From 1ea47e01ad6ea0fe99697c54c2413d81dd21fe32 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Thu, 14 Dec 2017 17:55:13 -0800 Subject: bpf: add support for bpf_call to interpreter though bpf_call is still the same call instruction and calling convention 'bpf to bpf' and 'bpf to helper' is the same the interpreter has to oparate on 'struct bpf_insn *'. To distinguish these two cases add a kernel internal opcode and mark call insns with it. This opcode is seen by interpreter only. JITs will never see it. Also add tiny bit of debug code to aid interpreter debugging. Signed-off-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: Daniel Borkmann --- include/linux/filter.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'include/linux/filter.h') diff --git a/include/linux/filter.h b/include/linux/filter.h index 5feb441d3dd9..f26e6da1007b 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -58,6 +58,9 @@ struct bpf_prog_aux; /* unused opcode to mark special call to bpf_tail_call() helper */ #define BPF_TAIL_CALL 0xf0 +/* unused opcode to mark call to interpreter with arguments */ +#define BPF_CALL_ARGS 0xe0 + /* As per nm, we expose JITed images as text (code) section for * kallsyms. That way, tools like perf can find it to match * addresses. @@ -710,6 +713,9 @@ bool sk_filter_charge(struct sock *sk, struct sk_filter *fp); void sk_filter_uncharge(struct sock *sk, struct sk_filter *fp); u64 __bpf_call_base(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5); +#define __bpf_call_base_args \ + ((u64 (*)(u64, u64, u64, u64, u64, const struct bpf_insn *)) \ + __bpf_call_base) struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog); void bpf_jit_compile(struct bpf_prog *prog); -- cgit v1.2.3 From 60b58afc96c9df71871df2dbad42037757ceef26 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Thu, 14 Dec 2017 17:55:14 -0800 Subject: bpf: fix net.core.bpf_jit_enable race global bpf_jit_enable variable is tested multiple times in JITs, blinding and verifier core. The malicious root can try to toggle it while loading the programs. This race condition was accounted for and there should be no issues, but it's safer to avoid this race condition. Signed-off-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: Daniel Borkmann --- arch/arm/net/bpf_jit_32.c | 2 +- arch/arm64/net/bpf_jit_comp.c | 2 +- arch/mips/net/ebpf_jit.c | 2 +- arch/powerpc/net/bpf_jit_comp64.c | 2 +- arch/s390/net/bpf_jit_comp.c | 2 +- arch/sparc/net/bpf_jit_comp_64.c | 2 +- arch/x86/net/bpf_jit_comp.c | 2 +- include/linux/filter.h | 5 +++-- kernel/bpf/core.c | 3 ++- kernel/bpf/verifier.c | 2 +- 10 files changed, 13 insertions(+), 11 deletions(-) (limited to 'include/linux/filter.h') diff --git a/arch/arm/net/bpf_jit_32.c b/arch/arm/net/bpf_jit_32.c index c199990e12b6..4425189bb24c 100644 --- a/arch/arm/net/bpf_jit_32.c +++ b/arch/arm/net/bpf_jit_32.c @@ -1824,7 +1824,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) /* If BPF JIT was not enabled then we must fall back to * the interpreter. */ - if (!bpf_jit_enable) + if (!prog->jit_requested) return orig_prog; /* If constant blinding was enabled and we failed during blinding diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c index ba38d403abb2..288137cb0871 100644 --- a/arch/arm64/net/bpf_jit_comp.c +++ b/arch/arm64/net/bpf_jit_comp.c @@ -844,7 +844,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) int image_size; u8 *image_ptr; - if (!bpf_jit_enable) + if (!prog->jit_requested) return orig_prog; tmp = bpf_jit_blind_constants(prog); diff --git a/arch/mips/net/ebpf_jit.c b/arch/mips/net/ebpf_jit.c index 962b0259b4b6..97069a1b6f43 100644 --- a/arch/mips/net/ebpf_jit.c +++ b/arch/mips/net/ebpf_jit.c @@ -1869,7 +1869,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) unsigned int image_size; u8 *image_ptr; - if (!bpf_jit_enable || !cpu_has_mips64r2) + if (!prog->jit_requested || !cpu_has_mips64r2) return prog; tmp = bpf_jit_blind_constants(prog); diff --git a/arch/powerpc/net/bpf_jit_comp64.c b/arch/powerpc/net/bpf_jit_comp64.c index 46d74e81aff1..d5a5bc43cf8f 100644 --- a/arch/powerpc/net/bpf_jit_comp64.c +++ b/arch/powerpc/net/bpf_jit_comp64.c @@ -993,7 +993,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp) struct bpf_prog *tmp_fp; bool bpf_blinded = false; - if (!bpf_jit_enable) + if (!fp->jit_requested) return org_fp; tmp_fp = bpf_jit_blind_constants(org_fp); diff --git a/arch/s390/net/bpf_jit_comp.c b/arch/s390/net/bpf_jit_comp.c index e81c16838b90..f4baa8c514d3 100644 --- a/arch/s390/net/bpf_jit_comp.c +++ b/arch/s390/net/bpf_jit_comp.c @@ -1300,7 +1300,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp) struct bpf_jit jit; int pass; - if (!bpf_jit_enable) + if (!fp->jit_requested) return orig_fp; tmp = bpf_jit_blind_constants(fp); diff --git a/arch/sparc/net/bpf_jit_comp_64.c b/arch/sparc/net/bpf_jit_comp_64.c index 5765e7e711f7..a2f1b5e774a7 100644 --- a/arch/sparc/net/bpf_jit_comp_64.c +++ b/arch/sparc/net/bpf_jit_comp_64.c @@ -1517,7 +1517,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) u8 *image_ptr; int pass; - if (!bpf_jit_enable) + if (!prog->jit_requested) return orig_prog; tmp = bpf_jit_blind_constants(prog); diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index 0554e8aef4d5..68859b58ab84 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1121,7 +1121,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) int pass; int i; - if (!bpf_jit_enable) + if (!prog->jit_requested) return orig_prog; tmp = bpf_jit_blind_constants(prog); diff --git a/include/linux/filter.h b/include/linux/filter.h index f26e6da1007b..3d6edc34932c 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -458,6 +458,7 @@ struct bpf_binary_header { struct bpf_prog { u16 pages; /* Number of allocated pages */ u16 jited:1, /* Is our filter JIT'ed? */ + jit_requested:1,/* archs need to JIT the prog */ locked:1, /* Program image locked? */ gpl_compatible:1, /* Is filter GPL compatible? */ cb_access:1, /* Is control block accessed? */ @@ -804,7 +805,7 @@ static inline bool bpf_prog_ebpf_jited(const struct bpf_prog *fp) return fp->jited && bpf_jit_is_ebpf(); } -static inline bool bpf_jit_blinding_enabled(void) +static inline bool bpf_jit_blinding_enabled(struct bpf_prog *prog) { /* These are the prerequisites, should someone ever have the * idea to call blinding outside of them, we make sure to @@ -812,7 +813,7 @@ static inline bool bpf_jit_blinding_enabled(void) */ if (!bpf_jit_is_ebpf()) return false; - if (!bpf_jit_enable) + if (!prog->jit_requested) return false; if (!bpf_jit_harden) return false; diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index dc12c4fd006e..bda911644b1c 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -94,6 +94,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) fp->pages = size / PAGE_SIZE; fp->aux = aux; fp->aux->prog = fp; + fp->jit_requested = ebpf_jit_enabled(); INIT_LIST_HEAD_RCU(&fp->aux->ksym_lnode); @@ -721,7 +722,7 @@ struct bpf_prog *bpf_jit_blind_constants(struct bpf_prog *prog) struct bpf_insn *insn; int i, rewritten; - if (!bpf_jit_blinding_enabled()) + if (!bpf_jit_blinding_enabled(prog)) return prog; clone = bpf_prog_clone_create(prog, GFP_USER); diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index cdc1f043c69b..8e0e4cd0d5e4 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -5080,7 +5080,7 @@ static int fixup_bpf_calls(struct bpf_verifier_env *env) /* BPF_EMIT_CALL() assumptions in some of the map_gen_lookup * handlers are currently limited to 64 bit only. */ - if (ebpf_jit_enabled() && BITS_PER_LONG == 64 && + if (prog->jit_requested && BITS_PER_LONG == 64 && insn->imm == BPF_FUNC_map_lookup_elem) { map_ptr = env->insn_aux_data[i + delta].map_ptr; if (map_ptr == BPF_MAP_PTR_POISON || -- cgit v1.2.3 From 