xref: /linux/drivers/platform/x86/intel/ifs/ifs.h (revision 3a39d672e7f48b8d6b91a09afa4b55352773b4b5)
1 /* SPDX-License-Identifier: GPL-2.0-only */
2 /* Copyright(c) 2022 Intel Corporation. */
3 
4 #ifndef _IFS_H_
5 #define _IFS_H_
6 
7 /**
8  * DOC: In-Field Scan
9  *
10  * =============
11  * In-Field Scan
12  * =============
13  *
14  * Introduction
15  * ------------
16  *
17  * In Field Scan (IFS) is a hardware feature to run circuit level tests on
18  * a CPU core to detect problems that are not caught by parity or ECC checks.
19  * Future CPUs will support more than one type of test which will show up
20  * with a new platform-device instance-id.
21  *
22  *
23  * IFS Image
24  * ---------
25  *
26  * Intel provides a firmware file containing the scan tests via
27  * github [#f1]_.  Similar to microcode there is a separate file for each
28  * family-model-stepping. IFS Images are not applicable for some test types.
29  * Wherever applicable the sysfs directory would provide a "current_batch" file
30  * (see below) for loading the image.
31  *
32  *
33  * IFS Image Loading
34  * -----------------
35  *
36  * The driver loads the tests into memory reserved BIOS local to each CPU
37  * socket in a two step process using writes to MSRs to first load the
38  * SHA hashes for the test. Then the tests themselves. Status MSRs provide
39  * feedback on the success/failure of these steps.
40  *
41  * The test files are kept in a fixed location: /lib/firmware/intel/ifs_<n>/
42  * For e.g if there are 3 test files, they would be named in the following
43  * fashion:
44  * ff-mm-ss-01.scan
45  * ff-mm-ss-02.scan
46  * ff-mm-ss-03.scan
47  * (where ff refers to family, mm indicates model and ss indicates stepping)
48  *
49  * A different test file can be loaded by writing the numerical portion
50  * (e.g 1, 2 or 3 in the above scenario) into the curent_batch file.
51  * To load ff-mm-ss-02.scan, the following command can be used::
52  *
53  *   # echo 2 > /sys/devices/virtual/misc/intel_ifs_<n>/current_batch
54  *
55  * The above file can also be read to know the currently loaded image.
56  *
57  * Running tests
58  * -------------
59  *
60  * Tests are run by the driver synchronizing execution of all threads on a
61  * core and then writing to the ACTIVATE_SCAN MSR on all threads. Instruction
62  * execution continues when:
63  *
64  * 1) All tests have completed.
65  * 2) Execution was interrupted.
66  * 3) A test detected a problem.
67  *
68  * Note that ALL THREADS ON THE CORE ARE EFFECTIVELY OFFLINE FOR THE
69  * DURATION OF THE TEST. This can be up to 200 milliseconds. If the system
70  * is running latency sensitive applications that cannot tolerate an
71  * interruption of this magnitude, the system administrator must arrange
72  * to migrate those applications to other cores before running a core test.
73  * It may also be necessary to redirect interrupts to other CPUs.
74  *
75  * In all cases reading the corresponding test's STATUS MSR provides details on what
76  * happened. The driver makes the value of this MSR visible to applications
77  * via the "details" file (see below). Interrupted tests may be restarted.
78  *
79  * The IFS driver provides sysfs interfaces via /sys/devices/virtual/misc/intel_ifs_<n>/
80  * to control execution:
81  *
82  * Test a specific core::
83  *
84  *   # echo <cpu#> > /sys/devices/virtual/misc/intel_ifs_<n>/run_test
85  *
86  * when HT is enabled any of the sibling cpu# can be specified to test
87  * its corresponding physical core. Since the tests are per physical core,
88  * the result of testing any thread is same. All siblings must be online
89  * to run a core test. It is only necessary to test one thread.
