Currently, CoCo VMs can perform conversion at the base page granularity,
which is the granularity that has to be tracked. In relevant setups, the
target page size is assumed to be equal to the host page size, thus
fixing the block size to the host page size.

However, since private memory and shared memory have different backend
at present, users can specify shared memory with a hugetlbfs backend
while private memory with guest_memfd backend only supports 4K page
size. In this scenario, ram_block->page_size is different from the host
page size which will trigger an assertion when retrieving the block
size.

To address this, return the host page size directly to relax the
restriction. This changes fixes a regression of using hugetlbfs backend
for shared memory within CoCo VMs, with or without VFIO devices' presence.

Acked-by: David Hildenbrand <david@redhat.com>
Tested-by: Farrah Chen <farrah.chen@intel.com>
Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>
---
Changes in v2:
  - Modify the commit message
  - Remove the argument in ram_block_attributes_get_block_size()
---
 system/ram-block-attributes.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/system/ram-block-attributes.c b/system/ram-block-attributes.c
index 68e8a027032..a7579de5b46 100644
--- a/system/ram-block-attributes.c
+++ b/system/ram-block-attributes.c
@@ -22,16 +22,14 @@ OBJECT_DEFINE_SIMPLE_TYPE_WITH_INTERFACES(RamBlockAttributes,
                                           { })
 
 static size_t
-ram_block_attributes_get_block_size(const RamBlockAttributes *attr)
+ram_block_attributes_get_block_size(void)
 {
     /*
      * Because page conversion could be manipulated in the size of at least 4K
      * or 4K aligned, Use the host page size as the granularity to track the
      * memory attribute.
      */
-    g_assert(attr && attr->ram_block);
-    g_assert(attr->ram_block->page_size == qemu_real_host_page_size());
-    return attr->ram_block->page_size;
+    return qemu_real_host_page_size();
 }
 
 
@@ -40,7 +38,7 @@ ram_block_attributes_rdm_is_populated(const RamDiscardManager *rdm,
                                       const MemoryRegionSection *section)
 {
     const RamBlockAttributes *attr = RAM_BLOCK_ATTRIBUTES(rdm);
-    const size_t block_size = ram_block_attributes_get_block_size(attr);
+    const size_t block_size = ram_block_attributes_get_block_size();
     const uint64_t first_bit = section->offset_within_region / block_size;
     const uint64_t last_bit =
         first_bit + int128_get64(section->size) / block_size - 1;
@@ -81,7 +79,7 @@ ram_block_attributes_for_each_populated_section(const RamBlockAttributes *attr,
 {
     unsigned long first_bit, last_bit;
     uint64_t offset, size;
-    const size_t block_size = ram_block_attributes_get_block_size(attr);
+    const size_t block_size = ram_block_attributes_get_block_size();
     int ret = 0;
 
     first_bit = section->offset_within_region / block_size;
@@ -122,7 +120,7 @@ ram_block_attributes_for_each_discarded_section(const RamBlockAttributes *attr,
 {
     unsigned long first_bit, last_bit;
     uint64_t offset, size;
-    const size_t block_size = ram_block_attributes_get_block_size(attr);
+    const size_t block_size = ram_block_attributes_get_block_size();
     int ret = 0;
 
     first_bit = section->offset_within_region / block_size;
@@ -163,7 +161,7 @@ ram_block_attributes_rdm_get_min_granularity(const RamDiscardManager *rdm,
     const RamBlockAttributes *attr = RAM_BLOCK_ATTRIBUTES(rdm);
 
     g_assert(mr == attr->ram_block->mr);
-    return ram_block_attributes_get_block_size(attr);
+    return ram_block_attributes_get_block_size();
 }
 
 static void
@@ -265,7 +263,7 @@ ram_block_attributes_is_valid_range(RamBlockAttributes *attr, uint64_t offset,
     g_assert(mr);
 
     uint64_t region_size = memory_region_size(mr);
-    const size_t block_size = ram_block_attributes_get_block_size(attr);
+    const size_t block_size = ram_block_attributes_get_block_size();
 
     if (!QEMU_IS_ALIGNED(offset, block_size) ||
         !QEMU_IS_ALIGNED(size, block_size)) {
@@ -322,7 +320,7 @@ int ram_block_attributes_state_change(RamBlockAttributes *attr,
                                       uint64_t offset, uint64_t size,
                                       bool to_discard)
 {
-    const size_t block_size = ram_block_attributes_get_block_size(attr);
+    const size_t block_size = ram_block_attributes_get_block_size();
     const unsigned long first_bit = offset / block_size;
     const unsigned long nbits = size / block_size;
     const unsigned long last_bit = first_bit + nbits - 1;
-- 
2.43.5

