[deliverable/linux.git] / arch / powerpc / kvm / book3s_64_mmu_hv.c

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License, version 2, as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
 *
 * Copyright 2010 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
 */

#include <linux/types.h>
#include <linux/string.h>
#include <linux/kvm.h>
#include <linux/kvm_host.h>
#include <linux/highmem.h>
#include <linux/gfp.h>
#include <linux/slab.h>
#include <linux/hugetlb.h>

#include <asm/tlbflush.h>
#include <asm/kvm_ppc.h>
#include <asm/kvm_book3s.h>
#include <asm/mmu-hash64.h>
#include <asm/hvcall.h>
#include <asm/synch.h>
#include <asm/ppc-opcode.h>
#include <asm/cputable.h>

/* For now use fixed-size 16MB page table */
#define HPT_ORDER	24
#define HPT_NPTEG	(1ul << (HPT_ORDER - 7))	/* 128B per pteg */
#define HPT_HASH_MASK	(HPT_NPTEG - 1)

/* Pages in the VRMA are 16MB pages */
#define VRMA_PAGE_ORDER	24
#define VRMA_VSID	0x1ffffffUL	/* 1TB VSID reserved for VRMA */

#define NR_LPIDS	(LPID_RSVD + 1)
unsigned long lpid_inuse[BITS_TO_LONGS(NR_LPIDS)];

long kvmppc_alloc_hpt(struct kvm *kvm)
{
	unsigned long hpt;
	unsigned long lpid;

	hpt = __get_free_pages(GFP_KERNEL|__GFP_ZERO|__GFP_REPEAT|__GFP_NOWARN,
			       HPT_ORDER - PAGE_SHIFT);
	if (!hpt) {
		pr_err("kvm_alloc_hpt: Couldn't alloc HPT\n");
		return -ENOMEM;
	}
	kvm->arch.hpt_virt = hpt;

	do {
		lpid = find_first_zero_bit(lpid_inuse, NR_LPIDS);
		if (lpid >= NR_LPIDS) {
			pr_err("kvm_alloc_hpt: No LPIDs free\n");
			free_pages(hpt, HPT_ORDER - PAGE_SHIFT);
			return -ENOMEM;
		}
	} while (test_and_set_bit(lpid, lpid_inuse));

	kvm->arch.sdr1 = __pa(hpt) | (HPT_ORDER - 18);
	kvm->arch.lpid = lpid;
	kvm->arch.host_sdr1 = mfspr(SPRN_SDR1);
	kvm->arch.host_lpid = mfspr(SPRN_LPID);
	kvm->arch.host_lpcr = mfspr(SPRN_LPCR);

	pr_info("KVM guest htab at %lx, LPID %lx\n", hpt, lpid);
	return 0;
}

void kvmppc_free_hpt(struct kvm *kvm)
{
	unsigned long i;
	struct kvmppc_pginfo *pginfo;

	clear_bit(kvm->arch.lpid, lpid_inuse);
	free_pages(kvm->arch.hpt_virt, HPT_ORDER - PAGE_SHIFT);

	if (kvm->arch.ram_pginfo) {
		pginfo = kvm->arch.ram_pginfo;
		kvm->arch.ram_pginfo = NULL;
		for (i = 0; i < kvm->arch.ram_npages; ++i)
			put_page(pfn_to_page(pginfo[i].pfn));
		kfree(pginfo);
	}
}

static unsigned long user_page_size(unsigned long addr)
{
	struct vm_area_struct *vma;
	unsigned long size = PAGE_SIZE;

	down_read(&current->mm->mmap_sem);
	vma = find_vma(current->mm, addr);
	if (vma)
		size = vma_kernel_pagesize(vma);
	up_read(&current->mm->mmap_sem);
	return size;
}

static pfn_t hva_to_pfn(unsigned long addr)
{
	struct page *page[1];
	int npages;

	might_sleep();

	npages = get_user_pages_fast(addr, 1, 1, page);

	if (unlikely(npages != 1))
		return 0;

	return page_to_pfn(page[0]);
}

long kvmppc_prepare_vrma(struct kvm *kvm,
			 struct kvm_userspace_memory_region *mem)
{
	unsigned long psize, porder;
	unsigned long i, npages;
	struct kvmppc_pginfo *pginfo;
	pfn_t pfn;
	unsigned long hva;