90  *
91  * For e.g. to test core corresponding to cpu5
92  *
93  *   # echo 5 > /sys/devices/virtual/misc/intel_ifs_<n>/run_test
94  *
95  * Results of the last test is provided in /sys::
96  *
97  *   $ cat /sys/devices/virtual/misc/intel_ifs_<n>/status
98  *   pass
99  *
100  * Status can be one of pass, fail, untested
101  *
102  * Additional details of the last test is provided by the details file::
103  *
104  *   $ cat /sys/devices/virtual/misc/intel_ifs_<n>/details
105  *   0x8081
106  *
107  * The details file reports the hex value of the test specific status MSR.
108  * Hardware defined error codes are documented in volume 4 of the Intel
109  * Software Developer's Manual but the error_code field may contain one of
110  * the following driver defined software codes:
111  *
112  * +------+--------------------+
113  * | 0xFD | Software timeout   |
114  * +------+--------------------+
115  * | 0xFE | Partial completion |
116  * +------+--------------------+
117  *
118  * Driver design choices
119  * ---------------------
120  *
121  * 1) The ACTIVATE_SCAN MSR allows for running any consecutive subrange of
122  * available tests. But the driver always tries to run all tests and only
123  * uses the subrange feature to restart an interrupted test.
124  *
125  * 2) Hardware allows for some number of cores to be tested in parallel.
126  * The driver does not make use of this, it only tests one core at a time.
127  *
128  * .. [#f1] https://github.com/intel/TBD
129  *
130  *
131  * Structural Based Functional Test at Field (SBAF):
132  * -------------------------------------------------
133  *
134  * SBAF is a new type of testing that provides comprehensive core test
135  * coverage complementing Scan at Field (SAF) testing. SBAF mimics the
136  * manufacturing screening environment and leverages the same test suite.
137  * It makes use of Design For Test (DFT) observation sites and features
138  * to maximize coverage in minimum time.
139  *
140  * Similar to the SAF test, SBAF isolates the core under test from the
141  * rest of the system during execution. Upon completion, the core
142  * seamlessly resets to its pre-test state and resumes normal operation.
143  * Any machine checks or hangs encountered during the test are confined to
144  * the isolated core, preventing disruption to the overall system.
145  *
146  * Like the SAF test, the SBAF test is also divided into multiple batches,
147  * and each batch test can take hundreds of milliseconds (100-200 ms) to
148  * complete. If such a lengthy interruption is undesirable, it is
149  * recommended to relocate the time-sensitive applications to other cores.
150  */
151 #include <linux/device.h>
152 #include <linux/miscdevice.h>
153 
154 #define MSR_ARRAY_BIST				0x00000105
155 
156 #define MSR_COPY_SBAF_HASHES			0x000002b8
157 #define MSR_SBAF_HASHES_STATUS			0x000002b9
158 #define MSR_AUTHENTICATE_AND_COPY_SBAF_CHUNK	0x000002ba
159 #define MSR_SBAF_CHUNKS_AUTHENTICATION_STATUS	0x000002bb
160 #define MSR_ACTIVATE_SBAF			0x000002bc
161 #define MSR_SBAF_STATUS				0x000002bd
162 
163 #define MSR_COPY_SCAN_HASHES			0x000002c2
164 #define MSR_SCAN_HASHES_STATUS			0x000002c3
165 #define MSR_AUTHENTICATE_AND_COPY_CHUNK		0x000002c4
