1 .. SPDX-License-Identifier: GPL-2.0
2
3 =======
4 HID-BPF
5 =======
6
7 HID is a standard protocol for input devices but some devices may require
8 custom tweaks, traditionally done with a kernel driver fix. Using the eBPF
9 capabilities instead speeds up development and adds new capabilities to the
10 existing HID interfaces.
11
12 .. contents::
13 :local:
14 :depth: 2
15
16
17 When (and why) to use HID-BPF
18 =============================
19
20 There are several use cases when using HID-BPF is better
21 than standard kernel driver fix:
22
23 Dead zone of a joystick
24 -----------------------
25
26 Assuming you have a joystick that is getting older, it is common to see it
27 wobbling around its neutral point. This is usually filtered at the application
28 level by adding a *dead zone* for this specific axis.
29
30 With HID-BPF, we can apply this filtering in the kernel directly so userspace
31 does not get woken up when nothing else is happening on the input controller.
32
33 Of course, given that this dead zone is specific to an individual device, we
34 can not create a generic fix for all of the same joysticks. Adding a custom
35 kernel API for this (e.g. by adding a sysfs entry) does not guarantee this new
36 kernel API will be broadly adopted and maintained.
37
38 HID-BPF allows the userspace program to load the program itself, ensuring we
39 only load the custom API when we have a user.
40
41 Simple fixup of report descriptor
42 ---------------------------------
43
44 In the HID tree, half of the drivers only fix one key or one byte
45 in the report descriptor. These fixes all require a kernel patch and the
46 subsequent shepherding into a release, a long and painful process for users.
47
48 We can reduce this burden by providing an eBPF program instead. Once such a
49 program has been verified by the user, we can embed the source code into the
50 kernel tree and ship the eBPF program and load it directly instead of loading
51 a specific kernel module for it.
52
53 Note: distribution of eBPF programs and their inclusion in the kernel is not
54 yet fully implemented
55
56 Add a new feature that requires a new kernel API
57 ------------------------------------------------
58
59 An example for such a feature are the Universal Stylus Interface (USI) pens.
60 Basically, USI pens require a new kernel API because there are new
61 channels of communication that our HID and input stack do not support.
62 Instead of using hidraw or creating new sysfs entries or ioctls, we can rely
63 on eBPF to have the kernel API controlled by the consumer and to not
64 impact the performances by waking up userspace every time there is an
65 event.
66
67 Morph a device into something else and control that from userspace
68 ------------------------------------------------------------------
69
70 The kernel has a relatively static mapping of HID items to evdev bits.
71 It cannot decide to dynamically transform a given device into something else
72 as it does not have the required context and any such transformation cannot be
73 undone (or even discovered) by userspace.
74
75 However, some devices are useless with that static way of defining devices. For
76 example, the Microsoft Surface Dial is a pushbutton with haptic feedback that
77 is barely usable as of today.
78
79 With eBPF, userspace can morph that device into a mouse, and convert the dial
80 events into wheel events. Also, the userspace program can set/unset the haptic
81 feedback depending on the context. For example, if a menu is visible on the
82 screen we likely need to have a haptic click every 15 degrees. But when
83 scrolling in a web page the user experience is better when the device emits
84 events at the highest resolution.
85
86 Firewall
87 --------
88
89 What if we want to prevent other users to access a specific feature of a
90 device? (think a possibly broken firmware update entry point)
91
92 With eBPF, we can intercept any HID command emitted to the device and
93 validate it or not.
94
95 This also allows to sync the state between the userspace and the
96 kernel/bpf program because we can intercept any incoming command.
97
98 Tracing
99 -------
100
101 The last usage is tracing events and all the fun we can do we BPF to summarize
102 and analyze events.
103
104 Right now, tracing relies on hidraw. It works well except for a couple
105 of issues:
106
107 1. if the driver doesn't export a hidraw node, we can't trace anything
108 (eBPF will be a "god-mode" there, so this may raise some eyebrows)
109 2. hidraw doesn't catch other processes' requests to the device, which
110 means that we have cases where we need to add printks to the kernel
111 to understand what is happening.
