[linux-2.6-block.git] / arch / powerpc / kvm / book3s_64_vio_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>
 * Copyright 2011 David Gibson, IBM Corporation <dwg@au1.ibm.com>
 * Copyright 2016 Alexey Kardashevskiy, IBM Corporation <aik@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 <linux/list.h>
#include <linux/stringify.h>

#include <asm/kvm_ppc.h>
#include <asm/kvm_book3s.h>
#include <asm/book3s/64/mmu-hash.h>
#include <asm/mmu_context.h>
#include <asm/hvcall.h>
#include <asm/synch.h>
#include <asm/ppc-opcode.h>
#include <asm/kvm_host.h>
#include <asm/udbg.h>
#include <asm/iommu.h>
#include <asm/tce.h>
#include <asm/pte-walk.h>

#ifdef CONFIG_BUG

#define WARN_ON_ONCE_RM(condition)	({			\
	static bool __section(.data.unlikely) __warned;		\
	int __ret_warn_once = !!(condition);			\
								\
	if (unlikely(__ret_warn_once && !__warned)) {		\
		__warned = true;				\
		pr_err("WARN_ON_ONCE_RM: (%s) at %s:%u\n",	\
				__stringify(condition),		\
				__func__, __LINE__);		\
		dump_stack();					\
	}							\
	unlikely(__ret_warn_once);				\
})

#else

#define WARN_ON_ONCE_RM(condition) ({				\
	int __ret_warn_on = !!(condition);			\
	unlikely(__ret_warn_on);				\
})

#endif

#define TCES_PER_PAGE	(PAGE_SIZE / sizeof(u64))

/*
 * Finds a TCE table descriptor by LIOBN.
 *
 * WARNING: This will be called in real or virtual mode on HV KVM and virtual
 *          mode on PR KVM
 */
struct kvmppc_spapr_tce_table *kvmppc_find_table(struct kvm *kvm,
		unsigned long liobn)
{
	struct kvmppc_spapr_tce_table *stt;

	list_for_each_entry_lockless(stt, &kvm->arch.spapr_tce_tables, list)
		if (stt->liobn == liobn)
			return stt;

	return NULL;
}
EXPORT_SYMBOL_GPL(kvmppc_find_table);

/*
 * Validates TCE address.
 * At the moment flags and page mask are validated.
 * As the host kernel does not access those addresses (just puts them
 * to the table and user space is supposed to process them), we can skip
 * checking other things (such as TCE is a guest RAM address or the page
 * was actually allocated).
 *
 * WARNING: This will be called in real-mode on HV KVM and virtual
 *          mode on PR KVM
 */
long kvmppc_tce_validate(struct kvmppc_spapr_tce_table *stt, unsigned long tce)
{
	unsigned long gpa = tce & ~(TCE_PCI_READ | TCE_PCI_WRITE);
	enum dma_data_direction dir = iommu_tce_direction(tce);

	/* Allow userspace to poison TCE table */
	if (dir == DMA_NONE)
		return H_SUCCESS;

	if (iommu_tce_check_gpa(stt->page_shift, gpa))
		return H_PARAMETER;

	return H_SUCCESS;
}
EXPORT_SYMBOL_GPL(kvmppc_tce_validate);

/* Note on the use of page_address() in real mode,
 *
 * It is safe to use page_address() in real mode on ppc64 because
 * page_address() is always defined as lowmem_page_address()
 * which returns __va(PFN_PHYS(page_to_pfn(page))) which is arithmetic
 * operation and does not access page struct.
 *
 * Theoretically page_address() could be defined different
 * but either WANT_PAGE_VIRTUAL or HASHED_PAGE_VIRTUAL
 * would have to be enabled.
 * WANT_PAGE_VIRTUAL is never enabled on ppc32/ppc64,
 * HASHED_PAGE_VIRTUAL could be enabled for ppc32 only and only
 * if CONFIG_HIGHMEM is defined. As CONFIG_SPARSEMEM_VMEMMAP
 * is not expected to be enabled on ppc32, page_address()
 * is safe for ppc32 as well.
 *
 * WARNING: This will be called in real-mode on HV KVM and virtual
 *          mode on PR KVM
 */
static u64 *kvmppc_page_address(struct page *page)
{
#if defined(HASHED_PAGE_VIRTUAL) || defined(WANT_PAGE_VIRTUAL)
#error TODO: fix to avoid page_address() here
#endif
	return (u64 *) page_address(page);
}

/*
 * Handles TCE requests for emulated devices.
 * Puts guest TCE values to the table and expects user space to convert them.
 * Called in both real and virtual modes.
 * Cannot fail so kvmppc_tce_validate must be called before it.
 *
 * WARNING: This will be called in real-mode on HV KVM and virtual
 *          mode on PR KVM
 */
void kvmppc_tce_put(struct kvmppc_spapr_tce_table *stt,
		unsigned long idx, unsigned long tce)
{
	struct page *page;
	u64 *tbl;

	idx -= stt->offset;
	page = stt->pages[idx / TCES_PER_PAGE];
	tbl = kvmppc_page_address(page);

	tbl[idx % TCES_PER_PAGE] = tce;
}
EXPORT_SYMBOL_GPL(kvmppc_tce_put);

long kvmppc_gpa_to_ua(struct kvm *kvm, unsigned long gpa,
		unsigned long *ua, unsigned long **prmap)
{
	unsigned long gfn = gpa >> PAGE_SHIFT;
	struct kvm_memory_slot *memslot;

	memslot = search_memslots(kvm_memslots(kvm), gfn);
	if (!memslot)
		return -EINVAL;

