[v8] Add NUMA mempolicy support for KVM guest-memfd

[RFC PATCH v8 5/7] KVM: guest_memfd: Add slab-allocated inode cache

Posted by Shivank Garg 3 months, 3 weeks ago

Add dedicated inode structure (kvm_gmem_inode_info) and slab-allocated
inode cache for guest memory backing, similar to how shmem handles inodes.

This adds the necessary allocation/destruction functions and prepares
for upcoming guest_memfd NUMA policy support changes.

Signed-off-by: Shivank Garg <shivankg@amd.com>
---
 virt/kvm/guest_memfd.c | 51 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/virt/kvm/guest_memfd.c b/virt/kvm/guest_memfd.c
index 159df462d193..5a1ce6f5e287 100644
--- a/virt/kvm/guest_memfd.c
+++ b/virt/kvm/guest_memfd.c
@@ -17,6 +17,15 @@ struct kvm_gmem {
 	struct list_head entry;
 };
 
+struct kvm_gmem_inode_info {
+	struct inode vfs_inode;
+};
+
+static inline struct kvm_gmem_inode_info *KVM_GMEM_I(struct inode *inode)
+{
+	return container_of(inode, struct kvm_gmem_inode_info, vfs_inode);
+}
+
 /**
  * folio_file_pfn - like folio_file_page, but return a pfn.
  * @folio: The folio which contains this index.
@@ -392,8 +401,33 @@ static struct file_operations kvm_gmem_fops = {
 	.fallocate	= kvm_gmem_fallocate,
 };
 
+static struct kmem_cache *kvm_gmem_inode_cachep;
+
+static struct inode *kvm_gmem_alloc_inode(struct super_block *sb)
+{
+	struct kvm_gmem_inode_info *info;
+
+	info = alloc_inode_sb(sb, kvm_gmem_inode_cachep, GFP_KERNEL);
+	if (!info)
+		return NULL;
+
+	return &info->vfs_inode;
+}
+
+static void kvm_gmem_destroy_inode(struct inode *inode)
+{
+}
+
+static void kvm_gmem_free_inode(struct inode *inode)
+{
+	kmem_cache_free(kvm_gmem_inode_cachep, KVM_GMEM_I(inode));
+}
+
 static const struct super_operations kvm_gmem_super_operations = {
 	.statfs		= simple_statfs,
+	.alloc_inode	= kvm_gmem_alloc_inode,
+	.destroy_inode	= kvm_gmem_destroy_inode,
+	.free_inode	= kvm_gmem_free_inode,
 };
 
 static int kvm_gmem_init_fs_context(struct fs_context *fc)
@@ -426,10 +460,26 @@ static int kvm_gmem_init_mount(void)
 	return 0;
 }
 
+static void kvm_gmem_init_inode(void *foo)
+{
+	struct kvm_gmem_inode_info *info = foo;
+
+	inode_init_once(&info->vfs_inode);
+}
+
+static void kvm_gmem_init_inodecache(void)
+{
+	kvm_gmem_inode_cachep = kmem_cache_create("kvm_gmem_inode_cache",
+						  sizeof(struct kvm_gmem_inode_info),
+						  0, SLAB_ACCOUNT,
+						  kvm_gmem_init_inode);
+}
+
 int kvm_gmem_init(struct module *module)
 {
 	kvm_gmem_fops.owner = module;
 
+	kvm_gmem_init_inodecache();
 	return kvm_gmem_init_mount();
 }
 
@@ -437,6 +487,7 @@ void kvm_gmem_exit(void)
 {
 	kern_unmount(kvm_gmem_mnt);
 	kvm_gmem_mnt = NULL;
+	kmem_cache_destroy(kvm_gmem_inode_cachep);
 }
 
 static int kvm_gmem_migrate_folio(struct address_space *mapping,
-- 
2.43.0

Re: [RFC PATCH v8 5/7] KVM: guest_memfd: Add slab-allocated inode cache

Posted by Huang, Ying 3 months, 2 weeks ago

Shivank Garg <shivankg@amd.com> writes:

