Merge tag 'for-linus-hmm' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma (39bed42d) · Commits · EulixOS / Software / Kernel

Documentation/vm/hmm.rst

+10 −10

Original line number	Original line	Diff line number	Diff line
	@@ -149,14 +149,14 @@ CPU page table into a device page table; HMM helps keep both synchronized. A
	device driver that wants to mirror a process address space must start with the		device driver that wants to mirror a process address space must start with the
	registration of a mmu_interval_notifier::		registration of a mmu_interval_notifier::

	mni->ops = &driver_ops;		int mmu_interval_notifier_insert(struct mmu_interval_notifier *interval_sub,
	int mmu_interval_notifier_insert(struct mmu_interval_notifier *mni,		struct mm_struct *mm, unsigned long start,
	unsigned long start, unsigned long length,		unsigned long length,
	struct mm_struct *mm);		const struct mmu_interval_notifier_ops *ops);

	During the driver_ops->invalidate() callback the device driver must perform		During the ops->invalidate() callback the device driver must perform the
	the update action to the range (mark range read only, or fully unmap,		update action to the range (mark range read only, or fully unmap, etc.). The
	etc.). The device must complete the update before the driver callback returns.		device must complete the update before the driver callback returns.

	When the device driver wants to populate a range of virtual addresses, it can		When the device driver wants to populate a range of virtual addresses, it can
	use::		use::
	@@ -183,7 +183,7 @@ The usage pattern is::
	struct hmm_range range;		struct hmm_range range;
	...		...

	range.notifier = &mni;		range.notifier = &interval_sub;
	range.start = ...;		range.start = ...;
	range.end = ...;		range.end = ...;
	range.pfns = ...;		range.pfns = ...;
	@@ -191,11 +191,11 @@ The usage pattern is::
	range.values = ...;		range.values = ...;
	range.pfn_shift = ...;		range.pfn_shift = ...;

	if (!mmget_not_zero(mni->notifier.mm))		if (!mmget_not_zero(interval_sub->notifier.mm))
	return -EFAULT;		return -EFAULT;

	again:		again:
	range.notifier_seq = mmu_interval_read_begin(&mni);		range.notifier_seq = mmu_interval_read_begin(&interval_sub);
	down_read(&mm->mmap_sem);		down_read(&mm->mmap_sem);
	ret = hmm_range_fault(&range, HMM_RANGE_SNAPSHOT);		ret = hmm_range_fault(&range, HMM_RANGE_SNAPSHOT);
	if (ret) {		if (ret) {

include/linux/mm_types.h

+1 −1

Original line number	Original line	Diff line number	Diff line
	@@ -490,7 +490,7 @@ struct mm_struct {
	/* store ref to file /proc/<pid>/exe symlink points to */		/* store ref to file /proc/<pid>/exe symlink points to */
	struct file __rcu *exe_file;		struct file __rcu *exe_file;
	#ifdef CONFIG_MMU_NOTIFIER		#ifdef CONFIG_MMU_NOTIFIER
	struct mmu_notifier_mm *mmu_notifier_mm;		struct mmu_notifier_subscriptions *notifier_subscriptions;
	#endif		#endif
	#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS		#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS
	pgtable_t pmd_huge_pte; /* protected by page_table_lock */		pgtable_t pmd_huge_pte; /* protected by page_table_lock */

include/linux/mmu_notifier.h

+46 −40

Original line number	Original line	Diff line number	Diff line
	@@ -8,7 +8,7 @@
	#include <linux/srcu.h>		#include <linux/srcu.h>
	#include <linux/interval_tree.h>		#include <linux/interval_tree.h>

	struct mmu_notifier_mm;		struct mmu_notifier_subscriptions;
	struct mmu_notifier;		struct mmu_notifier;
	struct mmu_notifier_range;		struct mmu_notifier_range;
	struct mmu_interval_notifier;		struct mmu_interval_notifier;
	@@ -73,7 +73,7 @@ struct mmu_notifier_ops {
	* through the gart alias address, so leading to memory		* through the gart alias address, so leading to memory
	* corruption.		* corruption.
	*/		*/
	void (release)(struct mmu_notifier mn,		void (release)(struct mmu_notifier subscription,
	struct mm_struct *mm);		struct mm_struct *mm);