	*ua = __gfn_to_hva_memslot(memslot, gfn) |
		(gpa & ~(PAGE_MASK | TCE_PCI_READ | TCE_PCI_WRITE));

#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
	if (prmap)
		*prmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
#endif

	return 0;
}
EXPORT_SYMBOL_GPL(kvmppc_gpa_to_ua);

#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
static void kvmppc_rm_clear_tce(struct iommu_table *tbl, unsigned long entry)
{
	unsigned long hpa = 0;
	enum dma_data_direction dir = DMA_NONE;

	iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);
}

static long kvmppc_rm_tce_iommu_mapped_dec(struct kvm *kvm,
		struct iommu_table *tbl, unsigned long entry)
{
	struct mm_iommu_table_group_mem_t *mem = NULL;
	const unsigned long pgsize = 1ULL << tbl->it_page_shift;
	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RM(tbl, entry);

	if (!pua)
		/* it_userspace allocation might be delayed */
		return H_TOO_HARD;

	mem = mm_iommu_lookup_rm(kvm->mm, be64_to_cpu(*pua), pgsize);
	if (!mem)
		return H_TOO_HARD;

	mm_iommu_mapped_dec(mem);

	*pua = cpu_to_be64(0);

	return H_SUCCESS;
}

static long kvmppc_rm_tce_iommu_do_unmap(struct kvm *kvm,
		struct iommu_table *tbl, unsigned long entry)
{
	enum dma_data_direction dir = DMA_NONE;
	unsigned long hpa = 0;
	long ret;

	if (iommu_tce_xchg_rm(tbl, entry, &hpa, &dir))
		/*
		 * real mode xchg can fail if struct page crosses
		 * a page boundary
		 */
		return H_TOO_HARD;

	if (dir == DMA_NONE)
		return H_SUCCESS;

	ret = kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);
	if (ret)
		iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);

	return ret;
}

static long kvmppc_rm_tce_iommu_unmap(struct kvm *kvm,
		struct kvmppc_spapr_tce_table *stt, struct iommu_table *tbl,
		unsigned long entry)
{
	unsigned long i, ret = H_SUCCESS;
	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	unsigned long io_entry = entry * subpages;

	for (i = 0; i < subpages; ++i) {
		ret = kvmppc_rm_tce_iommu_do_unmap(kvm, tbl, io_entry + i);
		if (ret != H_SUCCESS)
			break;
	}

	return ret;
}

static long kvmppc_rm_tce_iommu_do_map(struct kvm *kvm, struct iommu_table *tbl,
		unsigned long entry, unsigned long ua,
		enum dma_data_direction dir)
{
	long ret;
	unsigned long hpa = 0;
	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RM(tbl, entry);
	struct mm_iommu_table_group_mem_t *mem;

	if (!pua)
		/* it_userspace allocation might be delayed */
		return H_TOO_HARD;

	mem = mm_iommu_lookup_rm(kvm->mm, ua, 1ULL << tbl->it_page_shift);
	if (!mem)
		return H_TOO_HARD;

	if (WARN_ON_ONCE_RM(mm_iommu_ua_to_hpa_rm(mem, ua, tbl->it_page_shift,
			&hpa)))
		return H_HARDWARE;

	if (WARN_ON_ONCE_RM(mm_iommu_mapped_inc(mem)))
		return H_CLOSED;

	ret = iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);
	if (ret) {
		mm_iommu_mapped_dec(mem);
		/*
		 * real mode xchg can fail if struct page crosses
		 * a page boundary
		 */
		return H_TOO_HARD;
	}

	if (dir != DMA_NONE)
		kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);

	*pua = cpu_to_be64(ua);

	return 0;
}

static long kvmppc_rm_tce_iommu_map(struct kvm *kvm,
		struct kvmppc_spapr_tce_table *stt, struct iommu_table *tbl,
		unsigned long entry, unsigned long ua,
		enum dma_data_direction dir)
{
	unsigned long i, pgoff, ret = H_SUCCESS;
	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	unsigned long io_entry = entry * subpages;

	for (i = 0, pgoff = 0; i < subpages;
			++i, pgoff += IOMMU_PAGE_SIZE(tbl)) {

		ret = kvmppc_rm_tce_iommu_do_map(kvm, tbl,
				io_entry + i, ua + pgoff, dir);
		if (ret != H_SUCCESS)
			break;
	}

	return ret;
}

long kvmppc_rm_h_put_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
		unsigned long ioba, unsigned long tce)
{
	struct kvmppc_spapr_tce_table *stt;
	long ret;
	struct kvmppc_spapr_tce_iommu_table *stit;
	unsigned long entry, ua = 0;
	enum dma_data_direction dir;

	/* udbg_printf("H_PUT_TCE(): liobn=0x%lx ioba=0x%lx, tce=0x%lx\n", */
	/* 	    liobn, ioba, tce); */

	/* For radix, we might be in virtual mode, so punt */
	if (kvm_is_radix(vcpu->kvm))
		return H_TOO_HARD;

	stt = kvmppc_find_table(vcpu->kvm, liobn);
	if (!stt)
		return H_TOO_HARD;

	ret = kvmppc_ioba_validate(stt, ioba, 1);
	if (ret != H_SUCCESS)
		return ret;

	ret = kvmppc_tce_validate(stt, tce);
	if (ret != H_SUCCESS)
		return ret;

	dir = iommu_tce_direction(tce);
	if ((dir != DMA_NONE) && kvmppc_gpa_to_ua(vcpu->kvm,
			tce & ~(TCE_PCI_READ | TCE_PCI_WRITE), &ua, NULL))
		return H_PARAMETER;

	entry = ioba >> stt->page_shift;

