2018-05-31 11:23:42 -05:00
|
|
|
|
//! Memory allocation APIs
|
|
|
|
|
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#![stable(feature = "alloc_module", since = "1.28.0")]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
2019-04-14 21:23:21 -05:00
|
|
|
|
use crate::cmp;
|
|
|
|
|
use crate::fmt;
|
|
|
|
|
use crate::mem;
|
|
|
|
|
use crate::usize;
|
|
|
|
|
use crate::ptr::{self, NonNull};
|
|
|
|
|
use crate::num::NonZeroUsize;
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
/// Represents the combination of a starting address and
|
|
|
|
|
/// a total capacity of the returned block.
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
#[derive(Debug)]
|
2018-05-31 01:57:43 -05:00
|
|
|
|
pub struct Excess(pub NonNull<u8>, pub usize);
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
fn size_align<T>() -> (usize, usize) {
|
|
|
|
|
(mem::size_of::<T>(), mem::align_of::<T>())
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Layout of a block of memory.
|
|
|
|
|
///
|
|
|
|
|
/// An instance of `Layout` describes a particular layout of memory.
|
|
|
|
|
/// You build a `Layout` up as an input to give to an allocator.
|
|
|
|
|
///
|
|
|
|
|
/// All layouts have an associated non-negative size and a
|
|
|
|
|
/// power-of-two alignment.
|
|
|
|
|
///
|
|
|
|
|
/// (Note however that layouts are *not* required to have positive
|
|
|
|
|
/// size, even though many allocators require that all memory
|
2017-08-10 17:16:18 -05:00
|
|
|
|
/// requests have positive size. A caller to the `Alloc::alloc`
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// method must either ensure that conditions like this are met, or
|
|
|
|
|
/// use specific allocators with looser requirements.)
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2018-04-20 06:56:07 -05:00
|
|
|
|
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
|
2018-08-01 08:37:38 -05:00
|
|
|
|
#[lang = "alloc_layout"]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
pub struct Layout {
|
|
|
|
|
// size of the requested block of memory, measured in bytes.
|
2018-05-30 14:00:37 -05:00
|
|
|
|
size_: usize,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
// alignment of the requested block of memory, measured in bytes.
|
|
|
|
|
// we ensure that this is always a power-of-two, because API's
|
|
|
|
|
// like `posix_memalign` require it and it is a reasonable
|
|
|
|
|
// constraint to impose on Layout constructors.
|
|
|
|
|
//
|
|
|
|
|
// (However, we do not analogously require `align >= sizeof(void*)`,
|
|
|
|
|
// even though that is *also* a requirement of `posix_memalign`.)
|
2018-05-30 14:00:37 -05:00
|
|
|
|
align_: NonZeroUsize,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
impl Layout {
|
2017-06-15 15:44:56 -05:00
|
|
|
|
/// Constructs a `Layout` from a given `size` and `align`,
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// or returns `LayoutErr` if either of the following conditions
|
2017-06-15 15:44:56 -05:00
|
|
|
|
/// are not met:
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
2018-06-27 08:04:49 -05:00
|
|
|
|
/// * `align` must not be zero,
|
|
|
|
|
///
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// * `align` must be a power of two,
|
|
|
|
|
///
|
|
|
|
|
/// * `size`, when rounded up to the nearest multiple of `align`,
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// must not overflow (i.e., the rounded value must be less than
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// `usize::MAX`).
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub fn from_size_align(size: usize, align: usize) -> Result<Self, LayoutErr> {
|
2017-06-15 15:44:56 -05:00
|
|
|
|
if !align.is_power_of_two() {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
return Err(LayoutErr { private: () });
|
2017-06-15 15:44:56 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// (power-of-two implies align != 0.)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
// Rounded up size is:
|
|
|
|
|
// size_rounded_up = (size + align - 1) & !(align - 1);
|
|
|
|
|
//
|
|
|
|
|
// We know from above that align != 0. If adding (align - 1)
|
|
|
|
|
// does not overflow, then rounding up will be fine.
|
|
|
|
|
//
|
|
|
|
|
// Conversely, &-masking with !(align - 1) will subtract off
|
|
|
|
|
// only low-order-bits. Thus if overflow occurs with the sum,
|
|
|
|
|
// the &-mask cannot subtract enough to undo that overflow.
|
|
|
|
|
//
|
|
|
|
|
// Above implies that checking for summation overflow is both
|
|
|
|
|
// necessary and sufficient.
|
2017-06-15 15:44:56 -05:00
|
|
|
|
if size > usize::MAX - (align - 1) {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
return Err(LayoutErr { private: () });
|
2017-06-15 15:44:56 -05:00
|
|
|
|
}
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
2017-06-03 16:54:08 -05:00
|
|
|
|
unsafe {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
Ok(Layout::from_size_align_unchecked(size, align))
|
2017-06-03 16:54:08 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout, bypassing all checks.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
2017-06-03 16:54:08 -05:00
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// This function is unsafe as it does not verify the preconditions from
|
|
|
|
|
/// [`Layout::from_size_align`](#method.from_size_align).
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub unsafe fn from_size_align_unchecked(size: usize, align: usize) -> Self {
|
2018-05-30 14:00:37 -05:00
|
|
|
|
Layout { size_: size, align_: NonZeroUsize::new_unchecked(align) }
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// The minimum size in bytes for a memory block of this layout.
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
2018-05-30 14:00:37 -05:00
|
|
|
|
pub fn size(&self) -> usize { self.size_ }
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
/// The minimum byte alignment for a memory block of this layout.
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
2018-05-30 14:00:37 -05:00
|
|
|
|
pub fn align(&self) -> usize { self.align_.get() }
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
/// Constructs a `Layout` suitable for holding a value of type `T`.
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2018-05-30 14:00:37 -05:00
|
|
|
|
#[inline]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
pub fn new<T>() -> Self {
|
|
|
|
|
let (size, align) = size_align::<T>();
|
2018-04-11 12:47:16 -05:00
|
|
|
|
// Note that the align is guaranteed by rustc to be a power of two and
|
|
|
|
|
// the size+align combo is guaranteed to fit in our address space. As a
|
|
|
|
|
// result use the unchecked constructor here to avoid inserting code
|
|
|
|
|
// that panics if it isn't optimized well enough.
|
|
|
|
|
debug_assert!(Layout::from_size_align(size, align).is_ok());
|
|
|
|
|
unsafe {
|
|
|
|
|
Layout::from_size_align_unchecked(size, align)
|
|
|
|
|
}
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Produces layout describing a record that could be used to
|
|
|
|
|
/// allocate backing structure for `T` (which could be a trait
|
|
|
|
|
/// or other unsized type like a slice).
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2018-05-30 14:00:37 -05:00
|
|
|
|
#[inline]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
pub fn for_value<T: ?Sized>(t: &T) -> Self {
|
|
|
|
|
let (size, align) = (mem::size_of_val(t), mem::align_of_val(t));
|
2018-04-11 12:47:16 -05:00
|
|
|
|
// See rationale in `new` for why this us using an unsafe variant below
|
|
|
|
|
debug_assert!(Layout::from_size_align(size, align).is_ok());
|
|
|
|
|
unsafe {
|
|
|
|
|
Layout::from_size_align_unchecked(size, align)
|
|
|
|
|
}
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout describing the record that can hold a value
|
|
|
|
|
/// of the same layout as `self`, but that also is aligned to
|
|
|
|
|
/// alignment `align` (measured in bytes).
|
|
|
|
|
///
|
|
|
|
|
/// If `self` already meets the prescribed alignment, then returns
|
|
|
|
|
/// `self`.
|
|
|
|
|
///
|
|
|
|
|
/// Note that this method does not add any padding to the overall
|
|
|
|
|
/// size, regardless of whether the returned layout has a different
|
|
|
|
|
/// alignment. In other words, if `K` has size 16, `K.align_to(32)`
|
|
|
|
|
/// will *still* have size 16.
|
|
|
|
|
///
|
2018-10-30 16:36:43 -05:00
|
|
|
|
/// Returns an error if the combination of `self.size()` and the given
|
|
|
|
|
/// `align` violates the conditions listed in
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// [`Layout::from_size_align`](#method.from_size_align).