	/*		/*
	@@ -85,7 +85,7 @@ struct mmu_notifier_ops {
	* Start-end is necessary in case the secondary MMU is mapping the page		* Start-end is necessary in case the secondary MMU is mapping the page
	* at a smaller granularity than the primary MMU.		* at a smaller granularity than the primary MMU.
	*/		*/
	int (clear_flush_young)(struct mmu_notifier mn,		int (clear_flush_young)(struct mmu_notifier subscription,
	struct mm_struct *mm,		struct mm_struct *mm,
	unsigned long start,		unsigned long start,
	unsigned long end);		unsigned long end);
	@@ -95,7 +95,7 @@ struct mmu_notifier_ops {
	* latter, it is supposed to test-and-clear the young/accessed bitflag		* latter, it is supposed to test-and-clear the young/accessed bitflag
	* in the secondary pte, but it may omit flushing the secondary tlb.		* in the secondary pte, but it may omit flushing the secondary tlb.
	*/		*/
	int (clear_young)(struct mmu_notifier mn,		int (clear_young)(struct mmu_notifier subscription,
	struct mm_struct *mm,		struct mm_struct *mm,
	unsigned long start,		unsigned long start,
	unsigned long end);		unsigned long end);
	@@ -106,7 +106,7 @@ struct mmu_notifier_ops {
	* frequently used without actually clearing the flag or tearing		* frequently used without actually clearing the flag or tearing
	* down the secondary mapping on the page.		* down the secondary mapping on the page.
	*/		*/
	int (test_young)(struct mmu_notifier mn,		int (test_young)(struct mmu_notifier subscription,
	struct mm_struct *mm,		struct mm_struct *mm,
	unsigned long address);		unsigned long address);

	@@ -114,7 +114,7 @@ struct mmu_notifier_ops {
	* change_pte is called in cases that pte mapping to page is changed:		* change_pte is called in cases that pte mapping to page is changed:
	* for example, when ksm remaps pte to point to a new shared page.		* for example, when ksm remaps pte to point to a new shared page.
	*/		*/
	void (change_pte)(struct mmu_notifier mn,		void (change_pte)(struct mmu_notifier subscription,
	struct mm_struct *mm,		struct mm_struct *mm,
	unsigned long address,		unsigned long address,
	pte_t pte);		pte_t pte);
	@@ -169,9 +169,9 @@ struct mmu_notifier_ops {
	* invalidate_range_end.		* invalidate_range_end.
	*		*
	*/		*/
	int (invalidate_range_start)(struct mmu_notifier mn,		int (invalidate_range_start)(struct mmu_notifier subscription,
	const struct mmu_notifier_range *range);		const struct mmu_notifier_range *range);
	void (invalidate_range_end)(struct mmu_notifier mn,		void (invalidate_range_end)(struct mmu_notifier subscription,
	const struct mmu_notifier_range *range);		const struct mmu_notifier_range *range);

	/*		/*
	@@ -192,8 +192,10 @@ struct mmu_notifier_ops {
	* of what was passed to invalidate_range_start()/end(), if		* of what was passed to invalidate_range_start()/end(), if
	* called between those functions.		* called between those functions.
	*/		*/
	void (invalidate_range)(struct mmu_notifier mn, struct mm_struct *mm,		void (invalidate_range)(struct mmu_notifier subscription,
	unsigned long start, unsigned long end);		struct mm_struct *mm,
			unsigned long start,
			unsigned long end);

	/*		/*
	* These callbacks are used with the get/put interface to manage the		* These callbacks are used with the get/put interface to manage the
	@@ -206,7 +208,7 @@ struct mmu_notifier_ops {
	* and cannot sleep.		* and cannot sleep.
	*/		*/
	struct mmu_notifier (alloc_notifier)(struct mm_struct *mm);		struct mmu_notifier (alloc_notifier)(struct mm_struct *mm);
	void (free_notifier)(struct mmu_notifier mn);		void (free_notifier)(struct mmu_notifier subscription);
	};		};