	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
		if (dir == DMA_NONE)
			ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
					stit->tbl, entry);
		else
			ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
					stit->tbl, entry, ua, dir);

		if (ret == H_SUCCESS)
			continue;

		if (ret == H_TOO_HARD)
			return ret;

		WARN_ON_ONCE_RM(1);
		kvmppc_rm_clear_tce(stit->tbl, entry);
	}

	kvmppc_tce_put(stt, entry, tce);

	return H_SUCCESS;
}

static long kvmppc_rm_ua_to_hpa(struct kvm_vcpu *vcpu,
		unsigned long ua, unsigned long *phpa)
{
	pte_t *ptep, pte;
	unsigned shift = 0;

	/*
	 * Called in real mode with MSR_EE = 0. We are safe here.
	 * It is ok to do the lookup with arch.pgdir here, because
	 * we are doing this on secondary cpus and current task there
	 * is not the hypervisor. Also this is safe against THP in the
	 * host, because an IPI to primary thread will wait for the secondary
	 * to exit which will agains result in the below page table walk
	 * to finish.
	 */
	ptep = __find_linux_pte(vcpu->arch.pgdir, ua, NULL, &shift);
	if (!ptep || !pte_present(*ptep))
		return -ENXIO;
	pte = *ptep;

	if (!shift)
		shift = PAGE_SHIFT;

	/* Avoid handling anything potentially complicated in realmode */
	if (shift > PAGE_SHIFT)
		return -EAGAIN;

	if (!pte_young(pte))
		return -EAGAIN;

	*phpa = (pte_pfn(pte) << PAGE_SHIFT) | (ua & ((1ULL << shift) - 1)) |
			(ua & ~PAGE_MASK);

	return 0;
}

long kvmppc_rm_h_put_tce_indirect(struct kvm_vcpu *vcpu,
		unsigned long liobn, unsigned long ioba,
		unsigned long tce_list,	unsigned long npages)
{
	struct kvmppc_spapr_tce_table *stt;
	long i, ret = H_SUCCESS;
	unsigned long tces, entry, ua = 0;
	unsigned long *rmap = NULL;
	bool prereg = false;
	struct kvmppc_spapr_tce_iommu_table *stit;

	/* For radix, we might be in virtual mode, so punt */
	if (kvm_is_radix(vcpu->kvm))
		return H_TOO_HARD;

	stt = kvmppc_find_table(vcpu->kvm, liobn);
	if (!stt)
		return H_TOO_HARD;

	entry = ioba >> stt->page_shift;
	/*
	 * The spec says that the maximum size of the list is 512 TCEs
	 * so the whole table addressed resides in 4K page
	 */
	if (npages > 512)
		return H_PARAMETER;

	if (tce_list & (SZ_4K - 1))
		return H_PARAMETER;

	ret = kvmppc_ioba_validate(stt, ioba, npages);
	if (ret != H_SUCCESS)
		return ret;

	if (mm_iommu_preregistered(vcpu->kvm->mm)) {
		/*
		 * We get here if guest memory was pre-registered which
		 * is normally VFIO case and gpa->hpa translation does not
		 * depend on hpt.
		 */
		struct mm_iommu_table_group_mem_t *mem;

		if (kvmppc_gpa_to_ua(vcpu->kvm, tce_list, &ua, NULL))
			return H_TOO_HARD;

		mem = mm_iommu_lookup_rm(vcpu->kvm->mm, ua, IOMMU_PAGE_SIZE_4K);
		if (mem)
			prereg = mm_iommu_ua_to_hpa_rm(mem, ua,
					IOMMU_PAGE_SHIFT_4K, &tces) == 0;
	}

	if (!prereg) {
		/*
		 * This is usually a case of a guest with emulated devices only
		 * when TCE list is not in preregistered memory.
		 * We do not require memory to be preregistered in this case
		 * so lock rmap and do __find_linux_pte_or_hugepte().
		 */
		if (kvmppc_gpa_to_ua(vcpu->kvm, tce_list, &ua, &rmap))
			return H_TOO_HARD;

		rmap = (void *) vmalloc_to_phys(rmap);
		if (WARN_ON_ONCE_RM(!rmap))
			return H_HARDWARE;

		/*
		 * Synchronize with the MMU notifier callbacks in
		 * book3s_64_mmu_hv.c (kvm_unmap_hva_range_hv etc.).
		 * While we have the rmap lock, code running on other CPUs
		 * cannot finish unmapping the host real page that backs
		 * this guest real page, so we are OK to access the host
		 * real page.
		 */
		lock_rmap(rmap);
		if (kvmppc_rm_ua_to_hpa(vcpu, ua, &tces)) {
			ret = H_TOO_HARD;
			goto unlock_exit;
		}
	}

	for (i = 0; i < npages; ++i) {
		unsigned long tce = be64_to_cpu(((u64 *)tces)[i]);

		ret = kvmppc_tce_validate(stt, tce);
		if (ret != H_SUCCESS)
			goto unlock_exit;

		ua = 0;
		if (kvmppc_gpa_to_ua(vcpu->kvm,
				tce & ~(TCE_PCI_READ | TCE_PCI_WRITE),
				&ua, NULL))
			return H_PARAMETER;

		list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
			ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
					stit->tbl, entry + i, ua,
					iommu_tce_direction(tce));

			if (ret == H_SUCCESS)
				continue;

			if (ret == H_TOO_HARD)
				goto unlock_exit;

