Date: Wed, 15 Sep 2004 00:57:42 +0400 From: Roman Kurakin <rik@cronyx.ru> To: Scott Long <scottl@samsco.org> Cc: current@freebsd.org Subject: Re: Large mem disk with not very recent current Message-ID: <41475B46.8030002@cronyx.ru> In-Reply-To: <4145BB0D.20905@samsco.org> References: <414560D3.10108@cronyx.ru> <20040913105133.GA85717@hub.freebsd.org> <414581FE.6060702@cronyx.ru> <4145BB0D.20905@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Scott Long wrote: > