2019-05-27 13:55:06 +07:00
|
|
|
/* SPDX-License-Identifier: GPL-2.0-or-later */
|
2017-09-09 06:11:23 +07:00
|
|
|
/*
|
|
|
|
* Copyright 2013 Red Hat Inc.
|
|
|
|
*
|
2018-10-31 05:04:06 +07:00
|
|
|
* Authors: Jérôme Glisse <jglisse@redhat.com>
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
|
|
|
/*
|
|
|
|
* Heterogeneous Memory Management (HMM)
|
|
|
|
*
|
2018-03-22 02:22:47 +07:00
|
|
|
* See Documentation/vm/hmm.rst for reasons and overview of what HMM is and it
|
2017-09-09 06:11:23 +07:00
|
|
|
* is for. Here we focus on the HMM API description, with some explanation of
|
|
|
|
* the underlying implementation.
|
|
|
|
*
|
|
|
|
* Short description: HMM provides a set of helpers to share a virtual address
|
|
|
|
* space between CPU and a device, so that the device can access any valid
|
|
|
|
* address of the process (while still obeying memory protection). HMM also
|
|
|
|
* provides helpers to migrate process memory to device memory, and back. Each
|
|
|
|
* set of functionality (address space mirroring, and migration to and from
|
|
|
|
* device memory) can be used independently of the other.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* HMM address space mirroring API:
|
|
|
|
*
|
2019-05-07 06:29:39 +07:00
|
|
|
* Use HMM address space mirroring if you want to mirror a range of the CPU
|
|
|
|
* page tables of a process into a device page table. Here, "mirror" means "keep
|
2017-09-09 06:11:23 +07:00
|
|
|
* synchronized". Prerequisites: the device must provide the ability to write-
|
|
|
|
* protect its page tables (at PAGE_SIZE granularity), and must be able to
|
|
|
|
* recover from the resulting potential page faults.
|
|
|
|
*
|
|
|
|
* HMM guarantees that at any point in time, a given virtual address points to
|
|
|
|
* either the same memory in both CPU and device page tables (that is: CPU and
|
|
|
|
* device page tables each point to the same pages), or that one page table (CPU
|
|
|
|
* or device) points to no entry, while the other still points to the old page
|
|
|
|
* for the address. The latter case happens when the CPU page table update
|
|
|
|
* happens first, and then the update is mirrored over to the device page table.
|
|
|
|
* This does not cause any issue, because the CPU page table cannot start
|
|
|
|
* pointing to a new page until the device page table is invalidated.
|
|
|
|
*
|
|
|
|
* HMM uses mmu_notifiers to monitor the CPU page tables, and forwards any
|
|
|
|
* updates to each device driver that has registered a mirror. It also provides
|
|
|
|
* some API calls to help with taking a snapshot of the CPU page table, and to
|
|
|
|
* synchronize with any updates that might happen concurrently.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* HMM migration to and from device memory:
|
|
|
|
*
|
|
|
|
* HMM provides a set of helpers to hotplug device memory as ZONE_DEVICE, with
|
|
|
|
* a new MEMORY_DEVICE_PRIVATE type. This provides a struct page for each page
|
|
|
|
* of the device memory, and allows the device driver to manage its memory
|
|
|
|
* using those struct pages. Having struct pages for device memory makes
|
|
|
|
* migration easier. Because that memory is not addressable by the CPU it must
|
|
|
|
* never be pinned to the device; in other words, any CPU page fault can always
|
|
|
|
* cause the device memory to be migrated (copied/moved) back to regular memory.
|
|
|
|
*
|
|
|
|
* A new migrate helper (migrate_vma()) has been added (see mm/migrate.c) that
|
|
|
|
* allows use of a device DMA engine to perform the copy operation between
|
|
|
|
* regular system memory and device memory.
|
|
|
|
*/
|
|
|
|
#ifndef LINUX_HMM_H
|
|
|
|
#define LINUX_HMM_H
|
|
|
|
|
|
|
|
#include <linux/kconfig.h>
|
2018-12-28 15:39:46 +07:00
|
|
|
#include <asm/pgtable.h>
|
2017-09-09 06:11:23 +07:00
|
|
|
|
2017-09-09 06:12:02 +07:00
|
|
|
#include <linux/device.h>
|
2017-09-09 06:11:58 +07:00
|
|
|
#include <linux/migrate.h>
|
|
|
|
#include <linux/memremap.h>
|
|
|
|
#include <linux/completion.h>
|
2019-05-14 07:20:01 +07:00
|
|
|
#include <linux/mmu_notifier.h>
|
2017-09-09 06:11:58 +07:00
|
|
|
|
2017-09-09 06:11:23 +07:00
|
|
|
/*
|
2018-04-11 06:29:06 +07:00
|
|
|
* hmm_pfn_flag_e - HMM flag enums
|
|
|
|
*
|
2017-09-09 06:11:23 +07:00
|
|
|
* Flags:
|
2018-04-11 06:28:34 +07:00
|
|
|
* HMM_PFN_VALID: pfn is valid. It has, at least, read permission.
|
2017-09-09 06:11:23 +07:00
|
|
|
* HMM_PFN_WRITE: CPU page table has write permission set
|
2018-04-11 06:29:06 +07:00
|
|
|
* HMM_PFN_DEVICE_PRIVATE: private device memory (ZONE_DEVICE)
|
|
|
|
*
|
2019-05-07 06:29:39 +07:00
|
|
|
* The driver provides a flags array for mapping page protections to device
|
|
|
|
* PTE bits. If the driver valid bit for an entry is bit 3,
|
|
|
|
* i.e., (entry & (1 << 3)), then the driver must provide
|
2018-04-11 06:29:06 +07:00
|
|
|
* an array in hmm_range.flags with hmm_range.flags[HMM_PFN_VALID] == 1 << 3.
|
2019-05-07 06:29:39 +07:00
|
|
|
* Same logic apply to all flags. This is the same idea as vm_page_prot in vma
|
2018-04-11 06:29:06 +07:00
|
|
|
* except that this is per device driver rather than per architecture.
|
|
|
|
*/
|
|
|
|
enum hmm_pfn_flag_e {
|
|
|
|
HMM_PFN_VALID = 0,
|
|
|
|
HMM_PFN_WRITE,
|
|
|
|
HMM_PFN_DEVICE_PRIVATE,
|
|
|
|
HMM_PFN_FLAG_MAX
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* hmm_pfn_value_e - HMM pfn special value
|
|
|
|
*
|
|
|
|
* Flags:
|
2017-09-09 06:11:31 +07:00
|
|
|
* HMM_PFN_ERROR: corresponding CPU page table entry points to poisoned memory
|
2018-04-11 06:29:06 +07:00
|
|
|
* HMM_PFN_NONE: corresponding CPU page table entry is pte_none()
|
2017-09-09 06:11:31 +07:00
|
|
|
* HMM_PFN_SPECIAL: corresponding CPU page table entry is special; i.e., the
|
2018-10-27 05:04:26 +07:00
|
|
|
* result of vmf_insert_pfn() or vm_insert_page(). Therefore, it should not
|
2017-09-09 06:11:31 +07:00
|
|
|
* be mirrored by a device, because the entry will never have HMM_PFN_VALID
|
|
|
|
* set and the pfn value is undefined.
|
2018-04-11 06:29:06 +07:00
|
|
|
*
|
2019-05-07 06:29:39 +07:00
|
|
|
* Driver provides values for none entry, error entry, and special entry.
|
|
|
|
* Driver can alias (i.e., use same value) error and special, but
|
|
|
|
* it should not alias none with error or special.
|
2018-04-11 06:29:06 +07:00
|
|
|
*
|
|
|
|
* HMM pfn value returned by hmm_vma_get_pfns() or hmm_vma_fault() will be:
|
|
|
|
* hmm_range.values[HMM_PFN_ERROR] if CPU page table entry is poisonous,
|
2019-05-07 06:29:39 +07:00
|
|
|
* hmm_range.values[HMM_PFN_NONE] if there is no CPU page table entry,
|
2018-04-11 06:29:06 +07:00
|
|
|
* hmm_range.values[HMM_PFN_SPECIAL] if CPU page table entry is a special one
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
2018-04-11 06:29:06 +07:00
|
|
|
enum hmm_pfn_value_e {
|
|
|
|
HMM_PFN_ERROR,
|
|
|
|
HMM_PFN_NONE,
|
|
|
|
HMM_PFN_SPECIAL,
|
|
|
|
HMM_PFN_VALUE_MAX
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* struct hmm_range - track invalidation lock on virtual address range
|
|
|
|
*
|
2019-11-13 03:22:30 +07:00
|
|
|
* @notifier: a mmu_interval_notifier that includes the start/end
|
|
|
|
* @notifier_seq: result of mmu_interval_read_begin()
|
2019-05-14 07:19:48 +07:00
|
|
|
* @hmm: the core HMM structure this range is active against
|
2018-04-11 06:29:06 +07:00
|
|
|
* @vma: the vm area struct for the range
|
|
|
|
* @list: all range lock are on a list
|
|
|
|
* @start: range virtual start address (inclusive)
|
|
|
|
* @end: range virtual end address (exclusive)
|
|
|
|
* @pfns: array of pfns (big enough for the range)
|
|
|
|
* @flags: pfn flags to match device driver page table
|
|
|
|
* @values: pfn value for some special case (none, special, error, ...)