	/* First see what page size we have */
	psize = user_page_size(mem->userspace_addr);
	/* For now, only allow 16MB pages */
	if (psize != 1ul << VRMA_PAGE_ORDER || (mem->memory_size & (psize - 1))) {
		pr_err("bad psize=%lx memory_size=%llx @ %llx\n",
		       psize, mem->memory_size, mem->userspace_addr);
		return -EINVAL;
	}
	porder = __ilog2(psize);

	npages = mem->memory_size >> porder;
	pginfo = kzalloc(npages * sizeof(struct kvmppc_pginfo), GFP_KERNEL);
	if (!pginfo) {
		pr_err("kvmppc_prepare_vrma: couldn't alloc %lu bytes\n",
		       npages * sizeof(struct kvmppc_pginfo));
		return -ENOMEM;
	}

	for (i = 0; i < npages; ++i) {
		hva = mem->userspace_addr + (i << porder);
		if (user_page_size(hva) != psize)
			goto err;
		pfn = hva_to_pfn(hva);
		if (pfn == 0) {
			pr_err("oops, no pfn for hva %lx\n", hva);
			goto err;
		}
		if (pfn & ((1ul << (porder - PAGE_SHIFT)) - 1)) {
			pr_err("oops, unaligned pfn %llx\n", pfn);
			put_page(pfn_to_page(pfn));
			goto err;
		}
		pginfo[i].pfn = pfn;
	}

	kvm->arch.ram_npages = npages;
	kvm->arch.ram_psize = psize;
	kvm->arch.ram_porder = porder;
	kvm->arch.ram_pginfo = pginfo;

	return 0;

 err:
	kfree(pginfo);
	return -EINVAL;
}

void kvmppc_map_vrma(struct kvm *kvm, struct kvm_userspace_memory_region *mem)
{
	unsigned long i;
	unsigned long npages = kvm->arch.ram_npages;
	unsigned long pfn;
	unsigned long *hpte;
	unsigned long hash;
	struct kvmppc_pginfo *pginfo = kvm->arch.ram_pginfo;

	if (!pginfo)
		return;

	/* VRMA can't be > 1TB */
	if (npages > 1ul << (40 - kvm->arch.ram_porder))
		npages = 1ul << (40 - kvm->arch.ram_porder);
	/* Can't use more than 1 HPTE per HPTEG */
	if (npages > HPT_NPTEG)
		npages = HPT_NPTEG;

	for (i = 0; i < npages; ++i) {
		pfn = pginfo[i].pfn;
		/* can't use hpt_hash since va > 64 bits */
		hash = (i ^ (VRMA_VSID ^ (VRMA_VSID << 25))) & HPT_HASH_MASK;
		/*
		 * We assume that the hash table is empty and no
		 * vcpus are using it at this stage.  Since we create
		 * at most one HPTE per HPTEG, we just assume entry 7
		 * is available and use it.
		 */
		hpte = (unsigned long *) (kvm->arch.hpt_virt + (hash << 7));
		hpte += 7 * 2;
		/* HPTE low word - RPN, protection, etc. */
		hpte[1] = (pfn << PAGE_SHIFT) | HPTE_R_R | HPTE_R_C |
			HPTE_R_M | PP_RWXX;
		wmb();
		hpte[0] = HPTE_V_1TB_SEG | (VRMA_VSID << (40 - 16)) |
			(i << (VRMA_PAGE_ORDER - 16)) | HPTE_V_BOLTED |
			HPTE_V_LARGE | HPTE_V_VALID;
	}
}

