diff options
Diffstat (limited to 'kernel/bpf')
-rw-r--r-- | kernel/bpf/core.c | 34 | ||||
-rw-r--r-- | kernel/bpf/devmap.c | 6 | ||||
-rw-r--r-- | kernel/bpf/disasm.c | 16 | ||||
-rw-r--r-- | kernel/bpf/hashtab.c | 4 | ||||
-rw-r--r-- | kernel/bpf/helpers.c | 30 | ||||
-rw-r--r-- | kernel/bpf/verifier.c | 211 |
6 files changed, 139 insertions, 162 deletions
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 034ad93a1ad7..0a28a8095d3e 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -32,6 +32,8 @@ #include <linux/perf_event.h> #include <linux/extable.h> #include <linux/log2.h> + +#include <asm/barrier.h> #include <asm/unaligned.h> /* Registers */ @@ -1360,11 +1362,13 @@ u64 __weak bpf_probe_read_kernel(void *dst, u32 size, const void *unsafe_ptr) } /** - * __bpf_prog_run - run eBPF program on a given context + * ___bpf_prog_run - run eBPF program on a given context * @regs: is the array of MAX_BPF_EXT_REG eBPF pseudo-registers * @insn: is the array of eBPF instructions * * Decode and execute eBPF instructions. + * + * Return: whatever value is in %BPF_R0 at program exit */ static u64 ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn) { @@ -1377,6 +1381,7 @@ static u64 ___bpf_prog_run(u64 *regs, const struct bpf_insn *insn) /* Non-UAPI available opcodes. */ [BPF_JMP | BPF_CALL_ARGS] = &&JMP_CALL_ARGS, [BPF_JMP | BPF_TAIL_CALL] = &&JMP_TAIL_CALL, + [BPF_ST | BPF_NOSPEC] = &&ST_NOSPEC, [BPF_LDX | BPF_PROBE_MEM | BPF_B] = &&LDX_PROBE_MEM_B, [BPF_LDX | BPF_PROBE_MEM | BPF_H] = &&LDX_PROBE_MEM_H, [BPF_LDX | BPF_PROBE_MEM | BPF_W] = &&LDX_PROBE_MEM_W, @@ -1621,7 +1626,21 @@ out: COND_JMP(s, JSGE, >=) COND_JMP(s, JSLE, <=) #undef COND_JMP - /* STX and ST and LDX*/ + /* ST, STX and LDX*/ + ST_NOSPEC: + /* Speculation barrier for mitigating Speculative Store Bypass. + * In case of arm64, we rely on the firmware mitigation as + * controlled via the ssbd kernel parameter. Whenever the + * mitigation is enabled, it works for all of the kernel code + * with no need to provide any additional instructions here. + * In case of x86, we use 'lfence' insn for mitigation. We + * reuse preexisting logic from Spectre v1 mitigation that + * happens to produce the required code on x86 for v4 as well. + */ +#ifdef CONFIG_X86 + barrier_nospec(); +#endif + CONT; #define LDST(SIZEOP, SIZE) \ STX_MEM_##SIZEOP: \ *(SIZE *)(unsigned long) (DST + insn->off) = SRC; \ @@ -1861,6 +1880,9 @@ static void bpf_prog_select_func(struct bpf_prog *fp) * * Try to JIT eBPF program, if JIT is not available, use interpreter. * The BPF program will be executed via BPF_PROG_RUN() macro. + * + * Return: the &fp argument along with &err set to 0 for success or + * a negative errno code on failure */ struct bpf_prog *bpf_prog_select_runtime(struct bpf_prog *fp, int *err) { @@ -2236,8 +2258,14 @@ static void bpf_prog_free_deferred(struct work_struct *work) #endif if (aux->dst_trampoline) bpf_trampoline_put(aux->dst_trampoline); - for (i = 0; i < aux->func_cnt; i++) + for (i = 0; i < aux->func_cnt; i++) { + /* We can just unlink the subprog poke descriptor table as + * it was originally linked to the main program and is also + * released along with it. + */ + aux->func[i]->aux->poke_tab = NULL; bpf_jit_free(aux->func[i]); + } if (aux->func_cnt) { kfree(aux->func); bpf_prog_unlock_free(aux->prog); diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c