112
113 High-level view of HID-BPF
114 ==========================
115
116 The main idea behind HID-BPF is that it works at an array of bytes level.
117 Thus, all of the parsing of the HID report and the HID report descriptor
118 must be implemented in the userspace component that loads the eBPF
119 program.
120
121 For example, in the dead zone joystick from above, knowing which fields
122 in the data stream needs to be set to ``0`` needs to be computed by userspace.
123
124 A corollary of this is that HID-BPF doesn't know about the other subsystems
125 available in the kernel. *You can not directly emit input event through the
126 input API from eBPF*.
127
128 When a BPF program needs to emit input events, it needs to talk with the HID
129 protocol, and rely on the HID kernel processing to translate the HID data into
130 input events.
131
132 In-tree HID-BPF programs and ``udev-hid-bpf``
133 =============================================
134
135 Official device fixes are shipped in the kernel tree as source in the
136 ``drivers/hid/bpf/progs`` directory. This allows to add selftests to them in
137 ``tools/testing/selftests/hid``.
138
139 However, the compilation of these objects is not part of a regular kernel compilation
140 given that they need an external tool to be loaded. This tool is currently
141 `udev-hid-bpf <https://libevdev.pages.freedesktop.org/udev-hid-bpf/index.html>`_.
142
143 For convenience, that external repository duplicates the files from here in
144 ``drivers/hid/bpf/progs`` into its own ``src/bpf/stable`` directory. This allows
145 distributions to not have to pull the entire kernel source tree to ship and package
146 those HID-BPF fixes. ``udev-hid-bpf`` also has capabilities of handling multiple
147 objects files depending on the kernel the user is running.
148
149 Available types of programs
150 ===========================
151
152 HID-BPF is built "on top" of BPF, meaning that we use bpf struct_ops method to
153 declare our programs.
154
155 HID-BPF has the following attachment types available:
156
157 1. event processing/filtering with ``SEC("struct_ops/hid_device_event")`` in libbpf
158 2. actions coming from userspace with ``SEC("syscall")`` in libbpf
159 3. change of the report descriptor with ``SEC("struct_ops/hid_rdesc_fixup")`` or
160 ``SEC("struct_ops.s/hid_rdesc_fixup")`` in libbpf
161
162 A ``hid_device_event`` is calling a BPF program when an event is received from
163 the device. Thus we are in IRQ context and can act on the data or notify userspace.
164 And given that we are in IRQ context, we can not talk back to the device.
165
166 A ``syscall`` means that userspace called the syscall ``BPF_PROG_RUN`` facility.
167 This time, we can do any operations allowed by HID-BPF, and talking to the device is
168 allowed.
169
170 Last, ``hid_rdesc_fixup`` is different from the others as there can be only one
171 BPF program of this type. This is called on ``probe`` from the driver and allows to
172 change the report descriptor from the BPF program. Once a ``hid_rdesc_fixup``
173 program has been loaded, it is not possible to overwrite it unless the program which
174 inserted it allows us by pinning the program and closing all of its fds pointing to it.
175
176 Note that ``hid_rdesc_fixup`` can be declared as sleepable (``SEC("struct_ops.s/hid_rdesc_fixup")``).
177
178
179 Developer API:
180 ==============
181
182 Available ``struct_ops`` for HID-BPF:
183 -------------------------------------
184
185 .. kernel-doc:: include/linux/hid_bpf.h
186 :identifiers: hid_bpf_ops
187
188
189 User API data structures available in programs:
190 -----------------------------------------------
191
192 .. kernel-doc:: include/linux/hid_bpf.h
193 :identifiers: hid_bpf_ctx
194
195 Available API that can be used in all HID-BPF struct_ops programs:
196 ------------------------------------------------------------------
197
198 .. kernel-doc:: drivers/hid/bpf/hid_bpf_dispatch.c
199 :identifiers: hid_bpf_get_data
200
201 Available API that can be used in syscall HID-BPF programs or in sleepable HID-BPF struct_ops programs:
202 -------------------------------------------------------------------------------------------------------
203
204 .. kernel-doc:: drivers/hid/bpf/hid_bpf_dispatch.c
205 :identifiers: hid_bpf_hw_request hid_bpf_hw_output_report hid_bpf_input_report hid_bpf_try_input_report hid_bpf_allocate_context hid_bpf_release_context
206
207 General overview of a HID-BPF program
208 =====================================
209
210 Accessing the data attached to the context
211 ------------------------------------------
212
213 The ``struct hid_bpf_ctx`` doesn't export the ``data`` fields directly and to access
214 it, a bpf program needs to first call :c:func:`hid_bpf_get_data`.
