[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Chicken-users] clueless about memory consumtion
From: |
Felix |
Subject: |
Re: [Chicken-users] clueless about memory consumtion |
Date: |
Sun, 11 Jul 2010 22:00:41 +0200 (CEST) |
From: Jörg "F. Wittenberger" <address@hidden>
Subject: Re: [Chicken-users] clueless about memory consumtion
Date: Sat, 10 Jul 2010 18:46:10 +0200
> Am Mittwoch, den 07.07.2010, 14:31 +0200 schrieb Felix:
>> I have added a procedure for dumping the object-counts in the
>> heap (see "experimental" branch), called `##sys#dump-heap-state'.
>
> Thanks Felix!
>
> Now I've got quite some more data - and dunno how to make head or tail
> of it.
>
> Different results on different systems.
>
> On the old, busy deployment machine (where I noticed the consumption
> problem) the new runtime will hang in full cpu usage but respond to
> signals.
You could build with debugging settings (-g) and attach via gdb
to the process, interrupting it and inspecting the backtrace.
>
> On the less busy development system it shows steady increase of
> allocated objects - so far I could not find any with surprising high
> numbers or increases.
>
> However no test will run for a longer time.
>
> Usually the second, sometimes the third entry of the dump is 'fixnum'
> followed by 'bytevector' - or a segfault.
>
> So maybe it's not worth to write a script to dissect the dump, but go
> after this segfault first? How would I?
>
I fixed a bug in the dump procedure recently. You should try the newest
git HEAD (master branch).
cheers,
felix