1// SPDX-License-Identifier: GPL-2.0-only
  2/*:
  3 * Hibernate support specific for ARM64
  4 *
  5 * Derived from work on ARM hibernation support by:
  6 *
  7 * Ubuntu project, hibernation support for mach-dove
  8 * Copyright (C) 2010 Nokia Corporation (Hiroshi Doyu)
  9 * Copyright (C) 2010 Texas Instruments, Inc. (Teerth Reddy et al.)
 10 * Copyright (C) 2006 Rafael J. Wysocki <rjw@sisk.pl>
 11 */
 12#define pr_fmt(x) "hibernate: " x
 13#include <linux/cpu.h>
 14#include <linux/kvm_host.h>
 15#include <linux/pm.h>
 16#include <linux/sched.h>
 17#include <linux/suspend.h>
 18#include <linux/utsname.h>
 19
 20#include <asm/barrier.h>
 21#include <asm/cacheflush.h>
 22#include <asm/cputype.h>
 23#include <asm/daifflags.h>
 24#include <asm/irqflags.h>
 25#include <asm/kexec.h>
 26#include <asm/memory.h>
 27#include <asm/mmu_context.h>
 28#include <asm/mte.h>
 29#include <asm/sections.h>
 30#include <asm/smp.h>
 31#include <asm/smp_plat.h>
 32#include <asm/suspend.h>
 33#include <asm/sysreg.h>
 34#include <asm/trans_pgd.h>
 35#include <asm/virt.h>
 36
 37/*
 38 * Hibernate core relies on this value being 0 on resume, and marks it
 39 * __nosavedata assuming it will keep the resume kernel's '0' value. This
 40 * doesn't happen with either KASLR.
 41 *
 42 * defined as "__visible int in_suspend __nosavedata" in
 43 * kernel/power/hibernate.c
 44 */
 45extern int in_suspend;
 46
 47/* Do we need to reset el2? */
 48#define el2_reset_needed() (is_hyp_nvhe())
 49
 50/* hyp-stub vectors, used to restore el2 during resume from hibernate. */
 51extern char __hyp_stub_vectors[];
 52
 53/*
 54 * The logical cpu number we should resume on, initialised to a non-cpu
 55 * number.
 56 */
 57static int sleep_cpu = -EINVAL;
 58
 59/*
 60 * Values that may not change over hibernate/resume. We put the build number
 61 * and date in here so that we guarantee not to resume with a different
 62 * kernel.
 63 */
 64struct arch_hibernate_hdr_invariants {
 65	char		uts_version[__NEW_UTS_LEN + 1];
 66};
 67
 68/* These values need to be know across a hibernate/restore. */
 69static struct arch_hibernate_hdr {
 70	struct arch_hibernate_hdr_invariants invariants;
 71
 72	/* These are needed to find the relocated kernel if built with kaslr */
 73	phys_addr_t	ttbr1_el1;
 74	void		(*reenter_kernel)(void);
 75
 76	/*
 77	 * We need to know where the __hyp_stub_vectors are after restore to
 78	 * re-configure el2.