166 #define MSR_CHUNKS_AUTHENTICATION_STATUS	0x000002c5
167 #define MSR_ACTIVATE_SCAN			0x000002c6
168 #define MSR_SCAN_STATUS				0x000002c7
169 #define MSR_ARRAY_TRIGGER			0x000002d6
170 #define MSR_ARRAY_STATUS			0x000002d7
171 #define MSR_SAF_CTRL				0x000004f0
172 #define MSR_SBAF_CTRL				0x000004f8
173 
174 #define SCAN_NOT_TESTED				0
175 #define SCAN_TEST_PASS				1
176 #define SCAN_TEST_FAIL				2
177 
178 #define IFS_TYPE_SAF			0
179 #define IFS_TYPE_ARRAY_BIST		1
180 #define IFS_TYPE_SBAF			2
181 
182 #define ARRAY_GEN0			0
183 #define ARRAY_GEN1			1
184 
185 /* MSR_SCAN_HASHES_STATUS bit fields */
186 union ifs_scan_hashes_status {
187 	u64	data;
188 	struct {
189 		u32	chunk_size	:16;
190 		u32	num_chunks	:8;
191 		u32	rsvd1		:8;
192 		u32	error_code	:8;
193 		u32	rsvd2		:11;
194 		u32	max_core_limit	:12;
195 		u32	valid		:1;
196 	};
197 };
198 
199 union ifs_scan_hashes_status_gen2 {
200 	u64	data;
201 	struct {
202 		u16	chunk_size;
203 		u16	num_chunks;
204 		u32	error_code		:8;
205 		u32	chunks_in_stride	:9;
206 		u32	rsvd			:2;
207 		u32	max_core_limit		:12;
208 		u32	valid			:1;
209 	};
210 };
211 
212 /* MSR_CHUNKS_AUTH_STATUS bit fields */
213 union ifs_chunks_auth_status {
214 	u64	data;
215 	struct {
216 		u32	valid_chunks	:8;
217 		u32	total_chunks	:8;
218 		u32	rsvd1		:16;
219 		u32	error_code	:8;
220 		u32	rsvd2		:24;
221 	};
222 };
223 
224 union ifs_chunks_auth_status_gen2 {
225 	u64	data;
226 	struct {
227 		u16	valid_chunks;
228 		u16	total_chunks;
229 		u32	error_code	:8;
230 		u32	rsvd2		:8;
231 		u32	max_bundle	:16;
232 	};
233 };
234 
235 /* MSR_ACTIVATE_SCAN bit fields */
236 union ifs_scan {
237 	u64	data;
238 	struct {
239 		union {
240 			struct {
241 				u8	start;
242 				u8	stop;
243 				u16	rsvd;
244 			} gen0;
245 			struct {
246 				u16	start;
247 				u16	stop;
248 			} gen2;
249 		};
250 		u32	delay	:31;
251 		u32	sigmce	:1;
252 	};
253 };
254 
255 /* MSR_SCAN_STATUS bit fields */
256 union ifs_status {
257 	u64	data;
258 	struct {
259 		union {
260 			struct {
261 				u8	chunk_num;
262 				u8	chunk_stop_index;
263 				u16	rsvd1;
264 			} gen0;
265 			struct {
266 				u16	chunk_num;
267 				u16	chunk_stop_index;
268 			} gen2;
269 		};
270 		u32	error_code		:8;
271 		u32	rsvd2			:22;
272 		u32	control_error		:1;
273 		u32	signature_error		:1;
274 	};
275 };
276 
277 /* MSR_ARRAY_BIST bit fields */
278 union ifs_array {
279 	u64	data;
280 	struct {
281 		u32	array_bitmask;
282 		u16	array_bank;
283 		u16	rsvd			:15;
284 		u16	ctrl_result		:1;
285 	};
286 };
287 
288 /* MSR_ACTIVATE_SBAF bit fields */
289 union ifs_sbaf {
290 	u64	data;
291 	struct {
292 		u32	bundle_idx	:9;
293 		u32	rsvd1		:5;
294 		u32	pgm_idx		:2;
295 		u32	rsvd2		:16;
296 		u32	delay		:31;
297 		u32	sigmce		:1;
298 	};
299 };
300 
301 /* MSR_SBAF_STATUS bit fields */
302 union ifs_sbaf_status {
303 	u64	data;
304 	struct {
305 		u32	bundle_idx	:9;
306 		u32	rsvd1		:5;
307 		u32	pgm_idx		:2;
308 		u32	rsvd2		:16;
309 		u32	error_code	:8;
310 		u32	rsvd3		:21;
311 		u32	test_fail	:1;
312 		u32	sbaf_status	:2;
313 	};
314 };
315 
316 /*
317  * Driver populated error-codes
318  * 0xFD: Test timed out before completing all the chunks.
