From 772412176fb98493158929b220fe250127f611af Mon Sep 17 00:00:00 2001 From: Stanislav Fomichev Date: Wed, 27 Jan 2021 11:31:39 -0800 Subject: bpf: Allow rewriting to ports under ip_unprivileged_port_start At the moment, BPF_CGROUP_INET{4,6}_BIND hooks can rewrite user_port to the privileged ones (< ip_unprivileged_port_start), but it will be rejected later on in the __inet_bind or __inet6_bind. Let's add another return value to indicate that CAP_NET_BIND_SERVICE check should be ignored. Use the same idea as we currently use in cgroup/egress where bit #1 indicates CN. Instead, for cgroup/bind{4,6}, bit #1 indicates that CAP_NET_BIND_SERVICE should be bypassed. v5: - rename flags to be less confusing (Andrey Ignatov) - rework BPF_PROG_CGROUP_INET_EGRESS_RUN_ARRAY to work on flags and accept BPF_RET_SET_CN (no behavioral changes) v4: - Add missing IPv6 support (Martin KaFai Lau) v3: - Update description (Martin KaFai Lau) - Fix capability restore in selftest (Martin KaFai Lau) v2: - Switch to explicit return code (Martin KaFai Lau) Signed-off-by: Stanislav Fomichev Signed-off-by: Alexei Starovoitov Reviewed-by: Martin KaFai Lau Acked-by: Andrey Ignatov Link: https://lore.kernel.org/bpf/20210127193140.3170382-1-sdf@google.com --- include/linux/bpf.h | 52 +++++++++++++++++++++++++++++++++------------------- 1 file changed, 33 insertions(+), 19 deletions(-) (limited to 'include/linux/bpf.h') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 1aac2af12fed..321966fc35db 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1073,6 +1073,34 @@ int bpf_prog_array_copy(struct bpf_prog_array *old_array, struct bpf_prog *include_prog, struct bpf_prog_array **new_array); +/* BPF program asks to bypass CAP_NET_BIND_SERVICE in bind. */ +#define BPF_RET_BIND_NO_CAP_NET_BIND_SERVICE (1 << 0) +/* BPF program asks to set CN on the packet. */ +#define BPF_RET_SET_CN (1 << 0) + +#define BPF_PROG_RUN_ARRAY_FLAGS(array, ctx, func, ret_flags) \ + ({ \ + struct bpf_prog_array_item *_item; \ + struct bpf_prog *_prog; \ + struct bpf_prog_array *_array; \ + u32 _ret = 1; \ + u32 func_ret; \ + migrate_disable(); \ + rcu_read_lock(); \ + _array = rcu_dereference(array); \ + _item = &_array->items[0]; \ + while ((_prog = READ_ONCE(_item->prog))) { \ + bpf_cgroup_storage_set(_item->cgroup_storage); \ + func_ret = func(_prog, ctx); \ + _ret &= (func_ret & 1); \ + *(ret_flags) |= (func_ret >> 1); \ + _item++; \ + } \ + rcu_read_unlock(); \ + migrate_enable(); \ + _ret; \ + }) + #define __BPF_PROG_RUN_ARRAY(array, ctx, func, check_non_null) \ ({ \ struct bpf_prog_array_item *_item; \ @@ -1120,25 +1148,11 @@ _out: \ */ #define BPF_PROG_CGROUP_INET_EGRESS_RUN_ARRAY(array, ctx, func) \ ({ \ - struct bpf_prog_array_item *_item; \ - struct bpf_prog *_prog; \ - struct bpf_prog_array *_array; \ - u32 ret; \ - u32 _ret = 1; \ - u32 _cn = 0; \ - migrate_disable(); \ - rcu_read_lock(); \ - _array = rcu_dereference(array); \ - _item = &_array->items[0]; \ - while ((_prog = READ_ONCE(_item->prog))) { \ - bpf_cgroup_storage_set(_item->cgroup_storage); \ - ret = func(_prog, ctx); \ - _ret &= (ret & 1); \ - _cn |= (ret & 2); \ - _item++; \ - } \ - rcu_read_unlock(); \ - migrate_enable(); \ + u32 _flags = 0; \ + bool _cn; \ + u32 _ret; \ + _ret = BPF_PROG_RUN_ARRAY_FLAGS(array, ctx, func, &_flags); \ + _cn = _flags & BPF_RET_SET_CN; \ if (_ret) \ _ret = (_cn ? NET_XMIT_CN : NET_XMIT_SUCCESS); \ else \ -- cgit v1.2.3 From 01f810ace9ed37255f27608a0864abebccf0aab3 Mon Sep 17 00:00:00 2001 From: Andrei Matei Date: Sat, 6 Feb 2021 20:10:24 -0500 Subject: bpf: Allow variable-offset stack access Before this patch, variable offset access to the stack was dissalowed for regular instructions, but was allowed for "indirect" accesses (i.e. helpers). This patch removes the restriction, allowing reading and writing to the stack through stack pointers with variable offsets. This makes stack-allocated buffers more usable in programs, and brings stack pointers closer to other types of pointers. The motivation is being able to use stack-allocated buffers for data manipulation. When the stack size limit is sufficient, allocating buffers on the stack is simpler than per-cpu arrays, or other alternatives. In unpriviledged programs, variable-offset reads and writes are disallowed (they were already disallowed for the indirect access case) because the speculative execution checking code doesn't support them. Additionally, when writing through a variable-offset stack pointer, if any pointers are in the accessible range, there's possilibities of later leaking pointers because the write cannot be tracked precisely. Writes with variable offset mark the whole range as initialized, even though we don't know which stack slots are actually written. This is in order to not reject future reads to these slots. Note that this doesn't affect writes done through helpers; like before, helpers need the whole stack range to be initialized to begin with. All the stack slots are in range are considered scalars after the write; variable-offset register spills are not tracked. For reads, all the stack slots in the variable range needs to be initialized (but see above about what writes do), otherwise the read is rejected. All register spilled in stack slots that might be read are marked as having been read, however reads through such pointers don't do register filling; the target register will always be either a scalar or a constant zero. Signed-off-by: Andrei Matei Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/20210207011027.676572-2-andreimatei1@gmail.com --- include/linux/bpf.h | 5 + include/linux/bpf_verifier.h | 3 +- kernel/bpf/verifier.c | 657 +++++++++++++++++++++++++++++++++---------- 3 files changed, 518 insertions(+), 147 deletions(-) (limited to 'include/linux/bpf.h') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 321966fc35db..079162bbd387 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1290,6 +1290,11 @@ static inline bool bpf_allow_ptr_leaks(void) return perfmon_capable(); } +static inline bool bpf_allow_uninit_stack(void) +{ + return perfmon_capable(); +} + static inline bool bpf_allow_ptr_to_map_access(void) { return perfmon_capable(); diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h index dfe6f85d97dd..532c97836d0d 100644 --- a/include/linux/bpf_verifier.h +++ b/include/linux/bpf_verifier.h @@ -195,7 +195,7 @@ struct bpf_func_state { * 0 = main function, 1 = first callee. */ u32 frameno; - /* subprog number == index within subprog_stack_depth + /* subprog number == index within subprog_info * zero == main subprog */ u32 subprogno; @@ -404,6 +404,7 @@ struct bpf_verifier_env { u32 used_btf_cnt; /* number of used BTF objects */ u32 id_gen; /* used to generate unique reg IDs */ bool allow_ptr_leaks; + bool allow_uninit_stack; bool allow_ptr_to_map_access; bool bpf_capable; bool bypass_spec_v1; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 15694246f854..400d79e99fc8 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -2275,12 +2275,14 @@ static void save_register_state(struct bpf_func_state *state, state->stack[spi].slot_type[i] = STACK_SPILL; } -/* check_stack_read/write functions track spill/fill of registers, +/* check_stack_{read,write}_fixed_off functions track spill/fill of registers, * stack boundary and alignment are checked in check_mem_access() */ -static int check_stack_write(struct bpf_verifier_env *env, - struct bpf_func_state *state, /* func where register points to */ - int off, int size, int value_regno, int insn_idx) +static int check_stack_write_fixed_off(struct bpf_verifier_env *env, + /* stack frame we're writing to */ + struct bpf_func_state *state, + int off, int size, int value_regno, + int insn_idx) { struct bpf_func_state *cur; /* state of the current function */ int i, slot = -off - 1, spi = slot / BPF_REG_SIZE, err; @@ -2406,9 +2408,175 @@ static int check_stack_write(struct bpf_verifier_env *env, return 0; } -static int check_stack_read(struct bpf_verifier_env *env, - struct bpf_func_state *reg_state /* func where register points to */, - int off, int size, int value_regno) +/* Write the stack: 'stack[ptr_regno + off] = value_regno'. 