Date: Thu, 28 Jul 2005 00:08:07 +0100 (BST) From: Robert Watson <rwatson@FreeBSD.org> To: Pawel Jakub Dawidek <pjd@FreeBSD.org> Cc: freebsd-current@freebsd.org Subject: Re: malloc(9) statistics. Message-ID: <20050728000702.E54330@fledge.watson.org> In-Reply-To: <20050727175158.GP46538@darkness.comp.waw.pl> References: <20050727175158.GP46538@darkness.comp.waw.pl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 27 Jul 2005, Pawel Jakub Dawidek wrote: > I spend few hours today tracking memory leak. At the end I found, that > there were no memory leak actually, but memory statistics are confusing. > > Command 'vmstat -m' shows number of allocations in column named "InUse". > Coulmn name is wrong and confusing. The column only shows how many > allocation were there (successful or not), so when malloc(..., M_NOWAIT) > fails, it increases InUse value, which is never decreased, because there > will be no corresponding free(9). > > I've two proposals: > > 1. Don't increase allocations counter on failure: > > http://people.freebsd.org/~pjd/patches/kern_malloc.c.patch I think this makes more sense, and the problem is due to a mis-understanding of mine regarding how malloc(9) internally accounts for allocations. Feel free to commit, and MFC in short order to RELENG_6. > 2. Change "InUse" name to something else. (1) since InUse is a useful concept. Thanks! Robert N M Watson
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050728000702.E54330>