			WARN_ON_ONCE_RM(1);
			kvmppc_rm_clear_tce(stit->tbl, entry);
		}

		kvmppc_tce_put(stt, entry + i, tce);
	}

unlock_exit:
	if (rmap)
		unlock_rmap(rmap);

	return ret;
}

long kvmppc_rm_h_stuff_tce(struct kvm_vcpu *vcpu,
		unsigned long liobn, unsigned long ioba,
		unsigned long tce_value, unsigned long npages)
{
	struct kvmppc_spapr_tce_table *stt;
	long i, ret;
	struct kvmppc_spapr_tce_iommu_table *stit;

	/* For radix, we might be in virtual mode, so punt */
	if (kvm_is_radix(vcpu->kvm))
		return H_TOO_HARD;

	stt = kvmppc_find_table(vcpu->kvm, liobn);
	if (!stt)
		return H_TOO_HARD;

	ret = kvmppc_ioba_validate(stt, ioba, npages);
	if (ret != H_SUCCESS)
		return ret;

	/* Check permission bits only to allow userspace poison TCE for debug */
	if (tce_value & (TCE_PCI_WRITE | TCE_PCI_READ))
		return H_PARAMETER;

	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
		unsigned long entry = ioba >> stt->page_shift;

		for (i = 0; i < npages; ++i) {
			ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
					stit->tbl, entry + i);

			if (ret == H_SUCCESS)
				continue;

			if (ret == H_TOO_HARD)
				return ret;

			WARN_ON_ONCE_RM(1);
			kvmppc_rm_clear_tce(stit->tbl, entry);
		}
	}

	for (i = 0; i < npages; ++i, ioba += (1ULL << stt->page_shift))
		kvmppc_tce_put(stt, ioba >> stt->page_shift, tce_value);

	return H_SUCCESS;
}

/* This can be called in either virtual mode or real mode */
long kvmppc_h_get_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
		      unsigned long ioba)
{
	struct kvmppc_spapr_tce_table *stt;
	long ret;
	unsigned long idx;
	struct page *page;
	u64 *tbl;

	stt = kvmppc_find_table(vcpu->kvm, liobn);
	if (!stt)
		return H_TOO_HARD;

	ret = kvmppc_ioba_validate(stt, ioba, 1);
	if (ret != H_SUCCESS)
		return ret;

	idx = (ioba >> stt->page_shift) - stt->offset;
	page = stt->pages[idx / TCES_PER_PAGE];
	tbl = (u64 *)page_address(page);

	vcpu->arch.regs.gpr[4] = tbl[idx % TCES_PER_PAGE];

	return H_SUCCESS;
}
EXPORT_SYMBOL_GPL(kvmppc_h_get_tce);