 79	 */
 80	phys_addr_t	__hyp_stub_vectors;
 81
 82	u64		sleep_cpu_mpidr;
 83} resume_hdr;
 84
 85static inline void arch_hdr_invariants(struct arch_hibernate_hdr_invariants *i)
 86{
 87	memset(i, 0, sizeof(*i));
 88	memcpy(i->uts_version, init_utsname()->version, sizeof(i->uts_version));
 89}
 90
 91int pfn_is_nosave(unsigned long pfn)
 92{
 93	unsigned long nosave_begin_pfn = sym_to_pfn(&__nosave_begin);
 94	unsigned long nosave_end_pfn = sym_to_pfn(&__nosave_end - 1);
 95
 96	return ((pfn >= nosave_begin_pfn) && (pfn <= nosave_end_pfn)) ||
 97		crash_is_nosave(pfn);
 98}
 99
100void notrace save_processor_state(void)
101{
102	WARN_ON(num_online_cpus() != 1);
103}
104
105void notrace restore_processor_state(void)
106{
107}
108
109int arch_hibernation_header_save(void *addr, unsigned int max_size)
110{
111	struct arch_hibernate_hdr *hdr = addr;
112
113	if (max_size < sizeof(*hdr))
114		return -EOVERFLOW;
115
116	arch_hdr_invariants(&hdr->invariants);
117	hdr->ttbr1_el1		= __pa_symbol(swapper_pg_dir);
118	hdr->reenter_kernel	= _cpu_resume;
119
120	/* We can't use __hyp_get_vectors() because kvm may still be loaded */
121	if (el2_reset_needed())
122		hdr->__hyp_stub_vectors = __pa_symbol(__hyp_stub_vectors);
123	else
124		hdr->__hyp_stub_vectors = 0;
125
126	/* Save the mpidr of the cpu we called cpu_suspend() on... */
127	if (sleep_cpu < 0) {
128		pr_err("Failing to hibernate on an unknown CPU.\n");
129		return -ENODEV;
130	}
131	hdr->sleep_cpu_mpidr = cpu_logical_map(sleep_cpu);
132	pr_info("Hibernating on CPU %d [mpidr:0x%llx]\n", sleep_cpu,
133		hdr->sleep_cpu_mpidr);
134
135	return 0;
136}
137EXPORT_SYMBOL(arch_hibernation_header_save);
138
139int arch_hibernation_header_restore(void *addr)
140{
141	int ret;
142	struct arch_hibernate_hdr_invariants invariants;
143	struct arch_hibernate_hdr *hdr = addr;
144
145	arch_hdr_invariants(&invariants);
146	if (memcmp(&hdr->invariants, &invariants, sizeof(invariants))) {
147		pr_crit("Hibernate image not generated by this kernel!\n");
148		return -EINVAL;
149	}
150
151	sleep_cpu = get_logical_index(hdr->sleep_cpu_mpidr);
152	pr_info("Hibernated on CPU %d [mpidr:0x%llx]\n", sleep_cpu,
153		hdr->sleep_cpu_mpidr);
154	if (sleep_cpu < 0) {
155		pr_crit("Hibernated on a CPU not known to this kernel!\n");
156		sleep_cpu = -EINVAL;
157		return -EINVAL;
158	}
159
160	ret = bringup_hibernate_cpu(sleep_cpu);
161	if (ret) {
162		sleep_cpu = -EINVAL;
163		return ret;
164	}
165
166	resume_hdr = *hdr;
167
168	return 0;
169}
170EXPORT_SYMBOL(arch_hibernation_header_restore);
171
172static void *hibernate_page_alloc(void *arg)
173{
174	return (void *)get_safe_page((__force gfp_t)(unsigned long)arg);
175}
176
177/*
178 * Copies length bytes, starting at src_start into an new page,
179 * perform cache maintenance, then maps it at the specified address low
180 * address as executable.
181 *
182 * This is used by hibernate to copy the code it needs to execute when
183 * overwriting the kernel text. This function generates a new set of page
184 * tables, which it loads into ttbr0.
185 *
186 * Length is provided as we probably only want 4K of data, even on a 64K
187 * page system.
