kernel/trace/bpf_trace.c

   1 /* Copyright (c) 2011-2015 PLUMgrid, http://plumgrid.com
   2  *
   3  * This program is free software; you can redistribute it and/or
   4  * modify it under the terms of version 2 of the GNU General Public
   5  * License as published by the Free Software Foundation.
   6  */
   7 #include <linux/kernel.h>
   8 #include <linux/types.h>
   9 #include <linux/slab.h>
  10 #include <linux/bpf.h>
  11 #include <linux/filter.h>
  12 #include <linux/uaccess.h>
  13 #include <linux/ctype.h>
  14 #include "trace.h"
  15
  16 static DEFINE_PER_CPU(int, bpf_prog_active);
  17
  18 /**
  19  * trace_call_bpf - invoke BPF program
  20  * @prog: BPF program
  21  * @ctx: opaque context pointer
  22  *
  23  * kprobe handlers execute BPF programs via this helper.
  24  * Can be used from static tracepoints in the future.
  25  *
  26  * Return: BPF programs always return an integer which is interpreted by
  27  * kprobe handler as:
  28  * 0 - return from kprobe (event is filtered out)
  29  * 1 - store kprobe event into ring buffer
  30  * Other values are reserved and currently alias to 1
  31  */
  32 unsigned int trace_call_bpf(struct bpf_prog *prog, void *ctx)
  33 {
  34         unsigned int ret;
  35
  36         if (in_nmi()) /* not supported yet */
  37                 return 1;
  38
  39         preempt_disable();
  40
  41         if (unlikely(__this_cpu_inc_return(bpf_prog_active) != 1)) {
  42                 /*
  43                  * since some bpf program is already running on this cpu,
  44                  * don't call into another bpf program (same or different)
  45                  * and don't send kprobe event into ring-buffer,
  46                  * so return zero here
  47                  */
  48                 ret = 0;
  49                 goto out;
  50         }
  51
  52         rcu_read_lock();
  53         ret = BPF_PROG_RUN(prog, ctx);
  54         rcu_read_unlock();
  55
  56  out:
  57         __this_cpu_dec(bpf_prog_active);
  58         preempt_enable();
  59
  60         return ret;
  61 }
  62 EXPORT_SYMBOL_GPL(trace_call_bpf);
  63
  64 static u64 bpf_probe_read(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
  65 {
  66         void *dst = (void *) (long) r1;
  67         int size = (int) r2;
  68         void *unsafe_ptr = (void *) (long) r3;
  69
  70         return probe_kernel_read(dst, unsafe_ptr, size);
  71 }
  72
  73 static const struct bpf_func_proto bpf_probe_read_proto = {
  74         .func           = bpf_probe_read,
  75         .gpl_only       = true,
  76         .ret_type       = RET_INTEGER,
  77         .arg1_type      = ARG_PTR_TO_STACK,
  78         .arg2_type      = ARG_CONST_STACK_SIZE,
  79         .arg3_type      = ARG_ANYTHING,
  80 };
  81
  82 /*
  83  * limited trace_printk()
  84  * only %d %u %x %ld %lu %lx %lld %llu %llx %p %s conversion specifiers allowed
  85  */
  86 static u64 bpf_trace_printk(u64 r1, u64 fmt_size, u64 r3, u64 r4, u64 r5)
  87 {
  88         char *fmt = (char *) (long) r1;
  89         bool str_seen = false;
  90         int mod[3] = {};
  91         int fmt_cnt = 0;
  92         u64 unsafe_addr;
  93         char buf[64];
  94         int i;
  95
  96         /*
  97          * bpf_check()->check_func_arg()->check_stack_boundary()
  98          * guarantees that fmt points to bpf program stack,
  99          * fmt_size bytes of it were initialized and fmt_size > 0
 100          */
 101         if (fmt[--fmt_size] != 0)
 102                 return -EINVAL;
 103
 104         /* check format string for allowed specifiers */
 105         for (i = 0; i < fmt_size; i++) {
 106                 if ((!isprint(fmt[i]) && !isspace(fmt[i])) || !isascii(fmt[i]))
 107                         return -EINVAL;
 108
 109                 if (fmt[i] != '%')
 110                         continue;
 111
 112                 if (fmt_cnt >= 3)
 113                         return -EINVAL;
 114