#endif /* KVM_BOOK3S_HV_POSSIBLE */
Commit	Line	Data
54738c09 DG	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	* Copyright 2011 David Gibson, IBM Corporation <dwg@au1.ibm.com>
d3695aa4	17	* Copyright 2016 Alexey Kardashevskiy, IBM Corporation <aik@au1.ibm.com>
54738c09 DG	18	*/
	19
	20	#include <linux/types.h>
	21	#include <linux/string.h>
	22	#include <linux/kvm.h>
	23	#include <linux/kvm_host.h>
	24	#include <linux/highmem.h>
	25	#include <linux/gfp.h>
	26	#include <linux/slab.h>
	27	#include <linux/hugetlb.h>
	28	#include <linux/list.h>
5c35a02c	29	#include <linux/stringify.h>
54738c09	30
54738c09 DG	31	#include <asm/kvm_ppc.h>
54738c09 DG	32	#include <asm/kvm_book3s.h>
f64e8084	33	#include <asm/book3s/64/mmu-hash.h>
d3695aa4	34	#include <asm/mmu_context.h>
54738c09 DG	35	#include <asm/hvcall.h>
	36	#include <asm/synch.h>
	37	#include <asm/ppc-opcode.h>
	38	#include <asm/kvm_host.h>
	39	#include <asm/udbg.h>
fcbb2ce6	40	#include <asm/iommu.h>
5ee7af18	41	#include <asm/tce.h>
94171b19	42	#include <asm/pte-walk.h>
54738c09	43
121f80ba AK	44	#ifdef CONFIG_BUG
	45
	46	#define WARN_ON_ONCE_RM(condition) ({ \
	47	static bool __section(.data.unlikely) __warned; \
	48	int __ret_warn_once = !!(condition); \
	49	\
	50	if (unlikely(__ret_warn_once && !__warned)) { \
	51	__warned = true; \
	52	pr_err("WARN_ON_ONCE_RM: (%s) at %s:%u\n", \
	53	__stringify(condition), \
	54	__func__, __LINE__); \
	55	dump_stack(); \
	56	} \
	57	unlikely(__ret_warn_once); \
	58	})
	59
	60	#else
	61
	62	#define WARN_ON_ONCE_RM(condition) ({ \
	63	int __ret_warn_on = !!(condition); \
	64	unlikely(__ret_warn_on); \
	65	})
	66
	67	#endif
	68
54738c09 DG	69	#define TCES_PER_PAGE (PAGE_SIZE / sizeof(u64))
54738c09 DG	70
fcbb2ce6 AK	71	/*
	72	* Finds a TCE table descriptor by LIOBN.
	73	*
	74	* WARNING: This will be called in real or virtual mode on HV KVM and virtual
	75	* mode on PR KVM
	76	*/
503bfcbe	77	struct kvmppc_spapr_tce_table kvmppc_find_table(struct kvm kvm,
fcbb2ce6 AK	78	unsigned long liobn)
fcbb2ce6 AK	79	{
fcbb2ce6 AK	80	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6 AK	81
366baf28	82	list_for_each_entry_lockless(stt, &kvm->arch.spapr_tce_tables, list)
fcbb2ce6 AK	83	if (stt->liobn == liobn)
	84	return stt;
	85
	86	return NULL;
	87	}
d3695aa4	88	EXPORT_SYMBOL_GPL(kvmppc_find_table);
fcbb2ce6	89
5ee7af18 AK	90	/*
	91	* Validates TCE address.
	92	* At the moment flags and page mask are validated.
	93	* As the host kernel does not access those addresses (just puts them
	94	* to the table and user space is supposed to process them), we can skip
	95	* checking other things (such as TCE is a guest RAM address or the page
	96	* was actually allocated).
	97	*
	98	* WARNING: This will be called in real-mode on HV KVM and virtual
	99	* mode on PR KVM
	100	*/
	101	long kvmppc_tce_validate(struct kvmppc_spapr_tce_table *stt, unsigned long tce)
	102	{
b1af23d8 AK	103	unsigned long gpa = tce & ~(TCE_PCI_READ \| TCE_PCI_WRITE);
	104	enum dma_data_direction dir = iommu_tce_direction(tce);
	105
	106	/* Allow userspace to poison TCE table */
	107	if (dir == DMA_NONE)
	108	return H_SUCCESS;
5ee7af18	109
b1af23d8	110	if (iommu_tce_check_gpa(stt->page_shift, gpa))
5ee7af18 AK	111	return H_PARAMETER;
	112
	113	return H_SUCCESS;
	114	}
	115	EXPORT_SYMBOL_GPL(kvmppc_tce_validate);
	116
	117	/* Note on the use of page_address() in real mode,
	118	*
	119	* It is safe to use page_address() in real mode on ppc64 because
	120	* page_address() is always defined as lowmem_page_address()
	121	* which returns __va(PFN_PHYS(page_to_pfn(page))) which is arithmetic
	122	* operation and does not access page struct.
	123	*
	124	* Theoretically page_address() could be defined different
	125	* but either WANT_PAGE_VIRTUAL or HASHED_PAGE_VIRTUAL
	126	* would have to be enabled.
	127	* WANT_PAGE_VIRTUAL is never enabled on ppc32/ppc64,
	128	* HASHED_PAGE_VIRTUAL could be enabled for ppc32 only and only
	129	* if CONFIG_HIGHMEM is defined. As CONFIG_SPARSEMEM_VMEMMAP
	130	* is not expected to be enabled on ppc32, page_address()
	131	* is safe for ppc32 as well.
	132	*
	133	* WARNING: This will be called in real-mode on HV KVM and virtual
	134	* mode on PR KVM
	135	*/
	136	static u64 kvmppc_page_address(struct page page)
	137	{
	138	#if defined(HASHED_PAGE_VIRTUAL) \|\| defined(WANT_PAGE_VIRTUAL)
	139	#error TODO: fix to avoid page_address() here
	140	#endif
	141	return (u64 *) page_address(page);
	142	}
	143
	144	/*
	145	* Handles TCE requests for emulated devices.
	146	* Puts guest TCE values to the table and expects user space to convert them.
	147	* Called in both real and virtual modes.
	148	* Cannot fail so kvmppc_tce_validate must be called before it.
	149	*