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
2018-10-30 16:36:43 -05:00
|
|
|
|
pub fn align_to(&self, align: usize) -> Result<Self, LayoutErr> {
|
|
|
|
|
Layout::from_size_align(self.size(), cmp::max(self.align(), align))
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Returns the amount of padding we must insert after `self`
|
|
|
|
|
/// to ensure that the following address will satisfy `align`
|
|
|
|
|
/// (measured in bytes).
|
|
|
|
|
///
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// e.g., if `self.size()` is 9, then `self.padding_needed_for(4)`
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// returns 3, because that is the minimum number of bytes of
|
|
|
|
|
/// padding required to get a 4-aligned address (assuming that the
|
|
|
|
|
/// corresponding memory block starts at a 4-aligned address).
|
|
|
|
|
///
|
|
|
|
|
/// The return value of this function has no meaning if `align` is
|
|
|
|
|
/// not a power-of-two.
|
|
|
|
|
///
|
|
|
|
|
/// Note that the utility of the returned value requires `align`
|
|
|
|
|
/// to be less than or equal to the alignment of the starting
|
|
|
|
|
/// address for the whole allocated block of memory. One way to
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// satisfy this constraint is to ensure `align <= self.align()`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
pub fn padding_needed_for(&self, align: usize) -> usize {
|
|
|
|
|
let len = self.size();
|
|
|
|
|
|
|
|
|
|
// Rounded up value is:
|
|
|
|
|
// len_rounded_up = (len + align - 1) & !(align - 1);
|
|
|
|
|
// and then we return the padding difference: `len_rounded_up - len`.
|
|
|
|
|
//
|
|
|
|
|
// We use modular arithmetic throughout:
|
|
|
|
|
//
|
|
|
|
|
// 1. align is guaranteed to be > 0, so align - 1 is always
|
|
|
|
|
// valid.
|
|
|
|
|
//
|
|
|
|
|
// 2. `len + align - 1` can overflow by at most `align - 1`,
|
|
|
|
|
// so the &-mask wth `!(align - 1)` will ensure that in the
|
|
|
|
|
// case of overflow, `len_rounded_up` will itself be 0.
|
|
|
|
|
// Thus the returned padding, when added to `len`, yields 0,
|
|
|
|
|
// which trivially satisfies the alignment `align`.
|
|
|
|
|
//
|
|
|
|
|
// (Of course, attempts to allocate blocks of memory whose
|
|
|
|
|
// size and padding overflow in the above manner should cause
|
|
|
|
|
// the allocator to yield an error anyway.)
|
|
|
|
|
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let len_rounded_up = len.wrapping_add(align).wrapping_sub(1)
|
|
|
|
|
& !align.wrapping_sub(1);
|
2018-08-04 07:31:03 -05:00
|
|
|
|
len_rounded_up.wrapping_sub(len)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
2018-11-07 11:59:25 -06:00
|
|
|
|
/// Creates a layout by rounding the size of this layout up to a multiple
|
|
|
|
|
/// of the layout's alignment.
|
|
|
|
|
///
|
|
|
|
|
/// Returns `Err` if the padded size would overflow.
|
|
|
|
|
///
|
|
|
|
|
/// This is equivalent to adding the result of `padding_needed_for`
|
|
|
|
|
/// to the layout's current size.
|
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
|
|
|
|
#[inline]
|
|
|
|
|
pub fn pad_to_align(&self) -> Result<Layout, LayoutErr> {
|
|
|
|
|
let pad = self.padding_needed_for(self.align());
|
|
|
|
|
let new_size = self.size().checked_add(pad)
|
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
|
|
|
|
|
|
|
|
|
Layout::from_size_align(new_size, self.align())
|
|
|
|
|
}
|
|
|
|
|
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// Creates a layout describing the record for `n` instances of
|
|
|
|
|
/// `self`, with a suitable amount of padding between each to
|
|
|
|
|
/// ensure that each instance is given its requested size and
|
|
|
|
|
/// alignment. On success, returns `(k, offs)` where `k` is the
|
|
|
|
|
/// layout of the array and `offs` is the distance between the start
|
|
|
|
|
/// of each element in the array.
|
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// On arithmetic overflow, returns `LayoutErr`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2017-07-27 16:53:44 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub fn repeat(&self, n: usize) -> Result<(Self, usize), LayoutErr> {
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let padded_size = self.size().checked_add(self.padding_needed_for(self.align()))
|
2018-04-04 09:03:46 -05:00
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
|
|
|
|
let alloc_size = padded_size.checked_mul(n)
|
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
2018-06-02 09:11:57 -05:00
|
|
|
|
|
|
|
|
|
unsafe {
|
|
|
|
|
// self.align is already known to be valid and alloc_size has been
|
|
|
|
|
// padded already.
|
|
|
|
|
Ok((Layout::from_size_align_unchecked(alloc_size, self.align()), padded_size))
|
|
|
|
|
}
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout describing the record for `self` followed by
|
|
|
|
|
/// `next`, including any necessary padding to ensure that `next`
|
|
|
|
|
/// will be properly aligned. Note that the result layout will
|
|
|
|
|
/// satisfy the alignment properties of both `self` and `next`.
|
|
|
|
|
///
|
2018-10-27 16:38:37 -05:00
|
|
|
|
/// The resulting layout will be the same as that of a C struct containing
|
|
|
|
|
/// two fields with the layouts of `self` and `next`, in that order.
|
|
|
|
|
///
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// Returns `Some((k, offset))`, where `k` is layout of the concatenated
|
|
|
|
|
/// record and `offset` is the relative location, in bytes, of the
|
2017-08-11 13:34:14 -05:00
|
|
|
|
/// start of the `next` embedded within the concatenated record
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// (assuming that the record itself starts at offset 0).
|
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// On arithmetic overflow, returns `LayoutErr`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2018-06-02 09:11:57 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub fn extend(&self, next: Self) -> Result<(Self, usize), LayoutErr> {
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let new_align = cmp::max(self.align(), next.align());
|
2018-06-02 09:11:57 -05:00
|
|
|
|
let pad = self.padding_needed_for(next.align());
|
2017-06-15 15:44:56 -05:00
|
|
|
|
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let offset = self.size().checked_add(pad)
|
2018-04-04 09:03:46 -05:00
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let new_size = offset.checked_add(next.size())
|
2018-04-04 09:03:46 -05:00
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
2017-12-08 19:32:04 -06:00
|
|
|
|
|
|
|
|
|
let layout = Layout::from_size_align(new_size, new_align)?;
|
2018-04-04 09:03:46 -05:00
|
|
|
|
Ok((layout, offset))
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout describing the record for `n` instances of
|
|
|
|
|
/// `self`, with no padding between each instance.
|
|
|
|
|
///
|
|
|
|
|
/// Note that, unlike `repeat`, `repeat_packed` does not guarantee
|
|
|
|
|
/// that the repeated instances of `self` will be properly
|
|
|
|
|
/// aligned, even if a given instance of `self` is properly
|
|
|
|
|
/// aligned. In other words, if the layout returned by
|
|
|
|
|
/// `repeat_packed` is used to allocate an array, it is not
|
|
|
|
|
/// guaranteed that all elements in the array will be properly
|
|
|
|
|
/// aligned.
|
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// On arithmetic overflow, returns `LayoutErr`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2018-06-02 09:11:57 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub fn repeat_packed(&self, n: usize) -> Result<Self, LayoutErr> {
|
|
|
|
|
let size = self.size().checked_mul(n).ok_or(LayoutErr { private: () })?;
|
2018-05-30 14:00:37 -05:00
|
|
|
|
Layout::from_size_align(size, self.align())
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout describing the record for `self` followed by
|
|
|
|
|
/// `next` with no additional padding between the two. Since no
|
|
|
|
|
/// padding is inserted, the alignment of `next` is irrelevant,
|
2017-08-15 14:45:21 -05:00
|
|
|
|
/// and is not incorporated *at all* into the resulting layout.