1c2a088a6626d4f51d2f2c97b0cbedbfbf3637f6 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Thu, 14 Dec 2017 17:55:15 -0800 Subject: bpf: x64: add JIT support for multi-function programs Typical JIT does several passes over bpf instructions to compute total size and relative offsets of jumps and calls. With multitple bpf functions calling each other all relative calls will have invalid offsets intially therefore we need to additional last pass over the program to emit calls with correct offsets. For example in case of three bpf functions: main: call foo call bpf_map_lookup exit foo: call bar exit bar: exit We will call bpf_int_jit_compile() indepedently for main(), foo() and bar() x64 JIT typically does 4-5 passes to converge. After these initial passes the image for these 3 functions will be good except call targets, since start addresses of foo() and bar() are unknown when we were JITing main() (note that call bpf_map_lookup will be resolved properly during initial passes). Once start addresses of 3 functions are known we patch call_insn->imm to point to right functions and call bpf_int_jit_compile() again which needs only one pass. Additional safety checks are done to make sure this last pass doesn't produce image that is larger or smaller than previous pass. When constant blinding is on it's applied to all functions at the first pass, since doing it once again at the last pass can change size of the JITed code. Tested on x64 and arm64 hw with JIT on/off, blinding on/off. x64 jits bpf-to-bpf calls correctly while arm64 falls back to interpreter. All other JITs that support normal BPF_CALL will behave the same way since bpf-to-bpf call is equivalent to bpf-to-kernel call from JITs point of view. Signed-off-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: Daniel Borkmann --- arch/x86/net/bpf_jit_comp.c | 47 ++++++++++++++-- include/linux/bpf.h | 3 ++ include/linux/bpf_verifier.h | 1 + include/linux/filter.h | 2 + kernel/bpf/core.c | 13 ++++- kernel/bpf/syscall.c | 3 +- kernel/bpf/verifier.c | 126 +++++++++++++++++++++++++++++++++++++++++++ 7 files changed, 189 insertions(+), 6 deletions(-) (limited to 'include/linux/filter.h') diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index 68859b58ab84..87f214fbe66e 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1109,13 +1109,23 @@ common_load: return proglen; } +struct x64_jit_data { + struct bpf_binary_header *header; + int *addrs; + u8 *image; + int proglen; + struct jit_context ctx; +}; + struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) { struct bpf_binary_header *header = NULL; struct bpf_prog *tmp, *orig_prog = prog; + struct x64_jit_data *jit_data; int proglen, oldproglen = 0; struct jit_context ctx = {}; bool tmp_blinded = false; + bool extra_pass = false; u8 *image = NULL; int *addrs; int pass; @@ -1135,10 +1145,28 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) prog = tmp; } + jit_data = prog->aux->jit_data; + if (!jit_data) { + jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL); + if (!jit_data) { + prog = orig_prog; + goto out; + } + prog->aux->jit_data = jit_data; + } + addrs = jit_data->addrs; + if (addrs) { + ctx = jit_data->ctx; + oldproglen = jit_data->proglen; + image = jit_data->image; + header = jit_data->header; + extra_pass = true; + goto skip_init_addrs; + } addrs = kmalloc(prog->len * sizeof(*addrs), GFP_KERNEL); if (!addrs) { prog = orig_prog; - goto out; + goto out_addrs; } /* Before first pass, make a rough estimation of addrs[] @@ -1149,6 +1177,7 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) addrs[i] = proglen; } ctx.cleanup_addr = proglen; +skip_init_addrs: /* JITed image shrinks with every pass and the loop iterates * until the image stops shrinking. Very large bpf programs @@ -1189,7 +1218,15 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) if (image) { bpf_flush_icache(header, image + proglen); - bpf_jit_binary_lock_ro(header); + if (!prog->is_func || extra_pass) { + bpf_jit_binary_lock_ro(header); + } else { + jit_data->addrs = addrs; + jit_data->ctx = ctx; + jit_data->proglen = proglen; + jit_data->image = image; + jit_data->header = header; + } prog->bpf_func = (void *)image; prog->jited = 1; prog->jited_len = proglen; @@ -1197,8 +1234,12 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog) prog = orig_prog; } + if (!prog->is_func || extra_pass) { out_addrs: - kfree(addrs); + kfree(addrs); + kfree(jit_data); + prog->aux->jit_data = NULL; + } out: if (tmp_blinded) bpf_jit_prog_release_other(prog, prog == orig_prog ? diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 8935f6f63d5f..da54ef644fcd 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -200,6 +200,9 @@ struct bpf_prog_aux { u32 max_ctx_offset; u32 stack_depth; u32 id; + u32 func_cnt; + struct bpf_prog **func; + void *jit_data; /* JIT specific data. arch dependent */ struct latch_tree_node ksym_tnode; struct list_head ksym_lnode; const struct bpf_prog_ops *ops; diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h index 585d4e17ea88..aaac589e490c 100644 --- a/include/linux/bpf_verifier.h +++ b/include/linux/bpf_verifier.h @@ -143,6 +143,7 @@ struct bpf_insn_aux_data { union { enum bpf_reg_type ptr_type; /* pointer type for load/store insns */ struct bpf_map *map_ptr; /* pointer for call insn into lookup_elem */ + s32 call_imm; /* saved imm field of call insn */ }; int ctx_field_size; /* the ctx field size for load insn, maybe 0 */ bool seen; /* this insn was processed by the verifier */ diff --git a/include/linux/filter.h b/include/linux/filter.h index 3d6edc34932c..e872b4ebaa57 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -463,6 +463,8 @@ struct bpf_prog { gpl_compatible:1, /* Is filter GPL compatible? */ cb_access:1, /* Is control block accessed? */ dst_needed:1, /* Do we need dst entry? */ + blinded:1, /* Was blinded */ + is_func:1, /* program is a bpf function */ kprobe_override:1; /* Do we override a kprobe? */ enum bpf_prog_type type; /* Type of BPF program */ u32 len; /* Number of filter blocks */ diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index bda911644b1c..768e0a02d8c8 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -722,7 +722,7 @@ struct bpf_prog *bpf_jit_blind_constants(struct bpf_prog *prog) struct bpf_insn *insn; int i, rewritten; - if (!bpf_jit_blinding_enabled(prog)) + if (!bpf_jit_blinding_enabled(prog) || prog->blinded) return prog; clone = bpf_prog_clone_create(prog, GFP_USER); @@ -764,6 +764,7 @@ struct bpf_prog *bpf_jit_blind_constants(struct bpf_prog *prog) i += insn_delta; } + clone->blinded = 1; return clone; } #endif /* CONFIG_BPF_JIT */ @@ -1629,11 +1630,19 @@ int bpf_prog_array_copy_info(struct bpf_prog_array __rcu *array, static void bpf_prog_free_deferred(struct work_struct *work) { struct bpf_prog_aux *aux; + int i; aux = container_of(work, struct bpf_prog_aux, work); if (bpf_prog_is_dev_bound(aux)) bpf_prog_offload_destroy(aux->prog); - bpf_jit_free(aux->prog); + for (i = 0; i < aux->func_cnt; i++) + bpf_jit_free(aux->func[i]); + if (aux->func_cnt) { + kfree(aux->func); + bpf_prog_unlock_free(aux->prog); + } else { + bpf_jit_free(aux->prog); + } } /* Free internal BPF program */ diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index 2c4cfeaa8d5e..e2e1c78ce1dc 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1194,7 +1194,8 @@ static int bpf_prog_load(union bpf_attr *attr) goto free_used_maps; /* eBPF program is ready to be JITed */ - prog = bpf_prog_select_runtime(prog, &err); + if (!prog->bpf_func) + prog = bpf_prog_select_runtime(prog, &err); if (err < 0) goto free_used_maps; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 8e0e4cd0d5e4..48b2901cf483 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -5012,12 +5012,138 