[linux-2.6-block.git] / arch / s390 / kernel / vdso.c

// SPDX-License-Identifier: GPL-2.0
/*
 * vdso setup for s390
 *
 *  Copyright IBM Corp. 2008
 *  Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
 */

#include <linux/init.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/smp.h>
#include <linux/stddef.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/user.h>
#include <linux/elf.h>
#include <linux/security.h>
#include <linux/memblock.h>
#include <linux/compat.h>
#include <asm/asm-offsets.h>
#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/mmu.h>
#include <asm/mmu_context.h>
#include <asm/sections.h>
#include <asm/vdso.h>
#include <asm/facility.h>

#ifdef CONFIG_COMPAT_VDSO
extern char vdso32_start, vdso32_end;
static void *vdso32_kbase = &vdso32_start;
static unsigned int vdso32_pages;
static struct page **vdso32_pagelist;
#endif

extern char vdso64_start, vdso64_end;
static void *vdso64_kbase = &vdso64_start;
static unsigned int vdso64_pages;
static struct page **vdso64_pagelist;

/*
 * Should the kernel map a VDSO page into processes and pass its
 * address down to glibc upon exec()?
 */
unsigned int __read_mostly vdso_enabled = 1;

static vm_fault_t vdso_fault(const struct vm_special_mapping *sm,
		      struct vm_area_struct *vma, struct vm_fault *vmf)
{
	struct page **vdso_pagelist;
	unsigned long vdso_pages;

	vdso_pagelist = vdso64_pagelist;
	vdso_pages = vdso64_pages;
#ifdef CONFIG_COMPAT_VDSO
	if (vma->vm_mm->context.compat_mm) {
		vdso_pagelist = vdso32_pagelist;
		vdso_pages = vdso32_pages;
	}
#endif

	if (vmf->pgoff >= vdso_pages)
		return VM_FAULT_SIGBUS;

	vmf->page = vdso_pagelist[vmf->pgoff];
	get_page(vmf->page);
	return 0;
}

static int vdso_mremap(const struct vm_special_mapping *sm,
		       struct vm_area_struct *vma)
{
	unsigned long vdso_pages;

	vdso_pages = vdso64_pages;
#ifdef CONFIG_COMPAT_VDSO
	if (vma->vm_mm->context.compat_mm)
		vdso_pages = vdso32_pages;
#endif

	if ((vdso_pages << PAGE_SHIFT) != vma->vm_end - vma->vm_start)
		return -EINVAL;

	if (WARN_ON_ONCE(current->mm != vma->vm_mm))
		return -EFAULT;

	current->mm->context.vdso_base = vma->vm_start;
	return 0;
}

static const struct vm_special_mapping vdso_mapping = {
	.name = "[vdso]",
	.fault = vdso_fault,
	.mremap = vdso_mremap,
};

static int __init vdso_setup(char *s)
{
	unsigned long val;
	int rc;

	rc = 0;
	if (strncmp(s, "on", 3) == 0)
		vdso_enabled = 1;
	else if (strncmp(s, "off", 4) == 0)
		vdso_enabled = 0;
	else {
		rc = kstrtoul(s, 0, &val);
		vdso_enabled = rc ? 0 : !!val;
	}
	return !rc;
}
__setup("vdso=", vdso_setup);

/*
 * The vdso data page
 */
static union {
	struct vdso_data	data;
	u8			page[PAGE_SIZE];
} vdso_data_store __page_aligned_data;
struct vdso_data *vdso_data = &vdso_data_store.data;

/*
 * Setup vdso data page.
 */
static void __init vdso_init_data(struct vdso_data *vd)
{
	vd->ectg_available = test_facility(31);
}

