This document describes PartitionAlloc at a high level, with some architectural details. For implementation details, see the comments in
PartitionAlloc is a memory allocator optimized for space efficiency, allocation latency, and security.
A partition is a heap that is separated and protected from any other partitions, as well as from non-PartitionAlloc memory. The most typical use of partitions is to isolate certain object types. However, one can also isolate objects of certain sizes, or objects of a certain lifetime (as the caller prefers). Callers can create as many partitions as they need. The direct memory cost of partitions is minimal, but the implicit cost resulting from fragmentation is not to be underestimated.
Each partition holds multiple buckets. A bucket is a series of regions in a partition that contains similar-sized objects, e.g. one bucket holds sizes (240, 256], another (256, 288], and so on. Bucket sizes are geometrically-spaced, and go all the way up to
kMaxBucketed=960KiB (so called normal buckets). There are 8 buckets between each power of two. Note that buckets that aren‘t a multiple of
base::kAlignment can’t be used.
Larger allocations (>
kMaxBucketed) are realized by direct memory mapping (direct map).
PartitionAlloc is designed to be extremely fast in its fast paths. The fast paths of allocation and deallocation require very few (reasonably predictable) branches. The number of operations in the fast paths is minimal, leading to the possibility of inlining.
However, even the fast path isn't the fastest, because it requires taking a per-partition lock. Although we optimized the lock, there was still room for improvement. Therefore we introduced the thread cache, which holds a small amount of not-too-large memory chunks, ready to be allocated. Because these chunks are stored per-thread, they can be allocated without a lock, only requiring a faster thread-local storage (TLS) lookup, improving cache locality in the process. The thread cache has been tailored to satisfy a vast majority of requests by allocating from and releasing memory to the main allocator in batches, amortizing lock acquisition and further improving locality while not trapping excess memory.
Security is one of the important goals of PartitionAlloc.
PartitionAlloc guarantees that different partitions exist in different regions of the process's address space. When the caller has freed all objects contained in a page in a partition, PartitionAlloc returns the physical memory to the operating system, but continues to reserve the region of address space. PartitionAlloc will only reuse an address space region for the same partition.
Similarly, one page can contain only objects from the same bucket. When freed, PartitionAlloc returns the physical memory, but continues to reserve the region for this very bucket.
The above techniques help avoid type confusion attacks. Note, however, these apply only to normal buckets and not to direct map, as it'd waste too much address space.
PartitionAlloc also guarantees that:
Linear overflows/underflows cannot corrupt into, out of, or between partitions. There are guard pages at the beginning and the end of each memory region owned by a partition.
Linear overflows/underflows cannot corrupt the allocation metadata. PartitionAlloc records metadata in a dedicated, out-of-line region (not adjacent to objects), surrounded by guard pages. (Freelist pointers are an exception.)
Partial pointer overwrite of freelist pointer should fault.
Direct map allocations have guard pages at the beginning and the end.
PartitionAlloc guarantees that returned pointers are aligned on
base::kAlignment boundary (typically 16B on 64-bit systems, and 8B on 32-bit).
PartitionAlloc also supports higher levels of alignment, that can be requested via
PartitionAlloc::AlignedAllocFlags() or platform-specific APIs (such as
posix_memalign()). The requested alignment has to be a power of two. PartitionAlloc reserves the right to round up the requested size to the nearest power of two, greater than or equal to the requested alignment. This may be wasteful, but allows taking advantage of natural PartitionAlloc alignment guarantees. Allocations with an alignment requirement greater than
base::kAlignment are expected to be very rare.
Originally, PartitionAlloc was used only in Blink (Chromium’s rendering engine). It was invoked explicitly, by calling PartitionAlloc APIs directly.
PartitionAlloc-Everywhere is the name of the project that brought PartitionAlloc to the entire-ish codebase (exclusions apply). This was done by intercepting
posix_memalign(), etc. and routing them into PartitionAlloc. The shim located in
base/allocator/allocator_shim_default_dispatch_to_partition_alloc.h is responsible for intercepting. For more details, see base/allocator/README.md.
A special, catch-it-all Malloc partition has been created for the intercepted
malloc() et al. This is to isolate from already existing Blink partitions. The only exception from that is Blink‘s FastMalloc partition, which was also catch-it-all in nature, so it’s perfectly fine to merge these together, to minimize fragmentation.
PartitionAlloc-Everywhere was launched in M89 for Windows 64-bit and Android. Windows 32-bit and Linux followed it shortly after, in M90.
In PartitionAlloc, by system page we mean a memory page as defined by CPU/OS (often referred to as “virtual page” out there). It is most commonly 4KiB in size, but depending on CPU it can be larger (PartitionAlloc supports up to 64KiB).
The reason why we use the term “system page” is to disambiguate from partition page, which is the most common granularity used by PartitionAlloc. Each partition page consists of exactly 4 system pages.