	/*		/*
	@@ -235,7 +237,7 @@ struct mmu_notifier {
	* was required but mmu_notifier_range_blockable(range) is false.		* was required but mmu_notifier_range_blockable(range) is false.
	*/		*/
	struct mmu_interval_notifier_ops {		struct mmu_interval_notifier_ops {
	bool (invalidate)(struct mmu_interval_notifier mni,		bool (invalidate)(struct mmu_interval_notifier interval_sub,
	const struct mmu_notifier_range *range,		const struct mmu_notifier_range *range,
	unsigned long cur_seq);		unsigned long cur_seq);
	};		};
	@@ -265,7 +267,7 @@ struct mmu_notifier_range {

	static inline int mm_has_notifiers(struct mm_struct *mm)		static inline int mm_has_notifiers(struct mm_struct *mm)
	{		{
	return unlikely(mm->mmu_notifier_mm);		return unlikely(mm->notifier_subscriptions);
	}		}

	struct mmu_notifier mmu_notifier_get_locked(const struct mmu_notifier_ops ops,		struct mmu_notifier mmu_notifier_get_locked(const struct mmu_notifier_ops ops,
	@@ -280,30 +282,31 @@ mmu_notifier_get(const struct mmu_notifier_ops ops, struct mm_struct mm)
	up_write(&mm->mmap_sem);		up_write(&mm->mmap_sem);
	return ret;		return ret;
	}		}
	void mmu_notifier_put(struct mmu_notifier *mn);		void mmu_notifier_put(struct mmu_notifier *subscription);
	void mmu_notifier_synchronize(void);		void mmu_notifier_synchronize(void);

	extern int mmu_notifier_register(struct mmu_notifier *mn,		extern int mmu_notifier_register(struct mmu_notifier *subscription,
	struct mm_struct *mm);		struct mm_struct *mm);
	extern int __mmu_notifier_register(struct mmu_notifier *mn,		extern int __mmu_notifier_register(struct mmu_notifier *subscription,
	struct mm_struct *mm);		struct mm_struct *mm);
	extern void mmu_notifier_unregister(struct mmu_notifier *mn,		extern void mmu_notifier_unregister(struct mmu_notifier *subscription,
	struct mm_struct *mm);		struct mm_struct *mm);

	unsigned long mmu_interval_read_begin(struct mmu_interval_notifier *mni);		unsigned long
	int mmu_interval_notifier_insert(struct mmu_interval_notifier *mni,		mmu_interval_read_begin(struct mmu_interval_notifier *interval_sub);
			int mmu_interval_notifier_insert(struct mmu_interval_notifier *interval_sub,
	struct mm_struct *mm, unsigned long start,		struct mm_struct *mm, unsigned long start,
	unsigned long length,		unsigned long length,
	const struct mmu_interval_notifier_ops *ops);		const struct mmu_interval_notifier_ops *ops);
	int mmu_interval_notifier_insert_locked(		int mmu_interval_notifier_insert_locked(
	struct mmu_interval_notifier mni, struct mm_struct mm,		struct mmu_interval_notifier interval_sub, struct mm_struct mm,
	unsigned long start, unsigned long length,		unsigned long start, unsigned long length,
	const struct mmu_interval_notifier_ops *ops);		const struct mmu_interval_notifier_ops *ops);
	void mmu_interval_notifier_remove(struct mmu_interval_notifier *mni);		void mmu_interval_notifier_remove(struct mmu_interval_notifier *interval_sub);