	150	* WARNING: This will be called in real-mode on HV KVM and virtual
	151	* mode on PR KVM
	152	*/
	153	void kvmppc_tce_put(struct kvmppc_spapr_tce_table *stt,
	154	unsigned long idx, unsigned long tce)
	155	{
	156	struct page *page;
	157	u64 *tbl;
	158
14f853f1	159	idx -= stt->offset;
5ee7af18 AK	160	page = stt->pages[idx / TCES_PER_PAGE];
	161	tbl = kvmppc_page_address(page);
	162
	163	tbl[idx % TCES_PER_PAGE] = tce;
	164	}
	165	EXPORT_SYMBOL_GPL(kvmppc_tce_put);
fcbb2ce6	166
d3695aa4 AK	167	long kvmppc_gpa_to_ua(struct kvm *kvm, unsigned long gpa,
	168	unsigned long ua, unsigned long *prmap)
	169	{
	170	unsigned long gfn = gpa >> PAGE_SHIFT;
	171	struct kvm_memory_slot *memslot;
	172
	173	memslot = search_memslots(kvm_memslots(kvm), gfn);
	174	if (!memslot)
	175	return -EINVAL;
	176
	177	*ua = __gfn_to_hva_memslot(memslot, gfn) \|
	178	(gpa & ~(PAGE_MASK \| TCE_PCI_READ \| TCE_PCI_WRITE));
	179
	180	#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
	181	if (prmap)
	182	*prmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
	183	#endif
	184
	185	return 0;
	186	}
	187	EXPORT_SYMBOL_GPL(kvmppc_gpa_to_ua);
	188
	189	#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
121f80ba AK	190	static void kvmppc_rm_clear_tce(struct iommu_table *tbl, unsigned long entry)
	191	{
	192	unsigned long hpa = 0;
	193	enum dma_data_direction dir = DMA_NONE;
	194
	195	iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);
	196	}
	197
	198	static long kvmppc_rm_tce_iommu_mapped_dec(struct kvm *kvm,
	199	struct iommu_table *tbl, unsigned long entry)
	200	{
	201	struct mm_iommu_table_group_mem_t *mem = NULL;
	202	const unsigned long pgsize = 1ULL << tbl->it_page_shift;
a68bd126	203	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RM(tbl, entry);
121f80ba AK	204
	205	if (!pua)
	206	/* it_userspace allocation might be delayed */
	207	return H_TOO_HARD;
	208
00a5c58d	209	mem = mm_iommu_lookup_rm(kvm->mm, be64_to_cpu(*pua), pgsize);
121f80ba AK	210	if (!mem)
	211	return H_TOO_HARD;
	212
	213	mm_iommu_mapped_dec(mem);
	214
00a5c58d	215	*pua = cpu_to_be64(0);
121f80ba AK	216
	217	return H_SUCCESS;
	218	}
	219
ca1fc489	220	static long kvmppc_rm_tce_iommu_do_unmap(struct kvm *kvm,
121f80ba AK	221	struct iommu_table *tbl, unsigned long entry)
	222	{
	223	enum dma_data_direction dir = DMA_NONE;
	224	unsigned long hpa = 0;
	225	long ret;
	226
	227	if (iommu_tce_xchg_rm(tbl, entry, &hpa, &dir))
	228	/*
	229	* real mode xchg can fail if struct page crosses
	230	* a page boundary
	231	*/
	232	return H_TOO_HARD;
	233
	234	if (dir == DMA_NONE)
	235	return H_SUCCESS;
	236
	237	ret = kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);
	238	if (ret)
	239	iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);
	240
	241	return ret;
	242	}
	243
ca1fc489 AK	244	static long kvmppc_rm_tce_iommu_unmap(struct kvm *kvm,
	245	struct kvmppc_spapr_tce_table stt, struct iommu_table tbl,
	246	unsigned long entry)
	247	{
	248	unsigned long i, ret = H_SUCCESS;
	249	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	250	unsigned long io_entry = entry * subpages;
	251
	252	for (i = 0; i < subpages; ++i) {
	253	ret = kvmppc_rm_tce_iommu_do_unmap(kvm, tbl, io_entry + i);
	254	if (ret != H_SUCCESS)
	255	break;
	256	}
	257
	258	return ret;
	259	}
	260
	261	static long kvmppc_rm_tce_iommu_do_map(struct kvm kvm, struct iommu_table tbl,
121f80ba AK	262	unsigned long entry, unsigned long ua,
	263	enum dma_data_direction dir)
	264	{
	265	long ret;
	266	unsigned long hpa = 0;
a68bd126	267	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RM(tbl, entry);
121f80ba AK	268	struct mm_iommu_table_group_mem_t *mem;
	269
	270	if (!pua)
	271	/* it_userspace allocation might be delayed */
	272	return H_TOO_HARD;
	273
	274	mem = mm_iommu_lookup_rm(kvm->mm, ua, 1ULL << tbl->it_page_shift);
	275	if (!mem)
	276	return H_TOO_HARD;
	277
76fa4975 AK	278	if (WARN_ON_ONCE_RM(mm_iommu_ua_to_hpa_rm(mem, ua, tbl->it_page_shift,
76fa4975 AK	279	&hpa)))
121f80ba AK	280	return H_HARDWARE;
121f80ba AK	281
121f80ba AK	282	if (WARN_ON_ONCE_RM(mm_iommu_mapped_inc(mem)))
	283	return H_CLOSED;
	284
	285	ret = iommu_tce_xchg_rm(tbl, entry, &hpa, &dir);
	286	if (ret) {
	287	mm_iommu_mapped_dec(mem);
	288	/*
	289	* real mode xchg can fail if struct page crosses
	290	* a page boundary
	291	*/
	292	return H_TOO_HARD;
	293	}
	294
	295	if (dir != DMA_NONE)
	296	kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);
	297
00a5c58d	298	*pua = cpu_to_be64(ua);
121f80ba AK	299
	300	return 0;
	301	}
	302
ca1fc489 AK	303	static long kvmppc_rm_tce_iommu_map(struct kvm *kvm,
	304	struct kvmppc_spapr_tce_table stt, struct iommu_table tbl,
	305	unsigned long entry, unsigned long ua,
	306	enum dma_data_direction dir)
	307	{
	308	unsigned long i, pgoff, ret = H_SUCCESS;
	309	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	310	unsigned long io_entry = entry * subpages;
	311
	312	for (i = 0, pgoff = 0; i < subpages;
	313	++i, pgoff += IOMMU_PAGE_SIZE(tbl)) {
	314
	315	ret = kvmppc_rm_tce_iommu_do_map(kvm, tbl,
	316	io_entry + i, ua + pgoff, dir);
	317	if (ret != H_SUCCESS)
	318	break;
	319	}
	320
	321	return ret;
	322	}
	323
31217db7 AK	324	long kvmppc_rm_h_put_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
31217db7 AK	325	unsigned long ioba, unsigned long tce)
54738c09	326	{
503bfcbe	327	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6	328	long ret;
121f80ba AK	329	struct kvmppc_spapr_tce_iommu_table *stit;
	330	unsigned long entry, ua = 0;
	331	enum dma_data_direction dir;
54738c09 DG	332
	333	/* udbg_printf("H_PUT_TCE(): liobn=0x%lx ioba=0x%lx, tce=0x%lx\n", */
	334	/* liobn, ioba, tce); */
	335
acde2572 PM	336	/* For radix, we might be in virtual mode, so punt */
	337	if (kvm_is_radix(vcpu->kvm))
	338	return H_TOO_HARD;
	339
503bfcbe	340	stt = kvmppc_find_table(vcpu->kvm, liobn);
fcbb2ce6 AK	341	if (!stt)
	342	return H_TOO_HARD;
	343
	344	ret = kvmppc_ioba_validate(stt, ioba, 1);
	345	if (ret != H_SUCCESS)
	346	return ret;
	347
5ee7af18 AK	348	ret = kvmppc_tce_validate(stt, tce);
	349	if (ret != H_SUCCESS)
	350	return ret;
fcbb2ce6	351
121f80ba AK	352	dir = iommu_tce_direction(tce);
	353	if ((dir != DMA_NONE) && kvmppc_gpa_to_ua(vcpu->kvm,
	354	tce & ~(TCE_PCI_READ \| TCE_PCI_WRITE), &ua, NULL))
	355	return H_PARAMETER;
	356
	357	entry = ioba >> stt->page_shift;
	358
	359	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
	360	if (dir == DMA_NONE)
ca1fc489	361	ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
121f80ba AK	362	stit->tbl, entry);
121f80ba AK	363	else
ca1fc489	364	ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
121f80ba AK	365	stit->tbl, entry, ua, dir);
	366
	367	if (ret == H_SUCCESS)
	368	continue;
	369
	370	if (ret == H_TOO_HARD)
	371	return ret;
	372
	373	WARN_ON_ONCE_RM(1);
	374	kvmppc_rm_clear_tce(stit->tbl, entry);
	375	}
	376
	377	kvmppc_tce_put(stt, entry, tce);
fcbb2ce6 AK	378
fcbb2ce6 AK	379	return H_SUCCESS;
54738c09	380	}
69e9fbb2	381
d3695aa4 AK	382	static long kvmppc_rm_ua_to_hpa(struct kvm_vcpu *vcpu,
	383	unsigned long ua, unsigned long *phpa)
	384	{
	385	pte_t *ptep, pte;
	386	unsigned shift = 0;
	387
94171b19 AK	388	/*
	389	* Called in real mode with MSR_EE = 0. We are safe here.
	390	* It is ok to do the lookup with arch.pgdir here, because
	391	* we are doing this on secondary cpus and current task there
	392	* is not the hypervisor. Also this is safe against THP in the
	393	* host, because an IPI to primary thread will wait for the secondary
	394	* to exit which will agains result in the below page table walk
	395	* to finish.
	396	*/
	397	ptep = __find_linux_pte(vcpu->arch.pgdir, ua, NULL, &shift);
d3695aa4 AK	398	if (!ptep \|\| !pte_present(*ptep))
	399	return -ENXIO;
	400	pte = *ptep;
	401
	402	if (!shift)
	403	shift = PAGE_SHIFT;
	404
	405	/* Avoid handling anything potentially complicated in realmode */
	406	if (shift > PAGE_SHIFT)
	407	return -EAGAIN;
	408
	409	if (!pte_young(pte))
	410	return -EAGAIN;
	411
	412	*phpa = (pte_pfn(pte) << PAGE_SHIFT) \| (ua & ((1ULL << shift) - 1)) \|
	413	(ua & ~PAGE_MASK);
	414
	415	return 0;
	416	}
	417
	418	long kvmppc_rm_h_put_tce_indirect(struct kvm_vcpu *vcpu,
	419	unsigned long liobn, unsigned long ioba,
	420	unsigned long tce_list, unsigned long npages)
	421	{
	422	struct kvmppc_spapr_tce_table *stt;
	423	long i, ret = H_SUCCESS;
	424	unsigned long tces, entry, ua = 0;
	425	unsigned long *rmap = NULL;
da6f59e1	426	bool prereg = false;
121f80ba	427	struct kvmppc_spapr_tce_iommu_table *stit;
d3695aa4	428
acde2572 PM	429	/* For radix, we might be in virtual mode, so punt */
	430	if (kvm_is_radix(vcpu->kvm))
	431	return H_TOO_HARD;
	432
503bfcbe	433	stt = kvmppc_find_table(vcpu->kvm, liobn);
d3695aa4 AK	434	if (!stt)
	435	return H_TOO_HARD;
	436
fe26e527	437	entry = ioba >> stt->page_shift;
d3695aa4 AK	438	/*
	439	* The spec says that the maximum size of the list is 512 TCEs
	440	* so the whole table addressed resides in 4K page
	441	*/
	442	if (npages > 512)
	443	return H_PARAMETER;
	444
	445	if (tce_list & (SZ_4K - 1))
	446	return H_PARAMETER;
	447
	448	ret = kvmppc_ioba_validate(stt, ioba, npages);
	449	if (ret != H_SUCCESS)
	450	return ret;
	451
da6f59e1 AK	452	if (mm_iommu_preregistered(vcpu->kvm->mm)) {
	453	/*
	454	* We get here if guest memory was pre-registered which
	455	* is normally VFIO case and gpa->hpa translation does not
	456	* depend on hpt.
	457	*/
	458	struct mm_iommu_table_group_mem_t *mem;
d3695aa4	459
da6f59e1 AK	460	if (kvmppc_gpa_to_ua(vcpu->kvm, tce_list, &ua, NULL))
