HEADER_GLOBAL was using non-standard attributes to achieve what C++17
now permits with a keyword. Use the standard formulation.
Update the README to note that gcc is still not recommended, but because
of its poor codegen for 128-bit atomic compare and exchange, rather than
because it doesn't support the attribute used for HEADER_GLOBAL.
This commit changes the strategy for finding a free list from
a stack to a queue. This tends to avoid the slow path considerably more.
It has some memory overheads.
TOOD: We should move the bump allocation data out of the metaslab and
into the allocator. At the moment, the slab contains the bump allocation
data, we should move this into the allocator, as it only ever has one slab
it is bump allocating from per sizeclass.
Fix a false positive. The loops in the pagemap accessor are too
complicated for the static analyser to check, so it doesn't spot that a
non-null return is impossible.
Change the result of `operator=` on `Mod` to return `Mod&`. In this
example:
```c++
Mod<4, int> a;
int b = (a = 8);
```
The value of `b` will now be 0, not 8, and will equal the value of `a`.
Hopefully this is less confusing to users of this class.
This introduces a new `address_t` type and two new casts: `pointer_cast`
and `address_cast` for casting between an `address_t` and a pointer.
These should make it easier to audit the codebase for casts between
pointers and integers. In particular, the remaining `reinterpret_cast`s
and `pointer_cast`s should be the only places where we could perform
invalid pointer arithmetic.
Also adds a `pointer_offset` helper that adds an offset (in bytes) to a
pointer, preserving its original type. This is a sufficiently common
pattern that it seemed worthwhile to centralise it.
This is a generalisation of `std::nullptr_t` that allows non-0 values of
pointer-sized-but-not-a-pointer thing. `DLList` now takes an
`InvalidPointer`, `nullptr_t`, or some other compatible class as a
sentinel value instead of a `uintptr_t`.
The rounding code can be integrated directly into the sizeclass code,
which means, the code paths can be simplified slightly.
The special case for zero mantissa bits is not required, as the compiler
generates the same code from the more complex path due to
eliminating dead code after folding the constant zero through.
Introduce a wrapper that ensure various fields of meta data, represent
valid indexes into the the address space, and not beyond the current
slab/superslab.
If a sizeclass in the metadata is corrupted, then this can be used to
force an index beyond the end of these tables. This extends the tables
to the next power of two, and uses a mask on the index, so they are
always either a valid piece of data, or zero.