215
216 ``offset`` can be any integer, but ``size`` needs to be constant, known at compile
217 time.
218
219 This allows the following:
220
221 1. for a given device, if we know that the report length will always be of a certain value,
222 we can request the ``data`` pointer to point at the full report length.
223
224 The kernel will ensure we are using a correct size and offset and eBPF will ensure
225 the code will not attempt to read or write outside of the boundaries::
226
227 __u8 *data = hid_bpf_get_data(ctx, 0 /* offset */, 256 /* size */);
228
229 if (!data)
230 return 0; /* ensure data is correct, now the verifier knows we
231 * have 256 bytes available */
232
233 bpf_printk("hello world: %02x %02x %02x", data[0], data[128], data[255]);
234
235 2. if the report length is variable, but we know the value of ``X`` is always a 16-bit
236 integer, we can then have a pointer to that value only::
237
238 __u16 *x = hid_bpf_get_data(ctx, offset, sizeof(*x));
239
240 if (!x)
241 return 0; /* something went wrong */
242
243 *x += 1; /* increment X by one */
244
245 Effect of a HID-BPF program
246 ---------------------------
247
248 For all HID-BPF attachment types except for :c:func:`hid_rdesc_fixup`, several eBPF
249 programs can be attached to the same device. If a HID-BPF struct_ops has a
250 :c:func:`hid_rdesc_fixup` while another is already attached to the device, the
251 kernel will return `-EINVAL` when attaching the struct_ops.
252
253 Unless ``BPF_F_BEFORE`` is added to the flags while attaching the program, the new
254 program is appended at the end of the list.
255 ``BPF_F_BEFORE`` will insert the new program at the beginning of the list which is
256 useful for e.g. tracing where we need to get the unprocessed events from the device.
257
258 Note that if there are multiple programs using the ``BPF_F_BEFORE`` flag,
259 only the most recently loaded one is actually the first in the list.
260
261 ``SEC("struct_ops/hid_device_event")``
262 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
263
264 Whenever a matching event is raised, the eBPF programs are called one after the other
265 and are working on the same data buffer.
266
267 If a program changes the data associated with the context, the next one will see
268 the modified data but it will have *no* idea of what the original data was.
269
270 Once all the programs are run and return ``0`` or a positive value, the rest of the
271 HID stack will work on the modified data, with the ``size`` field of the last hid_bpf_ctx
272 being the new size of the input stream of data.
273
274 A BPF program returning a negative error discards the event, i.e. this event will not be
275 processed by the HID stack. Clients (hidraw, input, LEDs) will **not** see this event.
276
277 ``SEC("syscall")``
278 ~~~~~~~~~~~~~~~~~~
279
280 ``syscall`` are not attached to a given device. To tell which device we are working
281 with, userspace needs to refer to the device by its unique system id (the last 4 numbers
282 in the sysfs path: ``/sys/bus/hid/devices/xxxx:yyyy:zzzz:0000``).
283
284 To retrieve a context associated with the device, the program must call
285 hid_bpf_allocate_context() and must release it with hid_bpf_release_context()
286 before returning.
287 Once the context is retrieved, one can also request a pointer to kernel memory with
288 hid_bpf_get_data(). This memory is big enough to support all input/output/feature
289 reports of the given device.
290
291 ``SEC("struct_ops/hid_rdesc_fixup")``
292 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
293
294 The ``hid_rdesc_fixup`` program works in a similar manner to ``.report_fixup``
295 of ``struct hid_driver``.
296
297 When the device is probed, the kernel sets the data buffer of the context with the
298 content of the report descriptor. The memory associated with that buffer is
299 ``HID_MAX_DESCRIPTOR_SIZE`` (currently 4kB).
300
301 The eBPF program can modify the data buffer at-will and the kernel uses the
302 modified content and size as the report descriptor.