index 2546dafd6672..fdc20892837c 100644 --- a/kernel/bpf/devmap.c +++ b/kernel/bpf/devmap.c @@ -558,7 +558,8 @@ int dev_map_enqueue_multi(struct xdp_buff *xdp, struct net_device *dev_rx, if (map->map_type == BPF_MAP_TYPE_DEVMAP) { for (i = 0; i < map->max_entries; i++) { - dst = READ_ONCE(dtab->netdev_map[i]); + dst = rcu_dereference_check(dtab->netdev_map[i], + rcu_read_lock_bh_held()); if (!is_valid_dst(dst, xdp, exclude_ifindex)) continue; @@ -654,7 +655,8 @@ int dev_map_redirect_multi(struct net_device *dev, struct sk_buff *skb, if (map->map_type == BPF_MAP_TYPE_DEVMAP) { for (i = 0; i < map->max_entries; i++) { - dst = READ_ONCE(dtab->netdev_map[i]); + dst = rcu_dereference_check(dtab->netdev_map[i], + rcu_read_lock_bh_held()); if (!dst || dst->dev->ifindex == exclude_ifindex) continue; diff --git a/kernel/bpf/disasm.c b/kernel/bpf/disasm.c index bbfc6bb79240..ca3cd9aaa6ce 100644 --- a/kernel/bpf/disasm.c +++ b/kernel/bpf/disasm.c @@ -206,15 +206,17 @@ void print_bpf_insn(const struct bpf_insn_cbs *cbs, verbose(cbs->private_data, "BUG_%02x\n", insn->code); } } else if (class == BPF_ST) { - if (BPF_MODE(insn->code) != BPF_MEM) { + if (BPF_MODE(insn->code) == BPF_MEM) { + verbose(cbs->private_data, "(%02x) *(%s *)(r%d %+d) = %d\n", + insn->code, + bpf_ldst_string[BPF_SIZE(insn->code) >> 3], + insn->dst_reg, + insn->off, insn->imm); + } else if (BPF_MODE(insn->code) == 0xc0 /* BPF_NOSPEC, no UAPI */) { + verbose(cbs->private_data, "(%02x) nospec\n", insn->code); + } else { verbose(cbs->private_data, "BUG_st_%02x\n", insn->code); - return; } - verbose(cbs->private_data, "(%02x) *(%s *)(r%d %+d) = %d\n", - insn->code, - bpf_ldst_string[BPF_SIZE(insn->code) >> 3], - insn->dst_reg, - insn->off, insn->imm); } else if (class == BPF_LDX) { if (BPF_MODE(insn->code) != BPF_MEM) { verbose(cbs->private_data, "BUG_ldx_%02x\n", insn->code); diff --git a/kernel/bpf/hashtab.c b/kernel/bpf/hashtab.c index 72c58cc516a3..9c011f3a2687 100644 --- a/kernel/bpf/hashtab.c +++ b/kernel/bpf/hashtab.c @@ -1565,8 +1565,8 @@ alloc: /* We cannot do copy_from_user or copy_to_user inside * the rcu_read_lock. Allocate enough space here. */ - keys = kvmalloc(key_size * bucket_size, GFP_USER | __GFP_NOWARN); - values = kvmalloc(value_size * bucket_size, GFP_USER | __GFP_NOWARN); + keys = kvmalloc_array(key_size, bucket_size, GFP_USER | __GFP_NOWARN); + values = kvmalloc_array(value_size, bucket_size, GFP_USER | __GFP_NOWARN); if (!keys || !values) { ret = -ENOMEM; goto after_loop; diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c index 62cf00383910..55f83ea09dae 100644 --- a/kernel/bpf/helpers.c +++ b/kernel/bpf/helpers.c @@ -353,9 +353,15 @@ const struct bpf_func_proto bpf_jiffies64_proto = { #ifdef CONFIG_CGROUPS BPF_CALL_0(bpf_get_current_cgroup_id) { - struct cgroup *cgrp = task_dfl_cgroup(current); + struct cgroup *cgrp; + u64 cgrp_id; - return cgroup_id(cgrp); + rcu_read_lock(); + cgrp = task_dfl_cgroup(current); + cgrp_id = cgroup_id(cgrp); + rcu_read_unlock(); + + return cgrp_id; } const struct bpf_func_proto bpf_get_current_cgroup_id_proto = { @@ -366,13 +372,17 @@ const struct bpf_func_proto bpf_get_current_cgroup_id_proto = { BPF_CALL_1(bpf_get_current_ancestor_cgroup_id, int, ancestor_level) { - struct cgroup *cgrp = task_dfl_cgroup(current); + struct cgroup *cgrp; struct cgroup *ancestor; + u64 cgrp_id; + rcu_read_lock(); + cgrp = task_dfl_cgroup(current); ancestor = cgroup_ancestor(cgrp, ancestor_level); - if (!ancestor) - return 0; - return cgroup_id(ancestor); + cgrp_id = ancestor ? cgroup_id(ancestor) : 0; + rcu_read_unlock(); + + return cgrp_id; } const struct bpf_func_proto bpf_get_current_ancestor_cgroup_id_proto = { @@ -397,8 +407,8 @@ BPF_CALL_2(bpf_get_local_storage, struct bpf_map *, map, u64, flags) void *ptr; int i; - for (i = 0; i < BPF_CGROUP_STORAGE_NEST_MAX; i++) { - if (unlikely(this_cpu_read(bpf_cgroup_storage_info[i].task) != current)) + for (i = BPF_CGROUP_STORAGE_NEST_MAX - 1; i >= 0; i--) { + if (likely(this_cpu_read(bpf_cgroup_storage_info[i].task) != current)) continue; storage = this_cpu_read(bpf_cgroup_storage_info[i].storage[stype]); @@ -1070,12 +1080,12 @@ bpf_base_func_proto(enum bpf_func_id func_id) case BPF_FUNC_probe_read_user: return &bpf_probe_read_user_proto; case BPF_FUNC_probe_read_kernel: - return security_locked_down(LOCKDOWN_BPF_READ) < 0 ? + return security_locked_down(LOCKDOWN_BPF_READ_KERNEL) < 0 ? NULL : &bpf_probe_read_kernel_proto; case BPF_FUNC_probe_read_user_str: return &bpf_probe_read_user_str_proto; case BPF_FUNC_probe_read_kernel_str: - return security_locked_down(LOCKDOWN_BPF_READ) < 0 ? + return security_locked_down(LOCKDOWN_BPF_READ_KERNEL) < 0 ? NULL : &bpf_probe_read_kernel_str_proto; case BPF_FUNC_snprintf_btf: return &bpf_snprintf_btf_proto; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index be38bb930bf1..381d3d6f24bc 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -2610,6 +2610,19 @@ static int check_stack_write_fixed_off(struct bpf_verifier_env *env, cur = env->cur_state->frame[env->cur_state->curframe]; if (value_regno >= 0) reg = &cur->regs[value_regno]; + if (!env->bypass_spec_v4) { + bool sanitize = reg && is_spillable_regtype(reg->type); + + for (i = 0; i < size; i++) { + if (state->stack[spi].slot_type[i] == STACK_INVALID) { + sanitize = true; + break; + } + } + + if (sanitize) + env->insn_aux_data[insn_idx].sanitize_stack_spill = true; + } if (reg && size == BPF_REG_SIZE && register_is_bounded(reg) && !register_is_null(reg) && env->bpf_capable) { @@ -2632,47 +2645,10 @@ static int check_stack_write_fixed_off(struct bpf_verifier_env *env, verbose(env, "invalid size of register spill\n"); return -EACCES; } - if (state != cur && reg->type == PTR_TO_STACK) { verbose(env, "cannot spill pointers to stack into stack frame of the caller\n"); return -EINVAL; } - - if (!env->bypass_spec_v4) { - bool sanitize = false; - - if (state->stack[spi].slot_type[0] == STACK_SPILL && - register_is_const(&state->stack[spi].spilled_ptr)) - sanitize = true; - for (i = 0; i < BPF_REG_SIZE; i++) - if (state->stack[spi].slot_type[i] == STACK_MISC) { - sanitize = true; - break; - } - if (sanitize) { - int *poff = &env->insn_aux_data[insn_idx].sanitize_stack_off; - int soff = (-spi - 1) * BPF_REG_SIZE; - - /* detected reuse of integer stack slot with a pointer - * which means either llvm is reusing stack slot or - * an attacker is trying to exploit CVE-2018-3639 - * (speculative store bypass) - * Have to sanitize that slot with preemptive - * store of zero. - */ - if (*poff && *poff != soff) { - /* disallow