Struct mi_prof_stats_t 

#[repr(C)]
pub struct mi_prof_stats_t {

    pub size: usize,
    pub version: c_int,
    pub enabled: bool,
    pub accum: bool,
    pub sample_rate: usize,
    pub live_samples: usize,
    pub live_bytes: usize,
    pub accum_samples: usize,
    pub accum_bytes: usize,
    pub unique_stacks: usize,
    pub arena_committed: usize,
    pub stack_table_overflows: usize,
    pub dropped_samples: usize,
    pub heap_committed: usize,
    pub heap_reserved: usize,
    pub heap_malloc_requested: usize,
    pub heap_pages: usize,
    pub heap_pages_abandoned: usize,
    pub heap_count: usize,
    pub theap_count: usize,
    pub heap_purged: usize,
    pub heap_stats_detailed: bool,
}

Mirrors mi_prof_stats_t (include/mimalloc/profile.h) field-for-field.

Fields

size: usize

version: c_int

enabled: bool

accum: bool

sample_rate: usize

live_samples: usize

live_bytes: usize

accum_samples: usize

accum_bytes: usize

unique_stacks: usize

arena_committed: usize

stack_table_overflows: usize

dropped_samples: usize

v2. Mirrors mi_prof_stats_t.dropped_samples: count of ALL dropped samples (record-alloc failure, stack-intern failure, including the MI_PROF_STACK_CAP cap); stack_table_overflows is a subset, so dropped_samples >= stack_table_overflows always.

heap_committed: usize

v3. Allocator-level (“ground truth”) counters read from mi_stats_get(). Unlike every field above, these are exact rather than sampled.

heap_reserved: usize

heap_malloc_requested: usize

heap_pages: usize

heap_pages_abandoned: usize

heap_count: usize

theap_count: usize

Live thread-local heaps. The main thread’s statically-initialized theap is not counted, so a single-threaded process reports 0.

heap_purged: usize

heap_stats_detailed: bool

True when the C library was built with MI_STAT >= 2. heap_malloc_requested is only maintained at that level; a default release build reports 0.