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// On arithmetic overflow, returns `LayoutErr`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2018-06-02 09:11:57 -05:00
|
|
|
|
#[inline]
|
2018-10-30 16:36:43 -05:00
|
|
|
|
pub fn extend_packed(&self, next: Self) -> Result<Self, LayoutErr> {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
let new_size = self.size().checked_add(next.size())
|
|
|
|
|
.ok_or(LayoutErr { private: () })?;
|
2018-05-30 14:00:37 -05:00
|
|
|
|
let layout = Layout::from_size_align(new_size, self.align())?;
|
2018-10-30 16:36:43 -05:00
|
|
|
|
Ok(layout)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Creates a layout describing the record for a `[T; n]`.
|
|
|
|
|
///
|
2018-05-30 14:00:37 -05:00
|
|
|
|
/// On arithmetic overflow, returns `LayoutErr`.
|
2018-10-25 17:48:27 -05:00
|
|
|
|
#[unstable(feature = "alloc_layout_extra", issue = "55724")]
|
2018-06-02 09:11:57 -05:00
|
|
|
|
#[inline]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
pub fn array<T>(n: usize) -> Result<Self, LayoutErr> {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
Layout::new::<T>()
|
|
|
|
|
.repeat(n)
|
|
|
|
|
.map(|(k, offs)| {
|
|
|
|
|
debug_assert!(offs == mem::size_of::<T>());
|
|
|
|
|
k
|
|
|
|
|
})
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// The parameters given to `Layout::from_size_align`
|
|
|
|
|
/// or some other `Layout` constructor
|
|
|
|
|
/// do not satisfy its documented constraints.
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
#[derive(Clone, PartialEq, Eq, Debug)]
|
|
|
|
|
pub struct LayoutErr {
|
|
|
|
|
private: ()
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// (we need this for downstream impl of trait Error)
|
2018-05-31 12:10:08 -05:00
|
|
|
|
#[stable(feature = "alloc_layout", since = "1.28.0")]
|
2018-04-04 09:03:46 -05:00
|
|
|
|
impl fmt::Display for LayoutErr {
|
2019-04-18 18:37:12 -05:00
|
|
|
|
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
f.write_str("invalid parameters to Layout::from_size_align")
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// The `AllocErr` error indicates an allocation failure
|
|
|
|
|
/// that may be due to resource exhaustion or to
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// something wrong when combining the given input arguments with this
|
|
|
|
|
/// allocator.
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
#[derive(Clone, PartialEq, Eq, Debug)]
|
2018-04-03 08:41:09 -05:00
|
|
|
|
pub struct AllocErr;
|
2017-06-13 14:57:49 -05:00
|
|
|
|
|
|
|
|
|
// (we need this for downstream impl of trait Error)
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
2017-06-13 14:57:49 -05:00
|
|
|
|
impl fmt::Display for AllocErr {
|
2019-04-18 18:37:12 -05:00
|
|
|
|
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
2018-04-03 08:41:09 -05:00
|
|
|
|
f.write_str("memory allocation failed")
|
2017-06-13 14:57:49 -05:00
|
|
|
|
}
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// The `CannotReallocInPlace` error is used when `grow_in_place` or
|
|
|
|
|
/// `shrink_in_place` were unable to reuse the given memory block for
|
|
|
|
|
/// a requested layout.
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
#[derive(Clone, PartialEq, Eq, Debug)]
|
|
|
|
|
pub struct CannotReallocInPlace;
|
|
|
|
|
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
2017-06-13 14:57:49 -05:00
|
|
|
|
impl CannotReallocInPlace {
|
|
|
|
|
pub fn description(&self) -> &str {
|
|
|
|
|
"cannot reallocate allocator's memory in place"
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// (we need this for downstream impl of trait Error)
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
2017-06-13 14:57:49 -05:00
|
|
|
|
impl fmt::Display for CannotReallocInPlace {
|
2019-04-18 18:37:12 -05:00
|
|
|
|
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
2017-06-13 14:57:49 -05:00
|
|
|
|
write!(f, "{}", self.description())
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// A memory allocator that can be registered as the standard library’s default
|
2018-04-11 13:16:45 -05:00
|
|
|
|
/// though the `#[global_allocator]` attributes.
|
2018-05-31 11:23:42 -05:00
|
|
|
|
///
|
|
|
|
|
/// Some of the methods require that a memory block be *currently
|
|
|
|
|
/// allocated* via an allocator. This means that:
|
|
|
|
|
///
|
|
|
|
|
/// * the starting address for that memory block was previously
|
|
|
|
|
/// returned by a previous call to an allocation method
|
|
|
|
|
/// such as `alloc`, and
|
|
|
|
|
///
|
|
|
|
|
/// * the memory block has not been subsequently deallocated, where
|
|
|
|
|
/// blocks are deallocated either by being passed to a deallocation
|
|
|
|
|
/// method such as `dealloc` or by being
|
|
|
|
|
/// passed to a reallocation method that returns a non-null pointer.
|
|
|
|
|
///
|
|
|
|
|
///
|
|
|
|
|
/// # Example
|
|
|
|
|
///
|
|
|
|
|
/// ```no_run
|
|
|
|
|
/// use std::alloc::{GlobalAlloc, Layout, alloc};
|
|
|
|
|
/// use std::ptr::null_mut;
|
|
|
|
|
///
|
|
|
|
|
/// struct MyAllocator;
|
|
|
|
|
///
|
|
|
|
|
/// unsafe impl GlobalAlloc for MyAllocator {
|
|
|
|
|
/// unsafe fn alloc(&self, _layout: Layout) -> *mut u8 { null_mut() }
|
|
|
|
|
/// unsafe fn dealloc(&self, _ptr: *mut u8, _layout: Layout) {}
|
|
|
|
|
/// }
|
|
|
|
|
///
|
|
|
|
|
/// #[global_allocator]
|
|
|
|
|
/// static A: MyAllocator = MyAllocator;
|
|
|
|
|
///
|
|
|
|
|
/// fn main() {
|
|
|
|
|
/// unsafe {
|
|
|
|
|
/// assert!(alloc(Layout::new::<u32>()).is_null())
|
|
|
|
|
/// }
|
|
|
|
|
/// }
|
|
|
|
|
/// ```
|
|
|
|
|
///
|
2019-02-25 01:01:35 -06:00
|
|
|
|
/// # Safety
|
2018-05-31 11:23:42 -05:00
|
|
|
|
///
|
|
|
|
|
/// The `GlobalAlloc` trait is an `unsafe` trait for a number of reasons, and
|
|
|
|
|
/// implementors must ensure that they adhere to these contracts:
|
|
|
|
|
///
|
2019-02-09 15:23:30 -06:00
|
|
|
|
/// * It's undefined behavior if global allocators unwind. This restriction may
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// be lifted in the future, but currently a panic from any of these
|
|
|
|
|
/// functions may lead to memory unsafety.
|
|
|
|
|
///
|
|
|
|
|
/// * `Layout` queries and calculations in general must be correct. Callers of
|
|
|
|
|
/// this trait are allowed to rely on the contracts defined on each method,
|
|
|
|
|
/// and implementors must ensure such contracts remain true.
|
2018-05-31 12:13:18 -05:00
|
|
|
|
#[stable(feature = "global_alloc", since = "1.28.0")]
|
2018-04-03 07:07:06 -05:00
|
|
|
|
pub unsafe trait GlobalAlloc {
|
2018-04-11 13:16:45 -05:00
|
|
|
|
/// Allocate memory as described by the given `layout`.