188 */
189static int create_safe_exec_page(void *src_start, size_t length,
190				 phys_addr_t *phys_dst_addr)
191{
192	struct trans_pgd_info trans_info = {
193		.trans_alloc_page	= hibernate_page_alloc,
194		.trans_alloc_arg	= (__force void *)GFP_ATOMIC,
195	};
196
197	void *page = (void *)get_safe_page(GFP_ATOMIC);
198	phys_addr_t trans_ttbr0;
199	unsigned long t0sz;
200	int rc;
201
202	if (!page)
203		return -ENOMEM;
204
205	memcpy(page, src_start, length);
206	caches_clean_inval_pou((unsigned long)page, (unsigned long)page + length);
207	rc = trans_pgd_idmap_page(&trans_info, &trans_ttbr0, &t0sz, page);
208	if (rc)
209		return rc;
210
211	cpu_install_ttbr0(trans_ttbr0, t0sz);
212	*phys_dst_addr = virt_to_phys(page);
213
214	return 0;
215}
216
217#ifdef CONFIG_ARM64_MTE
218
219static DEFINE_XARRAY(mte_pages);
220
221static int save_tags(struct page *page, unsigned long pfn)
222{
223	void *tag_storage, *ret;
224
225	tag_storage = mte_allocate_tag_storage();
226	if (!tag_storage)
227		return -ENOMEM;
228
229	mte_save_page_tags(page_address(page), tag_storage);
230
231	ret = xa_store(&mte_pages, pfn, tag_storage, GFP_KERNEL);
232	if (WARN(xa_is_err(ret), "Failed to store MTE tags")) {
233		mte_free_tag_storage(tag_storage);
234		return xa_err(ret);
235	} else if (WARN(ret, "swsusp: %s: Duplicate entry", __func__)) {
236		mte_free_tag_storage(ret);
237	}
238
239	return 0;
240}
241
242static void swsusp_mte_free_storage(void)
243{
244	XA_STATE(xa_state, &mte_pages, 0);
245	void *tags;
246
247	xa_lock(&mte_pages);
248	xas_for_each(&xa_state, tags, ULONG_MAX) {
249		mte_free_tag_storage(tags);
250	}
251	xa_unlock(&mte_pages);
252
253	xa_destroy(&mte_pages);
254}
255
256static int swsusp_mte_save_tags(void)
257{
258	struct zone *zone;
259	unsigned long pfn, max_zone_pfn;
260	int ret = 0;
261	int n = 0;
262
263	if (!system_supports_mte())
264		return 0;
265
266	for_each_populated_zone(zone) {
267		max_zone_pfn = zone_end_pfn(zone);
268		for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++) {
269			struct page *page = pfn_to_online_page(pfn);
270
271			if (!page)
272				continue;
273
274			if (!page_mte_tagged(page))
275				continue;
276
277			ret = save_tags(page, pfn);
278			if (ret) {
279				swsusp_mte_free_storage();
280				goto out;
281			}
282
283			n++;
284		}
285	}
286	pr_info("Saved %d MTE pages\n", n);
287
288out:
289	return ret;
290}
291
292static void swsusp_mte_restore_tags(void)
293{
294	XA_STATE(xa_state, &mte_pages, 0);
295	int n = 0;
296	void *tags;
297
298	xa_lock(&mte_pages);
299	xas_for_each(&xa_state, tags, ULONG_MAX) {
300		unsigned long pfn = xa_state.xa_index;
301		struct page *page = pfn_to_online_page(pfn);
302
303		mte_restore_page_tags(page_address(page), tags);
304
305		mte_free_tag_storage(tags);
306		n++;
307	}
308	xa_unlock(&mte_pages);
309
310	pr_info("Restored %d MTE pages\n", n);
311
312	xa_destroy(&mte_pages);
313}
314
315#else	/* CONFIG_ARM64_MTE */
316
317static int swsusp_mte_save_tags(void)
318{
319	return 0;
320}
321
322static void swsusp_mte_restore_tags(void)
323{
324}
325
326#endif	/* CONFIG_ARM64_MTE */
327
328int swsusp_arch_suspend(void)
329{
330	int ret = 0;
331	unsigned long flags;
332	struct sleep_stack_data state;
333
334	if (cpus_are_stuck_in_kernel()) {
335		pr_err("Can't hibernate: no mechanism to offline secondary CPUs.\n");
336		return -EBUSY;
337	}
338
339	flags = local_daif_save();
340
341	if (__cpu_suspend_enter(&state)) {
342		/* make the crash dump kernel image visible/saveable */
343		crash_prepare_suspend();
344
345		ret = swsusp_mte_save_tags();
346		if (ret)
347			return ret;
348
349		sleep_cpu = smp_processor_id();
350		ret = swsusp_save();
351	} else {
352		/* Clean kernel core startup/idle code to PoC*/
353		dcache_clean_inval_poc((unsigned long)__mmuoff_data_start,
354				    (unsigned long)__mmuoff_data_end);
355		dcache_clean_inval_poc((unsigned long)__idmap_text_start,
356				    (unsigned long)__idmap_text_end);
357
358		/* Clean kvm setup code to PoC? */
359		if (el2_reset_needed()) {
360			dcache_clean_inval_poc(
361				(unsigned long)__hyp_idmap_text_start,
362				(unsigned long)__hyp_idmap_text_end);
363			dcache_clean_inval_poc((unsigned long)__hyp_text_start,
364					    (unsigned long)__hyp_text_end);
365		}
366
367		swsusp_mte_restore_tags();
368
369		/* make the crash dump kernel image protected again */
370		crash_post_resume();
371
372		/*
373		 * Tell the hibernation core that we've just restored
374		 * the memory
375		 */
376		in_suspend = 0;
377
378		sleep_cpu = -EINVAL;
379		__cpu_suspend_exit();
380
381		/*
382		 * Just in case the boot kernel did turn the SSBD
383		 * mitigation off behind our back, let's set the state
384		 * to what we expect it to be.