programs where single insn stores - * into two different stack slots, since verifier - * cannot sanitize them - */ - verbose(env, - "insn %d cannot access two stack slots fp%d and fp%d", - insn_idx, *poff, soff); - return -EINVAL; - } - *poff = soff; - } - } save_register_state(state, spi, reg); } else { u8 type = STACK_MISC; @@ -3677,6 +3653,8 @@ continue_func: if (tail_call_reachable) for (j = 0; j < frame; j++) subprog[ret_prog[j]].tail_call_reachable = true; + if (subprog[0].tail_call_reachable) + env->prog->aux->tail_call_reachable = true; /* end of for() loop means the last insn of the 'subprog' * was reached. Doesn't matter whether it was JA or EXIT @@ -6559,6 +6537,12 @@ static int sanitize_ptr_alu(struct bpf_verifier_env *env, alu_state |= off_is_imm ? BPF_ALU_IMMEDIATE : 0; alu_state |= ptr_is_dst_reg ? BPF_ALU_SANITIZE_SRC : BPF_ALU_SANITIZE_DST; + + /* Limit pruning on unknown scalars to enable deep search for + * potential masking differences from other program paths. + */ + if (!off_is_imm) + env->explore_alu_limits = true; } err = update_alu_sanitation_state(aux, alu_state, alu_limit); @@ -9934,8 +9918,8 @@ next: } /* Returns true if (rold safe implies rcur safe) */ -static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur, - struct bpf_id_pair *idmap) +static bool regsafe(struct bpf_verifier_env *env, struct bpf_reg_state *rold, + struct bpf_reg_state *rcur, struct bpf_id_pair *idmap) { bool equal; @@ -9961,6 +9945,8 @@ static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur, return false; switch (rold->type) { case SCALAR_VALUE: + if (env->explore_alu_limits) + return false; if (rcur->type == SCALAR_VALUE) { if (!rold->precise && !rcur->precise) return true; @@ -10051,9 +10037,8 @@ static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur, return false; } -static bool stacksafe(struct bpf_func_state *old, - struct bpf_func_state *cur, - struct bpf_id_pair *idmap) +static bool stacksafe(struct bpf_verifier_env *env, struct bpf_func_state *old, + struct bpf_func_state *cur, struct bpf_id_pair *idmap) { int i, spi; @@ -10098,9 +10083,8 @@ static bool stacksafe(struct bpf_func_state *old, continue; if (old->stack[spi].slot_type[0] != STACK_SPILL) continue; - if (!regsafe(&old->stack[spi].spilled_ptr, - &cur->stack[spi].spilled_ptr, - idmap)) + if (!regsafe(env, &old->stack[spi].spilled_ptr, + &cur->stack[spi].spilled_ptr, idmap)) /* when explored and current stack slot are both storing * spilled registers, check that stored pointers types * are the same as well. @@ -10157,10 +10141,11 @@ static bool func_states_equal(struct bpf_verifier_env *env, struct bpf_func_stat memset(env->idmap_scratch, 0, sizeof(env->idmap_scratch)); for (i = 0; i < MAX_BPF_REG; i++) - if (!regsafe(&old->regs[i], &cur->regs[i], env->idmap_scratch)) + if (!regsafe(env, &old->regs[i], &cur->regs[i], + env->idmap_scratch)) return false; - if (!stacksafe(old, cur, env->idmap_scratch)) + if (!stacksafe(env, old, cur, env->idmap_scratch)) return false; if (!refsafe(old, cur)) @@ -11678,6 +11663,7 @@ static void sanitize_dead_code(struct bpf_verifier_env *env) if (aux_data[i].seen) continue; memcpy(insn + i, &trap, sizeof(trap)); + aux_data[i].zext_dst = false; } } @@ -11904,35 +11890,33 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env) for (i = 