Commit | Line | Data |
---|---|---|
2541517c AS |
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> | |
9c959c86 | 13 | #include <linux/ctype.h> |
2541517c AS |
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 | ||
9c959c86 AS |
82 | /* |
83 | * limited trace_printk() | |
8d3b7dce | 84 | * only %d %u %x %ld %lu %lx %lld %llu %llx %p %s conversion specifiers allowed |
9c959c86 AS |
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; | |
8d3b7dce | 89 | bool str_seen = false; |
9c959c86 AS |
90 | int mod[3] = {}; |
91 | int fmt_cnt = 0; | |
8d3b7dce AS |
92 | u64 unsafe_addr; |
93 | char buf[64]; | |
9c959c86 AS |
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++; | |
8d3b7dce | 120 | } else if (fmt[i] == 'p' || fmt[i] == 's') { |
9c959c86 AS |
121 | mod[fmt_cnt]++; |
122 | i++; | |
123 | if (!isspace(fmt[i]) && !ispunct(fmt[i]) && fmt[i] != 0) | |
124 | return -EINVAL; | |
125 | fmt_cnt++; | |
8d3b7dce AS |
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 | } | |
9c959c86 AS |
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 | ||
0756ea3e AS |
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 | ||
35578d79 KX |
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; | |
e03e7ee3 | 194 | struct file *file; |
35578d79 KX |
195 | |
196 | if (unlikely(index >= array->map.max_entries)) | |
197 | return -E2BIG; | |
198 | ||
e03e7ee3 AS |
199 | file = (struct file *)array->ptrs[index]; |
200 | if (unlikely(!file)) | |
35578d79 KX |
201 | return -ENOENT; |
202 | ||
e03e7ee3 AS |
203 | event = file->private_data; |
204 | ||
62544ce8 AS |
205 | /* make sure event is local and doesn't have pmu::count */ |
206 | if (event->oncpu != smp_processor_id() || | |
207 | event->pmu->count) | |
208 | return -EINVAL; | |
209 | ||
35578d79 KX |
210 | /* |
211 | * we don't know if the function is run successfully by the | |
212 | * return value. It can be judged in other places, such as | |
213 | * eBPF programs. | |
214 | */ | |
215 | return perf_event_read_local(event); | |
216 | } | |
217 | ||
62544ce8 | 218 | static const struct bpf_func_proto bpf_perf_event_read_proto = { |
35578d79 | 219 | .func = bpf_perf_event_read, |
1075ef59 | 220 | .gpl_only = true, |
35578d79 KX |
221 | .ret_type = RET_INTEGER, |
222 | .arg1_type = ARG_CONST_MAP_PTR, | |
223 | .arg2_type = ARG_ANYTHING, | |
224 | }; | |
225 | ||
a43eec30 AS |
226 | static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 index, u64 r4, u64 size) |
227 | { | |
228 | struct pt_regs *regs = (struct pt_regs *) (long) r1; | |
229 | struct bpf_map *map = (struct bpf_map *) (long) r2; | |
230 | struct bpf_array *array = container_of(map, struct bpf_array, map); | |
231 | void *data = (void *) (long) r4; | |
232 | struct perf_sample_data sample_data; | |
233 | struct perf_event *event; | |
e03e7ee3 | 234 | struct file *file; |
a43eec30 AS |
235 | struct perf_raw_record raw = { |
236 | .size = size, | |
237 | .data = data, | |
238 | }; | |
239 | ||
240 | if (unlikely(index >= array->map.max_entries)) | |
241 | return -E2BIG; | |
242 | ||
e03e7ee3 AS |
243 | file = (struct file *)array->ptrs[index]; |
244 | if (unlikely(!file)) | |
a43eec30 AS |
245 | return -ENOENT; |
246 | ||
e03e7ee3 AS |
247 | event = file->private_data; |
248 | ||
a43eec30 AS |
249 | if (unlikely(event->attr.type != PERF_TYPE_SOFTWARE || |