 115                 /* fmt[i] != 0 && fmt[last] == 0, so we can access fmt[i + 1] */
 116                 i++;
 117                 if (fmt[i] == 'l') {
 118                         mod[fmt_cnt]++;
 119                         i++;
 120                 } else if (fmt[i] == 'p' || fmt[i] == 's') {
 121                         mod[fmt_cnt]++;
 122                         i++;
 123                         if (!isspace(fmt[i]) && !ispunct(fmt[i]) && fmt[i] != 0)
 124                                 return -EINVAL;
 125                         fmt_cnt++;
 126                         if (fmt[i - 1] == 's') {
 127                                 if (str_seen)
 128                                         /* allow only one '%s' per fmt string */
 129                                         return -EINVAL;
 130                                 str_seen = true;
 131
 132                                 switch (fmt_cnt) {
 133                                 case 1:
 134                                         unsafe_addr = r3;
 135                                         r3 = (long) buf;
 136                                         break;
 137                                 case 2:
 138                                         unsafe_addr = r4;
 139                                         r4 = (long) buf;
 140                                         break;
 141                                 case 3:
 142                                         unsafe_addr = r5;
 143                                         r5 = (long) buf;
 144                                         break;
 145                                 }
 146                                 buf[0] = 0;
 147                                 strncpy_from_unsafe(buf,
 148                                                     (void *) (long) unsafe_addr,
 149                                                     sizeof(buf));
 150                         }
 151                         continue;
 152                 }
 153
 154                 if (fmt[i] == 'l') {
 155                         mod[fmt_cnt]++;
 156                         i++;
 157                 }
 158
 159                 if (fmt[i] != 'd' && fmt[i] != 'u' && fmt[i] != 'x')
 160                         return -EINVAL;
 161                 fmt_cnt++;
 162         }
 163
 164         return __trace_printk(1/* fake ip will not be printed */, fmt,
 165                               mod[0] == 2 ? r3 : mod[0] == 1 ? (long) r3 : (u32) r3,
 166                               mod[1] == 2 ? r4 : mod[1] == 1 ? (long) r4 : (u32) r4,
 167                               mod[2] == 2 ? r5 : mod[2] == 1 ? (long) r5 : (u32) r5);
 168 }
 169
 170 static const struct bpf_func_proto bpf_trace_printk_proto = {
 171         .func           = bpf_trace_printk,
 172         .gpl_only       = true,
 173         .ret_type       = RET_INTEGER,
 174         .arg1_type      = ARG_PTR_TO_STACK,
 175         .arg2_type      = ARG_CONST_STACK_SIZE,
 176 };
 177
 178 const struct bpf_func_proto *bpf_get_trace_printk_proto(void)
 179 {
 180         /*
 181          * this program might be calling bpf_trace_printk,
 182          * so allocate per-cpu printk buffers
 183          */
 184         trace_printk_init_buffers();
 185
 186         return &bpf_trace_printk_proto;
 187 }
 188
 189 static u64 bpf_perf_event_read(u64 r1, u64 index, u64 r3, u64 r4, u64 r5)
 190 {
 191         struct bpf_map *map = (struct bpf_map *) (unsigned long) r1;
 192         struct bpf_array *array = container_of(map, struct bpf_array, map);
 193         struct perf_event *event;
 194
 195         if (unlikely(index >= array->map.max_entries))
 196                 return -E2BIG;
 197
 198         event = (struct perf_event *)array->ptrs[index];
 199         if (!event)
 200                 return -ENOENT;
 201
 202         /*
 203          * we don't know if the function is run successfully by the
 204          * return value. It can be judged in other places, such as
 205          * eBPF programs.