'ptr_regno' is + * known to contain a variable offset. + * This function checks whether the write is permitted and conservatively + * tracks the effects of the write, considering that each stack slot in the + * dynamic range is potentially written to. + * + * 'off' includes 'regno->off'. + * 'value_regno' can be -1, meaning that an unknown value is being written to + * the stack. + * + * Spilled pointers in range are not marked as written because we don't know + * what's going to be actually written. This means that read propagation for + * future reads cannot be terminated by this write. + * + * For privileged programs, uninitialized stack slots are considered + * initialized by this write (even though we don't know exactly what offsets + * are going to be written to). The idea is that we don't want the verifier to + * reject future reads that access slots written to through variable offsets. + */ +static int check_stack_write_var_off(struct bpf_verifier_env *env, + /* func where register points to */ + struct bpf_func_state *state, + int ptr_regno, int off, int size, + int value_regno, int insn_idx) +{ + struct bpf_func_state *cur; /* state of the current function */ + int min_off, max_off; + int i, err; + struct bpf_reg_state *ptr_reg = NULL, *value_reg = NULL; + bool writing_zero = false; + /* set if the fact that we're writing a zero is used to let any + * stack slots remain STACK_ZERO + */ + bool zero_used = false; + + cur = env->cur_state->frame[env->cur_state->curframe]; + ptr_reg = &cur->regs[ptr_regno]; + min_off = ptr_reg->smin_value + off; + max_off = ptr_reg->smax_value + off + size; + if (value_regno >= 0) + value_reg = &cur->regs[value_regno]; + if (value_reg && register_is_null(value_reg)) + writing_zero = true; + + err = realloc_func_state(state, round_up(-min_off, BPF_REG_SIZE), + state->acquired_refs, true); + if (err) + return err; + + + /* Variable offset writes destroy any spilled pointers in range. */ + for (i = min_off; i < max_off; i++) { + u8 new_type, *stype; + int slot, spi; + + slot = -i - 1; + spi = slot / BPF_REG_SIZE; + stype = &state->stack[spi].slot_type[slot % BPF_REG_SIZE]; + + if (!env->allow_ptr_leaks + && *stype != NOT_INIT + && *stype != SCALAR_VALUE) { + /* Reject the write if there's are spilled pointers in + * range. If we didn't reject here, the ptr status + * would be erased below (even though not all slots are + * actually overwritten), possibly opening the door to + * leaks. + */ + verbose(env, "spilled ptr in range of var-offset stack write; insn %d, ptr off: %d", + insn_idx, i); + return -EINVAL; + } + + /* Erase all spilled pointers. */ + state->stack[spi].spilled_ptr.type = NOT_INIT; + + /* Update the slot type. */ + new_type = STACK_MISC; + if (writing_zero && *stype == STACK_ZERO) { + new_type = STACK_ZERO; + zero_used = true; + } + /* If the slot is STACK_INVALID, we check whether it's OK to + * pretend that it will be initialized by this write. The slot + * might not actually be written to, and so if we mark it as + * initialized future reads might leak uninitialized memory. + * For privileged programs, we will accept such reads to slots + * that may or may not be written because, if we're reject + * them, the error would be too confusing. + */ + if (*stype == STACK_INVALID && !env->allow_uninit_stack) { + verbose(env, "uninit stack in range of var-offset write prohibited for !root; insn %d, off: %d", + insn_idx, i); + return -EINVAL; + } + *stype = new_type; + } + if (zero_used) { + /* backtracking doesn't work for STACK_ZERO yet. */ + err = mark_chain_precision(env, value_regno); + if (err) + return err; + } + return 0; +} + +/* When register 'dst_regno' is assigned some values from stack[min_off, + * max_off), we set the register's type according to the types of the + * respective stack slots. If all the stack values are known to be zeros, then + * so is the destination reg. Otherwise, the register is considered to be + * SCALAR. This function does not deal with register filling; the caller must + * ensure that all spilled registers in the stack range have been marked as + * read. + */ +static void mark_reg_stack_read(struct bpf_verifier_env *env, + /* func where src register points to */ + struct bpf_func_state *ptr_state, + int min_off, int max_off, int dst_regno) +{ + struct bpf_verifier_state *vstate = env->cur_state; + struct bpf_func_state *state = vstate->frame[vstate->curframe]; + int i, slot, spi; + u8 *stype; + int zeros = 0; + + for (i = min_off; i < max_off; i++) { + slot = -i - 1; + spi = slot / BPF_REG_SIZE; + stype = ptr_state->stack[spi].slot_type; + if (stype[slot % BPF_REG_SIZE] != STACK_ZERO) + break; + zeros++; + } + if (zeros == max_off - min_off) { + /* any access_size read into register is zero extended, + * so the whole register == const_zero + */ + __mark_reg_const_zero(&state->regs[dst_regno]); + /* backtracking doesn't support STACK_ZERO yet, + * so mark it precise here, so that later + * backtracking can stop here. + * Backtracking may not need this if this register + * doesn't participate in pointer adjustment. + * Forward propagation of precise flag is not + * necessary either. This mark is only to stop + * backtracking. Any register that contributed + * to const 0 was marked precise before spill. + */ + state->regs[dst_regno].precise = true; + } else { + /* have read misc data from the stack */ + mark_reg_unknown(env, state->regs, dst_regno); + } + state->regs[dst_regno].live |= REG_LIVE_WRITTEN; +} + +/* Read the stack at 'off' and put the results into the register indicated by + * 'dst_regno'. It handles reg filling if the addressed stack slot is a + * spilled reg. + * + * 'dst_regno' can be -1, meaning that the read value is not going to a + * register. + * + * The access is assumed to be within the current stack bounds. + */ +static int check_stack_read_fixed_off(struct bpf_verifier_env *env, + /* func where src register points to */ + struct bpf_func_state *reg_state, + int off, int size, int dst_regno) { struct bpf_verifier_state *vstate = env->cur_state; struct bpf_func_state *state = vstate->frame[vstate->curframe]; @@ -2416,11 +2584,6 @@ static int check_stack_read(struct bpf_verifier_env *env, struct bpf_reg_state *reg; u8 *stype; - if (reg_state->allocated_stack <= slot) { - verbose(env, "invalid read from stack off %d+0 size %d\n", - off, size); - return -EACCES; - } stype = reg_state->stack[spi].slot_type; reg = ®_state->stack[spi].spilled_ptr; @@ -2431,9 +2594,9 @@ static int check_stack_read(struct bpf_verifier_env *env, verbose(env, "invalid size of register fill\n"); return -EACCES; } - if (value_regno >= 0) { - mark_reg_unknown(env, state->regs, value_regno); - state->regs[value_regno].live |= REG_LIVE_WRITTEN; + if (dst_regno >= 0) { + mark_reg_unknown(env, state->regs, dst_regno); + state->regs[dst_regno].live |= REG_LIVE_WRITTEN; } mark_reg_read(env, reg, reg->parent, REG_LIVE_READ64); return 0; @@ -2445,16 +2608,16 @@ static int check_stack_read(struct bpf_verifier_env *env, } } - if (value_regno >= 0) { + if (dst_regno >= 0) { /* restore register state from stack */ - state->regs[value_regno] = *reg; + state->regs[dst_regno] = *reg; /* mark reg as written since spilled pointer state likely * has its liveness marks cleared by is_state_visited() * which resets stack/reg liveness for state transitions */ - state->regs[value_regno].live |= REG_LIVE_WRITTEN; + state->regs[dst_regno].live |= REG_LIVE_WRITTEN; } else if (__is_pointer_value(env->allow_ptr_leaks, reg)) { - /* If value_regno==-1, the caller is asking us whether + /* If dst_regno==-1, the caller is asking us whether * it is acceptable to use this value as a SCALAR_VALUE * (e.g. for XADD). * We must not allow unprivileged callers to do that @@ -2466,70 +2629,167 @@ static int check_stack_read(struct bpf_verifier_env *env, } mark_reg_read(env, reg, reg->parent, REG_LIVE_READ64); } else { - int zeros = 0; + u8 type; for (i = 0; i < size; i++) { - if (stype[(slot - i) % BPF_REG_SIZE] == STACK_MISC) + type = stype[(slot - i) % BPF_REG_SIZE]; + if (type == STACK_MISC) continue; - if (stype[(slot - i) % BPF_REG_SIZE] == STACK_ZERO) { - zeros++; + if (type == STACK_ZERO) continue; - } verbose(env, "invalid read from stack off %d+%d size %d\n", off, i, size); return -EACCES; } mark_reg_read(env, reg, reg->parent, REG_LIVE_READ64); - if (value_regno >= 0) { - if (zeros == size) { - /* any size read into register is zero extended, - * so the whole register == const_zero - */ - __mark_reg_const_zero(&state->regs[value_regno]); - /* backtracking doesn't support STACK_ZERO yet, - * so mark it precise here, so that later - * backtracking can stop here. - * Backtracking may not need this if this register - * doesn't participate in pointer adjustment. - * Forward propagation of precise flag is not - * necessary either. This mark is only to stop - * backtracking. Any register that contributed - * to const 0 was marked precise before spill. - */ - state->regs[value_regno].precise = true; - } else { - /* have read misc data from the stack */ - mark_reg_unknown(env, state->regs, value_regno); - } - state->regs[value_regno].live |= REG_LIVE_WRITTEN; - } + if (dst_regno >= 0) + mark_reg_stack_read(env, reg_state, off, off + size, dst_regno); } return 0; } -static int check_stack_access(struct bpf_verifier_env *env, - const struct bpf_reg_state *reg, - int off, int size) +enum stack_access_src { + ACCESS_DIRECT = 1, /* the access is performed by an instruction */ + ACCESS_HELPER = 2, /* the access is performed by a helper */ +}; + +static int check_stack_range_initialized(struct bpf_verifier_env *env, + int regno, int off, int access_size, + bool zero_size_allowed, + enum stack_access_src type, + struct bpf_call_arg_meta *meta); + +static struct bpf_reg_state *reg_state(struct bpf_verifier_env *env, int regno) +{ + return cur_regs(env) + regno; +} + +/* Read the stack at 'ptr_regno + off' and put the result into the register + * 'dst_regno'. + * 'off' includes the pointer register's fixed offset(i.e. 'ptr_regno.off'), + * but not its variable offset. + * 'size' is assumed to be <= reg size and the access is assumed to be aligned. + * + * As opposed to check_stack_read_fixed_off, this function doesn't deal with + * filling registers (i.e. reads of spilled register cannot be detected when + * the offset is not fixed). We conservatively mark 'dst_regno' as containing + * SCALAR_VALUE. That's why we assert that the 'ptr_regno' has a variable + * offset; for a fixed offset check_stack_read_fixed_off should be used + * instead. + */ +static int check_stack_read_var_off(struct bpf_verifier_env *env, + int ptr_regno, int off, int size, int dst_regno) { - /* Stack accesses must be at a fixed offset, so that we - * can determine what type of data were returned. See - * check_stack_read(). + /* The state of the source register. */ + struct bpf_reg_state *reg = reg_state(env, ptr_regno); + struct bpf_func_state *ptr_state = func(env, reg); + int err; + int min_off, max_off; + + /* Note that we pass a NULL meta, so raw access will not be permitted. */ - if (!tnum_is_const(reg->var_off)) { + err = check_stack_range_initialized(env, ptr_regno, off, size, + false, ACCESS_DIRECT, NULL); + if (err) + return err; + + min_off = reg->smin_value + off; + max_off = reg->smax_value + off; + mark_reg_stack_read(env, ptr_state, min_off, max_off + size, dst_regno); + return 0; +} + +/* check_stack_read dispatches to check_stack_read_fixed_off or + * check_stack_read_var_off. + * + * The caller must ensure that the offset falls within the allocated stack + * bounds. + * + * 'dst_regno' is a register which will receive the value from the stack. It + * can be -1, meaning that the read value is not going to a register. + */ +static int check_stack_read(struct bpf_verifier_env *env, + int ptr_regno, int off, int size, + int dst_regno) +{ + struct bpf_reg_state *reg = reg_state(env, ptr_regno); + struct bpf_func_state *state = func(env, reg); + int err; + /* Some accesses are only permitted with a static offset. */ + bool var_off = !tnum_is_const(reg->var_off); + + /* The offset is required to be static when reads don't go to a + * register, in order to not leak pointers (see + * check_stack_read_fixed_off). + */ + if (dst_regno < 0 && var_off) { char tn_buf[48]; tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); - verbose(env, "variable stack access var_off=%s off=%d size=%d\n", + verbose(env, "variable offset stack pointer cannot be passed into helper function; var_off=%s off=%d size=%d\n", tn_buf, off, size); return -EACCES; } + /* Variable offset is prohibited for unprivileged mode for simplicity + * since it requires corresponding support in Spectre masking for stack + * ALU. See also retrieve_ptr_limit(). + */ + if (!env->bypass_spec_v1 && var_off) { + char tn_buf[48]; - if (off >= 0 || off < -MAX_BPF_STACK) { - verbose(env, "invalid stack off=%d size=%d\n", off, size); + tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); + verbose(env, "R%d variable offset stack access prohibited for !root, var_off=%s\n", + ptr_regno, tn_buf); return -EACCES; } - return 0; + if (!var_off) { + off += reg->var_off.value; + err = check_stack_read_fixed_off(env, state, off, size, + dst_regno); + } else { + /* Variable offset stack reads need more conservative handling + * than fixed offset ones. Note that dst_regno >= 0 on this + * branch. + */ + err = check_stack_read_var_off(env, ptr_regno, off, size, + dst_regno); + } + return err; +} + + +/* check_stack_write