303
304 Whenever a struct_ops containing a ``SEC("struct_ops/hid_rdesc_fixup")`` program
305 is attached (if no program was attached before), the kernel immediately disconnects
306 the HID device and does a reprobe.
307
308 In the same way, when this struct_ops is detached, the kernel issues a disconnect
309 on the device.
310
311 There is no ``detach`` facility in HID-BPF. Detaching a program happens when
312 all the user space file descriptors pointing at a HID-BPF struct_ops link are closed.
313 Thus, if we need to replace a report descriptor fixup, some cooperation is
314 required from the owner of the original report descriptor fixup.
315 The previous owner will likely pin the struct_ops link in the bpffs, and we can then
316 replace it through normal bpf operations.
317
318 Attaching a bpf program to a device
319 ===================================
320
321 We now use standard struct_ops attachment through ``bpf_map__attach_struct_ops()``.
322 But given that we need to attach a struct_ops to a dedicated HID device, the caller
323 must set ``hid_id`` in the struct_ops map before loading the program in the kernel.
324
325 ``hid_id`` is the unique system ID of the HID device (the last 4 numbers in the
326 sysfs path: ``/sys/bus/hid/devices/xxxx:yyyy:zzzz:0000``)
327
328 One can also set ``flags``, which is of type ``enum hid_bpf_attach_flags``.
329
330 We can not rely on hidraw to bind a BPF program to a HID device. hidraw is an
331 artefact of the processing of the HID device, and is not stable. Some drivers
332 even disable it, so that removes the tracing capabilities on those devices
333 (where it is interesting to get the non-hidraw traces).
334
335 On the other hand, the ``hid_id`` is stable for the entire life of the HID device,
336 even if we change its report descriptor.
337
338 Given that hidraw is not stable when the device disconnects/reconnects, we recommend
339 accessing the current report descriptor of the device through the sysfs.
340 This is available at ``/sys/bus/hid/devices/BUS:VID:PID.000N/report_descriptor`` as a
341 binary stream.
342
343 Parsing the report descriptor is the responsibility of the BPF programmer or the userspace
344 component that loads the eBPF program.
345
346 An (almost) complete example of a BPF enhanced HID device
347 =========================================================
348
349 *Foreword: for most parts, this could be implemented as a kernel driver*
350
351 Let's imagine we have a new tablet device that has some haptic capabilities
352 to simulate the surface the user is scratching on. This device would also have
353 a specific 3 positions switch to toggle between *pencil on paper*, *cray on a wall*
354 and *brush on a painting canvas*. To make things even better, we can control the
355 physical position of the switch through a feature report.
356
357 And of course, the switch is relying on some userspace component to control the
358 haptic feature of the device itself.
359
360 Filtering events
361 ----------------
362
363 The first step consists in filtering events from the device. Given that the switch
364 position is actually reported in the flow of the pen events, using hidraw to implement
365 that filtering would mean that we wake up userspace for every single event.
366
367 This is OK for libinput, but having an external library that is just interested in
368 one byte in the report is less than ideal.
369
370 For that, we can create a basic skeleton for our BPF program::
371
372 #include "vmlinux.h"
373 #include <bpf/bpf_helpers.h>
374 #include <bpf/bpf_tracing.h>
375
376 /* HID programs need to be GPL */
377 char _license[] SEC("license") = "GPL";
378
379 /* HID-BPF kfunc API definitions */
380 extern __u8 *hid_bpf_get_data(struct hid_bpf_ctx *ctx,
381 unsigned int offset,
382 const size_t __sz) __ksym;
383
384 struct {
385 __uint(type, BPF_MAP_TYPE_RINGBUF);
386 __uint(max_entries, 4096 * 64);
387 } ringbuf SEC(".maps");
388
389 __u8 current_value = 0;
390
391 SEC("struct_ops/hid_device_event")
392 int BPF_PROG(filter_switch, struct hid_bpf_ctx *hid_ctx)
393 {
394 __u8 *data = hid_bpf_get_data(hid_ctx, 0 /* offset */, 192 /* size */);
395 __u8 *buf;
396
397 if (!data)
398 return 0; /* EPERM check */
399
400 if (current_value != data[152]) {
401 buf = bpf_ringbuf_reserve(&ringbuf, 1, 0);
402 if (!buf)
403 return 0;
404
405 *buf = data[152];
406
407 bpf_ringbuf_commit(buf, 0);
408
409 current_value = data[152];
410 }
411
412 return 0;
413 }
414
415 SEC(".struct_ops.link")
416 struct hid_bpf_ops haptic_tablet = {
417 .hid_device_event = (void *)filter_switch,
418 };
419
420
421 To attach ``haptic_tablet``, userspace needs to set ``hid_id`` first::
422
423 static int attach_filter(struct hid *hid_skel, int hid_id)
424 {
425 int err, link_fd;
426
427 hid_skel->struct_ops.haptic_tablet->hid_id = hid_id;
428 err = hid__load(skel);
429 if (err)
430 return err;
431
432 link_fd = bpf_map__attach_struct_ops(hid_skel->maps.haptic_tablet);
433 if (!link_fd) {
434 fprintf(stderr, "can not attach HID-BPF program: %m\n");
435 return -1;
436 }
437
438 return link_fd; /* the fd of the created bpf_link */
439 }
440
441 Our userspace program can now listen to notifications on the ring buffer, and
442 is awaken only when the value changes.