int kvmppc_mmu_hv_init(void)
{
	if (!cpu_has_feature(CPU_FTR_HVMODE_206))
		return -EINVAL;
	memset(lpid_inuse, 0, sizeof(lpid_inuse));
	set_bit(mfspr(SPRN_LPID), lpid_inuse);
	set_bit(LPID_RSVD, lpid_inuse);

	return 0;
}

void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
{
}

static void kvmppc_mmu_book3s_64_hv_reset_msr(struct kvm_vcpu *vcpu)
{
	kvmppc_set_msr(vcpu, MSR_SF | MSR_ME);
}

static int kvmppc_mmu_book3s_64_hv_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
				struct kvmppc_pte *gpte, bool data)
{
	return -ENOENT;
}

void kvmppc_mmu_book3s_hv_init(struct kvm_vcpu *vcpu)
{
	struct kvmppc_mmu *mmu = &vcpu->arch.mmu;

	vcpu->arch.slb_nr = 32;		/* Assume POWER7 for now */

	mmu->xlate = kvmppc_mmu_book3s_64_hv_xlate;
	mmu->reset_msr = kvmppc_mmu_book3s_64_hv_reset_msr;

	vcpu->arch.hflags |= BOOK3S_HFLAG_SLB;
}
Commit	Line	Data
de56a948 PM	1	/*
	2	* This program is free software; you can redistribute it and/or modify
	3	* it under the terms of the GNU General Public License, version 2, as
	4	* published by the Free Software Foundation.
	5	*
	6	* This program is distributed in the hope that it will be useful,
	7	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	8	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	9	* GNU General Public License for more details.
	10	*
	11	* You should have received a copy of the GNU General Public License
	12	* along with this program; if not, write to the Free Software
	13	* Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
	14	*
	15	* Copyright 2010 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
	16	*/
	17
	18	#include <linux/types.h>
	19	#include <linux/string.h>
	20	#include <linux/kvm.h>
	21	#include <linux/kvm_host.h>
	22	#include <linux/highmem.h>
	23	#include <linux/gfp.h>
	24	#include <linux/slab.h>
	25	#include <linux/hugetlb.h>
	26
	27	#include <asm/tlbflush.h>
	28	#include <asm/kvm_ppc.h>
	29	#include <asm/kvm_book3s.h>
	30	#include <asm/mmu-hash64.h>
	31	#include <asm/hvcall.h>
	32	#include <asm/synch.h>
	33	#include <asm/ppc-opcode.h>
	34	#include <asm/cputable.h>
	35
	36	/* For now use fixed-size 16MB page table */
	37	#define HPT_ORDER 24
	38	#define HPT_NPTEG (1ul << (HPT_ORDER - 7)) /* 128B per pteg */
	39	#define HPT_HASH_MASK (HPT_NPTEG - 1)
	40
	41	/* Pages in the VRMA are 16MB pages */
	42	#define VRMA_PAGE_ORDER 24
	43	#define VRMA_VSID 0x1ffffffUL /* 1TB VSID reserved for VRMA */
	44
	45	#define NR_LPIDS (LPID_RSVD + 1)
	46	unsigned long lpid_inuse[BITS_TO_LONGS(NR_LPIDS)];
	47
	48	long kvmppc_alloc_hpt(struct kvm *kvm)
	49	{
	50	unsigned long hpt;
	51	unsigned long lpid;
	52
	53	hpt = __get_free_pages(GFP_KERNEL\|__GFP_ZERO\|__GFP_REPEAT\|__GFP_NOWARN,
	54	HPT_ORDER - PAGE_SHIFT);
	55	if (!hpt) {