319  * 0xFE: not all scan chunks were executed. Maximum forward progress retries exceeded.
320  */
321 #define IFS_SW_TIMEOUT				0xFD
322 #define IFS_SW_PARTIAL_COMPLETION		0xFE
323 
324 #define IFS_SUFFIX_SZ		5
325 
326 struct ifs_test_caps {
327 	int	integrity_cap_bit;
328 	int	test_num;
329 	char	image_suffix[IFS_SUFFIX_SZ];
330 };
331 
332 /**
333  * struct ifs_test_msrs - MSRs used in IFS tests
334  * @copy_hashes: Copy test hash data
335  * @copy_hashes_status: Status of copied test hash data
336  * @copy_chunks: Copy chunks of the test data
337  * @copy_chunks_status: Status of the copied test data chunks
338  * @test_ctrl: Control the test attributes
339  */
340 struct ifs_test_msrs {
341 	u32	copy_hashes;
342 	u32	copy_hashes_status;
343 	u32	copy_chunks;
344 	u32	copy_chunks_status;
345 	u32	test_ctrl;
346 };
347 
348 /**
349  * struct ifs_data - attributes related to intel IFS driver
350  * @loaded_version: stores the currently loaded ifs image version.
351  * @loaded: If a valid test binary has been loaded into the memory
352  * @loading_error: Error occurred on another CPU while loading image
353  * @valid_chunks: number of chunks which could be validated.
354  * @status: it holds simple status pass/fail/untested
355  * @scan_details: opaque scan status code from h/w
356  * @cur_batch: number indicating the currently loaded test file
357  * @generation: IFS test generation enumerated by hardware
358  * @chunk_size: size of a test chunk
359  * @array_gen: test generation of array test
360  * @max_bundle: maximum bundle index
361  */
362 struct ifs_data {
363 	int	loaded_version;
364 	bool	loaded;
365 	bool	loading_error;
366 	int	valid_chunks;
367 	int	status;
368 	u64	scan_details;
369 	u32	cur_batch;
370 	u32	generation;
371 	u32	chunk_size;
372 	u32	array_gen;
373 	u32	max_bundle;
374 };
375 
376 struct ifs_work {
377 	struct work_struct w;
378 	struct device *dev;
379 };
380 
381 struct ifs_device {
382 	const struct ifs_test_caps *test_caps;
383 	const struct ifs_test_msrs *test_msrs;
384 	struct ifs_data rw_data;
385 	struct miscdevice misc;
386 };
387 
ifs_get_data(struct device * dev)388 static inline struct ifs_data *ifs_get_data(struct device *dev)
389 {
390 	struct miscdevice *m = dev_get_drvdata(dev);
391 	struct ifs_device *d = container_of(m, struct ifs_device, misc);
392 
393 	return &d->rw_data;
394 }
395 
ifs_get_test_caps(struct device * dev)396 static inline const struct ifs_test_caps *ifs_get_test_caps(struct device *dev)
397 {
398 	struct miscdevice *m = dev_get_drvdata(dev);
399 	struct ifs_device *d = container_of(m, struct ifs_device, misc);
400 
401 	return d->test_caps;
402 }
403 
ifs_get_test_msrs(struct device * dev)404 static inline const struct ifs_test_msrs *ifs_get_test_msrs(struct device *dev)
405 {
406 	struct miscdevice *m = dev_get_drvdata(dev);
407 	struct ifs_device *d = container_of(m, struct ifs_device, misc);
408 
409 	return d->test_msrs;
410 }
411 
412 extern bool *ifs_pkg_auth;
413 int ifs_load_firmware(struct device *dev);
414 int do_core_test(int cpu, struct device *dev);
415 extern struct attribute *plat_ifs_attrs[];
416 extern struct attribute *plat_ifs_array_attrs[];
417 
418 #endif
419