[linux-2.6-block.git] / arch / powerpc / kvm / book3s_64_vio_hv.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 *
 * 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

/*
 * 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);

#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
static long kvmppc_rm_tce_to_ua(struct kvm *kvm, unsigned long tce,
		unsigned long *ua, unsigned long **prmap)
{
	unsigned long gfn = tce >> PAGE_SHIFT;
	struct kvm_memory_slot *memslot;

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

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

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

	return 0;
}

/*
 * 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).
 */
static long kvmppc_rm_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);
	struct kvmppc_spapr_tce_iommu_table *stit;
	unsigned long ua = 0;

	/* 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;

	if (kvmppc_rm_tce_to_ua(stt->kvm, tce, &ua, NULL))
		return H_TOO_HARD;

	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
		unsigned long hpa = 0;
		struct mm_iommu_table_group_mem_t *mem;
		long shift = stit->tbl->it_page_shift;

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

		if (mm_iommu_ua_to_hpa_rm(mem, ua, shift, &hpa))
			return H_TOO_HARD;
	}

	return H_SUCCESS;
}

/* 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.
 * Cannot fail so kvmppc_rm_tce_validate must be called before it.
 */
static void kvmppc_rm_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];
	/*
	 * page must not be NULL in real mode,
	 * kvmppc_rm_ioba_validate() must have taken care of this.
	 */
	WARN_ON_ONCE_RM(!page);
	tbl = kvmppc_page_address(page);

	tbl[idx % TCES_PER_PAGE] = tce;
}

/*
 * TCEs pages are allocated in kvmppc_rm_tce_put() which won't be able to do so
 * in real mode.
 * Check if kvmppc_rm_tce_put() can succeed in real mode, i.e. a TCEs page is
 * allocated or not required (when clearing a tce entry).
 */
static long kvmppc_rm_ioba_validate(struct kvmppc_spapr_tce_table *stt,
		unsigned long ioba, unsigned long npages, bool clearing)
{
	unsigned long i, idx, sttpage, sttpages;
	unsigned long ret = kvmppc_ioba_validate(stt, ioba, npages);

	if (ret)
		return ret;
	/*
	 * clearing==true says kvmppc_rm_tce_put won't be allocating pages
	 * for empty tces.
	 */
	if (clearing)
		return H_SUCCESS;

	idx = (ioba >> stt->page_shift) - stt->offset;
	sttpage = idx / TCES_PER_PAGE;
	sttpages = _ALIGN_UP(idx % TCES_PER_PAGE + npages, TCES_PER_PAGE) /
			TCES_PER_PAGE;
	for (i = sttpage; i < sttpage + sttpages; ++i)
		if (!stt->pages[i])
			return H_TOO_HARD;

	return H_SUCCESS;
}

static long iommu_tce_xchg_rm(struct mm_struct *mm, struct iommu_table *tbl,
		unsigned long entry, unsigned long *hpa,
		enum dma_data_direction *direction)
{
	long ret;

	ret = tbl->it_ops->exchange_rm(tbl, entry, hpa, direction);

	if (!ret && ((*direction == DMA_FROM_DEVICE) ||
				(*direction == DMA_BIDIRECTIONAL))) {
		__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RO(tbl, entry);
		/*
		 * kvmppc_rm_tce_iommu_do_map() updates the UA cache after
		 * calling this so we still get here a valid UA.
		 */
		if (pua && *pua)
			mm_iommu_ua_mark_dirty_rm(mm, be64_to_cpu(*pua));
	}

	return ret;
}

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

	iommu_tce_xchg_rm(kvm->mm, 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_RO(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(kvm->mm, 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(kvm->mm, 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_RO(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_TOO_HARD;

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

	ret = iommu_tce_xchg_rm(kvm->mm, 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_rm_ioba_validate(stt, ioba, 1, tce == 0);
	if (ret != H_SUCCESS)
		return ret;

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

	dir = iommu_tce_direction(tce);
	if ((dir != DMA_NONE) && kvmppc_rm_tce_to_ua(vcpu->kvm, tce, &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) {
			kvmppc_rm_clear_tce(vcpu->kvm, stit->tbl, entry);
			return ret;
		}
	}

	kvmppc_rm_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_rm_ioba_validate(stt, ioba, npages, false);
	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_rm_tce_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_rm_tce_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_TOO_HARD;

		/*
		 * 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_rm_tce_validate(stt, tce);
		if (ret != H_SUCCESS)
			goto unlock_exit;
	}