	56	pr_err("kvm_alloc_hpt: Couldn't alloc HPT\n");
	57	return -ENOMEM;
	58	}
	59	kvm->arch.hpt_virt = hpt;
	60
	61	do {
	62	lpid = find_first_zero_bit(lpid_inuse, NR_LPIDS);
	63	if (lpid >= NR_LPIDS) {
	64	pr_err("kvm_alloc_hpt: No LPIDs free\n");
65	free_pages(hpt, HPT_ORDER - PAGE_SHIFT);
66	return -ENOMEM;
67	}
68	} while (test_and_set_bit(lpid, lpid_inuse));
69
70	kvm->arch.sdr1 = __pa(hpt) \| (HPT_ORDER - 18);
71	kvm->arch.lpid = lpid;
72	kvm->arch.host_sdr1 = mfspr(SPRN_SDR1);
73	kvm->arch.host_lpid = mfspr(SPRN_LPID);
74	kvm->arch.host_lpcr = mfspr(SPRN_LPCR);
75
76	pr_info("KVM guest htab at %lx, LPID %lx\n", hpt, lpid);
77	return 0;
78	}
79
80	void kvmppc_free_hpt(struct kvm *kvm)
81	{
82	unsigned long i;
83	struct kvmppc_pginfo *pginfo;
84
85	clear_bit(kvm->arch.lpid, lpid_inuse);
86	free_pages(kvm->arch.hpt_virt, HPT_ORDER - PAGE_SHIFT);
87
88	if (kvm->arch.ram_pginfo) {
89	pginfo = kvm->arch.ram_pginfo;
90	kvm->arch.ram_pginfo = NULL;
91	for (i = 0; i < kvm->arch.ram_npages; ++i)
92	put_page(pfn_to_page(pginfo[i].pfn));
93	kfree(pginfo);
94	}
95	}
96
97	static unsigned long user_page_size(unsigned long addr)
98	{
99	struct vm_area_struct *vma;
100	unsigned long size = PAGE_SIZE;
101
102	down_read(&current->mm->mmap_sem);
103	vma = find_vma(current->mm, addr);
104	if (vma)
105	size = vma_kernel_pagesize(vma);
106	up_read(&current->mm->mmap_sem);
107	return size;
108	}
109
110	static pfn_t hva_to_pfn(unsigned long addr)
111	{
112	struct page *page[1];
113	int npages;
114
115	might_sleep();
116
117	npages = get_user_pages_fast(addr, 1, 1, page);
118
119	if (unlikely(npages != 1))
120	return 0;
121
122	return page_to_pfn(page[0]);
123	}
124
125	long kvmppc_prepare_vrma(struct kvm *kvm,
126	struct kvm_userspace_memory_region *mem)
127	{
128	unsigned long psize, porder;
129	unsigned long i, npages;
130	struct kvmppc_pginfo *pginfo;
131	pfn_t pfn;
132	unsigned long hva;
133
134	/* First see what page size we have */
135	psize = user_page_size(mem->userspace_addr);
136	/* For now, only allow 16MB pages */
137	if (psize != 1ul << VRMA_PAGE_ORDER \|\| (mem->memory_size & (psize - 1))) {
138	pr_err("bad psize=%lx memory_size=%llx @ %llx\n",
139	psize, mem->memory_size, mem->userspace_addr);
140	return -EINVAL;
141	}
142	porder = __ilog2(psize);
143
144	npages = mem->memory_size >> porder;
145	pginfo = kzalloc(npages * sizeof(struct kvmppc_pginfo), GFP_KERNEL);
146	if (!pginfo) {
147	pr_err("kvmppc_prepare_vrma: couldn't alloc %lu bytes\n",
148	npages * sizeof(struct kvmppc_pginfo));
149	return -ENOMEM;
150	}
151
152	for (i = 0; i < npages; ++i) {
153	hva = mem->userspace_addr + (i << porder);
154	if (user_page_size(hva) != psize)
155	goto err;
156	pfn = hva_to_pfn(hva);
157	if (pfn == 0) {
158	pr_err("oops, no pfn for hva %lx\n", hva);