|
|
|
|
|
///
|
|
|
|
|
/// Returns a pointer to newly-allocated memory,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// or null to indicate allocation failure.
|
2018-04-11 13:16:45 -05:00
|
|
|
|
///
|
|
|
|
|
/// # Safety
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure that `layout` has non-zero size.
|
|
|
|
|
///
|
|
|
|
|
/// (Extension subtraits might provide more specific bounds on
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// behavior, e.g., guarantee a sentinel address or a null pointer
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// in response to a zero-size allocation request.)
|
|
|
|
|
///
|
|
|
|
|
/// The allocated block of memory may or may not be initialized.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning a null pointer indicates that either memory is exhausted
|
|
|
|
|
/// or `layout` does not meet allocator's size or alignment constraints.
|
|
|
|
|
///
|
|
|
|
|
/// Implementations are encouraged to return null on memory
|
|
|
|
|
/// exhaustion rather than aborting, but this is not
|
|
|
|
|
/// a strict requirement. (Specifically: it is *legal* to
|
|
|
|
|
/// implement this trait atop an underlying native allocation
|
|
|
|
|
/// library that aborts on memory exhaustion.)
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2018-05-31 12:13:18 -05:00
|
|
|
|
#[stable(feature = "global_alloc", since = "1.28.0")]
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn alloc(&self, layout: Layout) -> *mut u8;
|
2018-04-03 07:07:06 -05:00
|
|
|
|
|
2018-04-11 13:16:45 -05:00
|
|
|
|
/// Deallocate the block of memory at the given `ptr` pointer with the given `layout`.
|
|
|
|
|
///
|
|
|
|
|
/// # Safety
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must denote a block of memory currently allocated via
|
|
|
|
|
/// this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * `layout` must be the same layout that was used
|
|
|
|
|
/// to allocated that block of memory,
|
2018-05-31 12:13:18 -05:00
|
|
|
|
#[stable(feature = "global_alloc", since = "1.28.0")]
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn dealloc(&self, ptr: *mut u8, layout: Layout);
|
2018-04-03 07:07:06 -05:00
|
|
|
|
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// Behaves like `alloc`, but also ensures that the contents
|
|
|
|
|
/// are set to zero before being returned.
|
|
|
|
|
///
|
|
|
|
|
/// # Safety
|
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe for the same reasons that `alloc` is.
|
|
|
|
|
/// However the allocated block of memory is guaranteed to be initialized.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning a null pointer indicates that either memory is exhausted
|
|
|
|
|
/// or `layout` does not meet allocator's size or alignment constraints,
|
|
|
|
|
/// just as in `alloc`.
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2018-05-31 12:13:18 -05:00
|
|
|
|
#[stable(feature = "global_alloc", since = "1.28.0")]
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn alloc_zeroed(&self, layout: Layout) -> *mut u8 {
|
2018-04-03 07:07:06 -05:00
|
|
|
|
let size = layout.size();
|
|
|
|
|
let ptr = self.alloc(layout);
|
|
|
|
|
if !ptr.is_null() {
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr::write_bytes(ptr, 0, size);
|
2018-04-03 07:07:06 -05:00
|
|
|
|
}
|
|
|
|
|
ptr
|
|
|
|
|
}
|
|
|
|
|
|
2018-11-10 06:31:49 -06:00
|
|
|
|
/// Shrink or grow a block of memory to the given `new_size`.
|
2018-04-11 13:16:45 -05:00
|
|
|
|
/// The block is described by the given `ptr` pointer and `layout`.
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// If this returns a non-null pointer, then ownership of the memory block
|
2018-10-22 11:21:55 -05:00
|
|
|
|
/// referenced by `ptr` has been transferred to this allocator.
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// The memory may or may not have been deallocated,
|
|
|
|
|
/// and should be considered unusable (unless of course it was
|
|
|
|
|
/// transferred back to the caller again via the return value of
|
|
|
|
|
/// this method).
|
2018-04-11 13:16:45 -05:00
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// If this method returns null, then ownership of the memory
|
|
|
|
|
/// block has not been transferred to this allocator, and the
|
|
|
|
|
/// contents of the memory block are unaltered.
|
2018-04-11 13:16:45 -05:00
|
|
|
|
///
|
2018-04-03 10:51:03 -05:00
|
|
|
|
/// # Safety
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must be currently allocated via this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * `layout` must be the same layout that was used
|
|
|
|
|
/// to allocated that block of memory,
|
|
|
|
|
///
|
|
|
|
|
/// * `new_size` must be greater than zero.
|
|
|
|
|
///
|
|
|
|
|
/// * `new_size`, when rounded up to the nearest multiple of `layout.align()`,
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// must not overflow (i.e., the rounded value must be less than `usize::MAX`).
|
2018-05-31 11:23:42 -05:00
|
|
|
|
///
|
|
|
|
|
/// (Extension subtraits might provide more specific bounds on
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// behavior, e.g., guarantee a sentinel address or a null pointer
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// in response to a zero-size allocation request.)
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returns null if the new layout does not meet the size
|
|
|
|
|
/// and alignment constraints of the allocator, or if reallocation
|
|
|
|
|
/// otherwise fails.
|
|
|
|
|
///
|
|
|
|
|
/// Implementations are encouraged to return null on memory
|
|
|
|
|
/// exhaustion rather than panicking or aborting, but this is not
|
|
|
|
|
/// a strict requirement. (Specifically: it is *legal* to
|
|
|
|
|
/// implement this trait atop an underlying native allocation
|
|
|
|
|
/// library that aborts on memory exhaustion.)
|
2018-04-11 13:16:45 -05:00
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// Clients wishing to abort computation in response to a
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// reallocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2018-05-31 12:13:18 -05:00
|
|
|
|
#[stable(feature = "global_alloc", since = "1.28.0")]
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn realloc(&self, ptr: *mut u8, layout: Layout, new_size: usize) -> *mut u8 {
|
2018-04-11 13:16:45 -05:00
|
|
|
|
let new_layout = Layout::from_size_align_unchecked(new_size, layout.align());
|
2018-04-03 07:07:06 -05:00
|
|
|
|
let new_ptr = self.alloc(new_layout);
|
|
|
|
|
if !new_ptr.is_null() {
|
|
|
|
|
ptr::copy_nonoverlapping(
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr,
|
|
|
|
|
new_ptr,
|
2018-04-11 13:16:45 -05:00
|
|
|
|
cmp::min(layout.size(), new_size),
|
2018-04-03 07:07:06 -05:00
|
|
|
|
);
|
2018-04-11 13:16:45 -05:00
|
|
|
|
self.dealloc(ptr, layout);
|
2018-04-03 07:07:06 -05:00
|
|
|
|
}
|
|
|
|
|
new_ptr
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// An implementation of `Alloc` can allocate, reallocate, and
|
|
|
|
|
/// deallocate arbitrary blocks of data described via `Layout`.
|
|
|
|
|
///
|
|
|
|
|
/// Some of the methods require that a memory block be *currently
|
|
|
|
|
/// allocated* via an allocator. This means that:
|
|
|
|
|
///
|
|
|
|
|
/// * the starting address for that memory block was previously
|
|
|
|
|
/// returned by a previous call to an allocation method (`alloc`,
|
|
|
|
|
/// `alloc_zeroed`, `alloc_excess`, `alloc_one`, `alloc_array`) or
|
|
|
|
|
/// reallocation method (`realloc`, `realloc_excess`, or
|
|
|
|
|
/// `realloc_array`), and
|
|
|
|
|
///
|
|
|
|
|
/// * the memory block has not been subsequently deallocated, where
|
|
|
|
|
/// blocks are deallocated either by being passed to a deallocation
|
|
|
|
|
/// method (`dealloc`, `dealloc_one`, `dealloc_array`) or by being
|
|
|
|
|
/// passed to a reallocation method (see above) that returns `Ok`.