On 10/23/2025 5:55 PM, Chenyi Qiang wrote:
> Currently, CoCo VMs can perform conversion at the base page granularity,
> which is the granularity that has to be tracked. In relevant setups, the
> target page size is assumed to be equal to the host page size, thus
> fixing the block size to the host page size.
> 
> However, since private memory and shared memory have different backend
> at present, users can specify shared memory with a hugetlbfs backend
> while private memory with guest_memfd backend only supports 4K page
> size. In this scenario, ram_block->page_size is different from the host
> page size which will trigger an assertion when retrieving the block
> size.
> 
> To address this, return the host page size directly to relax the
> restriction. This changes fixes a regression of using hugetlbfs backend
> for shared memory within CoCo VMs, with or without VFIO devices' presence.
> 
> Acked-by: David Hildenbrand <david@redhat.com>
> Tested-by: Farrah Chen <farrah.chen@intel.com>
> Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>

The change looks good to me.

Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>

> ---
> Changes in v2:
>    - Modify the commit message
>    - Remove the argument in ram_block_attributes_get_block_size()
> ---
>   system/ram-block-attributes.c | 18 ++++++++----------
>   1 file changed, 8 insertions(+), 10 deletions(-)
> 
> diff --git a/system/ram-block-attributes.c b/system/ram-block-attributes.c
> index 68e8a027032..a7579de5b46 100644
> --- a/system/ram-block-attributes.c
> +++ b/system/ram-block-attributes.c
> @@ -22,16 +22,14 @@ OBJECT_DEFINE_SIMPLE_TYPE_WITH_INTERFACES(RamBlockAttributes,
>                                             { })
>   
>   static size_t
> -ram_block_attributes_get_block_size(const RamBlockAttributes *attr)
> +ram_block_attributes_get_block_size(void)
>   {
>       /*
>        * Because page conversion could be manipulated in the size of at least 4K
>        * or 4K aligned, Use the host page size as the granularity to track the
>        * memory attribute.
>        */
> -    g_assert(attr && attr->ram_block);
> -    g_assert(attr->ram_block->page_size == qemu_real_host_page_size());
> -    return attr->ram_block->page_size;
> +    return qemu_real_host_page_size();
>   }
>   
>   
> @@ -40,7 +38,7 @@ ram_block_attributes_rdm_is_populated(const RamDiscardManager *rdm,
>                                         const MemoryRegionSection *section)
>   {
>       const RamBlockAttributes *attr = RAM_BLOCK_ATTRIBUTES(rdm);
> -    const size_t block_size = ram_block_attributes_get_block_size(attr);
> +    const size_t block_size = ram_block_attributes_get_block_size();
>       const uint64_t first_bit = section->offset_within_region / block_size;
>       const uint64_t last_bit =
>           first_bit + int128_get64(section->size) / block_size - 1;
> @@ -81,7 +79,7 @@ ram_block_attributes_for_each_populated_section(const RamBlockAttributes *attr,
>   {
>       unsigned long first_bit, last_bit;
>       uint64_t offset, size;
> -    const size_t block_size = ram_block_attributes_get_block_size(attr);
> +    const size_t block_size = ram_block_attributes_get_block_size();
>       int ret = 0;
>   
>       first_bit = section->offset_within_region / block_size;
> @@ -122,7 +120,7 @@ ram_block_attributes_for_each_discarded_section(const RamBlockAttributes *attr,
>   {
>       unsigned long first_bit, last_bit;
>       uint64_t offset, size;
> -    const size_t block_size = ram_block_attributes_get_block_size(attr);
> +    const size_t block_size = ram_block_attributes_get_block_size();
>       int ret = 0;
>   
>       first_bit = section->offset_within_region / block_size;
> @@ -163,7 +161,7 @@ ram_block_attributes_rdm_get_min_granularity(const RamDiscardManager *rdm,
>       const RamBlockAttributes *attr = RAM_BLOCK_ATTRIBUTES(rdm);
>   
>       g_assert(mr == attr->ram_block->mr);
> -    return ram_block_attributes_get_block_size(attr);
> +    return ram_block_attributes_get_block_size();
>   }
>   
>   static void
> @@ -265,7 +263,7 @@ ram_block_attributes_is_valid_range(RamBlockAttributes *attr, uint64_t offset,
>       g_assert(mr);
>   
>       uint64_t region_size = memory_region_size(mr);
> -    const size_t block_size = ram_block_attributes_get_block_size(attr);
> +    const size_t block_size = ram_block_attributes_get_block_size();
>   
>       if (!QEMU_IS_ALIGNED(offset, block_size) ||
>           !QEMU_IS_ALIGNED(size, block_size)) {
> @@ -322,7 +320,7 @@ int ram_block_attributes_state_change(RamBlockAttributes *attr,
>                                         uint64_t offset, uint64_t size,
>                                         bool to_discard)
>   {
> -    const size_t block_size = ram_block_attributes_get_block_size(attr);
> +    const size_t block_size = ram_block_attributes_get_block_size();
>       const unsigned long first_bit = offset / block_size;
>       const unsigned long nbits = size / block_size;
>       const unsigned long last_bit = first_bit + nbits - 1;