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env) return 0; } +static int jit_subprogs(struct bpf_verifier_env *env) +{ + struct bpf_prog *prog = env->prog, **func, *tmp; + int i, j, subprog_start, subprog_end = 0, len, subprog; + struct bpf_insn *insn = prog->insnsi; + void *old_bpf_func; + int err = -ENOMEM; + + if (env->subprog_cnt == 0) + return 0; + + for (i = 0; i < prog->len; i++, insn++) { + if (insn->code != (BPF_JMP | BPF_CALL) || + insn->src_reg != BPF_PSEUDO_CALL) + continue; + subprog = find_subprog(env, i + insn->imm + 1); + if (subprog < 0) { + WARN_ONCE(1, "verifier bug. No program starts at insn %d\n", + i + insn->imm + 1); + return -EFAULT; + } + /* temporarily remember subprog id inside insn instead of + * aux_data, since next loop will split up all insns into funcs + */ + insn->off = subprog + 1; + /* remember original imm in case JIT fails and fallback + * to interpreter will be needed + */ + env->insn_aux_data[i].call_imm = insn->imm; + /* point imm to __bpf_call_base+1 from JITs point of view */ + insn->imm = 1; + } + + func = kzalloc(sizeof(prog) * (env->subprog_cnt + 1), GFP_KERNEL); + if (!func) + return -ENOMEM; + + for (i = 0; i <= env->subprog_cnt; i++) { + subprog_start = subprog_end; + if (env->subprog_cnt == i) + subprog_end = prog->len; + else + subprog_end = env->subprog_starts[i]; + + len = subprog_end - subprog_start; + func[i] = bpf_prog_alloc(bpf_prog_size(len), GFP_USER); + if (!func[i]) + goto out_free; + memcpy(func[i]->insnsi, &prog->insnsi[subprog_start], + len * sizeof(struct bpf_insn)); + func[i]->len = len; + func[i]->is_func = 1; + /* Use bpf_prog_F_tag to indicate functions in stack traces. + * Long term would need debug info to populate names + */ + func[i]->aux->name[0] = 'F'; + func[i]->aux->stack_depth = env->subprog_stack_depth[i]; + func[i]->jit_requested = 1; + func[i] = bpf_int_jit_compile(func[i]); + if (!func[i]->jited) { + err = -ENOTSUPP; + goto out_free; + } + cond_resched(); + } + /* at this point all bpf functions were successfully JITed + * now populate all bpf_calls with correct addresses and + * run last pass of JIT + */ + for (i = 0; i <= env->subprog_cnt; i++) { + insn = func[i]->insnsi; + for (j = 0; j < func[i]->len; j++, insn++) { + if (insn->code != (BPF_JMP | BPF_CALL) || + insn->src_reg != BPF_PSEUDO_CALL) + continue; + subprog = insn->off; + insn->off = 0; + insn->imm = (u64 (*)(u64, u64, u64, u64, u64)) + func[subprog]->bpf_func - + __bpf_call_base; + } + } + for (i = 0; i <= env->subprog_cnt; i++) { + old_bpf_func = func[i]->bpf_func; + tmp = bpf_int_jit_compile(func[i]); + if (tmp != func[i] || func[i]->bpf_func != old_bpf_func) { + verbose(env, "JIT doesn't support bpf-to-bpf calls\n"); + err = -EFAULT; + goto out_free; + } + cond_resched(); + } + + /* finally lock prog and jit images for all functions and + * populate kallsysm + */ + for (i = 0; i <= env->subprog_cnt; i++) { + bpf_prog_lock_ro(func[i]); + bpf_prog_kallsyms_add(func[i]); + } + prog->jited = 1; + prog->bpf_func = func[0]->bpf_func; + prog->aux->func = func; + prog->aux->func_cnt = env->subprog_cnt + 1; + return 0; +out_free: + for (i = 0; i <= env->subprog_cnt; i++) + if (func[i]) + bpf_jit_free(func[i]); + kfree(func); + /* cleanup main prog to be interpreted */ + prog->jit_requested = 0; + for (i = 0, insn = prog->insnsi; i < prog->len; i++, insn++) { + if (insn->code != (BPF_JMP | BPF_CALL) || + insn->src_reg != BPF_PSEUDO_CALL) + continue; + insn->off = 0; + insn->imm = env->insn_aux_data[i].call_imm; + } + return err; +} + static int fixup_call_args(struct bpf_verifier_env *env) { struct bpf_prog *prog = env->prog; struct bpf_insn *insn = prog->insnsi; int i, depth; + if (env->prog->jit_requested) + if (jit_subprogs(env) == 0) + return 0; + for (i = 0; i < prog->len; i++, insn++) { if (insn->code != (BPF_JMP | BPF_CALL) || insn->src_reg != BPF_PSEUDO_CALL) -- cgit v1.2.3