class BufferMemorySettings

This field will always be set by sysmem.

Rule for producers: For `BufferCollectionInfo` with

`ImageFormatConstraints`, storing an image of a given `ImageFormat` in a

buffer of that `BufferCollection` is only allowed if the result of

calling `ImageFormatImageSize` or a strictly equivalent computation is

less than or equal to `size_bytes`.

Producer participants _must_ stay within `size_bytes` when determining

whether an image will fit in a buffer, even if the VMO's size is larger

than this.

Failure of a producer to use `size_bytes` as the constraint (not VMO

size) when determining whether a given `ImageFormat` will fit in a

buffer will, in general, cause problems for any participant that has set

ImageFormatConstraints.pad_for_block_size to larger than {1, 1}, or

pad_beyond_image_size_bytes to greater than 0. Those participants rely

on the ability to access bytes in the VMO beyond the image size. Putting

an image in the VMO that exceeds `size_bytes` can potentially lead to

those participants trying to access an offset beyond the last page of

the VMO.

Similarly, any future padding-only constraint which is irrelevant to a

non-padding participant when locating each valid pixel's valid data will

also be accounted for by checking `ImageFormatImageSize()` against

`size_bytes` instead of the VMO size.

Following this rule allows producers to not need to worry about

arbitrary padding requirements of participants, including future-added

padding-only constraints.

The `bytes_per_row` and rounding up to a tile size boundary are included

in `ImageFormatImageSize` and accounted for in `size_bytes`, even though

these two aspects are partially about padding. These two aspects are

included in the `ImageFormatImageSize` calculation because these two

aspects are also relevant to all participants that need to locate pixel

data. The `fuchsia.images2.ImageFormat` intentionally doesn't contain

any padding-only fields.

Participants with a padding constraint (in addition to `bytes_per_row`

or tile size boundary), such as pad_for_block_size or

pad_beyond_image_size_bytes, are allowed to access, but not treat as

meaningful, bytes which are at offset `size_bytes` up to the max offset

implied by their own padding constraint, which will never exceed the VMO

size minus 1, as long as the `ImageFormatImageSize` is less than or

equal to `BufferMemorySettings.size_bytes` and other constraints under

`BufferCollectionInfo` aren't violated by the producer.

The same `size_bytes` value is sent to all participants.

This field is not page aligned. This value rounded up to

zx_system_page_size() boundary is guaranteed to be less than or equal

the buffer VMO size. In other words, the VMO can have more pages than

implied by this value, due to padding-only constraints from

participant(s).

This field will always be set by sysmem.

Rule for producers: For `BufferCollectionInfo` with

`ImageFormatConstraints`, storing an image of a given `ImageFormat` in a

buffer of that `BufferCollection` is only allowed if the result of

calling `ImageFormatImageSize` or a strictly equivalent computation is

less than or equal to `size_bytes`.

Producer participants _must_ stay within `size_bytes` when determining

whether an image will fit in a buffer, even if the VMO's size is larger

than this.

Failure of a producer to use `size_bytes` as the constraint (not VMO

size) when determining whether a given `ImageFormat` will fit in a

buffer will, in general, cause problems for any participant that has set

ImageFormatConstraints.pad_for_block_size to larger than {1, 1}, or

pad_beyond_image_size_bytes to greater than 0. Those participants rely

on the ability to access bytes in the VMO beyond the image size. Putting

an image in the VMO that exceeds `size_bytes` can potentially lead to

those participants trying to access an offset beyond the last page of

the VMO.

Similarly, any future padding-only constraint which is irrelevant to a

non-padding participant when locating each valid pixel's valid data will

also be accounted for by checking `ImageFormatImageSize()` against

`size_bytes` instead of the VMO size.

Following this rule allows producers to not need to worry about

arbitrary padding requirements of participants, including future-added

padding-only constraints.

The `bytes_per_row` and rounding up to a tile size boundary are included

in `ImageFormatImageSize` and accounted for in `size_bytes`, even though

these two aspects are partially about padding. These two aspects are

included in the `ImageFormatImageSize` calculation because these two

aspects are also relevant to all participants that need to locate pixel

data. The `fuchsia.images2.ImageFormat` intentionally doesn't contain

any padding-only fields.

Participants with a padding constraint (in addition to `bytes_per_row`

or tile size boundary), such as pad_for_block_size or

pad_beyond_image_size_bytes, are allowed to access, but not treat as

meaningful, bytes which are at offset `size_bytes` up to the max offset

implied by their own padding constraint, which will never exceed the VMO

size minus 1, as long as the `ImageFormatImageSize` is less than or

equal to `BufferMemorySettings.size_bytes` and other constraints under

`BufferCollectionInfo` aren't violated by the producer.

The same `size_bytes` value is sent to all participants.

This field is not page aligned. This value rounded up to

zx_system_page_size() boundary is guaranteed to be less than or equal

the buffer VMO size. In other words, the VMO can have more pages than

implied by this value, due to padding-only constraints from

participant(s).

This field will always be set by sysmem.

Rule for producers: For `BufferCollectionInfo` with

`ImageFormatConstraints`, storing an image of a given `ImageFormat` in a

buffer of that `BufferCollection` is only allowed if the result of

calling `ImageFormatImageSize` or a strictly equivalent computation is

less than or equal to `size_bytes`.

Producer participants _must_ stay within `size_bytes` when determining

whether an image will fit in a buffer, even if the VMO's size is larger

than this.