dispatches to check_stack_write_fixed_off or + * check_stack_write_var_off. + * + * 'ptr_regno' is the register used as a pointer into the stack. + * 'off' includes 'ptr_regno->off', but not its variable offset (if any). + * 'value_regno' is the register whose value we're writing to the stack. It can + * be -1, meaning that we're not writing from a register. + * + * The caller must ensure that the offset falls within the maximum stack size. + */ +static int check_stack_write(struct bpf_verifier_env *env, + int ptr_regno, int off, int size, + int value_regno, int insn_idx) +{ + struct bpf_reg_state *reg = reg_state(env, ptr_regno); + struct bpf_func_state *state = func(env, reg); + int err; + + if (tnum_is_const(reg->var_off)) { + off += reg->var_off.value; + err = check_stack_write_fixed_off(env, state, off, size, + value_regno, insn_idx); + } else { + /* Variable offset stack reads need more conservative handling + * than fixed offset ones. + */ + err = check_stack_write_var_off(env, state, + ptr_regno, off, size, + value_regno, insn_idx); + } + return err; } static int check_map_access_type(struct bpf_verifier_env *env, u32 regno, @@ -2862,11 +3122,6 @@ static int check_sock_access(struct bpf_verifier_env *env, int insn_idx, return -EACCES; } -static struct bpf_reg_state *reg_state(struct bpf_verifier_env *env, int regno) -{ - return cur_regs(env) + regno; -} - static bool is_pointer_value(struct bpf_verifier_env *env, int regno) { return __is_pointer_value(env->allow_ptr_leaks, reg_state(env, regno)); @@ -2985,8 +3240,8 @@ static int check_ptr_alignment(struct bpf_verifier_env *env, break; case PTR_TO_STACK: pointer_desc = "stack "; - /* The stack spill tracking logic in check_stack_write() - * and check_stack_read() relies on stack accesses being + /* The stack spill tracking logic in check_stack_write_fixed_off() + * and check_stack_read_fixed_off() relies on stack accesses being * aligned. */ strict = true; @@ -3402,6 +3657,91 @@ static int check_ptr_to_map_access(struct bpf_verifier_env *env, return 0; } +/* Check that the stack access at the given offset is within bounds. The + * maximum valid offset is -1. + * + * The minimum valid offset is -MAX_BPF_STACK for writes, and + * -state->allocated_stack for reads. + */ +static int check_stack_slot_within_bounds(int off, + struct bpf_func_state *state, + enum bpf_access_type t) +{ + int min_valid_off; + + if (t == BPF_WRITE) + min_valid_off = -MAX_BPF_STACK; + else + min_valid_off = -state->allocated_stack; + + if (off < min_valid_off || off > -1) + return -EACCES; + return 0; +} + +/* Check that the stack access at 'regno + off' falls within the maximum stack + * bounds. + * + * 'off' includes `regno->offset`, but not its dynamic part (if any). + */ +static int check_stack_access_within_bounds( + struct bpf_verifier_env *env, + int regno, int off, int access_size, + enum stack_access_src src, enum bpf_access_type type) +{ + struct bpf_reg_state *regs = cur_regs(env); + struct bpf_reg_state *reg = regs + regno; + struct bpf_func_state *state = func(env, reg); + int min_off, max_off; + int err; + char *err_extra; + + if (src == ACCESS_HELPER) + /* We don't know if helpers are reading or writing (or both). */ + err_extra = " indirect access to"; + else if (type == BPF_READ) + err_extra = " read from"; + else + err_extra = " write to"; + + if (tnum_is_const(reg->var_off)) { + min_off = reg->var_off.value + off; + if (access_size > 0) + max_off = min_off + access_size - 1; + else + max_off = min_off; + } else { + if (reg->smax_value >= BPF_MAX_VAR_OFF || + reg->smin_value <= -BPF_MAX_VAR_OFF) { + verbose(env, "invalid unbounded variable-offset%s stack R%d\n", + err_extra, regno); + return -EACCES; + } + min_off = reg->smin_value + off; + if (access_size > 0) + max_off = reg->smax_value + off + access_size - 1; + else + max_off = min_off; + } + + err = check_stack_slot_within_bounds(min_off, state, type); + if (!err) + err = check_stack_slot_within_bounds(max_off, state, type); + + if (err) { + if (tnum_is_const(reg->var_off)) { + verbose(env, "invalid%s stack R%d off=%d size=%d\n", + err_extra, regno, off, access_size); + } else { + char tn_buf[48]; + + tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); + verbose(env, "invalid variable-offset%s stack R%d var_off=%s size=%d\n", + err_extra, regno, tn_buf, access_size); + } + } + return err; +} /* check whether memory at (regno + off) is accessible for t = (read | write) * if t==write, value_regno is a register which value is stored into memory @@ -3517,8 +3857,8 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn } } else if (reg->type == PTR_TO_STACK) { - off += reg->var_off.value; - err = check_stack_access(env, reg, off, size); + /* Basic bounds checks. */ + err = check_stack_access_within_bounds(env, regno, off, size, ACCESS_DIRECT, t); if (err) return err; @@ -3527,12 +3867,12 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn if (err) return err; - if (t == BPF_WRITE) - err = check_stack_write(env, state, off, size, - value_regno, insn_idx); - else - err = check_stack_read(env, state, off, size, + if (t == BPF_READ) + err = check_stack_read(env, regno, off, size, value_regno); + else + err = check_stack_write(env, regno, off, size, + value_regno, insn_idx); } else if (reg_is_pkt_pointer(reg)) { if (t == BPF_WRITE && !may_access_direct_pkt_data(env, NULL, t)) { verbose(env, "cannot write into packet\n"); @@ -3699,49 +4039,53 @@ static int check_atomic(struct bpf_verifier_env *env, int insn_idx, struct bpf_i return 0; } -static int __check_stack_boundary(struct bpf_verifier_env *env, u32 regno, - int off, int access_size, - bool zero_size_allowed) +/* When register 'regno' is used to read the stack (either directly or through + * a helper function) make sure that it's within stack boundary and, depending + * on the access type, that all elements of the stack are initialized. + * + * 'off' includes 'regno->off', but not its dynamic part (if any). + * + * All registers that have been spilled on the stack in the slots within the + * read offsets are marked as read. + */ +static int check_stack_range_initialized( + struct bpf_verifier_env *env, int regno, int off, + int access_size, bool zero_size_allowed, + enum stack_access_src type, struct bpf_call_arg_meta *meta) { struct bpf_reg_state *reg = reg_state(env, regno); + struct bpf_func_state *state = func(env, reg); + int err, min_off, max_off, i, j, slot, spi; + char *err_extra = type == ACCESS_HELPER ? " indirect" : ""; + enum bpf_access_type bounds_check_type; + /* Some accesses can write anything into the stack, others are + * read-only. + */ + bool clobber = false; - if (off >= 0 || off < -MAX_BPF_STACK || off + access_size > 0 || - access_size < 0 || (access_size == 0 && !zero_size_allowed)) { - if (tnum_is_const(reg->var_off)) { - verbose(env, "invalid stack type R%d off=%d access_size=%d\n", - regno, off, access_size); - } else { - char tn_buf[48]; - - tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); - verbose(env, "invalid stack type R%d var_off=%s access_size=%d\n", - regno, tn_buf, access_size); - } + if (access_size == 0 && !zero_size_allowed) { + verbose(env, "invalid zero-sized read\n"); return -EACCES; } - return 0; -} -/* when register 'regno' is passed into function that will read 'access_size' - * bytes from that pointer, make sure that it's within stack boundary - * and all elements of stack are initialized. - * Unlike most pointer bounds-checking functions, this one doesn't take an - * 'off' argument, so it has to add in reg->off itself. - */ -static int check_stack_boundary(struct bpf_verifier_env *env, int regno, - int access_size, bool zero_size_allowed, - struct bpf_call_arg_meta *meta) -{ - struct bpf_reg_state *reg = reg_state(env, regno); - struct bpf_func_state *state = func(env, reg); - int err, min_off, max_off, i, j, slot, spi; + if (type == ACCESS_HELPER) { + /* The bounds checks for writes are more permissive than for + * reads. However, if raw_mode is not set, we'll do extra + * checks below. + */ + bounds_check_type = BPF_WRITE; + clobber = true; + } else { + bounds_check_type = BPF_READ; + } + err = check_stack_access_within_bounds(env, regno, off, access_size, + type, bounds_check_type); + if (err) + return err; + if (tnum_is_const(reg->var_off)) { - min_off = max_off = reg->var_off.value + reg->off; - err = __check_stack_boundary(env, regno, min_off, access_size, - zero_size_allowed); - if (err) - return err; + min_off = max_off = reg->var_off.value + off; } else { /* Variable offset is prohibited for unprivileged mode for * simplicity since it requires corresponding support in @@ -3752,8 +4096,8 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno, char tn_buf[48]; tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); - verbose(env, "R%d indirect variable offset stack access prohibited for !root, var_off=%s\n", - regno, tn_buf); + verbose(env, "R%d%s variable offset stack access prohibited for !root, var_off=%s\n", + regno, err_extra, tn_buf); return -EACCES; } /* Only initialized buffer on stack is allowed to be accessed @@ -3765,28 +4109,8 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno, if (meta && meta->raw_mode) meta = NULL; - if (reg->smax_value >= BPF_MAX_VAR_OFF || - reg->smax_value <= -BPF_MAX_VAR_OFF) { - verbose(env, "R%d unbounded indirect variable offset stack access\n", - regno); - return -EACCES; - } - min_off = reg->smin_value + reg->off; - max_off = reg->smax_value + reg->off; - err = __check_stack_boundary(env, regno, min_off, access_size, - zero_size_allowed); - if (err) { - verbose(env, "R%d min value is outside of stack bound\n", - regno); - return err; - } - err = __check_stack_boundary(env, regno, max_off, access_size, - zero_size_allowed); - if (err) { - verbose(env, "R%d max value is outside of stack bound\n", - regno); - return err; - } + min_off = reg->smin_value + off; + max_off = reg->smax_value + off; } if (meta && meta->raw_mode) { @@ -3806,8 +4130,10 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno, if (*stype == STACK_MISC) goto mark; if (*stype == STACK_ZERO) { - /* helper can write anything into the stack */ - *stype = STACK_MISC; + if (clobber) { + /* helper can write anything into the stack */ + *stype = STACK_MISC; + } goto mark; } @@ -3818,22 +4144,24 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno, if (state->stack[spi].slot_type[0] == STACK_SPILL && (state->stack[spi].spilled_ptr.type == SCALAR_VALUE || env->allow_ptr_leaks)) { - __mark_reg_unknown(env, &state->stack[spi].spilled_ptr); - for (j = 0; j < BPF_REG_SIZE; j++) - state->stack[spi].slot_type[j] = STACK_MISC; + if (clobber) { + __mark_reg_unknown(env, &state->stack[spi].spilled_ptr); + for (j = 0; j < BPF_REG_SIZE; j++) + state->stack[spi].slot_type[j] = STACK_MISC; + } goto mark; } err: if (tnum_is_const(reg->var_off)) { - verbose(env, "invalid indirect read from stack off %d+%d size %d\n", - min_off, i - min_off, access_size); + verbose(env, "invalid%s read from stack R%d off %d+%d size %d\n", + err_extra, regno, min_off, i - min_off, access_size); } else { char tn_buf[48]; tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); - verbose(env, "invalid indirect read from stack var_off %s+%d size %d\n", - tn_buf, i - min_off, access_size); + verbose(env, "invalid%s read from stack R%d var_off %s+%d size %d\n", + err_extra, regno, tn_buf, i - min_off, access_size); } return -EACCES; mark: @@ -3882,8 +4210,10 @@ static int check_helper_mem_access(struct bpf_verifier_env *env, int regno, "rdwr", &env->prog->aux->max_rdwr_access); case PTR_TO_STACK: - return check_stack_boundary(env, regno, access_size, - zero_size_allowed, meta); + return check_stack_range_initialized( + env, + regno, reg->off, access_size, + zero_size_allowed, ACCESS_HELPER, meta); default: /* scalar_value or invalid ptr */ /* Allow zero-byte read from NULL, regardless of pointer type */ if (zero_size_allowed && access_size == 0 && @@ -5547,6 +5877,41 @@ do_sim: return !ret ? -EFAULT : 0; } +/* check that stack access falls within stack limits and that 'reg' doesn't + * have a variable offset. + * + * Variable offset is prohibited for unprivileged mode for simplicity since it + * requires corresponding support in Spectre masking for stack ALU. See also + * retrieve_ptr_limit(). + * + * + * 'off' includes 'reg->off'. + */ +static int check_stack_access_for_ptr_arithmetic( + struct bpf_verifier_env *env, + int regno, + const struct bpf_reg_state *reg, + int off) +{ + if (!tnum_is_const(reg->var_off)) { + char tn_buf[48]; + + tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); + verbose(env, "R%d variable stack access prohibited for !root, var_off=%s off=%d\n", + regno, tn_buf, off); + return -EACCES; + } + + if (off >= 0 || off < -MAX_BPF_STACK) { + verbose(env, "R%d stack pointer arithmetic goes out of range, " + "prohibited for !root; off=%d\n", regno, off); + return -EACCES; + } + + return 0; +} + + /* Handles arithmetic on a pointer and a scalar: computes new min/max and var_off. * Caller should also handle BPF_MOV case separately. * If we return -EACCES, caller may want to try again treating pointer as a @@ -5790,10 +6155,9 @@ static int adjust_ptr_min_max_vals(struct bpf_verifier_env *env, "prohibited for !root\n", dst); return -EACCES; } else if (dst_reg->type == PTR_TO_STACK && - check_stack_access(env, dst_reg, dst_reg->off + - dst_reg->var_off.value, 1)) { - verbose(env, "R%d stack pointer arithmetic goes out of range, " - "prohibited for !root\n", dst); + check_stack_access_for_ptr_arithmetic( + env, dst, dst_reg, dst_reg->off + + dst_reg->var_off.value)) { return -EACCES; } } @@ -12129,6 +12493,7 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr, env->strict_alignment = false; env->allow_ptr_leaks = bpf_allow_ptr_leaks(); + env->allow_uninit_stack = bpf_allow_uninit_stack(); env->allow_ptr_to_map_access = bpf_allow_ptr_to_map_access(); env->bypass_spec_v1 = bpf_bypass_spec_v1(); env->bypass_spec_v4 = bpf_bypass_spec_v4(); -- cgit v1.2.3 From 700d4796ef59f5faf240d307839bd419e2b6bdff Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:26 -0800 Subject: bpf: Optimize program stats Move bpf_prog_stats from prog->aux into prog to avoid one extra load in critical path of program execution. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210033634.62081-2-alexei.starovoitov@gmail.com --- include/linux/bpf.h | 8 -------- include/linux/filter.h | 14 +++++++++++--- kernel/bpf/core.c | 8 ++++---- kernel/bpf/syscall.c | 2 +- kernel/bpf/trampoline.c | 2 +- kernel/bpf/verifier.c | 2 +- 6 files changed, 18 insertions(+), 18 deletions(-) (limited to 'include/linux/bpf.h') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 079162bbd387..5a388955e6b6 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -14,7 +14,6 @@ #include #include #include -#include #include #include #include @@ -507,12 +506,6 @@ enum bpf_cgroup_storage_type { */ #define MAX_BPF_FUNC_ARGS 12 -struct bpf_prog_stats { - u64 cnt; - u64 nsecs; - struct u64_stats_sync syncp; -} __aligned(2 * sizeof(u64)); - struct btf_func_model { u8 ret_size; u8 nr_args; @@ -845,7 +838,6 @@ struct bpf_prog_aux { u32 linfo_idx; u32 num_exentries; struct exception_table_entry *extable; - struct bpf_prog_stats __percpu *stats; union { struct work_struct work; struct rcu_head rcu; diff --git a/include/linux/filter.h b/include/linux/filter.h index 5b3137d7b690..cecb03c9d251 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -22,6 +22,7 @@ #include #include #include +#include #include @@ -539,6 +540,12 @@ struct bpf_binary_header { u8 image[] __aligned(BPF_IMAGE_ALIGNMENT); }; +struct bpf_prog_stats { + u64 cnt; + u64 nsecs; + struct u64_stats_sync syncp; +} __aligned(2 * sizeof(u64)); + struct bpf_prog { u16 pages; /* Number of allocated pages */ u16 jited:1, /* Is our filter JIT'ed? */ @@ -557,10 +564,11 @@ struct bpf_prog { u32 len; /* Number of filter blocks */ u32 jited_len; /* Size of jited insns in bytes */ u8 tag[BPF_TAG_SIZE]; - struct bpf_prog_aux *aux; /* Auxiliary fields */ - struct sock_fprog_kern *orig_prog; /* Original BPF program */ + struct bpf_prog_stats __percpu *stats; unsigned int (*bpf_func)(const void *ctx, const struct bpf_insn *insn); + struct bpf_prog_aux *aux; /* Auxiliary fields */ + struct sock_fprog_kern *orig_prog; /* Original BPF program */ /* Instructions for interpreter */ struct sock_filter insns[0]; struct bpf_insn insnsi[]; @@ -581,7 +589,7 @@ DECLARE_STATIC_KEY_FALSE(bpf_stats_enabled_key); struct bpf_prog_stats *__stats; \ u64 __start = sched_clock(); \ __ret = dfunc(ctx, (prog)->insnsi, (prog)->bpf_func); \ - __stats = this_cpu_ptr(prog->aux->stats); \ + __stats = this_cpu_ptr(prog->stats); \ u64_stats_update_begin(&__stats->syncp); \ __stats->cnt++; \ __stats->nsecs += sched_clock() - __start; \ diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 5bbd4884ff7a..2cf71fd39c22 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -114,8 +114,8 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) if (!prog) return NULL; - prog->aux->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); - if (!prog->aux->stats) { + prog->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); + if (!prog->stats) { kfree(prog->aux); vfree(prog); return NULL; @@ -124,7 +124,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) for_each_possible_cpu(cpu) { struct bpf_prog_stats *pstats; - pstats = per_cpu_ptr(prog->aux->stats, cpu); + pstats = per_cpu_ptr(prog->stats, cpu); u64_stats_init(&pstats->syncp); } return prog; @@ -249,10 +249,10 @@ void __bpf_prog_free(struct bpf_prog *fp) if (fp->aux) { mutex_destroy(&fp->aux->used_maps_mutex); mutex_destroy(&fp->aux->dst_mutex); - free_percpu(fp->aux->stats); kfree(fp->aux->poke_tab); kfree(fp->aux); } + free_percpu(fp->stats); vfree(fp); } diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index e5999d86c76e..f7df56a704de 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1739,7 +1739,7 @@ static void bpf_prog_get_stats(const struct bpf_prog *prog, unsigned int start; u64 tnsecs, tcnt; - st = per_cpu_ptr(prog->aux->stats, cpu); + st = per_cpu_ptr(prog->stats, cpu); do { start = u64_stats_fetch_begin_irq(&st->syncp); tnsecs = st->nsecs; diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 35c5887d82ff..5be3beeedd74 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -412,7 +412,7 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) * Hence check that 'start' is not zero. */ start) { - stats = this_cpu_ptr(prog->aux->stats); + stats = this_cpu_ptr(prog->stats); u64_stats_update_begin(&stats->syncp); stats->cnt++; stats->nsecs += sched_clock() - start; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 400d79e99fc8..424c1ba0f52f 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -11253,7 +11253,7 @@ static int jit_subprogs(struct bpf_verifier_env *env) /* BPF_PROG_RUN doesn't call subprogs directly, * hence main prog stats include the runtime of subprogs. * subprogs don't have IDs and not reachable via prog_get_next_id - * func[i]->aux->stats will never be accessed and stays NULL + * func[i]->stats will never be accessed and stays NULL */ func[i] = bpf_prog_alloc_no_stats(bpf_prog_size(len), GFP_USER); if (!func[i]) -- cgit v1.2.3 From f2dd3b39467411c53703125a111f45b3672c1771 Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:28 -0800 Subject: bpf: Compute program stats for sleepable programs Since sleepable programs don't migrate from the cpu the excution stats can be computed for them as well. Reuse the same infrastructure for both sleepable and non-sleepable programs. run_cnt -> the number of times the program was executed. run_time_ns -> the program execution time in nanoseconds including the off-cpu time when the program was sleeping. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Acked-by: KP Singh Link: https://lore.kernel.org/bpf/20210210033634.62081-4-alexei.starovoitov@gmail.com --- arch/x86/net/bpf_jit_comp.c | 31 ++++++++++++------------------- include/linux/bpf.h | 4 ++-- kernel/bpf/trampoline.c | 42 ++++++++++++++++++++++++++++-------------- 3 files changed, 42 insertions(+), 35 deletions(-) (limited to 'include/linux/bpf.h') diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index a3dc3bd154ac..d11b9bcebbea 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1742,15 +1742,12 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, u8 *prog = *pprog; int cnt = 0; - if (p->aux->sleepable) { - if (emit_call(&prog, __bpf_prog_enter_sleepable, prog)) + if (emit_call(&prog, + p->aux->sleepable ? __bpf_prog_enter_sleepable : + __bpf_prog_enter, prog)) return -EINVAL; - } else { - if (emit_call(&prog, __bpf_prog_enter, prog)) - return -EINVAL; - /* remember prog start time returned by __bpf_prog_enter */ - emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0); - } + /* remember prog start time returned by __bpf_prog_enter */ + emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0); /* arg1: lea rdi, [rbp - stack_size] */ EMIT4(0x48, 0x8D, 0x7D, -stack_size); @@ -1770,18 +1767,14 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, if (mod_ret) emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8); - if (p->aux->sleepable) { - if (emit_call(&prog, __bpf_prog_exit_sleepable, prog)) + /* arg1: mov rdi, progs[i] */ + emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p); + /* arg2: mov rsi, rbx <- start time in nsec */ + emit_mov_reg(&prog, true, BPF_REG_2, BPF_REG_6); + if (emit_call(&prog, + p->aux->sleepable ? __bpf_prog_exit_sleepable : + __bpf_prog_exit, prog)) return -EINVAL; - } else { - /* arg1: mov rdi, progs[i] */ - emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, - (u32) (long) p); - /* arg2: mov rsi, rbx <- start time in nsec */ - emit_mov_reg(&prog, true, BPF_REG_2, BPF_REG_6); - if (emit_call(&prog, __bpf_prog_exit, prog)) - return -EINVAL; - } *pprog = prog; return 0; diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 5a388955e6b6..e1840ceaf55f 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -563,8 +563,8 @@ int arch_prepare_bpf_trampoline(void *image, void *image_end, /* these two functions are called from generated trampoline */ u64 notrace __bpf_prog_enter(void); void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start); -void notrace __bpf_prog_enter_sleepable(void); -void notrace __bpf_prog_exit_sleepable(void); +u64 notrace __bpf_prog_enter_sleepable(void); +void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start); struct bpf_ksym { unsigned long start; diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 89fc849ba271..48eb021e1421 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -381,56 +381,70 @@ out: mutex_unlock(&trampoline_mutex); } +#define NO_START_TIME 0 +static u64 notrace bpf_prog_start_time(void) +{ + u64 start = NO_START_TIME; + + if (static_branch_unlikely(&bpf_stats_enabled_key)) + start = sched_clock(); + return start; +} + /* The logic is similar to BPF_PROG_RUN, but with an explicit * rcu_read_lock() and migrate_disable() which are required * for the trampoline. The macro is split into - * call _bpf_prog_enter + * call __bpf_prog_enter * call prog->bpf_func * call __bpf_prog_exit */ u64 notrace __bpf_prog_enter(void) __acquires(RCU) { - u64 start = 0; - rcu_read_lock(); migrate_disable(); - if (static_branch_unlikely(&bpf_stats_enabled_key)) - start = sched_clock(); - return start; + return bpf_prog_start_time(); } -void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) - __releases(RCU) +static void notrace update_prog_stats(struct bpf_prog *prog, + u64 start) { struct bpf_prog_stats *stats; if (static_branch_unlikely(&bpf_stats_enabled_key) && - /* static_key could be enabled in __bpf_prog_enter - * and disabled in __bpf_prog_exit. + /* static_key could be enabled in __bpf_prog_enter* + * and disabled in __bpf_prog_exit*. * And vice versa. - * Hence check that 'start' is not zero. + * Hence check that 'start' is valid. */ - start) { + start > NO_START_TIME) { stats = this_cpu_ptr(prog->stats); u64_stats_update_begin(&stats->syncp); stats->cnt++; stats->nsecs += sched_clock() - start; u64_stats_update_end(&stats->syncp); } +} + +void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) + __releases(RCU) +{ + update_prog_stats(prog, start); migrate_enable(); rcu_read_unlock(); } -void notrace __bpf_prog_enter_sleepable(void) +u64 notrace __bpf_prog_enter_sleepable(void) { rcu_read_lock_trace(); migrate_disable(); might_fault(); + return bpf_prog_start_time(); } -void notrace __bpf_prog_exit_sleepable(void) +void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start) { + update_prog_stats(prog, start); migrate_enable(); rcu_read_unlock_trace(); } -- cgit v1.2.3 From ca06f55b90020cd97f4cc6d52db95436162e7dcf Mon Sep 17 00:00:00 2001 From: Alexei Starovoitov Date: Tue, 9 Feb 2021 19:36:29 -0800 Subject: bpf: Add per-program recursion prevention mechanism Since both sleepable and non-sleepable programs execute under migrate_disable add recursion prevention mechanism to both types of programs when they're executed via bpf trampoline. Signed-off-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210033634.62081-5-alexei.starovoitov@gmail.com --- arch/x86/net/bpf_jit_comp.c | 15 ++++++++++++++ include/linux/bpf.h | 6 +++--- include/linux/filter.h | 1 + kernel/bpf/core.c | 8 ++++++++ kernel/bpf/trampoline.c | 23 ++++++++++++++++++---- .../selftests/bpf/prog_tests/fexit_stress.c | 4 ++-- .../selftests/bpf/prog_tests/trampoline_count.c | 4 ++-- 7 files changed, 50 insertions(+), 11 deletions(-) (limited to 'include/linux/bpf.h') diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c index d11b9bcebbea..79e7a0ec1da5 100644 --- a/arch/x86/net/bpf_jit_comp.c +++ b/arch/x86/net/bpf_jit_comp.c @@ -1740,8 +1740,11 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, struct bpf_prog *p, int stack_size, bool mod_ret) { u8 *prog = *pprog; + u8 *jmp_insn; int cnt = 0; + /* arg1: mov rdi, progs[i] */ + emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p); if (emit_call(&prog, p->aux->sleepable ? __bpf_prog_enter_sleepable : __bpf_prog_enter, prog)) @@ -1749,6 +1752,14 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, /* remember prog start time returned by __bpf_prog_enter */ emit_mov_reg(&prog, true, BPF_REG_6, BPF_REG_0); + /* if (__bpf_prog_enter*(prog) == 0) + * goto skip_exec_of_prog; + */ + EMIT3(0x48, 0x85, 0xC0); /* test rax,rax */ + /* emit 2 nops that will be replaced with JE insn */ + jmp_insn = prog; + emit_nops(&prog, 2); + /* arg1: lea rdi, [rbp - stack_size] */ EMIT4(0x48, 0x8D, 0x7D, -stack_size); /* arg2: progs[i]->insnsi for interpreter */ @@ -1767,6 +1778,10 @@ static int invoke_bpf_prog(const struct btf_func_model *m, u8 **pprog, if (mod_ret) emit_stx(&prog, BPF_DW, BPF_REG_FP, BPF_REG_0, -8); + /* replace 2 nops with JE insn, since jmp target is known */ + jmp_insn[0] = X86_JE; + jmp_insn[1] = prog - jmp_insn - 2; + /* arg1: mov rdi, progs[i] */ emit_mov_imm64(&prog, BPF_REG_1, (long) p >> 32, (u32) (long) p); /* arg2: mov rsi, rbx <- start time in nsec */ diff --git a/include/linux/bpf.h b/include/linux/bpf.h index e1840ceaf55f..1c8ea682c007 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -529,7 +529,7 @@ struct btf_func_model { /* Each call __bpf_prog_enter + call bpf_func + call __bpf_prog_exit is ~50 * bytes on x86. Pick a number to fit into BPF_IMAGE_SIZE / 2 */ -#define BPF_MAX_TRAMP_PROGS 40 +#define BPF_MAX_TRAMP_PROGS 38 struct bpf_tramp_progs { struct bpf_prog *progs[BPF_MAX_TRAMP_PROGS]; @@ -561,9 +561,9 @@ int arch_prepare_bpf_trampoline(void *image, void *image_end, struct bpf_tramp_progs *tprogs, void *orig_call); /* these two functions are called from generated trampoline */ -u64 notrace __bpf_prog_enter(void); +u64 notrace __bpf_prog_enter(struct bpf_prog *prog); void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start); -u64 notrace __bpf_prog_enter_sleepable(void); +u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog); void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start); struct bpf_ksym { diff --git a/include/linux/filter.h b/include/linux/filter.h index cecb03c9d251..6a06f3c69f4e 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -565,6 +565,7 @@ struct bpf_prog { u32 jited_len; /* Size of jited insns in bytes */ u8 tag[BPF_TAG_SIZE]; struct bpf_prog_stats __percpu *stats; + int __percpu *active; unsigned int (*bpf_func)(const void *ctx, const struct bpf_insn *insn); struct bpf_prog_aux *aux; /* Auxiliary fields */ diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index 2cf71fd39c22..334070c4b8a1 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -91,6 +91,12 @@ struct bpf_prog *bpf_prog_alloc_no_stats(unsigned int size, gfp_t gfp_extra_flag vfree(fp); return NULL; } + fp->active = alloc_percpu_gfp(int, GFP_KERNEL_ACCOUNT | gfp_extra_flags); + if (!fp->active) { + vfree(fp); + kfree(aux); + return NULL; + } fp->pages = size / PAGE_SIZE; fp->aux = aux; @@ -116,6 +122,7 @@ struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags) prog->stats = alloc_percpu_gfp(struct bpf_prog_stats, gfp_flags); if (!prog->stats) { + free_percpu(prog->active); kfree(prog->aux); vfree(prog); return NULL; @@ -253,6 +260,7 @@ void __bpf_prog_free(struct bpf_prog *fp) kfree(fp->aux); } free_percpu(fp->stats); + free_percpu(fp->active); vfree(fp); } diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index 48eb021e1421..89ef6320d19b 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -381,13 +381,16 @@ out: mutex_unlock(&trampoline_mutex); } -#define NO_START_TIME 0 +#define NO_START_TIME 1 static u64 notrace bpf_prog_start_time(void) { u64 start = NO_START_TIME; - if (static_branch_unlikely(&bpf_stats_enabled_key)) + if (static_branch_unlikely(&bpf_stats_enabled_key)) { start = sched_clock(); + if (unlikely(!start)) + start = NO_START_TIME; + } return start; } @@ -397,12 +400,20 @@ static u64 notrace bpf_prog_start_time(void) * call __bpf_prog_enter * call prog->bpf_func * call __bpf_prog_exit + * + * __bpf_prog_enter returns: + * 0 - skip execution of the bpf prog + * 1 - execute bpf prog + * [2..MAX_U64] - excute bpf prog and record execution time. + * This is start time. */ -u64 notrace __bpf_prog_enter(void) +u64 notrace __bpf_prog_enter(struct bpf_prog *prog) __acquires(RCU) { rcu_read_lock(); migrate_disable(); + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + return 0; return bpf_prog_start_time(); } @@ -430,21 +441,25 @@ void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start) __releases(RCU) { update_prog_stats(prog, start); + __this_cpu_dec(*(prog->active)); migrate_enable(); rcu_read_unlock(); } -u64 notrace __bpf_prog_enter_sleepable(void) +u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog) { rcu_read_lock_trace(); migrate_disable(); might_fault(); + if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) + return 0; return bpf_prog_start_time(); } void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start) { update_prog_stats(prog, start); + __this_cpu_dec(*(prog->active)); migrate_enable(); rcu_read_unlock_trace(); } diff --git a/tools/testing/selftests/bpf/prog_tests/fexit_stress.c b/tools/testing/selftests/bpf/prog_tests/fexit_stress.c index 3b9dbf7433f0..7c9b62e971f1 100644 --- a/tools/testing/selftests/bpf/prog_tests/fexit_stress.c +++ b/tools/testing/selftests/bpf/prog_tests/fexit_stress.c @@ -2,8 +2,8 @@ /* Copyright (c) 2019 Facebook */ #include -/* x86-64 fits 55 JITed and 43 interpreted progs into half page */ -#define CNT 40 +/* that's kernel internal BPF_MAX_TRAMP_PROGS define */ +#define CNT 38 void test_fexit_stress(void) { diff --git a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c index 781c8d11604b..f3022d934e2d 100644 --- a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c +++ b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c @@ -4,7 +4,7 @@ #include #include -#define MAX_TRAMP_PROGS 40 +#define MAX_TRAMP_PROGS 38 struct inst { struct bpf_object *obj; @@ -52,7 +52,7 @@ void test_trampoline_count(void) struct bpf_link *link; char comm[16] = {}; - /* attach 'allowed' 40 trampoline programs */ + /* attach 'allowed' trampoline programs */ for (i = 0; i < MAX_TRAMP_PROGS; i++) { obj = bpf_object__open_file(object, NULL); if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj))) { -- cgit v1.2.3 From c5dbb89fc2ac013afe67b9e4fcb3743c02b567cd Mon Sep 17 00:00:00 2001 From: Florent Revest Date: Wed, 10 Feb 2021 12:14:03 +0100 Subject: bpf: Expose bpf_get_socket_cookie to tracing programs This needs a new helper that: - can work in a sleepable context (using sock_gen_cookie) - takes a struct sock pointer and checks that it's not NULL Signed-off-by: Florent Revest Signed-off-by: Alexei Starovoitov Acked-by: KP Singh Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20210210111406.785541-2-revest@chromium.org --- include/linux/bpf.h | 1 + include/uapi/linux/bpf.h | 8 ++++++++ kernel/trace/bpf_trace.c | 2 ++ net/core/filter.c | 12 ++++++++++++ tools/include/uapi/linux/bpf.h | 8 ++++++++ 5 files changed, 31 insertions(+) (limited to 'include/linux/bpf.h') diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 1c8ea682c007..cccaef1088ea 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1885,6 +1885,7 @@ extern const struct bpf_func_proto bpf_per_cpu_ptr_proto; extern const struct bpf_func_proto bpf_this_cpu_ptr_proto; extern const struct bpf_func_proto bpf_ktime_get_coarse_ns_proto; extern const struct bpf_func_proto bpf_sock_from_file_proto; +extern const struct bpf_func_proto bpf_get_socket_ptr_cookie_proto; const struct bpf_func_proto *bpf_tracing_func_proto( enum bpf_func_id func_id, const struct bpf_prog *prog); diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index dbf10bf08582..07cc2e404291 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -1673,6 +1673,14 @@ union bpf_attr { * Return * A 8-byte long unique number. * + * u64 bpf_get_socket_cookie(struct sock *sk) + * Description + * Equivalent to **bpf_get_socket_cookie**\ () helper that accepts + * *sk*, but gets socket from a BTF **struct sock**. This helper + * also works for sleepable programs. + * Return + * A 8-byte long unique number or 0 if *sk* is NULL. + * * u32 bpf_get_socket_uid(struct sk_buff *skb) * Return * The owner UID of the socket associated to *skb*. If the socket diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 6c0018abe68a..845b2168e006 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -1760,6 +1760,8 @@ tracing_prog_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) return &bpf_sk_storage_delete_tracing_proto; case BPF_FUNC_sock_from_file: return &bpf_sock_from_file_proto; + case BPF_FUNC_get_socket_cookie: + return &bpf_get_socket_ptr_cookie_proto; #endif case BPF_FUNC_seq_printf: return prog->expected_attach_type == BPF_TRACE_ITER ? diff --git a/net/core/filter.c b/net/core/filter.c index e15d4741719a..57aaed478362 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -4631,6 +4631,18 @@ static const struct bpf_func_proto bpf_get_socket_cookie_sock_proto = { .arg1_type = ARG_PTR_TO_CTX, }; +BPF_CALL_1(bpf_get_socket_ptr_cookie, struct sock *, sk) +{ + return sk ? sock_gen_cookie(sk) : 0; +} + +const struct bpf_func_proto bpf_get_socket_ptr_cookie_proto = { + .func = bpf_get_socket_ptr_cookie, + .gpl_only = false, + .ret_type = RET_INTEGER, + .arg1_type = ARG_PTR_TO_BTF_ID_SOCK_COMMON, +}; + BPF_CALL_1(bpf_get_socket_cookie_sock_ops, struct bpf_sock_ops_kern *, ctx) { return __sock_gen_cookie(ctx->sk); diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index dbf10bf08582..07cc2e404291 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -1673,6 +1673,14 @@ union bpf_attr { * Return * A 8-byte long unique number. * + * u64 bpf_get_socket_cookie(struct sock *sk) + * Description + * Equivalent to **bpf_get_socket_cookie**\ () helper that accepts + * *sk*, but gets socket from a BTF **struct sock**. This helper + * also works for sleepable programs. + * Return + * A 8-byte long unique number or 0 if *sk* is NULL. + * * u32 bpf_get_socket_uid(struct sk_buff *skb) * Return * The owner UID of the socket associated to *skb*. If the socket -- cgit v1.2.3