/*
 * Allocate/free per cpu vdso data.
 */
#define SEGMENT_ORDER	2

/*
 * The initial vdso_data structure for the boot CPU. Eventually
 * it is replaced with a properly allocated structure in vdso_init.
 * This is necessary because a valid S390_lowcore.vdso_per_cpu_data
 * pointer is required to be able to return from an interrupt or
 * program check. See the exit paths in entry.S.
 */
struct vdso_data boot_vdso_data __initdata;

void __init vdso_alloc_boot_cpu(struct lowcore *lowcore)
{
	lowcore->vdso_per_cpu_data = (unsigned long) &boot_vdso_data;
}

int vdso_alloc_per_cpu(struct lowcore *lowcore)
{
	unsigned long segment_table, page_table, page_frame;
	struct vdso_per_cpu_data *vd;

	segment_table = __get_free_pages(GFP_KERNEL, SEGMENT_ORDER);
	page_table = get_zeroed_page(GFP_KERNEL);
	page_frame = get_zeroed_page(GFP_KERNEL);
	if (!segment_table || !page_table || !page_frame)
		goto out;
	arch_set_page_dat(virt_to_page(segment_table), SEGMENT_ORDER);
	arch_set_page_dat(virt_to_page(page_table), 0);

	/* Initialize per-cpu vdso data page */
	vd = (struct vdso_per_cpu_data *) page_frame;
	vd->cpu_nr = lowcore->cpu_nr;
	vd->node_id = cpu_to_node(vd->cpu_nr);

	/* Set up page table for the vdso address space */
	memset64((u64 *)segment_table, _SEGMENT_ENTRY_EMPTY, _CRST_ENTRIES);
	memset64((u64 *)page_table, _PAGE_INVALID, PTRS_PER_PTE);

	*(unsigned long *) segment_table = _SEGMENT_ENTRY + page_table;
	*(unsigned long *) page_table = _PAGE_PROTECT + page_frame;

	lowcore->vdso_asce = segment_table +
		_ASCE_TABLE_LENGTH + _ASCE_USER_BITS + _ASCE_TYPE_SEGMENT;
	lowcore->vdso_per_cpu_data = page_frame;

	return 0;

out:
	free_page(page_frame);
	free_page(page_table);
	free_pages(segment_table, SEGMENT_ORDER);
	return -ENOMEM;
}

void vdso_free_per_cpu(struct lowcore *lowcore)
{
	unsigned long segment_table, page_table, page_frame;

	segment_table = lowcore->vdso_asce & PAGE_MASK;
	page_table = *(unsigned long *) segment_table;
	page_frame = *(unsigned long *) page_table;

	free_page(page_frame);
	free_page(page_table);
	free_pages(segment_table, SEGMENT_ORDER);
}

/*
 * This is called from binfmt_elf, we create the special vma for the
 * vDSO and insert it into the mm struct tree
 */
int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long vdso_pages;
	unsigned long vdso_base;
	int rc;

	if (!vdso_enabled)
		return 0;

	vdso_pages = vdso64_pages;
#ifdef CONFIG_COMPAT_VDSO
	mm->context.compat_mm = is_compat_task();
	if (mm->context.compat_mm)
		vdso_pages = vdso32_pages;
#endif
	/*
	 * vDSO has a problem and was disabled, just don't "enable" it for
	 * the process
	 */
	if (vdso_pages == 0)
		return 0;

	/*
	 * pick a base address for the vDSO in process space. We try to put
	 * it at vdso_base which is the "natural" base for it, but we might
	 * fail and end up putting it elsewhere.
	 */
	if (down_write_killable(&mm->mmap_sem))
		return -EINTR;
	vdso_base = get_unmapped_area(NULL, 0, vdso_pages << PAGE_SHIFT, 0, 0);
	if (IS_ERR_VALUE(vdso_base)) {
		rc = vdso_base;
		goto out_up;
	}

	/*
	 * our vma flags don't have VM_WRITE so by default, the process
	 * isn't allowed to write those pages.
	 * gdb can break that with ptrace interface, and thus trigger COW
	 * on those pages but it's then your responsibility to never do that
	 * on the "data" page of the vDSO or you'll stop getting kernel
	 * updates and your nice userland gettimeofday will be totally dead.
	 * It's fine to use that for setting breakpoints in the vDSO code
	 * pages though.
	 */
	vma = _install_special_mapping(mm, vdso_base, vdso_pages << PAGE_SHIFT,
				       VM_READ|VM_EXEC|
				       VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,
				       &vdso_mapping);
	if (IS_ERR(vma)) {
		rc = PTR_ERR(vma);
		goto out_up;
	}

	current->mm->context.vdso_base = vdso_base;
	rc = 0;

out_up:
	up_write(&mm->mmap_sem);
	return rc;
}

static int __init vdso_init(void)
{
	int i;

	vdso_init_data(vdso_data);
#ifdef CONFIG_COMPAT_VDSO
	/* Calculate the size of the 32 bit vDSO */
	vdso32_pages = ((&vdso32_end - &vdso32_start
			 + PAGE_SIZE - 1) >> PAGE_SHIFT) + 1;