0; i < insn_cnt; i++, insn++) { bpf_convert_ctx_access_t convert_ctx_access; + bool ctx_access; if (insn->code == (BPF_LDX | BPF_MEM | BPF_B) || insn->code == (BPF_LDX | BPF_MEM | BPF_H) || insn->code == (BPF_LDX | BPF_MEM | BPF_W) || - insn->code == (BPF_LDX | BPF_MEM | BPF_DW)) + insn->code == (BPF_LDX | BPF_MEM | BPF_DW)) { type = BPF_READ; - else if (insn->code == (BPF_STX | BPF_MEM | BPF_B) || - insn->code == (BPF_STX | BPF_MEM | BPF_H) || - insn->code == (BPF_STX | BPF_MEM | BPF_W) || - insn->code == (BPF_STX | BPF_MEM | BPF_DW)) + ctx_access = true; + } else if (insn->code == (BPF_STX | BPF_MEM | BPF_B) || + insn->code == (BPF_STX | BPF_MEM | BPF_H) || + insn->code == (BPF_STX | BPF_MEM | BPF_W) || + insn->code == (BPF_STX | BPF_MEM | BPF_DW) || + insn->code == (BPF_ST | BPF_MEM | BPF_B) || + insn->code == (BPF_ST | BPF_MEM | BPF_H) || + insn->code == (BPF_ST | BPF_MEM | BPF_W) || + insn->code == (BPF_ST | BPF_MEM | BPF_DW)) { type = BPF_WRITE; - else + ctx_access = BPF_CLASS(insn->code) == BPF_STX; + } else { continue; + } if (type == BPF_WRITE && - env->insn_aux_data[i + delta].sanitize_stack_off) { + env->insn_aux_data[i + delta].sanitize_stack_spill) { struct bpf_insn patch[] = { - /* Sanitize suspicious stack slot with zero. - * There are no memory dependencies for this store, - * since it's only using frame pointer and immediate - * constant of zero - */ - BPF_ST_MEM(BPF_DW, BPF_REG_FP, - env->insn_aux_data[i + delta].sanitize_stack_off, - 0), - /* the original STX instruction will immediately - * overwrite the same stack slot with appropriate value - */ *insn, + BPF_ST_NOSPEC(), }; cnt = ARRAY_SIZE(patch); @@ -11946,6 +11930,9 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env) continue; } + if (!ctx_access) + continue; + switch (env->insn_aux_data[i + delta].ptr_type) { case PTR_TO_CTX: if (!ops->convert_ctx_access) @@ -12121,33 +12108,19 @@ static int jit_subprogs(struct bpf_verifier_env *env) goto out_free; func[i]->is_func = 1; func[i]->aux->func_idx = i; - /* the btf and func_info will be freed only at prog->aux */ + /* Below members will be freed only at prog->aux */ func[i]->aux->btf = prog->aux->btf; func[i]->aux->func_info = prog->aux->func_info; + func[i]->aux->poke_tab = prog->aux->poke_tab; + func[i]->aux->size_poke_tab = prog->aux->size_poke_tab; for (j = 0; j < prog->aux->size_poke_tab; j++) { - u32 insn_idx = prog->aux->poke_tab[j].insn_idx; - int ret; + struct bpf_jit_poke_descriptor *poke; - if (!(insn_idx >= subprog_start && - insn_idx <= subprog_end)) - continue; - - ret = bpf_jit_add_poke_descriptor(func[i], - &prog->aux->poke_tab[j]); - if (ret < 0) { - verbose(env, "adding tail call poke descriptor failed\n"); - goto out_free; - } - - func[i]->insnsi[insn_idx - subprog_start].imm = ret + 1; - - map_ptr = func[i]->aux->poke_tab[ret].tail_call.map; - ret = map_ptr->ops->map_poke_track(map_ptr, func[i]->aux); - if (ret < 0) { - verbose(env, "tracking tail call prog failed\n"); - goto out_free; - } + poke = &prog->aux->poke_tab[j]; + if (poke->insn_idx < subprog_end && + poke->insn_idx >= subprog_start) + poke->aux = func[i]->aux; } /* Use bpf_prog_F_tag to indicate functions in stack traces. @@ -12178,18 +12151,6 @@ static int jit_subprogs(struct bpf_verifier_env *env) cond_resched(); } - /* Untrack main program's