|
|
|
|
|
///
|
|
|
|
|
/// A note regarding zero-sized types and zero-sized layouts: many
|
|
|
|
|
/// methods in the `Alloc` trait state that allocation requests
|
|
|
|
|
/// must be non-zero size, or else undefined behavior can result.
|
|
|
|
|
///
|
|
|
|
|
/// * However, some higher-level allocation methods (`alloc_one`,
|
|
|
|
|
/// `alloc_array`) are well-defined on zero-sized types and can
|
|
|
|
|
/// optionally support them: it is left up to the implementor
|
|
|
|
|
/// whether to return `Err`, or to return `Ok` with some pointer.
|
|
|
|
|
///
|
|
|
|
|
/// * If an `Alloc` implementation chooses to return `Ok` in this
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// case (i.e., the pointer denotes a zero-sized inaccessible block)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// then that returned pointer must be considered "currently
|
|
|
|
|
/// allocated". On such an allocator, *all* methods that take
|
|
|
|
|
/// currently-allocated pointers as inputs must accept these
|
|
|
|
|
/// zero-sized pointers, *without* causing undefined behavior.
|
|
|
|
|
///
|
|
|
|
|
/// * In other words, if a zero-sized pointer can flow out of an
|
|
|
|
|
/// allocator, then that allocator must likewise accept that pointer
|
|
|
|
|
/// flowing back into its deallocation and reallocation methods.
|
|
|
|
|
///
|
|
|
|
|
/// Some of the methods require that a layout *fit* a memory block.
|
|
|
|
|
/// What it means for a layout to "fit" a memory block means (or
|
|
|
|
|
/// equivalently, for a memory block to "fit" a layout) is that the
|
|
|
|
|
/// following two conditions must hold:
|
|
|
|
|
///
|
|
|
|
|
/// 1. The block's starting address must be aligned to `layout.align()`.
|
|
|
|
|
///
|
|
|
|
|
/// 2. The block's size must fall in the range `[use_min, use_max]`, where:
|
|
|
|
|
///
|
|
|
|
|
/// * `use_min` is `self.usable_size(layout).0`, and
|
|
|
|
|
///
|
|
|
|
|
/// * `use_max` is the capacity that was (or would have been)
|
|
|
|
|
/// returned when (if) the block was allocated via a call to
|
|
|
|
|
/// `alloc_excess` or `realloc_excess`.
|
|
|
|
|
///
|
|
|
|
|
/// Note that:
|
|
|
|
|
///
|
|
|
|
|
/// * the size of the layout most recently used to allocate the block
|
|
|
|
|
/// is guaranteed to be in the range `[use_min, use_max]`, and
|
|
|
|
|
///
|
|
|
|
|
/// * a lower-bound on `use_max` can be safely approximated by a call to
|
|
|
|
|
/// `usable_size`.
|
|
|
|
|
///
|
|
|
|
|
/// * if a layout `k` fits a memory block (denoted by `ptr`)
|
|
|
|
|
/// currently allocated via an allocator `a`, then it is legal to
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// use that layout to deallocate it, i.e., `a.dealloc(ptr, k);`.
|
2017-08-22 16:36:49 -05:00
|
|
|
|
///
|
2019-02-25 01:01:35 -06:00
|
|
|
|
/// # Safety
|
2017-08-22 16:36:49 -05:00
|
|
|
|
///
|
|
|
|
|
/// The `Alloc` trait is an `unsafe` trait for a number of reasons, and
|
|
|
|
|
/// implementors must ensure that they adhere to these contracts:
|
|
|
|
|
///
|
|
|
|
|
/// * Pointers returned from allocation functions must point to valid memory and
|
|
|
|
|
/// retain their validity until at least the instance of `Alloc` is dropped
|
|
|
|
|
/// itself.
|
|
|
|
|
///
|
|
|
|
|
/// * `Layout` queries and calculations in general must be correct. Callers of
|
|
|
|
|
/// this trait are allowed to rely on the contracts defined on each method,
|
|
|
|
|
/// and implementors must ensure such contracts remain true.
|
|
|
|
|
///
|
|
|
|
|
/// Note that this list may get tweaked over time as clarifications are made in
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// the future.
|
2018-05-31 11:36:51 -05:00
|
|
|
|
#[unstable(feature = "allocator_api", issue = "32838")]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
pub unsafe trait Alloc {
|
|
|
|
|
|
2018-05-31 11:23:42 -05:00
|
|
|
|
// (Note: some existing allocators have unspecified but well-defined
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
// behavior in response to a zero size allocation request ;
|
2018-11-26 20:59:49 -06:00
|
|
|
|
// e.g., in C, `malloc` of 0 will either return a null pointer or a
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
// unique pointer, but will not have arbitrary undefined
|
2018-05-31 11:23:42 -05:00
|
|
|
|
// behavior.
|
|
|
|
|
// However in jemalloc for example,
|
|
|
|
|
// `mallocx(0)` is documented as undefined behavior.)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
/// Returns a pointer meeting the size and alignment guarantees of
|
|
|
|
|
/// `layout`.
|
|
|
|
|
///
|
|
|
|
|
/// If this method returns an `Ok(addr)`, then the `addr` returned
|
|
|
|
|
/// will be non-null address pointing to a block of storage
|
|
|
|
|
/// suitable for holding an instance of `layout`.
|
|
|
|
|
///
|
|
|
|
|
/// The returned block of storage may or may not have its contents
|
|
|
|
|
/// initialized. (Extension subtraits might restrict this
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// behavior, e.g., to ensure initialization to particular sets of
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// bit patterns.)
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure that `layout` has non-zero size.
|
|
|
|
|
///
|
|
|
|
|
/// (Extension subtraits might provide more specific bounds on
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// behavior, e.g., guarantee a sentinel address or a null pointer
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// in response to a zero-size allocation request.)
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `layout` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints.
|
|
|
|
|
///
|
|
|
|
|
/// Implementations are encouraged to return `Err` on memory
|
|
|
|
|
/// exhaustion rather than panicking or aborting, but this is not
|
|
|
|
|
/// a strict requirement. (Specifically: it is *legal* to
|
|
|
|
|
/// implement this trait atop an underlying native allocation
|
|
|
|
|
/// library that aborts on memory exhaustion.)
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn alloc(&mut self, layout: Layout) -> Result<NonNull<u8>, AllocErr>;
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
/// Deallocate the memory referenced by `ptr`.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must denote a block of memory currently allocated via
|
|
|
|
|
/// this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * `layout` must *fit* that block of memory,
|
|
|
|
|
///
|
|
|
|
|
/// * In addition to fitting the block of memory `layout`, the
|
|
|
|
|
/// alignment of the `layout` must match the alignment used
|
|
|
|
|
/// to allocate that block of memory.