443
444 When the userspace program doesn't need to listen to events anymore, it can just
445 close the returned bpf link from :c:func:`attach_filter`, which will tell the kernel to
446 detach the program from the HID device.
447
448 Of course, in other use cases, the userspace program can also pin the fd to the
449 BPF filesystem through a call to :c:func:`bpf_obj_pin`, as with any bpf_link.
450
451 Controlling the device
452 ----------------------
453
454 To be able to change the haptic feedback from the tablet, the userspace program
455 needs to emit a feature report on the device itself.
456
457 Instead of using hidraw for that, we can create a ``SEC("syscall")`` program
458 that talks to the device::
459
460 /* some more HID-BPF kfunc API definitions */
461 extern struct hid_bpf_ctx *hid_bpf_allocate_context(unsigned int hid_id) __ksym;
462 extern void hid_bpf_release_context(struct hid_bpf_ctx *ctx) __ksym;
463 extern int hid_bpf_hw_request(struct hid_bpf_ctx *ctx,
464 __u8* data,
465 size_t len,
466 enum hid_report_type type,
467 enum hid_class_request reqtype) __ksym;
468
469
470 struct hid_send_haptics_args {
471 /* data needs to come at offset 0 so we can do a memcpy into it */
472 __u8 data[10];
473 unsigned int hid;
474 };
475
476 SEC("syscall")
477 int send_haptic(struct hid_send_haptics_args *args)
478 {
479 struct hid_bpf_ctx *ctx;
480 int ret = 0;
481
482 ctx = hid_bpf_allocate_context(args->hid);
483 if (!ctx)
484 return 0; /* EPERM check */
485
486 ret = hid_bpf_hw_request(ctx,
487 args->data,
488 10,
489 HID_FEATURE_REPORT,
490 HID_REQ_SET_REPORT);
491
492 hid_bpf_release_context(ctx);
493
494 return ret;
495 }
496
497 And then userspace needs to call that program directly::
498
499 static int set_haptic(struct hid *hid_skel, int hid_id, __u8 haptic_value)
500 {
501 int err, prog_fd;
502 int ret = -1;
503 struct hid_send_haptics_args args = {
504 .hid = hid_id,
505 };
506 DECLARE_LIBBPF_OPTS(bpf_test_run_opts, tattrs,
507 .ctx_in = &args,
508 .ctx_size_in = sizeof(args),
509 );
510
511 args.data[0] = 0x02; /* report ID of the feature on our device */
512 args.data[1] = haptic_value;
513
514 prog_fd = bpf_program__fd(hid_skel->progs.set_haptic);
515
516 err = bpf_prog_test_run_opts(prog_fd, &tattrs);
517 return err;
518 }
519
520 Now our userspace program is aware of the haptic state and can control it. The
521 program could make this state further available to other userspace programs
522 (e.g. via a DBus API).
523
524 The interesting bit here is that we did not created a new kernel API for this.
525 Which means that if there is a bug in our implementation, we can change the
526 interface with the kernel at-will, because the userspace application is
527 responsible for its own usage.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.