385		 */
386		spectre_v4_enable_mitigation(NULL);
387	}
388
389	local_daif_restore(flags);
390
391	return ret;
392}
393
394/*
395 * Setup then Resume from the hibernate image using swsusp_arch_suspend_exit().
396 *
397 * Memory allocated by get_safe_page() will be dealt with by the hibernate code,
398 * we don't need to free it here.
399 */
400int swsusp_arch_resume(void)
401{
402	int rc;
403	void *zero_page;
404	size_t exit_size;
405	pgd_t *tmp_pg_dir;
406	phys_addr_t el2_vectors;
407	void __noreturn (*hibernate_exit)(phys_addr_t, phys_addr_t, void *,
408					  void *, phys_addr_t, phys_addr_t);
409	struct trans_pgd_info trans_info = {
410		.trans_alloc_page	= hibernate_page_alloc,
411		.trans_alloc_arg	= (void *)GFP_ATOMIC,
412	};
413
414	/*
415	 * Restoring the memory image will overwrite the ttbr1 page tables.
416	 * Create a second copy of just the linear map, and use this when
417	 * restoring.
418	 */
419	rc = trans_pgd_create_copy(&trans_info, &tmp_pg_dir, PAGE_OFFSET,
420				   PAGE_END);
421	if (rc)
422		return rc;
423
424	/*
425	 * We need a zero page that is zero before & after resume in order
426	 * to break before make on the ttbr1 page tables.
427	 */
428	zero_page = (void *)get_safe_page(GFP_ATOMIC);
429	if (!zero_page) {
430		pr_err("Failed to allocate zero page.\n");
431		return -ENOMEM;
432	}
433
434	if (el2_reset_needed()) {
435		rc = trans_pgd_copy_el2_vectors(&trans_info, &el2_vectors);
436		if (rc) {
437			pr_err("Failed to setup el2 vectors\n");
438			return rc;
439		}
440	}
441
442	exit_size = __hibernate_exit_text_end - __hibernate_exit_text_start;
443	/*
444	 * Copy swsusp_arch_suspend_exit() to a safe page. This will generate
445	 * a new set of ttbr0 page tables and load them.
446	 */
447	rc = create_safe_exec_page(__hibernate_exit_text_start, exit_size,
448				   (phys_addr_t *)&hibernate_exit);
449	if (rc) {
450		pr_err("Failed to create safe executable page for hibernate_exit code.\n");
451		return rc;
452	}
453
454	/*
455	 * KASLR will cause the el2 vectors to be in a different location in
456	 * the resumed kernel. Load hibernate's temporary copy into el2.
457	 *
458	 * We can skip this step if we booted at EL1, or are running with VHE.
459	 */
460	if (el2_reset_needed())
461		__hyp_set_vectors(el2_vectors);
462
463	hibernate_exit(virt_to_phys(tmp_pg_dir), resume_hdr.ttbr1_el1,
464		       resume_hdr.reenter_kernel, restore_pblist,
465		       resume_hdr.__hyp_stub_vectors, virt_to_phys(zero_page));
466
467	return 0;
468}
469
470int hibernate_resume_nonboot_cpu_disable(void)
471{
472	if (sleep_cpu < 0) {
473		pr_err("Failing to resume from hibernate on an unknown CPU.\n");
474		return -ENODEV;
475	}
476
477	return freeze_secondary_cpus(sleep_cpu);
478}