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

		ua = 0;
		if (kvmppc_rm_tce_to_ua(vcpu->kvm, tce, &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) {
				kvmppc_rm_clear_tce(vcpu->kvm, stit->tbl,
						entry);
				goto unlock_exit;
			}
		}

		kvmppc_rm_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_rm_ioba_validate(stt, ioba, npages, tce_value == 0);
	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(vcpu->kvm, stit->tbl, entry);
		}
	}

	for (i = 0; i < npages; ++i, ioba += (1ULL << stt->page_shift))
		kvmppc_rm_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];
	if (!page) {
		vcpu->arch.regs.gpr[4] = 0;
		return H_SUCCESS;
	}
	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
d94d71cb	1	// SPDX-License-Identifier: GPL-2.0-only
54738c09	2	/*
54738c09 DG	3	*
	4	* Copyright 2010 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
	5	* Copyright 2011 David Gibson, IBM Corporation <dwg@au1.ibm.com>
d3695aa4	6	* Copyright 2016 Alexey Kardashevskiy, IBM Corporation <aik@au1.ibm.com>
54738c09 DG	7	*/
	8
	9	#include <linux/types.h>
	10	#include <linux/string.h>
	11	#include <linux/kvm.h>
	12	#include <linux/kvm_host.h>
	13	#include <linux/highmem.h>
	14	#include <linux/gfp.h>
	15	#include <linux/slab.h>
	16	#include <linux/hugetlb.h>
	17	#include <linux/list.h>
5c35a02c	18	#include <linux/stringify.h>
54738c09	19
54738c09 DG	20	#include <asm/kvm_ppc.h>
54738c09 DG	21	#include <asm/kvm_book3s.h>
f64e8084	22	#include <asm/book3s/64/mmu-hash.h>
d3695aa4	23	#include <asm/mmu_context.h>
54738c09 DG	24	#include <asm/hvcall.h>
	25	#include <asm/synch.h>
	26	#include <asm/ppc-opcode.h>
	27	#include <asm/kvm_host.h>
	28	#include <asm/udbg.h>
fcbb2ce6	29	#include <asm/iommu.h>
5ee7af18	30	#include <asm/tce.h>
94171b19	31	#include <asm/pte-walk.h>
54738c09	32
121f80ba AK	33	#ifdef CONFIG_BUG
	34
	35	#define WARN_ON_ONCE_RM(condition) ({ \
	36	static bool __section(.data.unlikely) __warned; \
	37	int __ret_warn_once = !!(condition); \
	38	\
	39	if (unlikely(__ret_warn_once && !__warned)) { \
	40	__warned = true; \
	41	pr_err("WARN_ON_ONCE_RM: (%s) at %s:%u\n", \
	42	__stringify(condition), \
	43	__func__, __LINE__); \
	44	dump_stack(); \
	45	} \
	46	unlikely(__ret_warn_once); \
	47	})
	48
	49	#else
	50
	51	#define WARN_ON_ONCE_RM(condition) ({ \
	52	int __ret_warn_on = !!(condition); \
	53	unlikely(__ret_warn_on); \
	54	})
	55
	56	#endif
	57
fcbb2ce6 AK	58	/*
	59	* Finds a TCE table descriptor by LIOBN.
	60	*
	61	* WARNING: This will be called in real or virtual mode on HV KVM and virtual
	62	* mode on PR KVM
	63	*/
503bfcbe	64	struct kvmppc_spapr_tce_table kvmppc_find_table(struct kvm kvm,
fcbb2ce6 AK	65	unsigned long liobn)
fcbb2ce6 AK	66	{
fcbb2ce6 AK	67	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6 AK	68
366baf28	69	list_for_each_entry_lockless(stt, &kvm->arch.spapr_tce_tables, list)
fcbb2ce6 AK	70	if (stt->liobn == liobn)
	71	return stt;
	72
	73	return NULL;
	74	}
d3695aa4	75	EXPORT_SYMBOL_GPL(kvmppc_find_table);
fcbb2ce6	76
42de7b9e	77	#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
2001825e AK	78	static long kvmppc_rm_tce_to_ua(struct kvm *kvm, unsigned long tce,
	79	unsigned long ua, unsigned long *prmap)