	/* Make sure pages are in the correct state */
	vdso32_pagelist = kcalloc(vdso32_pages + 1, sizeof(struct page *),
				  GFP_KERNEL);
	BUG_ON(vdso32_pagelist == NULL);
	for (i = 0; i < vdso32_pages - 1; i++) {
		struct page *pg = virt_to_page(vdso32_kbase + i*PAGE_SIZE);
		get_page(pg);
		vdso32_pagelist[i] = pg;
	}
	vdso32_pagelist[vdso32_pages - 1] = virt_to_page(vdso_data);
	vdso32_pagelist[vdso32_pages] = NULL;
#endif

	/* Calculate the size of the 64 bit vDSO */
	vdso64_pages = ((&vdso64_end - &vdso64_start
			 + PAGE_SIZE - 1) >> PAGE_SHIFT) + 1;

	/* Make sure pages are in the correct state */
	vdso64_pagelist = kcalloc(vdso64_pages + 1, sizeof(struct page *),
				  GFP_KERNEL);
	BUG_ON(vdso64_pagelist == NULL);
	for (i = 0; i < vdso64_pages - 1; i++) {
		struct page *pg = virt_to_page(vdso64_kbase + i*PAGE_SIZE);
		get_page(pg);
		vdso64_pagelist[i] = pg;
	}
	vdso64_pagelist[vdso64_pages - 1] = virt_to_page(vdso_data);
	vdso64_pagelist[vdso64_pages] = NULL;
	if (vdso_alloc_per_cpu(&S390_lowcore))
		BUG();

	get_page(virt_to_page(vdso_data));