159	goto err;
160	}
161	if (pfn & ((1ul << (porder - PAGE_SHIFT)) - 1)) {
162	pr_err("oops, unaligned pfn %llx\n", pfn);
163	put_page(pfn_to_page(pfn));
164	goto err;
165	}
166	pginfo[i].pfn = pfn;
167	}
168
169	kvm->arch.ram_npages = npages;
170	kvm->arch.ram_psize = psize;
171	kvm->arch.ram_porder = porder;
172	kvm->arch.ram_pginfo = pginfo;
173
174	return 0;
175
176	err:
177	kfree(pginfo);
178	return -EINVAL;
179	}
180
181	void kvmppc_map_vrma(struct kvm kvm, struct kvm_userspace_memory_region mem)
182	{
183	unsigned long i;
184	unsigned long npages = kvm->arch.ram_npages;
185	unsigned long pfn;
186	unsigned long *hpte;
187	unsigned long hash;
188	struct kvmppc_pginfo *pginfo = kvm->arch.ram_pginfo;
189
190	if (!pginfo)
191	return;
192
193	/* VRMA can't be > 1TB */
194	if (npages > 1ul << (40 - kvm->arch.ram_porder))
195	npages = 1ul << (40 - kvm->arch.ram_porder);
196	/* Can't use more than 1 HPTE per HPTEG */
197	if (npages > HPT_NPTEG)
198	npages = HPT_NPTEG;
199
200	for (i = 0; i < npages; ++i) {
201	pfn = pginfo[i].pfn;
202	/* can't use hpt_hash since va > 64 bits */
203	hash = (i ^ (VRMA_VSID ^ (VRMA_VSID << 25))) & HPT_HASH_MASK;
204	/*
205	* We assume that the hash table is empty and no
206	* vcpus are using it at this stage. Since we create
207	* at most one HPTE per HPTEG, we just assume entry 7
208	* is available and use it.
209	*/
210	hpte = (unsigned long *) (kvm->arch.hpt_virt + (hash << 7));
211	hpte += 7 * 2;
212	/* HPTE low word - RPN, protection, etc. */
213	hpte[1] = (pfn << PAGE_SHIFT) \| HPTE_R_R \| HPTE_R_C \|
214	HPTE_R_M \| PP_RWXX;
215	wmb();
216	hpte[0] = HPTE_V_1TB_SEG \| (VRMA_VSID << (40 - 16)) \|
217	(i << (VRMA_PAGE_ORDER - 16)) \| HPTE_V_BOLTED \|
218	HPTE_V_LARGE \| HPTE_V_VALID;
219	}
220	}
221
222	int kvmppc_mmu_hv_init(void)
223	{
224	if (!cpu_has_feature(CPU_FTR_HVMODE_206))
225	return -EINVAL;
226	memset(lpid_inuse, 0, sizeof(lpid_inuse));
227	set_bit(mfspr(SPRN_LPID), lpid_inuse);
228	set_bit(LPID_RSVD, lpid_inuse);
229
230	return 0;
231	}
232
233	void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
234	{
235	}
236
237	static void kvmppc_mmu_book3s_64_hv_reset_msr(struct kvm_vcpu *vcpu)
238	{
239	kvmppc_set_msr(vcpu, MSR_SF \| MSR_ME);
240	}
241
242	static int kvmppc_mmu_book3s_64_hv_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
243	struct kvmppc_pte *gpte, bool data)
244	{
245	return -ENOENT;
246	}
247
248	void kvmppc_mmu_book3s_hv_init(struct kvm_vcpu *vcpu)
249	{
250	struct kvmppc_mmu *mmu = &vcpu->arch.mmu;
251
252	vcpu->arch.slb_nr = 32; /* Assume POWER7 for now */
253
254	mmu->xlate = kvmppc_mmu_book3s_64_hv_xlate;
255	mmu->reset_msr = kvmppc_mmu_book3s_64_hv_reset_msr;
256
257	vcpu->arch.hflags \|= BOOK3S_HFLAG_SLB;
258	}