 206          */
 207         return perf_event_read_local(event);
 208 }
 209
 210 const struct bpf_func_proto bpf_perf_event_read_proto = {
 211         .func           = bpf_perf_event_read,
 212         .gpl_only       = false,
 213         .ret_type       = RET_INTEGER,
 214         .arg1_type      = ARG_CONST_MAP_PTR,
 215         .arg2_type      = ARG_ANYTHING,
 216 };
 217
 218 static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 index, u64 r4, u64 size)
 219 {
 220         struct pt_regs *regs = (struct pt_regs *) (long) r1;
 221         struct bpf_map *map = (struct bpf_map *) (long) r2;
 222         struct bpf_array *array = container_of(map, struct bpf_array, map);
 223         void *data = (void *) (long) r4;
 224         struct perf_sample_data sample_data;
 225         struct perf_event *event;
 226         struct perf_raw_record raw = {
 227                 .size = size,
 228                 .data = data,
 229         };
 230
 231         if (unlikely(index >= array->map.max_entries))
 232                 return -E2BIG;
 233
 234         event = (struct perf_event *)array->ptrs[index];
 235         if (unlikely(!event))
 236                 return -ENOENT;
 237
 238         if (unlikely(event->attr.type != PERF_TYPE_SOFTWARE ||
 239                      event->attr.config != PERF_COUNT_SW_BPF_OUTPUT))
 240                 return -EINVAL;
 241
 242         if (unlikely(event->oncpu != smp_processor_id()))
 243                 return -EOPNOTSUPP;
 244
 245         perf_sample_data_init(&sample_data, 0, 0);
 246         sample_data.raw = &raw;
 247         perf_event_output(event, &sample_data, regs);
 248         return 0;
 249 }
 250
 251 static const struct bpf_func_proto bpf_perf_event_output_proto = {
 252         .func           = bpf_perf_event_output,
 253         .gpl_only       = false,
 254         .ret_type       = RET_INTEGER,
 255         .arg1_type      = ARG_PTR_TO_CTX,
 256         .arg2_type      = ARG_CONST_MAP_PTR,
 257         .arg3_type      = ARG_ANYTHING,
 258         .arg4_type      = ARG_PTR_TO_STACK,
 259         .arg5_type      = ARG_CONST_STACK_SIZE,
 260 };
 261
 262 static const struct bpf_func_proto *kprobe_prog_func_proto(enum bpf_func_id func_id)
 263 {
 264         switch (func_id) {
 265         case BPF_FUNC_map_lookup_elem:
 266                 return &bpf_map_lookup_elem_proto;
 267         case BPF_FUNC_map_update_elem:
 268                 return &bpf_map_update_elem_proto;
 269         case BPF_FUNC_map_delete_elem:
 270                 return &bpf_map_delete_elem_proto;
 271         case BPF_FUNC_probe_read:
 272                 return &bpf_probe_read_proto;
 273         case BPF_FUNC_ktime_get_ns:
 274                 return &bpf_ktime_get_ns_proto;
 275         case BPF_FUNC_tail_call:
 276                 return &bpf_tail_call_proto;
 277         case BPF_FUNC_get_current_pid_tgid:
 278                 return &bpf_get_current_pid_tgid_proto;
 279         case BPF_FUNC_get_current_uid_gid:
 280                 return &bpf_get_current_uid_gid_proto;
 281         case BPF_FUNC_get_current_comm:
 282                 return &bpf_get_current_comm_proto;
 283         case BPF_FUNC_trace_printk:
 284                 return bpf_get_trace_printk_proto();
 285         case BPF_FUNC_get_smp_processor_id:
 286                 return &bpf_get_smp_processor_id_proto;
 287         case BPF_FUNC_perf_event_read:
 288                 return &bpf_perf_event_read_proto;
 289         case BPF_FUNC_perf_event_output:
 290                 return &bpf_perf_event_output_proto;
 291         default:
 292                 return NULL;
 293         }
 294 }
 295
 296 /* bpf+kprobe programs can access fields of 'struct pt_regs' */
 297 static bool kprobe_prog_is_valid_access(int off, int size, enum bpf_access_type type)
 298 {
 299         /* check bounds */
 300         if (off < 0 || off >= sizeof(struct pt_regs))
 301                 return false;
 302
 303         /* only read is allowed */
 304         if (type != BPF_READ)
 305                 return false;
 306
 307         /* disallow misaligned access */
 308         if (off % size != 0)
 309                 return false;
 310
 311         return true;
 312 }
 313
 314 static struct bpf_verifier_ops kprobe_prog_ops = {
 315         .get_func_proto  = kprobe_prog_func_proto,
 316         .is_valid_access = kprobe_prog_is_valid_access,
 317 };
 318
 319 static struct bpf_prog_type_list kprobe_tl = {
 320         .ops    = &kprobe_prog_ops,
 321         .type   = BPF_PROG_TYPE_KPROBE,
 322 };
 323
 324 static int __init register_kprobe_prog_ops(void)
 325 {
 326         bpf_register_prog_type(&kprobe_tl);
 327         return 0;
 328 }
 329 late_initcall(register_kprobe_prog_ops);