	/**		/**
	* mmu_interval_set_seq - Save the invalidation sequence		* mmu_interval_set_seq - Save the invalidation sequence
	* @mni - The mni passed to invalidate		* @interval_sub - The subscription passed to invalidate
	* @cur_seq - The cur_seq passed to the invalidate() callback		* @cur_seq - The cur_seq passed to the invalidate() callback
	*		*
	* This must be called unconditionally from the invalidate callback of a		* This must be called unconditionally from the invalidate callback of a
	@@ -314,15 +317,16 @@ void mmu_interval_notifier_remove(struct mmu_interval_notifier *mni);
	* If the caller does not call mmu_interval_read_begin() or		* If the caller does not call mmu_interval_read_begin() or
	* mmu_interval_read_retry() then this call is not required.		* mmu_interval_read_retry() then this call is not required.
	*/		*/
	static inline void mmu_interval_set_seq(struct mmu_interval_notifier *mni,		static inline void
			mmu_interval_set_seq(struct mmu_interval_notifier *interval_sub,
	unsigned long cur_seq)		unsigned long cur_seq)
	{		{
	WRITE_ONCE(mni->invalidate_seq, cur_seq);		WRITE_ONCE(interval_sub->invalidate_seq, cur_seq);
	}		}

	/**		/**
	* mmu_interval_read_retry - End a read side critical section against a VA range		* mmu_interval_read_retry - End a read side critical section against a VA range
	* mni: The range		* interval_sub: The subscription
	* seq: The return of the paired mmu_interval_read_begin()		* seq: The return of the paired mmu_interval_read_begin()
	*		*
	* This MUST be called under a user provided lock that is also held		* This MUST be called under a user provided lock that is also held
	@@ -334,15 +338,16 @@ static inline void mmu_interval_set_seq(struct mmu_interval_notifier *mni,
	* Returns true if an invalidation collided with this critical section, and		* Returns true if an invalidation collided with this critical section, and
	* the caller should retry.		* the caller should retry.
	*/		*/
	static inline bool mmu_interval_read_retry(struct mmu_interval_notifier *mni,		static inline bool
			mmu_interval_read_retry(struct mmu_interval_notifier *interval_sub,
	unsigned long seq)		unsigned long seq)
	{		{
	return mni->invalidate_seq != seq;		return interval_sub->invalidate_seq != seq;
	}		}

	/**		/**
	* mmu_interval_check_retry - Test if a collision has occurred		* mmu_interval_check_retry - Test if a collision has occurred
	* mni: The range		* interval_sub: The subscription
	* seq: The return of the matching mmu_interval_read_begin()		* seq: The return of the matching mmu_interval_read_begin()
	*		*
	* This can be used in the critical section between mmu_interval_read_begin()		* This can be used in the critical section between mmu_interval_read_begin()
	@@ -357,14 +362,15 @@ static inline bool mmu_interval_read_retry(struct mmu_interval_notifier *mni,
	* This call can be used as part of loops and other expensive operations to		* This call can be used as part of loops and other expensive operations to
	* expedite a retry.		* expedite a retry.
	*/		*/
	static inline bool mmu_interval_check_retry(struct mmu_interval_notifier *mni,		static inline bool
			mmu_interval_check_retry(struct mmu_interval_notifier *interval_sub,
	unsigned long seq)		unsigned long seq)
	{		{
	/* Pairs with the WRITE_ONCE in mmu_interval_set_seq() */		/* Pairs with the WRITE_ONCE in mmu_interval_set_seq() */
	return READ_ONCE(mni->invalidate_seq) != seq;		return READ_ONCE(interval_sub->invalidate_seq) != seq;
	}		}

	extern void __mmu_notifier_mm_destroy(struct mm_struct *mm);		extern void __mmu_notifier_subscriptions_destroy(struct mm_struct *mm);
	extern void __mmu_notifier_release(struct mm_struct *mm);		extern void __mmu_notifier_release(struct mm_struct *mm);
	extern int __mmu_notifier_clear_flush_young(struct mm_struct *mm,		extern int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
	unsigned long start,		unsigned long start,
	@@ -480,15 +486,15 @@ static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
	__mmu_notifier_invalidate_range(mm, start, end);		__mmu_notifier_invalidate_range(mm, start, end);
	}		}