> Add dedicated inode structure (kvm_gmem_inode_info) and slab-allocated
> inode cache for guest memory backing, similar to how shmem handles inodes.
>
> This adds the necessary allocation/destruction functions and prepares
> for upcoming guest_memfd NUMA policy support changes.
>
> Signed-off-by: Shivank Garg <shivankg@amd.com>
> ---
>  virt/kvm/guest_memfd.c | 51 ++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 51 insertions(+)
>
> diff --git a/virt/kvm/guest_memfd.c b/virt/kvm/guest_memfd.c
> index 159df462d193..5a1ce6f5e287 100644
> --- a/virt/kvm/guest_memfd.c
> +++ b/virt/kvm/guest_memfd.c
> @@ -17,6 +17,15 @@ struct kvm_gmem {
>  	struct list_head entry;
>  };
>  
> +struct kvm_gmem_inode_info {
> +	struct inode vfs_inode;
> +};
> +
> +static inline struct kvm_gmem_inode_info *KVM_GMEM_I(struct inode *inode)
> +{
> +	return container_of(inode, struct kvm_gmem_inode_info, vfs_inode);
> +}
> +
>  /**
>   * folio_file_pfn - like folio_file_page, but return a pfn.
>   * @folio: The folio which contains this index.
> @@ -392,8 +401,33 @@ static struct file_operations kvm_gmem_fops = {
>  	.fallocate	= kvm_gmem_fallocate,
>  };
>  
> +static struct kmem_cache *kvm_gmem_inode_cachep;
> +
> +static struct inode *kvm_gmem_alloc_inode(struct super_block *sb)
> +{
> +	struct kvm_gmem_inode_info *info;
> +
> +	info = alloc_inode_sb(sb, kvm_gmem_inode_cachep, GFP_KERNEL);
> +	if (!info)
> +		return NULL;
> +
> +	return &info->vfs_inode;
> +}
> +
> +static void kvm_gmem_destroy_inode(struct inode *inode)
> +{
> +}
> +
> +static void kvm_gmem_free_inode(struct inode *inode)
> +{
> +	kmem_cache_free(kvm_gmem_inode_cachep, KVM_GMEM_I(inode));
> +}
> +
>  static const struct super_operations kvm_gmem_super_operations = {
>  	.statfs		= simple_statfs,
> +	.alloc_inode	= kvm_gmem_alloc_inode,
> +	.destroy_inode	= kvm_gmem_destroy_inode,
> +	.free_inode	= kvm_gmem_free_inode,
>  };
>  
>  static int kvm_gmem_init_fs_context(struct fs_context *fc)
> @@ -426,10 +460,26 @@ static int kvm_gmem_init_mount(void)
>  	return 0;
>  }
>  
> +static void kvm_gmem_init_inode(void *foo)
> +{
> +	struct kvm_gmem_inode_info *info = foo;
> +
> +	inode_init_once(&info->vfs_inode);
> +}
> +
> +static void kvm_gmem_init_inodecache(void)
> +{
> +	kvm_gmem_inode_cachep = kmem_cache_create("kvm_gmem_inode_cache",
> +						  sizeof(struct kvm_gmem_inode_info),
> +						  0, SLAB_ACCOUNT,
> +						  kvm_gmem_init_inode);

Check the return value?

And, I'm not a big fan of (logically) one line function encapsulation.

> +}
> +
>  int kvm_gmem_init(struct module *module)
>  {
>  	kvm_gmem_fops.owner = module;
>  
> +	kvm_gmem_init_inodecache();
>  	return kvm_gmem_init_mount();

kmem_cache_destroy(kvm_gmem_inode_cachep) if kvm_gmem_init_mount()
return with error?

>  }
>  
> @@ -437,6 +487,7 @@ void kvm_gmem_exit(void)
>  {
>  	kern_unmount(kvm_gmem_mnt);
>  	kvm_gmem_mnt = NULL;
> +	kmem_cache_destroy(kvm_gmem_inode_cachep);
>  }
>  
>  static int kvm_gmem_migrate_folio(struct address_space *mapping,

---
Best Regards,
Huang, Ying

Re: [RFC PATCH v8 5/7] KVM: guest_memfd: Add slab-allocated inode cache

Posted by Shivank Garg 3 months, 1 week ago

On 6/24/2025 9:46 AM, Huang, Ying wrote:
> Shivank Garg <shivankg@amd.com> writes:
> 
>> +static void kvm_gmem_init_inodecache(void)
>> +{
>> +	kvm_gmem_inode_cachep = kmem_cache_create("kvm_gmem_inode_cache",
>> +						  sizeof(struct kvm_gmem_inode_info),
>> +						  0, SLAB_ACCOUNT,
>> +						  kvm_gmem_init_inode);
> 
> Check the return value?
> 
> And, I'm not a big fan of (logically) one line function encapsulation.
> 
>> +}
>> +
>>  int kvm_gmem_init(struct module *module)
>>  {
>>  	kvm_gmem_fops.owner = module;
>>  
>> +	kvm_gmem_init_inodecache();
>>  	return kvm_gmem_init_mount();
> 
> kmem_cache_destroy(kvm_gmem_inode_cachep) if kvm_gmem_init_mount()
> return with error?
> 

Thanks for the feedback, Ying.
Good catch on the leak!

Regarding the missing error check, I noticed while looking at examples that 
kernel code is sometimes inconsistent with kmem_cache_create() error handling, 
but you're right about checking for failures, So I'll handle them properly.

diff --git a/virt/kvm/guest_memfd.c b/virt/kvm/guest_memfd.c
index 7b8c548986d4..8cb83692f7a0 100644
--- a/virt/kvm/guest_memfd.c
+++ b/virt/kvm/guest_memfd.c
@@ -530,20 +530,23 @@ static void kvm_gmem_init_inode(void *foo)
        inode_init_once(&info->vfs_inode);
 }

-static void kvm_gmem_init_inodecache(void)
+int kvm_gmem_init(struct module *module)
 {
+       int ret;
+
+       kvm_gmem_fops.owner = module;
        kvm_gmem_inode_cachep = kmem_cache_create("kvm_gmem_inode_cache",
                                                  sizeof(struct kvm_gmem_inode_info),
                                                  0, SLAB_ACCOUNT,
                                                  kvm_gmem_init_inode);
-}
-
-int kvm_gmem_init(struct module *module)
-{
-       kvm_gmem_fops.owner = module;
-
-       kvm_gmem_init_inodecache();
-       return kvm_gmem_init_mount();
+       if (!kvm_gmem_inode_cachep)
+               return -ENOMEM;
+       ret = kvm_gmem_init_mount();
+       if (ret) {
+               kmem_cache_destroy(kvm_gmem_inode_cachep);
+               return ret;
+       }
+       return 0;
 }

Best Regards,
Shivank