da6f59e1 AK	461	return H_TOO_HARD;
d3695aa4	462
da6f59e1 AK	463	mem = mm_iommu_lookup_rm(vcpu->kvm->mm, ua, IOMMU_PAGE_SIZE_4K);
da6f59e1 AK	464	if (mem)
76fa4975 AK	465	prereg = mm_iommu_ua_to_hpa_rm(mem, ua,
76fa4975 AK	466	IOMMU_PAGE_SHIFT_4K, &tces) == 0;
da6f59e1 AK	467	}
	468
	469	if (!prereg) {
	470	/*
	471	* This is usually a case of a guest with emulated devices only
	472	* when TCE list is not in preregistered memory.
	473	* We do not require memory to be preregistered in this case
	474	* so lock rmap and do __find_linux_pte_or_hugepte().
	475	*/
	476	if (kvmppc_gpa_to_ua(vcpu->kvm, tce_list, &ua, &rmap))
	477	return H_TOO_HARD;
	478
	479	rmap = (void *) vmalloc_to_phys(rmap);
121f80ba AK	480	if (WARN_ON_ONCE_RM(!rmap))
121f80ba AK	481	return H_HARDWARE;
da6f59e1 AK	482
	483	/*
	484	* Synchronize with the MMU notifier callbacks in
39c983ea	485	* book3s_64_mmu_hv.c (kvm_unmap_hva_range_hv etc.).
da6f59e1 AK	486	* While we have the rmap lock, code running on other CPUs
	487	* cannot finish unmapping the host real page that backs
	488	* this guest real page, so we are OK to access the host
	489	* real page.
	490	*/
	491	lock_rmap(rmap);
	492	if (kvmppc_rm_ua_to_hpa(vcpu, ua, &tces)) {
	493	ret = H_TOO_HARD;
	494	goto unlock_exit;
	495	}
d3695aa4 AK	496	}
	497
	498	for (i = 0; i < npages; ++i) {
	499	unsigned long tce = be64_to_cpu(((u64 *)tces)[i]);
	500
	501	ret = kvmppc_tce_validate(stt, tce);
	502	if (ret != H_SUCCESS)
	503	goto unlock_exit;
	504
121f80ba AK	505	ua = 0;
	506	if (kvmppc_gpa_to_ua(vcpu->kvm,
	507	tce & ~(TCE_PCI_READ \| TCE_PCI_WRITE),
	508	&ua, NULL))
	509	return H_PARAMETER;
	510
	511	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
ca1fc489	512	ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
121f80ba AK	513	stit->tbl, entry + i, ua,
	514	iommu_tce_direction(tce));
	515
	516	if (ret == H_SUCCESS)
	517	continue;
	518
	519	if (ret == H_TOO_HARD)
	520	goto unlock_exit;
	521
	522	WARN_ON_ONCE_RM(1);
	523	kvmppc_rm_clear_tce(stit->tbl, entry);
	524	}
	525
d3695aa4 AK	526	kvmppc_tce_put(stt, entry + i, tce);
	527	}
	528
	529	unlock_exit:
da6f59e1 AK	530	if (rmap)
da6f59e1 AK	531	unlock_rmap(rmap);
d3695aa4 AK	532
	533	return ret;
	534	}
	535
31217db7	536	long kvmppc_rm_h_stuff_tce(struct kvm_vcpu *vcpu,
d3695aa4 AK	537	unsigned long liobn, unsigned long ioba,
	538	unsigned long tce_value, unsigned long npages)
	539	{
	540	struct kvmppc_spapr_tce_table *stt;
	541	long i, ret;
121f80ba	542	struct kvmppc_spapr_tce_iommu_table *stit;
d3695aa4	543
acde2572 PM	544	/* For radix, we might be in virtual mode, so punt */
	545	if (kvm_is_radix(vcpu->kvm))
	546	return H_TOO_HARD;
	547
503bfcbe	548	stt = kvmppc_find_table(vcpu->kvm, liobn);
d3695aa4 AK	549	if (!stt)
	550	return H_TOO_HARD;
	551
	552	ret = kvmppc_ioba_validate(stt, ioba, npages);
	553	if (ret != H_SUCCESS)
	554	return ret;
	555
	556	/* Check permission bits only to allow userspace poison TCE for debug */
	557	if (tce_value & (TCE_PCI_WRITE \| TCE_PCI_READ))
	558	return H_PARAMETER;
	559
121f80ba	560	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
c6b61661	561	unsigned long entry = ioba >> stt->page_shift;
121f80ba AK	562
121f80ba AK	563	for (i = 0; i < npages; ++i) {
ca1fc489	564	ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
121f80ba AK	565	stit->tbl, entry + i);
	566
	567	if (ret == H_SUCCESS)
	568	continue;
	569
	570	if (ret == H_TOO_HARD)
	571	return ret;
	572
	573	WARN_ON_ONCE_RM(1);
	574	kvmppc_rm_clear_tce(stit->tbl, entry);
	575	}
	576	}
	577
fe26e527 AK	578	for (i = 0; i < npages; ++i, ioba += (1ULL << stt->page_shift))
fe26e527 AK	579	kvmppc_tce_put(stt, ioba >> stt->page_shift, tce_value);
d3695aa4 AK	580
	581	return H_SUCCESS;
	582	}
d3695aa4	583
acde2572	584	/* This can be called in either virtual mode or real mode */
69e9fbb2 LD	585	long kvmppc_h_get_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
	586	unsigned long ioba)
	587	{
503bfcbe	588	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6 AK	589	long ret;
	590	unsigned long idx;
	591	struct page *page;
	592	u64 *tbl;
69e9fbb2	593
503bfcbe	594	stt = kvmppc_find_table(vcpu->kvm, liobn);
fcbb2ce6 AK	595	if (!stt)
fcbb2ce6 AK	596	return H_TOO_HARD;
69e9fbb2	597
fcbb2ce6 AK	598	ret = kvmppc_ioba_validate(stt, ioba, 1);
	599	if (ret != H_SUCCESS)
	600	return ret;
69e9fbb2	601
14f853f1	602	idx = (ioba >> stt->page_shift) - stt->offset;
fcbb2ce6 AK	603	page = stt->pages[idx / TCES_PER_PAGE];
fcbb2ce6 AK	604	tbl = (u64 *)page_address(page);
69e9fbb2	605
1143a706	606	vcpu->arch.regs.gpr[4] = tbl[idx % TCES_PER_PAGE];
69e9fbb2	607
fcbb2ce6	608	return H_SUCCESS;
69e9fbb2 LD	609	}
69e9fbb2 LD	610	EXPORT_SYMBOL_GPL(kvmppc_h_get_tce);
d3695aa4 AK	611
d3695aa4 AK	612	#endif /* KVM_BOOK3S_HV_POSSIBLE */