	80	{
	81	unsigned long gfn = tce >> PAGE_SHIFT;
	82	struct kvm_memory_slot *memslot;
	83
	84	memslot = search_memslots(kvm_memslots_raw(kvm), gfn);
	85	if (!memslot)
	86	return -EINVAL;
	87
	88	*ua = __gfn_to_hva_memslot(memslot, gfn) \|
	89	(tce & ~(PAGE_MASK \| TCE_PCI_READ \| TCE_PCI_WRITE));
	90
	91	if (prmap)
	92	*prmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
	93
	94	return 0;
	95	}
	96
5ee7af18 AK	97	/*
	98	* Validates TCE address.
	99	* At the moment flags and page mask are validated.
	100	* As the host kernel does not access those addresses (just puts them
	101	* to the table and user space is supposed to process them), we can skip
	102	* checking other things (such as TCE is a guest RAM address or the page
	103	* was actually allocated).
5ee7af18	104	*/
42de7b9e AK	105	static long kvmppc_rm_tce_validate(struct kvmppc_spapr_tce_table *stt,
42de7b9e AK	106	unsigned long tce)
5ee7af18	107	{
b1af23d8 AK	108	unsigned long gpa = tce & ~(TCE_PCI_READ \| TCE_PCI_WRITE);
b1af23d8 AK	109	enum dma_data_direction dir = iommu_tce_direction(tce);
42de7b9e AK	110	struct kvmppc_spapr_tce_iommu_table *stit;
42de7b9e AK	111	unsigned long ua = 0;
b1af23d8 AK	112
	113	/* Allow userspace to poison TCE table */
	114	if (dir == DMA_NONE)
	115	return H_SUCCESS;
5ee7af18	116
b1af23d8	117	if (iommu_tce_check_gpa(stt->page_shift, gpa))
5ee7af18 AK	118	return H_PARAMETER;
5ee7af18 AK	119
2001825e	120	if (kvmppc_rm_tce_to_ua(stt->kvm, tce, &ua, NULL))
42de7b9e AK	121	return H_TOO_HARD;
	122
	123	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
	124	unsigned long hpa = 0;
	125	struct mm_iommu_table_group_mem_t *mem;
	126	long shift = stit->tbl->it_page_shift;
	127
	128	mem = mm_iommu_lookup_rm(stt->kvm->mm, ua, 1ULL << shift);
	129	if (!mem)
	130	return H_TOO_HARD;
	131
	132	if (mm_iommu_ua_to_hpa_rm(mem, ua, shift, &hpa))
	133	return H_TOO_HARD;
	134	}
	135
5ee7af18 AK	136	return H_SUCCESS;
5ee7af18 AK	137	}
5ee7af18 AK	138
	139	/* Note on the use of page_address() in real mode,
	140	*
	141	* It is safe to use page_address() in real mode on ppc64 because
	142	* page_address() is always defined as lowmem_page_address()
	143	* which returns __va(PFN_PHYS(page_to_pfn(page))) which is arithmetic
	144	* operation and does not access page struct.
	145	*
	146	* Theoretically page_address() could be defined different
	147	* but either WANT_PAGE_VIRTUAL or HASHED_PAGE_VIRTUAL
	148	* would have to be enabled.
	149	* WANT_PAGE_VIRTUAL is never enabled on ppc32/ppc64,
	150	* HASHED_PAGE_VIRTUAL could be enabled for ppc32 only and only
	151	* if CONFIG_HIGHMEM is defined. As CONFIG_SPARSEMEM_VMEMMAP
	152	* is not expected to be enabled on ppc32, page_address()
	153	* is safe for ppc32 as well.
	154	*
	155	* WARNING: This will be called in real-mode on HV KVM and virtual
	156	* mode on PR KVM
	157	*/
	158	static u64 kvmppc_page_address(struct page page)
	159	{
	160	#if defined(HASHED_PAGE_VIRTUAL) \|\| defined(WANT_PAGE_VIRTUAL)
	161	#error TODO: fix to avoid page_address() here
	162	#endif
	163	return (u64 *) page_address(page);
	164	}
	165
	166	/*
	167	* Handles TCE requests for emulated devices.
	168	* Puts guest TCE values to the table and expects user space to convert them.