	return 0;
}
early_initcall(vdso_init);
Commit	Line	Data
a17ae4c3	1	// SPDX-License-Identifier: GPL-2.0
b020632e MS	2	/*
	3	* vdso setup for s390
	4	*
	5	* Copyright IBM Corp. 2008
	6	* Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
b020632e MS	7	*/
b020632e MS	8
3994a52b	9	#include <linux/init.h>
b020632e MS	10	#include <linux/errno.h>
	11	#include <linux/sched.h>
	12	#include <linux/kernel.h>
	13	#include <linux/mm.h>
	14	#include <linux/smp.h>
	15	#include <linux/stddef.h>
	16	#include <linux/unistd.h>
	17	#include <linux/slab.h>
	18	#include <linux/user.h>
	19	#include <linux/elf.h>
	20	#include <linux/security.h>
57c8a661	21	#include <linux/memblock.h>
7757591a	22	#include <linux/compat.h>
cbb870c8	23	#include <asm/asm-offsets.h>
b020632e	24	#include <asm/pgtable.h>
b020632e MS	25	#include <asm/processor.h>
	26	#include <asm/mmu.h>
	27	#include <asm/mmu_context.h>
	28	#include <asm/sections.h>
	29	#include <asm/vdso.h>
a0616cde	30	#include <asm/facility.h>
b020632e	31
96ca7674	32	#ifdef CONFIG_COMPAT_VDSO
b020632e MS	33	extern char vdso32_start, vdso32_end;
	34	static void *vdso32_kbase = &vdso32_start;
	35	static unsigned int vdso32_pages;
	36	static struct page **vdso32_pagelist;
	37	#endif
	38
b020632e MS	39	extern char vdso64_start, vdso64_end;
	40	static void *vdso64_kbase = &vdso64_start;
	41	static unsigned int vdso64_pages;
	42	static struct page **vdso64_pagelist;
b020632e MS	43
	44	/*
	45	* Should the kernel map a VDSO page into processes and pass its
	46	* address down to glibc upon exec()?
	47	*/
	48	unsigned int __read_mostly vdso_enabled = 1;
	49
ef4b891f	50	static vm_fault_t vdso_fault(const struct vm_special_mapping *sm,
35bb092a MS	51	struct vm_area_struct vma, struct vm_fault vmf)
	52	{
	53	struct page **vdso_pagelist;
	54	unsigned long vdso_pages;
	55
	56	vdso_pagelist = vdso64_pagelist;
	57	vdso_pages = vdso64_pages;
96ca7674	58	#ifdef CONFIG_COMPAT_VDSO
d1befa65	59	if (vma->vm_mm->context.compat_mm) {
35bb092a MS	60	vdso_pagelist = vdso32_pagelist;
	61	vdso_pages = vdso32_pages;
	62	}
	63	#endif
	64
	65	if (vmf->pgoff >= vdso_pages)
	66	return VM_FAULT_SIGBUS;
	67
	68	vmf->page = vdso_pagelist[vmf->pgoff];
	69	get_page(vmf->page);
	70	return 0;
	71	}
	72
	73	static int vdso_mremap(const struct vm_special_mapping *sm,
	74	struct vm_area_struct *vma)
	75	{
	76	unsigned long vdso_pages;
	77
	78	vdso_pages = vdso64_pages;
96ca7674	79	#ifdef CONFIG_COMPAT_VDSO
d1befa65	80	if (vma->vm_mm->context.compat_mm)
35bb092a MS	81	vdso_pages = vdso32_pages;
	82	#endif
	83
	84	if ((vdso_pages << PAGE_SHIFT) != vma->vm_end - vma->vm_start)
	85	return -EINVAL;
	86
	87	if (WARN_ON_ONCE(current->mm != vma->vm_mm))
	88	return -EFAULT;
	89
	90	current->mm->context.vdso_base = vma->vm_start;
	91	return 0;
	92	}
	93
	94	static const struct vm_special_mapping vdso_mapping = {
	95	.name = "[vdso]",
	96	.fault = vdso_fault,
	97	.mremap = vdso_mremap,
	98	};
	99
b020632e MS	100	static int __init vdso_setup(char *s)
b020632e MS	101	{
7aa79f94 MS	102	unsigned long val;
	103	int rc;
	104
	105	rc = 0;
	106	if (strncmp(s, "on", 3) == 0)
	107	vdso_enabled = 1;
	108	else if (strncmp(s, "off", 4) == 0)
	109	vdso_enabled = 0;
	110	else {
958d9072	111	rc = kstrtoul(s, 0, &val);
7aa79f94 MS	112	vdso_enabled = rc ? 0 : !!val;
	113	}
	114	return !rc;
b020632e MS	115	}
	116	__setup("vdso=", vdso_setup);
	117
	118	/*
	119	* The vdso data page
	120	*/
	121	static union {
	122	struct vdso_data data;
	123	u8 page[PAGE_SIZE];
