diff options
| author | Daniel Borkmann <daniel@iogearbox.net> | 2018-01-16 03:46:08 +0100 |
|---|---|---|
| committer | Alexei Starovoitov <ast@kernel.org> | 2018-01-16 20:29:15 +0100 |
| commit | a2284d912bfc865cdca4c00488e08a3550f9a405 (patch) | |
| tree | 03aea207e0887441cad8e42550fc6fbf3475ae8c /arch/arm64/net | |
| parent | bpf: fix 32-bit divide by zero (diff) | |
| download | linux-a2284d912bfc865cdca4c00488e08a3550f9a405.tar.xz linux-a2284d912bfc865cdca4c00488e08a3550f9a405.zip | |
bpf, arm64: fix stack_depth tracking in combination with tail calls
Using dynamic stack_depth tracking in arm64 JIT is currently broken in
combination with tail calls. In prologue, we cache ctx->stack_size and
adjust SP reg for setting up function call stack, and tearing it down
again in epilogue. Problem is that when doing a tail call, the cached
ctx->stack_size might not be the same.
One way to fix the problem with minimal overhead is to re-adjust SP in
emit_bpf_tail_call() and properly adjust it to the current program's
ctx->stack_size. Tested on Cavium ThunderX ARMv8.
Fixes: f1c9eed7f437 ("bpf, arm64: take advantage of stack_depth tracking")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Diffstat (limited to 'arch/arm64/net')
| -rw-r--r-- | arch/arm64/net/bpf_jit_comp.c | 20 |
1 files changed, 11 insertions, 9 deletions
diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c index ba38d403abb2..bb32f7f6dd0f 100644 --- a/arch/arm64/net/bpf_jit_comp.c +++ b/arch/arm64/net/bpf_jit_comp.c @@ -148,7 +148,8 @@ static inline int epilogue_offset(const struct jit_ctx *ctx) /* Stack must be multiples of 16B */ #define STACK_ALIGN(sz) (((sz) + 15) & ~15) -#define PROLOGUE_OFFSET 8 +/* Tail call offset to jump into */ +#define PROLOGUE_OFFSET 7 static int build_prologue(struct jit_ctx *ctx) { @@ -200,19 +201,19 @@ static int build_prologue(struct jit_ctx *ctx) /* Initialize tail_call_cnt */ emit(A64_MOVZ(1, tcc, 0, 0), ctx); - /* 4 byte extra for skb_copy_bits buffer */ - ctx->stack_size = prog->aux->stack_depth + 4; - ctx->stack_size = STACK_ALIGN(ctx->stack_size); - - /* Set up function call stack */ - emit(A64_SUB_I(1, A64_SP, A64_SP, ctx->stack_size), ctx); - cur_offset = ctx->idx - idx0; if (cur_offset != PROLOGUE_OFFSET) { pr_err_once("PROLOGUE_OFFSET = %d, expected %d!\n", cur_offset, PROLOGUE_OFFSET); return -1; } + + /* 4 byte extra for skb_copy_bits buffer */ + ctx->stack_size = prog->aux->stack_depth + 4; + ctx->stack_size = STACK_ALIGN(ctx->stack_size); + + /* Set up function call stack */ + emit(A64_SUB_I(1, A64_SP, A64_SP, ctx->stack_size), ctx); return 0; } @@ -260,11 +261,12 @@ static int emit_bpf_tail_call(struct jit_ctx *ctx) emit(A64_LDR64(prg, tmp, prg), ctx); emit(A64_CBZ(1, prg, jmp_offset), ctx); - /* goto *(prog->bpf_func + prologue_size); */ + /* goto *(prog->bpf_func + prologue_offset); */ off = offsetof(struct bpf_prog, bpf_func); emit_a64_mov_i64(tmp, off, ctx); emit(A64_LDR64(tmp, prg, tmp), ctx); emit(A64_ADD_I(1, tmp, tmp, sizeof(u32) * PROLOGUE_OFFSET), ctx); + emit(A64_ADD_I(1, A64_SP, A64_SP, ctx->stack_size), ctx); emit(A64_BR(tmp), ctx); /* out: */ |