Re: [PATCH v3 01/28] KVM: x86/mmu: Use common iterator for walking invalid TDP MMU roots

[Mingwei Zhang]

On Sat, Feb 26, 2022, Sean Christopherson wrote:
> Now that tdp_mmu_next_root() can process both valid and invalid roots,
> extend it to be able to process _only_ invalid roots, add yet another
> iterator macro for walking invalid roots, and use the new macro in
> kvm_tdp_mmu_zap_invalidated_roots().
> 
> No functional change intended.
> 
> Reviewed-by: David Matlack <dmatlack@xxxxxxxxxx>
> Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> ---
>  arch/x86/kvm/mmu/tdp_mmu.c | 74 ++++++++++++++------------------------
>  1 file changed, 26 insertions(+), 48 deletions(-)
> 
> diff --git a/arch/x86/kvm/mmu/tdp_mmu.c b/arch/x86/kvm/mmu/tdp_mmu.c
> index debf08212f12..25148e8b711d 100644
> --- a/arch/x86/kvm/mmu/tdp_mmu.c
> +++ b/arch/x86/kvm/mmu/tdp_mmu.c
> @@ -98,6 +98,12 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root,
>  	call_rcu(&root->rcu_head, tdp_mmu_free_sp_rcu_callback);
>  }
>  
> +enum tdp_mmu_roots_iter_type {
> +	ALL_ROOTS = -1,
> +	VALID_ROOTS = 0,
> +	INVALID_ROOTS = 1,
> +};

I am wondering what the trick is to start from -1?
> +
>  /*
>   * Returns the next root after @prev_root (or the first root if @prev_root is
>   * NULL).  A reference to the returned root is acquired, and the reference to
> @@ -110,10 +116,16 @@ void kvm_tdp_mmu_put_root(struct kvm *kvm, struct kvm_mmu_page *root,
>   */
>  static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
>  					      struct kvm_mmu_page *prev_root,
> -					      bool shared, bool only_valid)
> +					      bool shared,
> +					      enum tdp_mmu_roots_iter_type type)
>  {
>  	struct kvm_mmu_page *next_root;
>  
> +	kvm_lockdep_assert_mmu_lock_held(kvm, shared);
> +
> +	/* Ensure correctness for the below comparison against role.invalid. */
> +	BUILD_BUG_ON(!!VALID_ROOTS || !INVALID_ROOTS);
> +
>  	rcu_read_lock();
>  
>  	if (prev_root)
> @@ -125,7 +137,7 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
>  						   typeof(*next_root), link);
>  
>  	while (next_root) {
> -		if ((!only_valid || !next_root->role.invalid) &&
> +		if ((type == ALL_ROOTS || (type == !!next_root->role.invalid)) &&
>  		    kvm_tdp_mmu_get_root(next_root))
>  			break;
>  
> @@ -151,18 +163,21 @@ static struct kvm_mmu_page *tdp_mmu_next_root(struct kvm *kvm,
>   * mode. In the unlikely event that this thread must free a root, the lock
>   * will be temporarily dropped and reacquired in write mode.
>   */
> -#define __for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, _only_valid)\
> -	for (_root = tdp_mmu_next_root(_kvm, NULL, _shared, _only_valid);	\
> +#define __for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, _type) \
> +	for (_root = tdp_mmu_next_root(_kvm, NULL, _shared, _type);		\
>  	     _root;								\
> -	     _root = tdp_mmu_next_root(_kvm, _root, _shared, _only_valid))	\
> -		if (kvm_mmu_page_as_id(_root) != _as_id) {			\
> +	     _root = tdp_mmu_next_root(_kvm, _root, _shared, _type))		\
> +		if (_as_id > 0 && kvm_mmu_page_as_id(_root) != _as_id) {	\
>  		} else
>  
> +#define for_each_invalid_tdp_mmu_root_yield_safe(_kvm, _root)			\
> +	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, -1, true, INVALID_ROOTS)
> +
>  #define for_each_valid_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared)	\
> -	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, true)
> +	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, VALID_ROOTS)
>  
>  #define for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared)		\
> -	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, false)
> +	__for_each_tdp_mmu_root_yield_safe(_kvm, _root, _as_id, _shared, ALL_ROOTS)
>  
>  #define for_each_tdp_mmu_root(_kvm, _root, _as_id)				\
>  	list_for_each_entry_rcu(_root, &_kvm->arch.tdp_mmu_roots, link,		\
> @@ -810,28 +825,6 @@ void kvm_tdp_mmu_zap_all(struct kvm *kvm)
>  		kvm_flush_remote_tlbs(kvm);
>  }
>  
> -static struct kvm_mmu_page *next_invalidated_root(struct kvm *kvm,
> -						  struct kvm_mmu_page *prev_root)
> -{
> -	struct kvm_mmu_page *next_root;
> -
> -	if (prev_root)
> -		next_root = list_next_or_null_rcu(&kvm->arch.tdp_mmu_roots,
> -						  &prev_root->link,
> -						  typeof(*prev_root), link);
> -	else
> -		next_root = list_first_or_null_rcu(&kvm->arch.tdp_mmu_roots,
> -						   typeof(*next_root), link);
> -
> -	while (next_root && !(next_root->role.invalid &&
> -			      refcount_read(&next_root->tdp_mmu_root_count)))
> -		next_root = list_next_or_null_rcu(&kvm->arch.tdp_mmu_roots,
> -						  &next_root->link,
> -						  typeof(*next_root), link);
> -
> -	return next_root;
> -}
> -
>  /*
>   * Since kvm_tdp_mmu_zap_all_fast has acquired a reference to each
>   * invalidated root, they will not be freed until this function drops the
> @@ -842,36 +835,21 @@ static struct kvm_mmu_page *next_invalidated_root(struct kvm *kvm,
>   */
>  void kvm_tdp_mmu_zap_invalidated_roots(struct kvm *kvm)
>  {
> -	struct kvm_mmu_page *next_root;
>  	struct kvm_mmu_page *root;
>  	bool flush = false;
>  
>  	lockdep_assert_held_read(&kvm->mmu_lock);
>  
> -	rcu_read_lock();
> -
> -	root = next_invalidated_root(kvm, NULL);
> -
> -	while (root) {
> -		next_root = next_invalidated_root(kvm, root);
> -
> -		rcu_read_unlock();
> -
> +	for_each_invalid_tdp_mmu_root_yield_safe(kvm, root) {
>  		flush = zap_gfn_range(kvm, root, 0, -1ull, true, flush, true);
>  
>  		/*
> -		 * Put the reference acquired in
> -		 * kvm_tdp_mmu_invalidate_roots
> +		 * Put the reference acquired in kvm_tdp_mmu_invalidate_roots().
> +		 * Note, the iterator holds its own reference.
>  		 */
>  		kvm_tdp_mmu_put_root(kvm, root, true);
> -
> -		root = next_root;
> -
> -		rcu_read_lock();
>  	}
>  
> -	rcu_read_unlock();
> -
>  	if (flush)
>  		kvm_flush_remote_tlbs(kvm);
>  }
> -- 
> 2.35.1.574.g5d30c73bfb-goog
>