|
2019-05-14 07:20:05 +07:00
|
|
|
* @default_flags: default flags for the range (write, read, ... see hmm doc)
|
|
|
|
* @pfn_flags_mask: allows to mask pfn flags so that only default_flags matter
|
2018-04-11 06:29:06 +07:00
|
|
|
* @pfn_shifts: pfn shift value (should be <= PAGE_SHIFT)
|
|
|
|
* @valid: pfns array did not change since it has been fill by an HMM function
|
|
|
|
*/
|
|
|
|
struct hmm_range {
|
2019-11-13 03:22:20 +07:00
|
|
|
struct mmu_interval_notifier *notifier;
|
|
|
|
unsigned long notifier_seq;
|
2018-04-11 06:29:06 +07:00
|
|
|
unsigned long start;
|
|
|
|
unsigned long end;
|
|
|
|
uint64_t *pfns;
|
|
|
|
const uint64_t *flags;
|
|
|
|
const uint64_t *values;
|
2019-05-14 07:20:05 +07:00
|
|
|
uint64_t default_flags;
|
|
|
|
uint64_t pfn_flags_mask;
|
2018-04-11 06:29:06 +07:00
|
|
|
uint8_t pfn_shift;
|
|
|
|
};
|
2017-09-09 06:11:23 +07:00
|
|
|
|
|
|
|
/*
|
2019-05-14 07:20:31 +07:00
|
|
|
* hmm_device_entry_to_page() - return struct page pointed to by a device entry
|
|
|
|
* @range: range use to decode device entry value
|
|
|
|
* @entry: device entry value to get corresponding struct page from
|
2019-05-07 06:29:39 +07:00
|
|
|
* Return: struct page pointer if entry is a valid, NULL otherwise
|
2017-09-09 06:11:23 +07:00
|
|
|
*
|
2019-05-14 07:20:31 +07:00
|
|
|
* If the device entry is valid (ie valid flag set) then return the struct page
|
|
|
|
* matching the entry value. Otherwise return NULL.
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
2019-05-14 07:20:31 +07:00
|
|
|
static inline struct page *hmm_device_entry_to_page(const struct hmm_range *range,
|
|
|
|
uint64_t entry)
|
2017-09-09 06:11:23 +07:00
|
|
|
{
|
2019-05-14 07:20:31 +07:00
|
|
|
if (entry == range->values[HMM_PFN_NONE])
|
2018-04-11 06:29:06 +07:00
|
|
|
return NULL;
|
2019-05-14 07:20:31 +07:00
|
|
|
if (entry == range->values[HMM_PFN_ERROR])
|
2018-04-11 06:29:06 +07:00
|
|
|
return NULL;
|
2019-05-14 07:20:31 +07:00
|
|
|
if (entry == range->values[HMM_PFN_SPECIAL])
|
2017-09-09 06:11:23 +07:00
|
|
|
return NULL;
|
2019-05-14 07:20:31 +07:00
|
|
|
if (!(entry & range->flags[HMM_PFN_VALID]))
|
2018-04-11 06:29:06 +07:00
|
|
|
return NULL;
|
2019-05-14 07:20:31 +07:00
|
|
|
return pfn_to_page(entry >> range->pfn_shift);
|
2017-09-09 06:11:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2019-05-14 07:20:31 +07:00
|
|
|
* hmm_device_entry_to_pfn() - return pfn value store in a device entry
|
|
|
|
* @range: range use to decode device entry value
|
|
|
|
* @entry: device entry to extract pfn from
|
2019-05-07 06:29:39 +07:00
|
|
|
* Return: pfn value if device entry is valid, -1UL otherwise
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
2019-05-14 07:20:31 +07:00
|
|
|
static inline unsigned long
|
|
|
|
hmm_device_entry_to_pfn(const struct hmm_range *range, uint64_t pfn)
|
2017-09-09 06:11:23 +07:00
|
|
|
{
|
2018-04-11 06:29:06 +07:00
|
|
|
if (pfn == range->values[HMM_PFN_NONE])
|
|
|
|
return -1UL;
|
|
|
|
if (pfn == range->values[HMM_PFN_ERROR])
|
|
|
|
return -1UL;
|
|
|
|
if (pfn == range->values[HMM_PFN_SPECIAL])
|
2017-09-09 06:11:23 +07:00
|
|
|