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn dealloc(&mut self, ptr: NonNull<u8>, layout: Layout);
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
|
|
|
|
|
// == ALLOCATOR-SPECIFIC QUANTITIES AND LIMITS ==
|
|
|
|
|
// usable_size
|
|
|
|
|
|
|
|
|
|
/// Returns bounds on the guaranteed usable size of a successful
|
|
|
|
|
/// allocation created with the specified `layout`.
|
|
|
|
|
///
|
|
|
|
|
/// In particular, if one has a memory block allocated via a given
|
|
|
|
|
/// allocator `a` and layout `k` where `a.usable_size(k)` returns
|
|
|
|
|
/// `(l, u)`, then one can pass that block to `a.dealloc()` with a
|
|
|
|
|
/// layout in the size range [l, u].
|
|
|
|
|
///
|
|
|
|
|
/// (All implementors of `usable_size` must ensure that
|
|
|
|
|
/// `l <= k.size() <= u`)
|
|
|
|
|
///
|
|
|
|
|
/// Both the lower- and upper-bounds (`l` and `u` respectively)
|
|
|
|
|
/// are provided, because an allocator based on size classes could
|
|
|
|
|
/// misbehave if one attempts to deallocate a block without
|
|
|
|
|
/// providing a correct value for its size (i.e., one within the
|
|
|
|
|
/// range `[l, u]`).
|
|
|
|
|
///
|
|
|
|
|
/// Clients who wish to make use of excess capacity are encouraged
|
|
|
|
|
/// to use the `alloc_excess` and `realloc_excess` instead, as
|
|
|
|
|
/// this method is constrained to report conservative values that
|
|
|
|
|
/// serve as valid bounds for *all possible* allocation method
|
|
|
|
|
/// calls.
|
|
|
|
|
///
|
|
|
|
|
/// However, for clients that do not wish to track the capacity
|
|
|
|
|
/// returned by `alloc_excess` locally, this method is likely to
|
|
|
|
|
/// produce useful results.
|
2017-06-03 16:54:08 -05:00
|
|
|
|
#[inline]
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
fn usable_size(&self, layout: &Layout) -> (usize, usize) {
|
|
|
|
|
(layout.size(), layout.size())
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// == METHODS FOR MEMORY REUSE ==
|
|
|
|
|
// realloc. alloc_excess, realloc_excess
|
|
|
|
|
|
|
|
|
|
/// Returns a pointer suitable for holding data described by
|
2018-11-10 06:31:49 -06:00
|
|
|
|
/// a new layout with `layout`’s alignment and a size given
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// by `new_size`. To
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// accomplish this, this may extend or shrink the allocation
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// referenced by `ptr` to fit the new layout.
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// If this returns `Ok`, then ownership of the memory block
|
|
|
|
|
/// referenced by `ptr` has been transferred to this
|
|
|
|
|
/// allocator. The memory may or may not have been freed, and
|
|
|
|
|
/// should be considered unusable (unless of course it was
|
|
|
|
|
/// transferred back to the caller again via the return value of
|
|
|
|
|
/// this method).
|
|
|
|
|
///
|
|
|
|
|
/// If this method returns `Err`, then ownership of the memory
|
|
|
|
|
/// block has not been transferred to this allocator, and the
|
|
|
|
|
/// contents of the memory block are unaltered.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must be currently allocated via this allocator,
|
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// * `layout` must *fit* the `ptr` (see above). (The `new_size`
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// argument need not fit it.)
|
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// * `new_size` must be greater than zero.
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// * `new_size`, when rounded up to the nearest multiple of `layout.align()`,
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// must not overflow (i.e., the rounded value must be less than `usize::MAX`).
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// (Extension subtraits might provide more specific bounds on
|
2018-11-26 20:59:49 -06:00
|
|
|
|
/// behavior, e.g., guarantee a sentinel address or a null pointer
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// in response to a zero-size allocation request.)
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// Returns `Err` only if the new layout
|
|
|
|
|
/// does not meet the allocator's size
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// and alignment constraints of the allocator, or if reallocation
|
|
|
|
|
/// otherwise fails.
|
|
|
|
|
///
|
|
|
|
|
/// Implementations are encouraged to return `Err` on memory
|
|
|
|
|
/// exhaustion rather than panicking or aborting, but this is not
|
|
|
|
|
/// a strict requirement. (Specifically: it is *legal* to
|
|
|
|
|
/// implement this trait atop an underlying native allocation
|
|
|
|
|
/// library that aborts on memory exhaustion.)
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// Clients wishing to abort computation in response to a
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// reallocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
unsafe fn realloc(&mut self,
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr: NonNull<u8>,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
layout: Layout,
|
2018-05-31 01:57:43 -05:00
|
|
|
|
new_size: usize) -> Result<NonNull<u8>, AllocErr> {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let old_size = layout.size();
|
|
|
|
|
|
2018-04-04 10:19:16 -05:00
|
|
|
|
if new_size >= old_size {
|
|
|
|
|
if let Ok(()) = self.grow_in_place(ptr, layout.clone(), new_size) {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
return Ok(ptr);
|
|
|
|
|
}
|
2018-04-04 10:19:16 -05:00
|
|
|
|
} else if new_size < old_size {
|
|
|
|
|
if let Ok(()) = self.shrink_in_place(ptr, layout.clone(), new_size) {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
return Ok(ptr);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// otherwise, fall back on alloc + copy + dealloc.
|
2018-04-04 10:19:16 -05:00
|
|
|
|
let new_layout = Layout::from_size_align_unchecked(new_size, layout.align());
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let result = self.alloc(new_layout);
|
|
|
|
|
if let Ok(new_ptr) = result {
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr::copy_nonoverlapping(ptr.as_ptr(),
|
|
|
|
|
new_ptr.as_ptr(),
|
2018-04-02 18:51:02 -05:00
|
|
|
|
cmp::min(old_size, new_size));
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
self.dealloc(ptr, layout);
|
|
|
|
|
}
|
|
|
|
|
result
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Behaves like `alloc`, but also ensures that the contents
|
|
|
|
|
/// are set to zero before being returned.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe for the same reasons that `alloc` is.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `layout` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints, just as in `alloc`.
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2018-05-31 01:57:43 -05:00
|
|
|
|
unsafe fn alloc_zeroed(&mut self, layout: Layout) -> Result<NonNull<u8>, AllocErr> {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let size = layout.size();
|
|
|
|
|
let p = self.alloc(layout);
|
|
|
|
|
if let Ok(p) = p {
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr::write_bytes(p.as_ptr(), 0, size);
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
p
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Behaves like `alloc`, but also returns the whole size of
|
|
|
|
|
/// the returned block. For some `layout` inputs, like arrays, this
|
|
|
|
|
/// may include extra storage usable for additional data.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe for the same reasons that `alloc` is.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `layout` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints, just as in `alloc`.
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
unsafe fn alloc_excess(&mut self, layout: Layout) -> Result<Excess, AllocErr> {
|
|
|
|
|
let usable_size = self.usable_size(&layout);
|
|
|
|
|
self.alloc(layout).map(|p| Excess(p, usable_size.1))
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Behaves like `realloc`, but also returns the whole size of
|
|
|
|
|
/// the returned block. For some `layout` inputs, like arrays, this
|
|
|
|
|
/// may include extra storage usable for additional data.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe for the same reasons that `realloc` is.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `layout` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints, just as in `realloc`.
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// Clients wishing to abort computation in response to a
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// reallocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
unsafe fn realloc_excess(&mut self,
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr: NonNull<u8>,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
layout: Layout,
|
2018-04-04 10:19:16 -05:00
|
|
|
|
new_size: usize) -> Result<Excess, AllocErr> {
|
|
|
|
|
let new_layout = Layout::from_size_align_unchecked(new_size, layout.align());
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let usable_size = self.usable_size(&new_layout);
|
2018-04-04 10:19:16 -05:00
|
|
|
|
self.realloc(ptr, layout, new_size)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
.map(|p| Excess(p, usable_size.1))
|
|
|
|
|
}
|
|
|
|
|
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// Attempts to extend the allocation referenced by `ptr` to fit `new_size`.