e1a1ef84	169	* Cannot fail so kvmppc_rm_tce_validate must be called before it.
5ee7af18	170	*/
e1a1ef84	171	static void kvmppc_rm_tce_put(struct kvmppc_spapr_tce_table *stt,
5ee7af18 AK	172	unsigned long idx, unsigned long tce)
	173	{
	174	struct page *page;
	175	u64 *tbl;
	176
14f853f1	177	idx -= stt->offset;
5ee7af18	178	page = stt->pages[idx / TCES_PER_PAGE];
e1a1ef84 AK	179	/*
	180	* page must not be NULL in real mode,
	181	* kvmppc_rm_ioba_validate() must have taken care of this.
	182	*/
	183	WARN_ON_ONCE_RM(!page);
5ee7af18 AK	184	tbl = kvmppc_page_address(page);
	185
	186	tbl[idx % TCES_PER_PAGE] = tce;
	187	}
fcbb2ce6	188
e1a1ef84 AK	189	/*
	190	* TCEs pages are allocated in kvmppc_rm_tce_put() which won't be able to do so
	191	* in real mode.
	192	* Check if kvmppc_rm_tce_put() can succeed in real mode, i.e. a TCEs page is
	193	* allocated or not required (when clearing a tce entry).
	194	*/
	195	static long kvmppc_rm_ioba_validate(struct kvmppc_spapr_tce_table *stt,
	196	unsigned long ioba, unsigned long npages, bool clearing)
	197	{
	198	unsigned long i, idx, sttpage, sttpages;
	199	unsigned long ret = kvmppc_ioba_validate(stt, ioba, npages);
	200
	201	if (ret)
	202	return ret;
	203	/*
	204	* clearing==true says kvmppc_rm_tce_put won't be allocating pages
	205	* for empty tces.
	206	*/
	207	if (clearing)
	208	return H_SUCCESS;
	209
	210	idx = (ioba >> stt->page_shift) - stt->offset;
	211	sttpage = idx / TCES_PER_PAGE;
	212	sttpages = _ALIGN_UP(idx % TCES_PER_PAGE + npages, TCES_PER_PAGE) /
	213	TCES_PER_PAGE;
	214	for (i = sttpage; i < sttpage + sttpages; ++i)
	215	if (!stt->pages[i])
	216	return H_TOO_HARD;
	217
	218	return H_SUCCESS;
	219	}
	220
425333bf AK	221	static long iommu_tce_xchg_rm(struct mm_struct mm, struct iommu_table tbl,
	222	unsigned long entry, unsigned long *hpa,
	223	enum dma_data_direction *direction)
	224	{
	225	long ret;
	226
	227	ret = tbl->it_ops->exchange_rm(tbl, entry, hpa, direction);
	228
	229	if (!ret && ((*direction == DMA_FROM_DEVICE) \|\|
	230	(*direction == DMA_BIDIRECTIONAL))) {
6e301a8e	231	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RO(tbl, entry);
425333bf AK	232	/*
	233	* kvmppc_rm_tce_iommu_do_map() updates the UA cache after
	234	* calling this so we still get here a valid UA.
	235	*/
	236	if (pua && *pua)
	237	mm_iommu_ua_mark_dirty_rm(mm, be64_to_cpu(*pua));
	238	}
	239
	240	return ret;
	241	}
	242
	243	static void kvmppc_rm_clear_tce(struct kvm kvm, struct iommu_table tbl,
	244	unsigned long entry)
121f80ba AK	245	{
	246	unsigned long hpa = 0;
	247	enum dma_data_direction dir = DMA_NONE;
	248
425333bf	249	iommu_tce_xchg_rm(kvm->mm, tbl, entry, &hpa, &dir);
121f80ba AK	250	}
	251
	252	static long kvmppc_rm_tce_iommu_mapped_dec(struct kvm *kvm,
	253	struct iommu_table *tbl, unsigned long entry)
	254	{
	255	struct mm_iommu_table_group_mem_t *mem = NULL;
	256	const unsigned long pgsize = 1ULL << tbl->it_page_shift;
6e301a8e	257	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RO(tbl, entry);
121f80ba AK	258
	259	if (!pua)
	260	/* it_userspace allocation might be delayed */
	261	return H_TOO_HARD;
	262