On 23.10.25 11:55, Chenyi Qiang wrote:

Subject should probably rather be:

"ram-block-attributes: fix interaction with hugetlb memory backends"

Maybe that can be fixed up when applying.

> Currently, CoCo VMs can perform conversion at the base page granularity,
> which is the granularity that has to be tracked. In relevant setups, the
> target page size is assumed to be equal to the host page size, thus
> fixing the block size to the host page size.
> 
> However, since private memory and shared memory have different backend
> at present, users can specify shared memory with a hugetlbfs backend
> while private memory with guest_memfd backend only supports 4K page
> size. In this scenario, ram_block->page_size is different from the host
> page size which will trigger an assertion when retrieving the block
> size.
> 
> To address this, return the host page size directly to relax the
> restriction. This changes fixes a regression of using hugetlbfs backend
> for shared memory within CoCo VMs, with or without VFIO devices' presence.
> 
> Acked-by: David Hildenbrand <david@redhat.com>
> Tested-by: Farrah Chen <farrah.chen@intel.com>
> Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>
> ---


-- 
Cheers

David / dhildenb

On Thu, Oct 23, 2025 at 12:16:46PM +0200, David Hildenbrand wrote:
> On 23.10.25 11:55, Chenyi Qiang wrote:
> 
> Subject should probably rather be:
> 
> "ram-block-attributes: fix interaction with hugetlb memory backends"
> 
> Maybe that can be fixed up when applying.

I also agree the old subject is slightly confusing..

I queued the two patches with all the small fixups.

Thanks,

> 
> > Currently, CoCo VMs can perform conversion at the base page granularity,
> > which is the granularity that has to be tracked. In relevant setups, the
> > target page size is assumed to be equal to the host page size, thus
> > fixing the block size to the host page size.
> > 
> > However, since private memory and shared memory have different backend
> > at present, users can specify shared memory with a hugetlbfs backend
> > while private memory with guest_memfd backend only supports 4K page
> > size. In this scenario, ram_block->page_size is different from the host
> > page size which will trigger an assertion when retrieving the block
> > size.
> > 
> > To address this, return the host page size directly to relax the
> > restriction. This changes fixes a regression of using hugetlbfs backend
> > for shared memory within CoCo VMs, with or without VFIO devices' presence.
> > 
> > Acked-by: David Hildenbrand <david@redhat.com>
> > Tested-by: Farrah Chen <farrah.chen@intel.com>
> > Signed-off-by: Chenyi Qiang <chenyi.qiang@intel.com>
> > ---
> 
> 
> -- 
> Cheers
> 
> David / dhildenb
> 

-- 
Peter Xu