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// If this returns `Ok`, then the allocator has asserted that the
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// memory block referenced by `ptr` now fits `new_size`, and thus can
|
|
|
|
|
/// be used to carry data of a layout of that size and same alignment as
|
|
|
|
|
/// `layout`. (The allocator is allowed to
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// expend effort to accomplish this, such as extending the memory block to
|
|
|
|
|
/// include successor blocks, or virtual memory tricks.)
|
|
|
|
|
///
|
|
|
|
|
/// Regardless of what this method returns, ownership of the
|
|
|
|
|
/// memory block referenced by `ptr` has not been transferred, and
|
|
|
|
|
/// the contents of the memory block are unaltered.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must be currently allocated via this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * `layout` must *fit* the `ptr` (see above); note the
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// `new_size` argument need not fit it,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// * `new_size` must not be less than `layout.size()`,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returns `Err(CannotReallocInPlace)` when the allocator is
|
|
|
|
|
/// unable to assert that the memory block referenced by `ptr`
|
|
|
|
|
/// could fit `layout`.
|
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// Note that one cannot pass `CannotReallocInPlace` to the `handle_alloc_error`
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// function; clients are expected either to be able to recover from
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// `grow_in_place` failures without aborting, or to fall back on
|
|
|
|
|
/// another reallocation method before resorting to an abort.
|
|
|
|
|
unsafe fn grow_in_place(&mut self,
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr: NonNull<u8>,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
layout: Layout,
|
2018-04-04 10:19:16 -05:00
|
|
|
|
new_size: usize) -> Result<(), CannotReallocInPlace> {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let _ = ptr; // this default implementation doesn't care about the actual address.
|
2018-05-30 14:00:37 -05:00
|
|
|
|
debug_assert!(new_size >= layout.size());
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let (_l, u) = self.usable_size(&layout);
|
|
|
|
|
// _l <= layout.size() [guaranteed by usable_size()]
|
|
|
|
|
// layout.size() <= new_layout.size() [required by this method]
|
2018-04-04 10:19:16 -05:00
|
|
|
|
if new_size <= u {
|
2018-08-04 07:31:03 -05:00
|
|
|
|
Ok(())
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
} else {
|
2018-08-04 07:31:03 -05:00
|
|
|
|
Err(CannotReallocInPlace)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// Attempts to shrink the allocation referenced by `ptr` to fit `new_size`.
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// If this returns `Ok`, then the allocator has asserted that the
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// memory block referenced by `ptr` now fits `new_size`, and
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// thus can only be used to carry data of that smaller
|
|
|
|
|
/// layout. (The allocator is allowed to take advantage of this,
|
|
|
|
|
/// carving off portions of the block for reuse elsewhere.) The
|
|
|
|
|
/// truncated contents of the block within the smaller layout are
|
|
|
|
|
/// unaltered, and ownership of block has not been transferred.
|
|
|
|
|
///
|
|
|
|
|
/// If this returns `Err`, then the memory block is considered to
|
|
|
|
|
/// still represent the original (larger) `layout`. None of the
|
|
|
|
|
/// block has been carved off for reuse elsewhere, ownership of
|
|
|
|
|
/// the memory block has not been transferred, and the contents of
|
|
|
|
|
/// the memory block are unaltered.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must be currently allocated via this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * `layout` must *fit* the `ptr` (see above); note the
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// `new_size` argument need not fit it,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
2018-04-04 10:19:16 -05:00
|
|
|
|
/// * `new_size` must not be greater than `layout.size()`
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// (and must be greater than zero),
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returns `Err(CannotReallocInPlace)` when the allocator is
|
|
|
|
|
/// unable to assert that the memory block referenced by `ptr`
|
|
|
|
|
/// could fit `layout`.
|
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// Note that one cannot pass `CannotReallocInPlace` to the `handle_alloc_error`
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// function; clients are expected either to be able to recover from
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
/// `shrink_in_place` failures without aborting, or to fall back
|
|
|
|
|
/// on another reallocation method before resorting to an abort.
|
|
|
|
|
unsafe fn shrink_in_place(&mut self,
|
2018-05-31 01:57:43 -05:00
|
|
|
|
ptr: NonNull<u8>,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
layout: Layout,
|
2018-04-04 10:19:16 -05:00
|
|
|
|
new_size: usize) -> Result<(), CannotReallocInPlace> {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let _ = ptr; // this default implementation doesn't care about the actual address.
|
2018-05-30 14:00:37 -05:00
|
|
|
|
debug_assert!(new_size <= layout.size());
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
let (l, _u) = self.usable_size(&layout);
|
|
|
|
|
// layout.size() <= _u [guaranteed by usable_size()]
|
|
|
|
|
// new_layout.size() <= layout.size() [required by this method]
|
2018-04-04 10:19:16 -05:00
|
|
|
|
if l <= new_size {
|
2018-08-04 07:31:03 -05:00
|
|
|
|
Ok(())
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
} else {
|
2018-08-04 07:31:03 -05:00
|
|
|
|
Err(CannotReallocInPlace)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// == COMMON USAGE PATTERNS ==
|
|
|
|
|
// alloc_one, dealloc_one, alloc_array, realloc_array. dealloc_array
|
|
|
|
|
|
|
|
|
|
/// Allocates a block suitable for holding an instance of `T`.
|
|
|
|
|
///
|
|
|
|
|
/// Captures a common usage pattern for allocators.
|
|
|
|
|
///
|
|
|
|
|
/// The returned block is suitable for passing to the
|
|
|
|
|
/// `alloc`/`realloc` methods of this allocator.
|
|
|
|
|
///
|
|
|
|
|
/// Note to implementors: If this returns `Ok(ptr)`, then `ptr`
|
|
|
|
|
/// must be considered "currently allocated" and must be
|
|
|
|
|
/// acceptable input to methods such as `realloc` or `dealloc`,
|
|
|
|
|
/// *even if* `T` is a zero-sized type. In other words, if your
|
|
|
|
|
/// `Alloc` implementation overrides this method in a manner
|
|
|
|
|
/// that can return a zero-sized `ptr`, then all reallocation and
|
|
|
|
|
/// deallocation methods need to be similarly overridden to accept
|
|
|
|
|
/// such values as input.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `T` does not meet allocator's size or alignment constraints.
|
|
|
|
|
///
|
|
|
|
|
/// For zero-sized `T`, may return either of `Ok` or `Err`, but
|
|
|
|
|
/// will *not* yield undefined behavior.
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2017-12-22 12:12:22 -06:00
|
|
|
|
fn alloc_one<T>(&mut self) -> Result<NonNull<T>, AllocErr>
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
where Self: Sized
|
|
|
|
|
{
|
|
|
|
|
let k = Layout::new::<T>();
|
|
|
|
|
if k.size() > 0 {
|
2018-04-02 18:51:02 -05:00
|
|
|
|
unsafe { self.alloc(k).map(|p| p.cast()) }
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
} else {
|
2018-04-03 08:41:09 -05:00
|
|
|
|
Err(AllocErr)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Deallocates a block suitable for holding an instance of `T`.
|
|
|
|
|
///
|
|
|
|
|
/// The given block must have been produced by this allocator,
|
|
|
|
|
/// and must be suitable for storing a `T` (in terms of alignment
|
|
|
|
|
/// as well as minimum and maximum size); otherwise yields
|
|
|
|
|
/// undefined behavior.
|
|
|
|
|
///
|
|
|
|
|
/// Captures a common usage pattern for allocators.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure both:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must denote a block of memory currently allocated via this allocator
|
|
|
|
|
///
|
|
|
|
|
/// * the layout of `T` must *fit* that block of memory.
|
2017-12-22 12:12:22 -06:00
|
|
|
|
unsafe fn dealloc_one<T>(&mut self, ptr: NonNull<T>)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
where Self: Sized
|
|
|
|
|
{
|
|
|
|
|
let k = Layout::new::<T>();
|
|
|
|
|
if k.size() > 0 {
|
2018-05-31 01:57:43 -05:00
|
|
|
|
self.dealloc(ptr.cast(), k);
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Allocates a block suitable for holding `n` instances of `T`.
|
|
|
|
|
///
|
|
|
|
|
/// Captures a common usage pattern for allocators.
|
|
|
|
|
///
|
|
|
|
|
/// The returned block is suitable for passing to the
|
|
|
|
|
/// `alloc`/`realloc` methods of this allocator.