00a5c58d	263	mem = mm_iommu_lookup_rm(kvm->mm, be64_to_cpu(*pua), pgsize);
121f80ba AK	264	if (!mem)
	265	return H_TOO_HARD;
	266
	267	mm_iommu_mapped_dec(mem);
	268
00a5c58d	269	*pua = cpu_to_be64(0);
121f80ba AK	270
	271	return H_SUCCESS;
	272	}
	273
ca1fc489	274	static long kvmppc_rm_tce_iommu_do_unmap(struct kvm *kvm,
121f80ba AK	275	struct iommu_table *tbl, unsigned long entry)
	276	{
	277	enum dma_data_direction dir = DMA_NONE;
	278	unsigned long hpa = 0;
	279	long ret;
	280
425333bf	281	if (iommu_tce_xchg_rm(kvm->mm, tbl, entry, &hpa, &dir))
121f80ba AK	282	/*
	283	* real mode xchg can fail if struct page crosses
	284	* a page boundary
	285	*/
	286	return H_TOO_HARD;
	287
	288	if (dir == DMA_NONE)
	289	return H_SUCCESS;
	290
	291	ret = kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);
	292	if (ret)
425333bf	293	iommu_tce_xchg_rm(kvm->mm, tbl, entry, &hpa, &dir);
121f80ba AK	294
	295	return ret;
	296	}
	297
ca1fc489 AK	298	static long kvmppc_rm_tce_iommu_unmap(struct kvm *kvm,
	299	struct kvmppc_spapr_tce_table stt, struct iommu_table tbl,
	300	unsigned long entry)
	301	{
	302	unsigned long i, ret = H_SUCCESS;
	303	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	304	unsigned long io_entry = entry * subpages;
	305
	306	for (i = 0; i < subpages; ++i) {
	307	ret = kvmppc_rm_tce_iommu_do_unmap(kvm, tbl, io_entry + i);
	308	if (ret != H_SUCCESS)
	309	break;
	310	}
	311
	312	return ret;
	313	}
	314
	315	static long kvmppc_rm_tce_iommu_do_map(struct kvm kvm, struct iommu_table tbl,
121f80ba AK	316	unsigned long entry, unsigned long ua,
	317	enum dma_data_direction dir)
	318	{
	319	long ret;
	320	unsigned long hpa = 0;
6e301a8e	321	__be64 *pua = IOMMU_TABLE_USERSPACE_ENTRY_RO(tbl, entry);
121f80ba AK	322	struct mm_iommu_table_group_mem_t *mem;
	323
	324	if (!pua)
	325	/* it_userspace allocation might be delayed */
	326	return H_TOO_HARD;
	327
	328	mem = mm_iommu_lookup_rm(kvm->mm, ua, 1ULL << tbl->it_page_shift);
	329	if (!mem)
	330	return H_TOO_HARD;
	331
76fa4975 AK	332	if (WARN_ON_ONCE_RM(mm_iommu_ua_to_hpa_rm(mem, ua, tbl->it_page_shift,
76fa4975 AK	333	&hpa)))
f7960e29	334	return H_TOO_HARD;
121f80ba	335
121f80ba	336	if (WARN_ON_ONCE_RM(mm_iommu_mapped_inc(mem)))
f7960e29	337	return H_TOO_HARD;
121f80ba	338
425333bf	339	ret = iommu_tce_xchg_rm(kvm->mm, tbl, entry, &hpa, &dir);
121f80ba AK	340	if (ret) {
	341	mm_iommu_mapped_dec(mem);
	342	/*
	343	* real mode xchg can fail if struct page crosses
	344	* a page boundary
	345	*/
	346	return H_TOO_HARD;
	347	}
	348
	349	if (dir != DMA_NONE)
	350	kvmppc_rm_tce_iommu_mapped_dec(kvm, tbl, entry);
	351
00a5c58d	352	*pua = cpu_to_be64(ua);
121f80ba AK	353
	354	return 0;
	355	}
	356
ca1fc489 AK	357	static long kvmppc_rm_tce_iommu_map(struct kvm *kvm,
	358	struct kvmppc_spapr_tce_table stt, struct iommu_table tbl,
	359	unsigned long entry, unsigned long ua,
	360	enum dma_data_direction dir)
	361	{
	362	unsigned long i, pgoff, ret = H_SUCCESS;
	363	unsigned long subpages = 1ULL << (stt->page_shift - tbl->it_page_shift);
	364	unsigned long io_entry = entry * subpages;
	365
	366	for (i = 0, pgoff = 0; i < subpages;
	367	++i, pgoff += IOMMU_PAGE_SIZE(tbl)) {
	368
	369	ret = kvmppc_rm_tce_iommu_do_map(kvm, tbl,
