org Signed-off-by: Alexei Starovoitov Acked-by: Eduard Zingerman

libbpf: remove unused libbpf_strerror_r and STRERR_BUFSIZE

2025-10-01T22:27:25Z

libbpf_strerror_r() is not exposed as public API and neither is it used inside libbpf itself. Remove it altogether. Same for STRERR_BUFSIZE, it's just an orphaned leftover constant which we missed to clean up some time earlier. Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20251001171326.3883055-3-andrii@kernel.org Signed-off-by: Alexei Starovoitov Acked-by: Eduard Zingerman

libbpf: make libbpf_errno.c into more generic libbpf_utils.c

2025-10-01T22:27:24Z

Libbpf is missing one convenient place to put common "utils"-like code that is generic and usable from multiple places. Use libbpf_errno.c as the base for more generic libbpf_utils.c. Signed-off-by: Andrii Nakryiko Link: https://lore.kernel.org/r/20251001171326.3883055-2-andrii@kernel.org Signed-off-by: Alexei Starovoitov Acked-by: Eduard Zingerman

libbpf: Replace AF_ALG with open coded SHA-256

2025-09-28T11:25:31Z

Reimplement libbpf_sha256() using some basic SHA-256 C code. This eliminates the newly-added dependency on AF_ALG, which is a problematic UAPI that is not supported by all kernels. Make libbpf_sha256() return void, since it can no longer fail. This simplifies some callers. Also drop the unnecessary 'sha_out_sz' parameter. Finally, also fix the typo in "compute_sha_udpate_offsets". Fixes: c297fe3e9f99 ("libbpf: Implement SHA256 internal helper") Signed-off-by: Eric Biggers Link: https://lore.kernel.org/r/20250928003833.138407-1-ebiggers@kernel.org Signed-off-by: Alexei Starovoitov

libbpf: Fix error when st-prefix_ops and ops from differ btf

2025-09-26T20:03:19Z

When a module registers a struct_ops, the struct_ops type and its corresponding map_value type ("bpf_struct_ops_") may reside in different btf objects, here are four possible case: +--------+---------------+-------------+---------------------------------+ | |bpf_struct_ops_| xxx_ops | | +--------+---------------+-------------+---------------------------------+ | case 0 | btf_vmlinux | btf_vmlinux | be used and reg only in vmlinux | +--------+---------------+-------------+---------------------------------+ | case 1 | btf_vmlinux | mod_btf | INVALID | +--------+---------------+-------------+---------------------------------+ | case 2 | mod_btf | btf_vmlinux | reg in mod but be used both in | | | | | vmlinux and mod. | +--------+---------------+-------------+---------------------------------+ | case 3 | mod_btf | mod_btf | be used and reg only in mod | +--------+---------------+-------------+---------------------------------+ Currently we figure out the mod_btf by searching with the struct_ops type, which makes it impossible to figure out the mod_btf when the struct_ops type is in btf_vmlinux while it's corresponding map_value type is in mod_btf (case 2). The fix is to use the corresponding map_value type ("bpf_struct_ops_") as the lookup anchor instead of the struct_ops type to figure out the `btf` and `mod_btf` via find_ksym_btf_id(), and then we can locate the kern_type_id via btf__find_by_name_kind() with the `btf` we just obtained from find_ksym_btf_id(). With this change the lookup obtains the correct btf and mod_btf for case 2, preserves correct behavior for other valid cases, and still fails as expected for the invalid scenario (case 1). Fixes: 590a00888250 ("bpf: libbpf: Add STRUCT_OPS support") Signed-off-by: D. Wythe Signed-off-by: Andrii Nakryiko Acked-by: Andrii Nakryiko Acked-by: Martin KaFai Lau Link: https://lore.kernel.org/bpf/20250926071751.108293-1-alibuda@linux.alibaba.com