|
|
|
|
|
///
|
|
|
|
|
/// Note to implementors: If this returns `Ok(ptr)`, then `ptr`
|
|
|
|
|
/// must be considered "currently allocated" and must be
|
|
|
|
|
/// acceptable input to methods such as `realloc` or `dealloc`,
|
|
|
|
|
/// *even if* `T` is a zero-sized type. In other words, if your
|
|
|
|
|
/// `Alloc` implementation overrides this method in a manner
|
|
|
|
|
/// that can return a zero-sized `ptr`, then all reallocation and
|
|
|
|
|
/// deallocation methods need to be similarly overridden to accept
|
|
|
|
|
/// such values as input.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `[T; n]` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints.
|
|
|
|
|
///
|
|
|
|
|
/// For zero-sized `T` or `n == 0`, may return either of `Ok` or
|
|
|
|
|
/// `Err`, but will *not* yield undefined behavior.
|
|
|
|
|
///
|
|
|
|
|
/// Always returns `Err` on arithmetic overflow.
|
|
|
|
|
///
|
|
|
|
|
/// Clients wishing to abort computation in response to an
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// allocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
2017-12-22 12:12:22 -06:00
|
|
|
|
fn alloc_array<T>(&mut self, n: usize) -> Result<NonNull<T>, AllocErr>
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
where Self: Sized
|
|
|
|
|
{
|
|
|
|
|
match Layout::array::<T>(n) {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
Ok(ref layout) if layout.size() > 0 => {
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
unsafe {
|
2018-04-02 18:51:02 -05:00
|
|
|
|
self.alloc(layout.clone()).map(|p| p.cast())
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
2018-04-03 08:41:09 -05:00
|
|
|
|
_ => Err(AllocErr),
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Reallocates a block previously suitable for holding `n_old`
|
|
|
|
|
/// instances of `T`, returning a block suitable for holding
|
|
|
|
|
/// `n_new` instances of `T`.
|
|
|
|
|
///
|
|
|
|
|
/// Captures a common usage pattern for allocators.
|
|
|
|
|
///
|
|
|
|
|
/// The returned block is suitable for passing to the
|
|
|
|
|
/// `alloc`/`realloc` methods of this allocator.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure all of the following:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must be currently allocated via this allocator,
|
|
|
|
|
///
|
|
|
|
|
/// * the layout of `[T; n_old]` must *fit* that block of memory.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either memory is exhausted or
|
|
|
|
|
/// `[T; n_new]` does not meet allocator's size or alignment
|
|
|
|
|
/// constraints.
|
|
|
|
|
///
|
|
|
|
|
/// For zero-sized `T` or `n_new == 0`, may return either of `Ok` or
|
|
|
|
|
/// `Err`, but will *not* yield undefined behavior.
|
|
|
|
|
///
|
|
|
|
|
/// Always returns `Err` on arithmetic overflow.
|
|
|
|
|
///
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// Clients wishing to abort computation in response to a
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// reallocation error are encouraged to call the [`handle_alloc_error`] function,
|
2018-05-31 11:23:42 -05:00
|
|
|
|
/// rather than directly invoking `panic!` or similar.
|
2018-06-01 02:18:25 -05:00
|
|
|
|
///
|
2018-06-13 17:32:30 -05:00
|
|
|
|
/// [`handle_alloc_error`]: ../../alloc/alloc/fn.handle_alloc_error.html
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
unsafe fn realloc_array<T>(&mut self,
|
2017-12-22 12:12:22 -06:00
|
|
|
|
ptr: NonNull<T>,
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
n_old: usize,
|
2017-12-22 12:12:22 -06:00
|
|
|
|
n_new: usize) -> Result<NonNull<T>, AllocErr>
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
where Self: Sized
|
|
|
|
|
{
|
2018-04-02 18:51:02 -05:00
|
|
|
|
match (Layout::array::<T>(n_old), Layout::array::<T>(n_new)) {
|
|
|
|
|
(Ok(ref k_old), Ok(ref k_new)) if k_old.size() > 0 && k_new.size() > 0 => {
|
2018-04-04 10:19:16 -05:00
|
|
|
|
debug_assert!(k_old.align() == k_new.align());
|
2018-05-31 01:57:43 -05:00
|
|
|
|
self.realloc(ptr.cast(), k_old.clone(), k_new.size()).map(NonNull::cast)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
_ => {
|
2018-04-03 08:41:09 -05:00
|
|
|
|
Err(AllocErr)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// Deallocates a block suitable for holding `n` instances of `T`.
|
|
|
|
|
///
|
|
|
|
|
/// Captures a common usage pattern for allocators.
|
|
|
|
|
///
|
2017-08-24 10:33:36 -05:00
|
|
|
|
/// # Safety
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
///
|
|
|
|
|
/// This function is unsafe because undefined behavior can result
|
|
|
|
|
/// if the caller does not ensure both:
|
|
|
|
|
///
|
|
|
|
|
/// * `ptr` must denote a block of memory currently allocated via this allocator
|
|
|
|
|
///
|
|
|
|
|
/// * the layout of `[T; n]` must *fit* that block of memory.
|
|
|
|
|
///
|
|
|
|
|
/// # Errors
|
|
|
|
|
///
|
|
|
|
|
/// Returning `Err` indicates that either `[T; n]` or the given
|
|
|
|
|
/// memory block does not meet allocator's size or alignment
|
|
|
|
|
/// constraints.
|
|
|
|
|
///
|
|
|
|
|
/// Always returns `Err` on arithmetic overflow.
|
2017-12-22 12:12:22 -06:00
|
|
|
|
unsafe fn dealloc_array<T>(&mut self, ptr: NonNull<T>, n: usize) -> Result<(), AllocErr>
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
where Self: Sized
|
|
|
|
|
{
|
|
|
|
|
match Layout::array::<T>(n) {
|
2018-04-04 09:03:46 -05:00
|
|
|
|
Ok(ref k) if k.size() > 0 => {
|
2018-05-31 01:57:43 -05:00
|
|
|
|
Ok(self.dealloc(ptr.cast(), k.clone()))
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
_ => {
|
2018-04-03 08:41:09 -05:00
|
|
|
|
Err(AllocErr)
|
Add API for `Alloc` trait.
Includes `alloc_zeroed` method that `RawVec` has come to depend on.
Exposed private `Layout::from_size_align` ctor to be `pub`, and added
explicit conditions for when it will panic (namely, when `align` is
not power-of-two, or if rounding up `size` to a multiple of `align`
overflows). Normalized all `Layout` construction to go through
`Layout::from_size_align`.
Addressed review feedback regarding `struct Layout` and zero-sized
layouts.
Restrict specification for `dealloc`, adding additional constraint
that the given alignment has to match that used to allocate the block.
(This is a maximally conservative constraint on the alignment. An open
question to resolve (before stabilization) is whether we can return to
a looser constraint such as the one previously specified.)
Split `fn realloc_in_place` into separate `fn grow_in_place` and `fn
shrink_in_place` methods, which have default impls that check against
usable_size for reuse. Make `realloc` default impl try `grow_in_place`
or `shrink_in_place` as appropriate before fallback on
alloc+copy+dealloc.
Drive-by: When reviewing calls to `padding_needed_for`, discovered
what I think was an over-conservative choice for its argument
alignment. Namely, in `fn extend`, we automatically realign the whole
resulting layout to satisfy both old (self) and new alignments. When
the old alignment exceeds the new, this means we would insert
unnecessary padding. So I changed the code to pass in `next.align`
instead of `new_align` to `padding_needed_for`.
Replaced ref to `realloc_in_place` with `grow_in_place`/`shrink_in_place`.
Revised docs replacing my idiosyncratic style of `fn foo` with just
`foo` when referring to the function or method `foo`.
(Alpha-renamed `Allocator` to `Alloc`.)
Post-rebased, added `Debug` derive for `allocator::Excess` to satisfy
`missing_debug_implementations`.
2017-05-23 07:47:09 -05:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|