abe1ee3a	124	} vdso_data_store __page_aligned_data;
b020632e MS	125	struct vdso_data *vdso_data = &vdso_data_store.data;
b020632e MS	126
c742b31c MS	127	/*
	128	* Setup vdso data page.
	129	*/
249c543b	130	static void __init vdso_init_data(struct vdso_data *vd)
c742b31c	131	{
e258d719	132	vd->ectg_available = test_facility(31);
c742b31c MS	133	}
c742b31c MS	134
c742b31c MS	135	/*
	136	* Allocate/free per cpu vdso data.
	137	*/
c742b31c	138	#define SEGMENT_ORDER 2
c742b31c	139
608796ff MS	140	/*
	141	* The initial vdso_data structure for the boot CPU. Eventually
	142	* it is replaced with a properly allocated structure in vdso_init.
	143	* This is necessary because a valid S390_lowcore.vdso_per_cpu_data
	144	* pointer is required to be able to return from an interrupt or
	145	* program check. See the exit paths in entry.S.
	146	*/
	147	struct vdso_data boot_vdso_data __initdata;
	148
	149	void __init vdso_alloc_boot_cpu(struct lowcore *lowcore)
	150	{
	151	lowcore->vdso_per_cpu_data = (unsigned long) &boot_vdso_data;
	152	}
	153
c667aeac	154	int vdso_alloc_per_cpu(struct lowcore *lowcore)
c742b31c MS	155	{
c742b31c MS	156	unsigned long segment_table, page_table, page_frame;
249c543b	157	struct vdso_per_cpu_data *vd;
c742b31c MS	158
c742b31c MS	159	segment_table = __get_free_pages(GFP_KERNEL, SEGMENT_ORDER);
0aaba41b	160	page_table = get_zeroed_page(GFP_KERNEL);
c742b31c MS	161	page_frame = get_zeroed_page(GFP_KERNEL);
	162	if (!segment_table \|\| !page_table \|\| !page_frame)
	163	goto out;
c9b5ad54 MS	164	arch_set_page_dat(virt_to_page(segment_table), SEGMENT_ORDER);
c9b5ad54 MS	165	arch_set_page_dat(virt_to_page(page_table), 0);
c742b31c	166
249c543b MS	167	/* Initialize per-cpu vdso data page */
	168	vd = (struct vdso_per_cpu_data *) page_frame;
	169	vd->cpu_nr = lowcore->cpu_nr;
	170	vd->node_id = cpu_to_node(vd->cpu_nr);
	171
0aaba41b	172	/* Set up page table for the vdso address space */
41879ff6 HC	173	memset64((u64 *)segment_table, _SEGMENT_ENTRY_EMPTY, _CRST_ENTRIES);
41879ff6 HC	174	memset64((u64 *)page_table, _PAGE_INVALID, PTRS_PER_PTE);
c742b31c MS	175
c742b31c MS	176	(unsigned long ) segment_table = _SEGMENT_ENTRY + page_table;
e5098611	177	(unsigned long ) page_table = _PAGE_PROTECT + page_frame;
c742b31c	178
0aaba41b	179	lowcore->vdso_asce = segment_table +
c742b31c	180	_ASCE_TABLE_LENGTH + _ASCE_USER_BITS + _ASCE_TYPE_SEGMENT;
c742b31c MS	181	lowcore->vdso_per_cpu_data = page_frame;
c742b31c MS	182
c742b31c MS	183	return 0;
	184
	185	out:
	186	free_page(page_frame);
	187	free_page(page_table);
	188	free_pages(segment_table, SEGMENT_ORDER);
	189	return -ENOMEM;
	190	}
	191
c667aeac	192	void vdso_free_per_cpu(struct lowcore *lowcore)
c742b31c MS	193	{
c742b31c MS	194	unsigned long segment_table, page_table, page_frame;
c742b31c	195
0aaba41b	196	segment_table = lowcore->vdso_asce & PAGE_MASK;
c742b31c MS	197	page_table = (unsigned long ) segment_table;
	198	page_frame = (unsigned long ) page_table;
	199
	200	free_page(page_frame);
	201	free_page(page_table);
	202	free_pages(segment_table, SEGMENT_ORDER);
	203	}
c742b31c	204
b020632e MS	205	/*
	206	* This is called from binfmt_elf, we create the special vma for the
	207	* vDSO and insert it into the mm struct tree
	208	*/
	209	int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
	210	{
	211	struct mm_struct *mm = current->mm;
35bb092a	212	struct vm_area_struct *vma;
b020632e MS	213	unsigned long vdso_pages;
	214	unsigned long vdso_base;
	215	int rc;
	216
	217	if (!vdso_enabled)
	218	return 0;
