> even after changing to a power-of-two allocation and starting
> with 8k items, aux/acidleak still takes 400mb on a 40mb proc
> with only 155278 bytes actually allocated (in the target process).
>
> is the a chance that pool is not packing the small
> allocations together well?

i wouldn't judge pool based on acidleak.
have you checked how much data acidleak
is allocating in that 400mb?  i expect the
data array to be huge: it knows about every
word in memory that looks vaguely like a pointer.
acidleak was never intended to be lightweight.

>> if upas/fs is allocating arbitrarily large buffers,
>> then its allocation behavior is broken too.
>
> there are a fixed number of large buffers.  up to 15 messages
> and mdir uses dirreadall so it can qsort.  the qsort is easy enough
> to eliminate, but it's more difficult to bound message buffers.

i never said fixing it was easy.  ;-)

russ

Reply via email to