aux structs so that during map_poke_run() - * we will not stumble upon the unfilled poke descriptors; each - * of the main program's poke descs got distributed across subprogs - * and got tracked onto map, so we are sure that none of them will - * be missed after the operation below - */ - for (i = 0; i < prog->aux->size_poke_tab; i++) { - map_ptr = prog->aux->poke_tab[i].tail_call.map; - - map_ptr->ops->map_poke_untrack(map_ptr, prog->aux); - } - /* at this point all bpf functions were successfully JITed * now populate all bpf_calls with correct addresses and * run last pass of JIT @@ -12267,14 +12228,22 @@ static int jit_subprogs(struct bpf_verifier_env *env) bpf_prog_jit_attempt_done(prog); return 0; out_free: + /* We failed JIT'ing, so at this point we need to unregister poke + * descriptors from subprogs, so that kernel is not attempting to + * patch it anymore as we're freeing the subprog JIT memory. + */ + for (i = 0; i < prog->aux->size_poke_tab; i++) { + map_ptr = prog->aux->poke_tab[i].tail_call.map; + map_ptr->ops->map_poke_untrack(map_ptr, prog->aux); + } + /* At this point we're guaranteed that poke descriptors are not + * live anymore. We can just unlink its descriptor table as it's + * released with the main prog. + */ for (i = 0; i < env->subprog_cnt; i++) { if (!func[i]) continue; - - for (j = 0; j < func[i]->aux->size_poke_tab; j++) { - map_ptr = func[i]->aux->poke_tab[j].tail_call.map; - map_ptr->ops->map_poke_untrack(map_ptr, func[i]->aux); - } + func[i]->aux->poke_tab = NULL; bpf_jit_free(func[i]); } kfree(func); @@ -12768,37 +12737,6 @@ static void free_states(struct bpf_verifier_env *env) } } -/* The verifier is using insn_aux_data[] to store temporary data during - * verification and to store information for passes that run after the - * verification like dead code sanitization. do_check_common() for subprogram N - * may analyze many other subprograms. sanitize_insn_aux_data() clears all - * temporary data after do_check_common() finds that subprogram N cannot be - * verified independently. pass_cnt counts the number of times - * do_check_common() was run and insn->aux->seen tells the pass number - * insn_aux_data was touched. These variables are compared to clear temporary - * data from failed pass. For testing and experiments do_check_common() can be - * run multiple times even when prior attempt to verify is unsuccessful. - * - * Note that special handling is needed on !env->bypass_spec_v1 if this is - * ever called outside of error path with subsequent program rejection. - */ -static void sanitize_insn_aux_data(struct bpf_verifier_env *env) -{ - struct bpf_insn *insn = env->prog->insnsi; - struct bpf_insn_aux_data *aux; - int i, class; - - for (i = 0; i < env->prog->len; i++) { - class = BPF_CLASS(insn[i].code); - if (class != BPF_LDX && class != BPF_STX) - continue; - aux = &env->insn_aux_data[i]; - if (aux->seen != env->pass_cnt) - continue; - memset(aux, 0, offsetof(typeof(*aux), orig_idx)); - } -} - static int do_check_common(struct bpf_verifier_env *env, int subprog) { bool pop_log = !(env->log.level & BPF_LOG_LEVEL2); @@ -12875,9 +12813,6 @@ out: if (!ret && pop_log) bpf_vlog_reset(&env->log, 0); free_states(env); - if (ret) - /* clean aux data in case subprog was rejected */ - sanitize_insn_aux_data(env); return ret; } |