Failure of a producer to use `size_bytes` as the constraint (not VMO

size) when determining whether a given `ImageFormat` will fit in a

buffer will, in general, cause problems for any participant that has set

ImageFormatConstraints.pad_for_block_size to larger than {1, 1}, or

pad_beyond_image_size_bytes to greater than 0. Those participants rely

on the ability to access bytes in the VMO beyond the image size. Putting

an image in the VMO that exceeds `size_bytes` can potentially lead to

those participants trying to access an offset beyond the last page of

the VMO.

Similarly, any future padding-only constraint which is irrelevant to a

non-padding participant when locating each valid pixel's valid data will

also be accounted for by checking `ImageFormatImageSize()` against

`size_bytes` instead of the VMO size.

Following this rule allows producers to not need to worry about

arbitrary padding requirements of participants, including future-added

padding-only constraints.

The `bytes_per_row` and rounding up to a tile size boundary are included

in `ImageFormatImageSize` and accounted for in `size_bytes`, even though

these two aspects are partially about padding. These two aspects are

included in the `ImageFormatImageSize` calculation because these two

aspects are also relevant to all participants that need to locate pixel

data. The `fuchsia.images2.ImageFormat` intentionally doesn't contain

any padding-only fields.

Participants with a padding constraint (in addition to `bytes_per_row`

or tile size boundary), such as pad_for_block_size or

pad_beyond_image_size_bytes, are allowed to access, but not treat as

meaningful, bytes which are at offset `size_bytes` up to the max offset

implied by their own padding constraint, which will never exceed the VMO

size minus 1, as long as the `ImageFormatImageSize` is less than or

equal to `BufferMemorySettings.size_bytes` and other constraints under

`BufferCollectionInfo` aren't violated by the producer.

The same `size_bytes` value is sent to all participants.

This field is not page aligned. This value rounded up to

zx_system_page_size() boundary is guaranteed to be less than or equal

the buffer VMO size. In other words, the VMO can have more pages than

implied by this value, due to padding-only constraints from

participant(s).

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

This field will always be set by sysmem.

The specific heap from which buffers are allocated.

This field will always be set by sysmem.

The specific heap from which buffers are allocated.

This field will always be set by sysmem.

The specific heap from which buffers are allocated.

This field will always be set by sysmem.

This is guaranteed to be equal to the VMO size obtained via

`zx_vmo_get_size`. This field is here because `pad_*` constraints can

result in the VMO size being larger than implied by `size_bytes`, and

for participants using `pad_*` constraints, it can be convenient to have

this field available rather than having to query for the VMO size.

When checking with `ImageFormatImageSize` to see if an image will fit in

a buffer, use `size_bytes` as the threshold, not `raw_vmo_size`.

Producers using `pad_*` constraints must still only put images in the

buffer which have `ImageFormatImageSize()

<

= size_bytes`. Participants

using `pad_*` constraints are allowed to access bytes beyond

`size_bytes` consistent with what the participant set for the pad_*

constriant(s), which will always fit within `raw_vmo_size` assuming

correct operation of the participant and sysmem. This field can be

useful for double-checking that accesses will be within the VMO.

When setting up VMO mappings and VMO pins, a participant using `pad_*`

constraint(s) should use `raw_vmo_size` for the size of the mapping. In

contrast, a participant not using any pad_* constraint can use

`size_bytes` rounded up to a page size boundary or said participant can

use raw_vmo_size, but for that participant, using `size_bytes` rounded

up to the page size can avoid mapping a few pages.

This is guaranteed to be equal to the VMO size obtained via

`zx_vmo_get_size`. This field is here because `pad_*` constraints can

result in the VMO size being larger than implied by `size_bytes`, and

for participants using `pad_*` constraints, it can be convenient to have

this field available rather than having to query for the VMO size.

When checking with `ImageFormatImageSize` to see if an image will fit in

a buffer, use `size_bytes` as the threshold, not `raw_vmo_size`.

Producers using `pad_*` constraints must still only put images in the

buffer which have `ImageFormatImageSize()

<

= size_bytes`. Participants

using `pad_*` constraints are allowed to access bytes beyond

`size_bytes` consistent with what the participant set for the pad_*

constriant(s), which will always fit within `raw_vmo_size` assuming

correct operation of the participant and sysmem. This field can be

useful for double-checking that accesses will be within the VMO.

When setting up VMO mappings and VMO pins, a participant using `pad_*`

constraint(s) should use `raw_vmo_size` for the size of the mapping. In

contrast, a participant not using any pad_* constraint can use

`size_bytes` rounded up to a page size boundary or said participant can

use raw_vmo_size, but for that participant, using `size_bytes` rounded

up to the page size can avoid mapping a few pages.

This is guaranteed to be equal to the VMO size obtained via

`zx_vmo_get_size`. This field is here because `pad_*` constraints can

result in the VMO size being larger than implied by `size_bytes`, and

for participants using `pad_*` constraints, it can be convenient to have

this field available rather than having to query for the VMO size.

When checking with `ImageFormatImageSize` to see if an image will fit in

a buffer, use `size_bytes` as the threshold, not `raw_vmo_size`.

Producers using `pad_*` constraints must still only put images in the

buffer which have `ImageFormatImageSize()

<

= size_bytes`. Participants

using `pad_*` constraints are allowed to access bytes beyond

`size_bytes` consistent with what the participant set for the pad_*

constriant(s), which will always fit within `raw_vmo_size` assuming

correct operation of the participant and sysmem. This field can be

useful for double-checking that accesses will be within the VMO.

When setting up VMO mappings and VMO pins, a participant using `pad_*`

constraint(s) should use `raw_vmo_size` for the size of the mapping. In

contrast, a participant not using any pad_* constraint can use

`size_bytes` rounded up to a page size boundary or said participant can

use raw_vmo_size, but for that participant, using `size_bytes` rounded

up to the page size can avoid mapping a few pages.