5 Commits

Author SHA1 Message Date
Nikolas Klauser
e99c4906e4
[libc++] Granularize <cstddef> includes (#108696) 2024-10-31 02:20:10 +01:00
Louis Dionne
848b20de76
[libc++] Correctly handle custom deleters in hardened unique_ptr<T[]> (#110685)
It turns out that we can never do bounds-checking for unique_ptrs with
custom deleters, except when converting from a unique_ptr with a default
deleter to one with a custom deleter.

If we had an API like `std::make_unique` that allowed passing a custom
deleter, we could at least get bounds checking when the unique_ptr is
created through those APIs, but for now that is not possible.

Fixes #110683
2024-10-03 11:45:58 -04:00
Louis Dionne
18df9d23ea [libc++] Add an ABI setting to harden unique_ptr<T[]>::operator[] (#91798)
This allows catching OOB accesses inside `unique_ptr<T[]>` when the size
of the allocation is known. The size of the allocation can be known when
the unique_ptr has been created with make_unique & friends or when the
type necessitates an array cookie before the allocation.

This is a re-aplpication of 45a09d181 which had been reverted in
f11abac6 due to unrelated CI failures.
2024-09-30 08:32:35 -04:00
Chris B
f11abac652
Revert "[libc++][modules] Rewrite the modulemap to have fewer top-level modules (#107638)" (#110384)
This reverts 3 commits:
45a09d1811d5d6597385ef02ecf2d4b7320c37c5
24bc3244d4e221f4e6740f45e2bf15a1441a3076
bc6bd3bc1e99c7ec9e22dff23b4f4373fa02cae3

The GitHub pre-merge CI has been broken since this PR went in. This
change reverts it to see if I can get the pre-merge CI working again.
2024-09-28 21:47:09 -05:00
Louis Dionne
45a09d1811
[libc++] Add an ABI setting to harden unique_ptr<T[]>::operator[] (#91798)
This allows catching OOB accesses inside `unique_ptr<T[]>` when the size
of the allocation is known. The size of the allocation can be known when
the unique_ptr has been created with make_unique & friends or when the
type necessitates an array cookie before the allocation.
2024-09-27 08:49:22 -04:00