	static inline void mmu_notifier_mm_init(struct mm_struct *mm)		static inline void mmu_notifier_subscriptions_init(struct mm_struct *mm)
	{		{
	mm->mmu_notifier_mm = NULL;		mm->notifier_subscriptions = NULL;
	}		}

	static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)		static inline void mmu_notifier_subscriptions_destroy(struct mm_struct *mm)
	{		{
	if (mm_has_notifiers(mm))		if (mm_has_notifiers(mm))
	__mmu_notifier_mm_destroy(mm);		__mmu_notifier_subscriptions_destroy(mm);
	}		}


	@@ -692,11 +698,11 @@ static inline void mmu_notifier_invalidate_range(struct mm_struct *mm,
	{		{
	}		}

	static inline void mmu_notifier_mm_init(struct mm_struct *mm)		static inline void mmu_notifier_subscriptions_init(struct mm_struct *mm)
	{		{
	}		}

	static inline void mmu_notifier_mm_destroy(struct mm_struct *mm)		static inline void mmu_notifier_subscriptions_destroy(struct mm_struct *mm)
	{		{
	}		}

kernel/fork.c

+2 −2

Original line number	Original line	Diff line number	Diff line
	@@ -692,7 +692,7 @@ void __mmdrop(struct mm_struct *mm)
	WARN_ON_ONCE(mm == current->active_mm);		WARN_ON_ONCE(mm == current->active_mm);
	mm_free_pgd(mm);		mm_free_pgd(mm);
	destroy_context(mm);		destroy_context(mm);
	mmu_notifier_mm_destroy(mm);		mmu_notifier_subscriptions_destroy(mm);
	check_mm(mm);		check_mm(mm);
	put_user_ns(mm->user_ns);		put_user_ns(mm->user_ns);
	free_mm(mm);		free_mm(mm);
	@@ -1025,7 +1025,7 @@ static struct mm_struct mm_init(struct mm_struct mm, struct task_struct *p,
	mm_init_aio(mm);		mm_init_aio(mm);
	mm_init_owner(mm, p);		mm_init_owner(mm, p);
	RCU_INIT_POINTER(mm->exe_file, NULL);		RCU_INIT_POINTER(mm->exe_file, NULL);
	mmu_notifier_mm_init(mm);		mmu_notifier_subscriptions_init(mm);
	init_tlb_flush_pending(mm);		init_tlb_flush_pending(mm);
	#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS		#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS
	mm->pmd_huge_pte = NULL;		mm->pmd_huge_pte = NULL;

mm/debug.c

+2 −2

Original line number	Original line	Diff line number	Diff line
	@@ -153,7 +153,7 @@ void dump_mm(const struct mm_struct *mm)
	#endif		#endif
	"exe_file %px\n"		"exe_file %px\n"
	#ifdef CONFIG_MMU_NOTIFIER		#ifdef CONFIG_MMU_NOTIFIER
	"mmu_notifier_mm %px\n"		"notifier_subscriptions %px\n"
	#endif		#endif
	#ifdef CONFIG_NUMA_BALANCING		#ifdef CONFIG_NUMA_BALANCING
	"numa_next_scan %lu numa_scan_offset %lu numa_scan_seq %d\n"		"numa_next_scan %lu numa_scan_offset %lu numa_scan_seq %d\n"
	@@ -185,7 +185,7 @@ void dump_mm(const struct mm_struct *mm)
	#endif		#endif
	mm->exe_file,		mm->exe_file,
	#ifdef CONFIG_MMU_NOTIFIER		#ifdef CONFIG_MMU_NOTIFIER
	mm->mmu_notifier_mm,		mm->notifier_subscriptions,
	#endif		#endif
	#ifdef CONFIG_NUMA_BALANCING		#ifdef CONFIG_NUMA_BALANCING
	mm->numa_next_scan, mm->numa_scan_offset, mm->numa_scan_seq,		mm->numa_next_scan, mm->numa_scan_offset, mm->numa_scan_seq,