A super page is a 2MiB region, aligned on a 2MiB boundary. Don't confuse it with CPU/OS terms like “large page” or “huge page”, which are also commonly 2MiB in size. These have to be fully committed/uncommitted in memory, whereas super pages can be partially committed, with system page granularity.
A slot is an indivisible allocation unit. Slot sizes are tied to buckets. For example each allocation that falls into the bucket (240; 256] would be satisfied with a slot of size 256. This applies only to normal buckets, not to direct map.
A slot span is just a grouping of slots of the same size next to each other in memory. Slot span size is a multiple of a partition page.
A bucket is a collection of slot spans containing slots of the same size, organized as linked-lists.
Allocations up to 4 partition pages are referred to as small buckets. In these cases, slot spans are always between 1 and 4 partition pages in size. The size is chosen based on the slot size, such that the rounding waste is minimized. For example, if the slot size was 96B and slot span was 1 partition page of 16KiB, 64B would be wasted at the end, but nothing is wasted if 3 partition pages totalling 48KiB are used. Furthermore, PartitionAlloc may avoid waste by lowering the number of committed system pages compared to the number of reserved pages. For example, for the slot size of 80B we'd use a slot span of 4 partition pages of 16KiB, i.e. 16 system pages of 4KiB, but commit only up to 15, thus resulting in perfect packing.
Allocations above 4 partition pages (but ≤
kMaxBucketed) are referred to as single slot spans. That‘s because each slot span is guaranteed to hold exactly one slot. Fun fact: there are sizes ≤4 partition pages that result in a slot span having exactly 1 slot, but nonetheless they’re still classified as small buckets. The reason is that single slot spans are often handled by a different code path, and that distinction is made purely based on slot size, for simplicity and efficiency.
PartitionAlloc handles normal buckets by reserving (not committing) 2MiB super pages. Each super page is split into partition pages. The first and the last partition page are permanently inaccessible and serve as guard pages, with the exception of one system page in the middle of the first partition page that holds metadata (32B struct per partition page).
As allocation requests arrive, there is eventually a need to allocate a new slot span. Address space for such a slot span is carved out from the last super page. If not enough space, a new super page is allocated. Due to varying sizes of slot span, this may lead to leaving space unused (we never go back to fill previous super pages), which is fine because this memory is merely reserved, which is far less precious than committed memory. Note also that address space reserved for a slot span is never released, even if the slot span isn't used for a long time.
All slots in a newly allocated slot span are free, i.e. available for allocation.
All free slots within a slot span are chained into a singly-linked free-list, by writing the next pointer at the beginning of each slot, and the head of the list is written in the metadata struct.
However, writing a pointer in each free slot of a newly allocated span would require committing and faulting in physical pages upfront, which would be unacceptable. Therefore, PartitionAlloc has a concept of provisioning slots. Only provisioned slots are chained into the freelist. Once provisioned slots in a span are depleted, then another page worth of slots is provisioned (note, a slot that crosses a page boundary only gets provisioned with slots of the next page). See
PartitionBucket::ProvisionMoreSlotsAndAllocOne() for more details.
Freelist pointers are stored at the beginning of each free slot. As such, they are the only metadata that is inline, i.e. stored among the objects. This makes them prone to overruns. On little-endian systems, the pointers are encoded by reversing byte order, so that partial overruns will very likely result in destroying the pointer, as opposed to forming a valid pointer to a nearby location.
Furthermore, a shadow of a freelist pointer is stored next to it, encoded in a different manner. This helps PartitionAlloc detect corruptions.
A slot span can be in any of 4 states:
PartitionAlloc prioritizes getting an available slot from an active span, over an empty one, in hope that the latter can be soon transitioned into a decommitted state, thus releasing memory. There is no mechanism, however, to prioritize selection of a slot span based on the number of already allocated slots.
An empty span becomes decommitted either when there are too many empty spans (FIFO), or when
PartitionRoot::PurgeMemory() gets invoked periodically (or in low memory pressure conditions). An allocation can be satisfied from a decommitted span if there are no active or empty spans available. The slot provisioning mechanism kicks back in, committing the pages gradually as needed, and the span becomes active. (There is currently no other way to unprovision slots than decommitting the entire span).
As mentioned above, a bucket is a collection of slot spans containing slots of the same size. In fact, each bucket has 3 linked-lists, chaining active, empty and decommitted spans (see
PartitionBucket::*_slot_spans_head). There is no need for a full span list. The lists are updated lazily. An empty, decommitted or full span may stay on the active list for some time, until
PartitionBucket::SetNewActiveSlotSpan() encounters it. A decommitted span may stay on the empty list for some time, until
PartitionBucket<thread_safe>::SlowPathAlloc() encounters it. However, the inaccuracy can't happen in the other direction, i.e. an active span can only be on the active list, and an empty span can only be on the active or empty list.