	370	io_entry + i, ua + pgoff, dir);
	371	if (ret != H_SUCCESS)
	372	break;
	373	}
	374
	375	return ret;
	376	}
	377
31217db7 AK	378	long kvmppc_rm_h_put_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
31217db7 AK	379	unsigned long ioba, unsigned long tce)
54738c09	380	{
503bfcbe	381	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6	382	long ret;
121f80ba AK	383	struct kvmppc_spapr_tce_iommu_table *stit;
	384	unsigned long entry, ua = 0;
	385	enum dma_data_direction dir;
54738c09 DG	386
	387	/* udbg_printf("H_PUT_TCE(): liobn=0x%lx ioba=0x%lx, tce=0x%lx\n", */
	388	/* liobn, ioba, tce); */
	389
acde2572 PM	390	/* For radix, we might be in virtual mode, so punt */
	391	if (kvm_is_radix(vcpu->kvm))
	392	return H_TOO_HARD;
	393
503bfcbe	394	stt = kvmppc_find_table(vcpu->kvm, liobn);
fcbb2ce6 AK	395	if (!stt)
	396	return H_TOO_HARD;
	397
e1a1ef84	398	ret = kvmppc_rm_ioba_validate(stt, ioba, 1, tce == 0);
fcbb2ce6 AK	399	if (ret != H_SUCCESS)
	400	return ret;
	401
42de7b9e	402	ret = kvmppc_rm_tce_validate(stt, tce);
5ee7af18 AK	403	if (ret != H_SUCCESS)
5ee7af18 AK	404	return ret;
fcbb2ce6	405
121f80ba	406	dir = iommu_tce_direction(tce);
2001825e	407	if ((dir != DMA_NONE) && kvmppc_rm_tce_to_ua(vcpu->kvm, tce, &ua, NULL))
121f80ba AK	408	return H_PARAMETER;
	409
	410	entry = ioba >> stt->page_shift;
	411
	412	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
	413	if (dir == DMA_NONE)
ca1fc489	414	ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
121f80ba AK	415	stit->tbl, entry);
121f80ba AK	416	else
ca1fc489	417	ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
121f80ba AK	418	stit->tbl, entry, ua, dir);
121f80ba AK	419
2691f0ff AK	420	if (ret != H_SUCCESS) {
2691f0ff AK	421	kvmppc_rm_clear_tce(vcpu->kvm, stit->tbl, entry);
121f80ba	422	return ret;
2691f0ff	423	}
121f80ba AK	424	}
121f80ba AK	425
e1a1ef84	426	kvmppc_rm_tce_put(stt, entry, tce);
fcbb2ce6 AK	427
fcbb2ce6 AK	428	return H_SUCCESS;
54738c09	429	}
69e9fbb2	430
d3695aa4 AK	431	static long kvmppc_rm_ua_to_hpa(struct kvm_vcpu *vcpu,
	432	unsigned long ua, unsigned long *phpa)
	433	{
	434	pte_t *ptep, pte;
	435	unsigned shift = 0;
	436
94171b19 AK	437	/*
	438	* Called in real mode with MSR_EE = 0. We are safe here.
	439	* It is ok to do the lookup with arch.pgdir here, because
	440	* we are doing this on secondary cpus and current task there
	441	* is not the hypervisor. Also this is safe against THP in the
	442	* host, because an IPI to primary thread will wait for the secondary
	443	* to exit which will agains result in the below page table walk
	444	* to finish.
	445	*/
	446	ptep = __find_linux_pte(vcpu->arch.pgdir, ua, NULL, &shift);
d3695aa4 AK	447	if (!ptep \|\| !pte_present(*ptep))
	448	return -ENXIO;
	449	pte = *ptep;
	450
	451	if (!shift)
	452	shift = PAGE_SHIFT;
	453
	454	/* Avoid handling anything potentially complicated in realmode */
	455	if (shift > PAGE_SHIFT)
	456	return -EAGAIN;
	457
	458	if (!pte_young(pte))
	459	return -EAGAIN;
	460
	461	*phpa = (pte_pfn(pte) << PAGE_SHIFT) \| (ua & ((1ULL << shift) - 1)) \|
	462	(ua & ~PAGE_MASK);
	463
	464	return 0;
	465	}
	466
	467	long kvmppc_rm_h_put_tce_indirect(struct kvm_vcpu *vcpu,
	468	unsigned long liobn, unsigned long ioba,
