Date: Wed, 3 Jul 2002 18:09:54 -0700 (PDT) From: Julian Elischer <julian@elischer.org> To: Peter Wemm <peter@wemm.org> Cc: Alan Cox <alc@cs.rice.edu>, Julian Elischer <julian@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/i386/i386 pmap.c Message-ID: <Pine.BSF.4.21.0207031803350.3993-100000@InterJet.elischer.org> In-Reply-To: <20020704005141.5C4FE390F@overcee.wemm.org>
next in thread | previous in thread | raw e-mail | index | archive | help
OK so the problem was that I was calling pmap_lookup() with indexes 1 -> KSTACK_PAGES but that object only had pages registered for indexes 0->(KSTACK_PAGES - 1) this bug was only shown up now because Jeff's uma code was broken and it never called this function.. (actually it never called thread_fini() which is the only caller of this function). He fixed it yesterday when I figured this out and suddenly we have untested code being run.. :-/ Hopefully this explains a few things that we have been seeing.. 1/ a memeory leak of 8k, times about 600 per buildworld I thin this is why someone reported their machine getting slower and slower. 2/ odd crashes with the smell of burning pmap. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0207031803350.3993-100000>