]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
bpf, arm64: Fix fp initialization for exception boundary
authorPuranjay Mohan <puranjay@kernel.org>
Tue, 22 Jul 2025 13:34:09 +0000 (13:34 +0000)
committerDaniel Borkmann <daniel@iogearbox.net>
Sat, 26 Jul 2025 19:23:38 +0000 (21:23 +0200)
commitb114fcee766d5101eada1aca7bb5fd0a86c89b35
tree45b128b801b10442f085b33be3a2b182159b6dcd
parentb7b3500bd4eef2c3b5124ed195f26eb048407d9b
bpf, arm64: Fix fp initialization for exception boundary

In the ARM64 BPF JIT when prog->aux->exception_boundary is set for a BPF
program, find_used_callee_regs() is not called because for a program
acting as exception boundary, all callee saved registers are saved.
find_used_callee_regs() sets `ctx->fp_used = true;` when it sees FP
being used in any of the instructions.

For programs acting as exception boundary, ctx->fp_used remains false
even if frame pointer is used by the program and therefore, FP is not
set-up for such programs in the prologue. This can cause the kernel to
crash due to a pagefault.

Fix it by setting ctx->fp_used = true for exception boundary programs as
fp is always saved in such programs.

Fixes: 5d4fa9ec5643 ("bpf, arm64: Avoid blindly saving/restoring all callee-saved registers")
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Xu Kuohai <xukuohai@huawei.com>
Link: https://lore.kernel.org/bpf/20250722133410.54161-2-puranjay@kernel.org
arch/arm64/net/bpf_jit_comp.c