	469	unsigned long tce_list, unsigned long npages)
	470	{
	471	struct kvmppc_spapr_tce_table *stt;
	472	long i, ret = H_SUCCESS;
	473	unsigned long tces, entry, ua = 0;
	474	unsigned long *rmap = NULL;
da6f59e1	475	bool prereg = false;
121f80ba	476	struct kvmppc_spapr_tce_iommu_table *stit;
d3695aa4	477
acde2572 PM	478	/* For radix, we might be in virtual mode, so punt */
	479	if (kvm_is_radix(vcpu->kvm))
	480	return H_TOO_HARD;
	481
503bfcbe	482	stt = kvmppc_find_table(vcpu->kvm, liobn);
d3695aa4 AK	483	if (!stt)
	484	return H_TOO_HARD;
	485
fe26e527	486	entry = ioba >> stt->page_shift;
d3695aa4 AK	487	/*
	488	* The spec says that the maximum size of the list is 512 TCEs
	489	* so the whole table addressed resides in 4K page
	490	*/
	491	if (npages > 512)
	492	return H_PARAMETER;
	493
	494	if (tce_list & (SZ_4K - 1))
	495	return H_PARAMETER;
	496
e1a1ef84	497	ret = kvmppc_rm_ioba_validate(stt, ioba, npages, false);
d3695aa4 AK	498	if (ret != H_SUCCESS)
	499	return ret;
	500
da6f59e1 AK	501	if (mm_iommu_preregistered(vcpu->kvm->mm)) {
	502	/*
	503	* We get here if guest memory was pre-registered which
	504	* is normally VFIO case and gpa->hpa translation does not
	505	* depend on hpt.
	506	*/
	507	struct mm_iommu_table_group_mem_t *mem;
d3695aa4	508
2001825e	509	if (kvmppc_rm_tce_to_ua(vcpu->kvm, tce_list, &ua, NULL))
da6f59e1	510	return H_TOO_HARD;
d3695aa4	511
da6f59e1 AK	512	mem = mm_iommu_lookup_rm(vcpu->kvm->mm, ua, IOMMU_PAGE_SIZE_4K);
da6f59e1 AK	513	if (mem)
76fa4975 AK	514	prereg = mm_iommu_ua_to_hpa_rm(mem, ua,
76fa4975 AK	515	IOMMU_PAGE_SHIFT_4K, &tces) == 0;
da6f59e1 AK	516	}
	517
	518	if (!prereg) {
	519	/*
	520	* This is usually a case of a guest with emulated devices only
	521	* when TCE list is not in preregistered memory.
	522	* We do not require memory to be preregistered in this case
	523	* so lock rmap and do __find_linux_pte_or_hugepte().
	524	*/
2001825e	525	if (kvmppc_rm_tce_to_ua(vcpu->kvm, tce_list, &ua, &rmap))
da6f59e1 AK	526	return H_TOO_HARD;
	527
	528	rmap = (void *) vmalloc_to_phys(rmap);
121f80ba	529	if (WARN_ON_ONCE_RM(!rmap))
f7960e29	530	return H_TOO_HARD;
da6f59e1 AK	531
	532	/*
	533	* Synchronize with the MMU notifier callbacks in
39c983ea	534	* book3s_64_mmu_hv.c (kvm_unmap_hva_range_hv etc.).
da6f59e1 AK	535	* While we have the rmap lock, code running on other CPUs
	536	* cannot finish unmapping the host real page that backs
	537	* this guest real page, so we are OK to access the host
	538	* real page.
	539	*/
	540	lock_rmap(rmap);
	541	if (kvmppc_rm_ua_to_hpa(vcpu, ua, &tces)) {
	542	ret = H_TOO_HARD;
	543	goto unlock_exit;
	544	}
d3695aa4 AK	545	}
	546
	547	for (i = 0; i < npages; ++i) {
	548	unsigned long tce = be64_to_cpu(((u64 *)tces)[i]);
	549
42de7b9e	550	ret = kvmppc_rm_tce_validate(stt, tce);
d3695aa4 AK	551	if (ret != H_SUCCESS)
d3695aa4 AK	552	goto unlock_exit;
e199ad2b AK	553	}
	554
	555	for (i = 0; i < npages; ++i) {
	556	unsigned long tce = be64_to_cpu(((u64 *)tces)[i]);
d3695aa4	557
121f80ba	558	ua = 0;
2001825e	559	if (kvmppc_rm_tce_to_ua(vcpu->kvm, tce, &ua, NULL))
121f80ba AK	560	return H_PARAMETER;
	561
	562	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