250 | event->attr.config != PERF_COUNT_SW_BPF_OUTPUT)) | |
251 | return -EINVAL; | |
252 | ||
253 | if (unlikely(event->oncpu != smp_processor_id())) | |
254 | return -EOPNOTSUPP; | |
255 | ||
256 | perf_sample_data_init(&sample_data, 0, 0); | |
257 | sample_data.raw = &raw; | |
258 | perf_event_output(event, &sample_data, regs); | |
259 | return 0; | |
260 | } | |
261 | ||
262 | static const struct bpf_func_proto bpf_perf_event_output_proto = { | |
263 | .func = bpf_perf_event_output, | |
1075ef59 | 264 | .gpl_only = true, |
a43eec30 AS |
265 | .ret_type = RET_INTEGER, |
266 | .arg1_type = ARG_PTR_TO_CTX, | |
267 | .arg2_type = ARG_CONST_MAP_PTR, | |
268 | .arg3_type = ARG_ANYTHING, | |
269 | .arg4_type = ARG_PTR_TO_STACK, | |
270 | .arg5_type = ARG_CONST_STACK_SIZE, | |
271 | }; | |
272 | ||
2541517c AS |
273 | static const struct bpf_func_proto *kprobe_prog_func_proto(enum bpf_func_id func_id) |
274 | { | |
275 | switch (func_id) { | |
276 | case BPF_FUNC_map_lookup_elem: | |
277 | return &bpf_map_lookup_elem_proto; | |
278 | case BPF_FUNC_map_update_elem: | |
279 | return &bpf_map_update_elem_proto; | |
280 | case BPF_FUNC_map_delete_elem: | |
281 | return &bpf_map_delete_elem_proto; | |
282 | case BPF_FUNC_probe_read: | |
283 | return &bpf_probe_read_proto; | |
d9847d31 AS |
284 | case BPF_FUNC_ktime_get_ns: |
285 | return &bpf_ktime_get_ns_proto; | |
04fd61ab AS |
286 | case BPF_FUNC_tail_call: |
287 | return &bpf_tail_call_proto; | |
ffeedafb AS |
288 | case BPF_FUNC_get_current_pid_tgid: |
289 | return &bpf_get_current_pid_tgid_proto; | |
290 | case BPF_FUNC_get_current_uid_gid: | |
291 | return &bpf_get_current_uid_gid_proto; | |
292 | case BPF_FUNC_get_current_comm: | |
293 | return &bpf_get_current_comm_proto; | |
9c959c86 | 294 | case BPF_FUNC_trace_printk: |
0756ea3e | 295 | return bpf_get_trace_printk_proto(); |
ab1973d3 AS |
296 | case BPF_FUNC_get_smp_processor_id: |
297 | return &bpf_get_smp_processor_id_proto; | |
35578d79 KX |
298 | case BPF_FUNC_perf_event_read: |
299 | return &bpf_perf_event_read_proto; | |
a43eec30 AS |
300 | case BPF_FUNC_perf_event_output: |
301 | return &bpf_perf_event_output_proto; | |
2541517c AS |
302 | default: |
303 | return NULL; | |
304 | } | |
305 | } | |
306 | ||
307 | /* bpf+kprobe programs can access fields of 'struct pt_regs' */ | |
308 | static bool kprobe_prog_is_valid_access(int off, int size, enum bpf_access_type type) | |
309 | { | |
310 | /* check bounds */ | |
311 | if (off < 0 || off >= sizeof(struct pt_regs)) | |
312 | return false; | |
313 | ||
314 | /* only read is allowed */ | |
315 | if (type != BPF_READ) | |
316 | return false; | |
317 | ||
318 | /* disallow misaligned access */ | |
319 | if (off % size != 0) | |
320 | return false; | |
321 | ||
322 | return true; | |
323 | } | |
324 | ||
27dff4e0 | 325 | static const struct bpf_verifier_ops kprobe_prog_ops = { |
2541517c AS |
326 | .get_func_proto = kprobe_prog_func_proto, |
327 | .is_valid_access = kprobe_prog_is_valid_access, | |
328 | }; | |
329 | ||
330 | static struct bpf_prog_type_list kprobe_tl = { | |
331 | .ops = &kprobe_prog_ops, | |
332 | .type = BPF_PROG_TYPE_KPROBE, | |
333 | }; | |
334 | ||
335 | static int __init register_kprobe_prog_ops(void) | |
336 | { | |
337 | bpf_register_prog_type(&kprobe_tl); | |
338 | return 0; | |
339 | } | |
340 | late_initcall(register_kprobe_prog_ops); |