b020632e	219
b020632e	220	vdso_pages = vdso64_pages;
96ca7674	221	#ifdef CONFIG_COMPAT_VDSO
190f056f VG	222	mm->context.compat_mm = is_compat_task();
190f056f VG	223	if (mm->context.compat_mm)
b020632e	224	vdso_pages = vdso32_pages;
b020632e	225	#endif
b020632e MS	226	/*
	227	* vDSO has a problem and was disabled, just don't "enable" it for
	228	* the process
	229	*/
	230	if (vdso_pages == 0)
	231	return 0;
	232
b020632e MS	233	/*
	234	* pick a base address for the vDSO in process space. We try to put
	235	* it at vdso_base which is the "natural" base for it, but we might
	236	* fail and end up putting it elsewhere.
	237	*/
69048176 MH	238	if (down_write_killable(&mm->mmap_sem))
69048176 MH	239	return -EINTR;
e7828bbd	240	vdso_base = get_unmapped_area(NULL, 0, vdso_pages << PAGE_SHIFT, 0, 0);
b020632e MS	241	if (IS_ERR_VALUE(vdso_base)) {
	242	rc = vdso_base;
	243	goto out_up;
	244	}
	245
	246	/*
	247	* our vma flags don't have VM_WRITE so by default, the process
	248	* isn't allowed to write those pages.
	249	* gdb can break that with ptrace interface, and thus trigger COW
	250	* on those pages but it's then your responsibility to never do that
	251	* on the "data" page of the vDSO or you'll stop getting kernel
	252	* updates and your nice userland gettimeofday will be totally dead.
	253	* It's fine to use that for setting breakpoints in the vDSO code
909af768	254	* pages though.
b020632e	255	*/
35bb092a MS	256	vma = _install_special_mapping(mm, vdso_base, vdso_pages << PAGE_SHIFT,
	257	VM_READ\|VM_EXEC\|
	258	VM_MAYREAD\|VM_MAYWRITE\|VM_MAYEXEC,
	259	&vdso_mapping);
	260	if (IS_ERR(vma)) {
	261	rc = PTR_ERR(vma);
	262	goto out_up;
	263	}
	264
	265	current->mm->context.vdso_base = vdso_base;
	266	rc = 0;
	267
b020632e MS	268	out_up:
	269	up_write(&mm->mmap_sem);
	270	return rc;
	271	}
	272
b020632e MS	273	static int __init vdso_init(void)
	274	{
	275	int i;
	276
c742b31c	277	vdso_init_data(vdso_data);
96ca7674	278	#ifdef CONFIG_COMPAT_VDSO
b020632e MS	279	/* Calculate the size of the 32 bit vDSO */
	280	vdso32_pages = ((&vdso32_end - &vdso32_start
	281	+ PAGE_SIZE - 1) >> PAGE_SHIFT) + 1;
	282
	283	/* Make sure pages are in the correct state */
6396bb22	284	vdso32_pagelist = kcalloc(vdso32_pages + 1, sizeof(struct page *),
b020632e MS	285	GFP_KERNEL);
	286	BUG_ON(vdso32_pagelist == NULL);
	287	for (i = 0; i < vdso32_pages - 1; i++) {
	288	struct page pg = virt_to_page(vdso32_kbase + iPAGE_SIZE);
b020632e MS	289	get_page(pg);
	290	vdso32_pagelist[i] = pg;
	291	}
	292	vdso32_pagelist[vdso32_pages - 1] = virt_to_page(vdso_data);
	293	vdso32_pagelist[vdso32_pages] = NULL;
	294	#endif
	295
b020632e MS	296	/* Calculate the size of the 64 bit vDSO */
	297	vdso64_pages = ((&vdso64_end - &vdso64_start
	298	+ PAGE_SIZE - 1) >> PAGE_SHIFT) + 1;
	299
	300	/* Make sure pages are in the correct state */
6396bb22	301	vdso64_pagelist = kcalloc(vdso64_pages + 1, sizeof(struct page *),
b020632e MS	302	GFP_KERNEL);
	303	BUG_ON(vdso64_pagelist == NULL);
	304	for (i = 0; i < vdso64_pages - 1; i++) {
	305	struct page pg = virt_to_page(vdso64_kbase + iPAGE_SIZE);
b020632e MS	306	get_page(pg);
	307	vdso64_pagelist[i] = pg;
	308	}
	309	vdso64_pagelist[vdso64_pages - 1] = virt_to_page(vdso_data);
	310	vdso64_pagelist[vdso64_pages] = NULL;
8b646bd7	311	if (vdso_alloc_per_cpu(&S390_lowcore))
81ffa041	312	BUG();
b020632e MS	313
	314	get_page(virt_to_page(vdso_data));
	315
b020632e MS	316	return 0;
b020632e MS	317	}
8b646bd7	318	early_initcall(vdso_init);