ca1fc489	563	ret = kvmppc_rm_tce_iommu_map(vcpu->kvm, stt,
121f80ba AK	564	stit->tbl, entry + i, ua,
	565	iommu_tce_direction(tce));
	566
2691f0ff AK	567	if (ret != H_SUCCESS) {
	568	kvmppc_rm_clear_tce(vcpu->kvm, stit->tbl,
	569	entry);
121f80ba	570	goto unlock_exit;
2691f0ff	571	}
121f80ba AK	572	}
121f80ba AK	573
e1a1ef84	574	kvmppc_rm_tce_put(stt, entry + i, tce);
d3695aa4 AK	575	}
	576
	577	unlock_exit:
da6f59e1 AK	578	if (rmap)
da6f59e1 AK	579	unlock_rmap(rmap);
d3695aa4 AK	580
	581	return ret;
	582	}
	583
31217db7	584	long kvmppc_rm_h_stuff_tce(struct kvm_vcpu *vcpu,
d3695aa4 AK	585	unsigned long liobn, unsigned long ioba,
	586	unsigned long tce_value, unsigned long npages)
	587	{
	588	struct kvmppc_spapr_tce_table *stt;
	589	long i, ret;
121f80ba	590	struct kvmppc_spapr_tce_iommu_table *stit;
d3695aa4	591
acde2572 PM	592	/* For radix, we might be in virtual mode, so punt */
	593	if (kvm_is_radix(vcpu->kvm))
	594	return H_TOO_HARD;
	595
503bfcbe	596	stt = kvmppc_find_table(vcpu->kvm, liobn);
d3695aa4 AK	597	if (!stt)
	598	return H_TOO_HARD;
	599
e1a1ef84	600	ret = kvmppc_rm_ioba_validate(stt, ioba, npages, tce_value == 0);
d3695aa4 AK	601	if (ret != H_SUCCESS)
	602	return ret;
	603
	604	/* Check permission bits only to allow userspace poison TCE for debug */
	605	if (tce_value & (TCE_PCI_WRITE \| TCE_PCI_READ))
	606	return H_PARAMETER;
	607
121f80ba	608	list_for_each_entry_lockless(stit, &stt->iommu_tables, next) {
c6b61661	609	unsigned long entry = ioba >> stt->page_shift;
121f80ba AK	610
121f80ba AK	611	for (i = 0; i < npages; ++i) {
ca1fc489	612	ret = kvmppc_rm_tce_iommu_unmap(vcpu->kvm, stt,
121f80ba AK	613	stit->tbl, entry + i);
	614
	615	if (ret == H_SUCCESS)
	616	continue;
	617
	618	if (ret == H_TOO_HARD)
	619	return ret;
	620
	621	WARN_ON_ONCE_RM(1);
425333bf	622	kvmppc_rm_clear_tce(vcpu->kvm, stit->tbl, entry);
121f80ba AK	623	}
	624	}
	625
fe26e527	626	for (i = 0; i < npages; ++i, ioba += (1ULL << stt->page_shift))
e1a1ef84	627	kvmppc_rm_tce_put(stt, ioba >> stt->page_shift, tce_value);
d3695aa4 AK	628
	629	return H_SUCCESS;
	630	}
d3695aa4	631
acde2572	632	/* This can be called in either virtual mode or real mode */
69e9fbb2 LD	633	long kvmppc_h_get_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
	634	unsigned long ioba)
	635	{
503bfcbe	636	struct kvmppc_spapr_tce_table *stt;
fcbb2ce6 AK	637	long ret;
	638	unsigned long idx;
	639	struct page *page;
	640	u64 *tbl;
69e9fbb2	641
503bfcbe	642	stt = kvmppc_find_table(vcpu->kvm, liobn);
fcbb2ce6 AK	643	if (!stt)
fcbb2ce6 AK	644	return H_TOO_HARD;
69e9fbb2	645
fcbb2ce6 AK	646	ret = kvmppc_ioba_validate(stt, ioba, 1);
	647	if (ret != H_SUCCESS)
	648	return ret;
69e9fbb2	649
14f853f1	650	idx = (ioba >> stt->page_shift) - stt->offset;
fcbb2ce6	651	page = stt->pages[idx / TCES_PER_PAGE];
e1a1ef84 AK	652	if (!page) {
	653	vcpu->arch.regs.gpr[4] = 0;
	654	return H_SUCCESS;
	655	}
fcbb2ce6	656	tbl = (u64 *)page_address(page);
69e9fbb2	657
1143a706	658	vcpu->arch.regs.gpr[4] = tbl[idx % TCES_PER_PAGE];
69e9fbb2	659
fcbb2ce6	660	return H_SUCCESS;
69e9fbb2 LD	661	}
69e9fbb2 LD	662	EXPORT_SYMBOL_GPL(kvmppc_h_get_tce);
d3695aa4 AK	663
d3695aa4 AK	664	#endif /* KVM_BOOK3S_HV_POSSIBLE */