return -1UL;
|
2018-04-11 06:29:06 +07:00
|
|
|
if (!(pfn & range->flags[HMM_PFN_VALID]))
|
|
|
|
return -1UL;
|
|
|
|
return (pfn >> range->pfn_shift);
|
2017-09-09 06:11:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2019-05-14 07:20:31 +07:00
|
|
|
* hmm_device_entry_from_page() - create a valid device entry for a page
|
2018-04-11 06:29:06 +07:00
|
|
|
* @range: range use to encode HMM pfn value
|
2019-05-14 07:20:31 +07:00
|
|
|
* @page: page for which to create the device entry
|
2019-05-07 06:29:39 +07:00
|
|
|
* Return: valid device entry for the page
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
2019-05-14 07:20:31 +07:00
|
|
|
static inline uint64_t hmm_device_entry_from_page(const struct hmm_range *range,
|
|
|
|
struct page *page)
|
2017-09-09 06:11:23 +07:00
|
|
|
{
|
2018-04-11 06:29:06 +07:00
|
|
|
return (page_to_pfn(page) << range->pfn_shift) |
|
|
|
|
range->flags[HMM_PFN_VALID];
|
2017-09-09 06:11:23 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2019-05-14 07:20:31 +07:00
|
|
|
* hmm_device_entry_from_pfn() - create a valid device entry value from pfn
|
2018-04-11 06:29:06 +07:00
|
|
|
* @range: range use to encode HMM pfn value
|
2019-05-14 07:20:31 +07:00
|
|
|
* @pfn: pfn value for which to create the device entry
|
2019-05-07 06:29:39 +07:00
|
|
|
* Return: valid device entry for the pfn
|
2017-09-09 06:11:23 +07:00
|
|
|
*/
|
2019-05-14 07:20:31 +07:00
|
|
|
static inline uint64_t hmm_device_entry_from_pfn(const struct hmm_range *range,
|
|
|
|
unsigned long pfn)
|
2017-09-09 06:11:23 +07:00
|
|
|
{
|
2018-04-11 06:29:06 +07:00
|
|
|
return (pfn << range->pfn_shift) |
|
|
|
|
range->flags[HMM_PFN_VALID];
|
2017-09-09 06:11:23 +07:00
|
|
|
}
|
|
|
|
|
2019-11-13 03:22:21 +07:00
|
|
|
/*
|
|
|
|
* Retry fault if non-blocking, drop mmap_sem and return -EAGAIN in that case.
|
|
|
|
*/
|
|
|
|
#define HMM_FAULT_ALLOW_RETRY (1 << 0)
|
|
|
|
|
|
|
|
/* Don't fault in missing PTEs, just snapshot the current state. */
|
|
|
|
#define HMM_FAULT_SNAPSHOT (1 << 1)
|
|
|
|
|
|
|
|
#ifdef CONFIG_HMM_MIRROR
|
2017-09-09 06:11:31 +07:00
|
|
|
/*
|
2019-05-14 07:20:01 +07:00
|
|
|
* Please see Documentation/vm/hmm.rst for how to use the range API.
|
2017-09-09 06:11:31 +07:00
|
|
|
*/
|
2019-07-26 07:56:46 +07:00
|
|
|
long hmm_range_fault(struct hmm_range *range, unsigned int flags);
|
2019-11-13 03:22:21 +07:00
|
|
|
#else
|
|
|
|
static inline long hmm_range_fault(struct hmm_range *range, unsigned int flags)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
|
|
|
#endif
|
2017-09-09 06:11:35 +07:00
|
|
|
|
|
|
|
/*
|
2019-05-14 07:20:01 +07:00
|
|
|
* HMM_RANGE_DEFAULT_TIMEOUT - default timeout (ms) when waiting for a range
|
2017-09-09 06:11:35 +07:00
|
|
|
*
|
2019-05-14 07:20:01 +07:00
|
|
|
* When waiting for mmu notifiers we need some kind of time out otherwise we
|
|
|
|
* could potentialy wait for ever, 1000ms ie 1s sounds like a long time to
|
|
|
|
* wait already.
|
2017-09-09 06:11:35 +07:00
|
|
|
*/
|
2019-05-14 07:20:01 +07:00
|
|
|
#define HMM_RANGE_DEFAULT_TIMEOUT 1000
|
|
|
|
|
2017-09-09 06:11:23 +07:00
|
|
|
#endif /* LINUX_HMM_H */
|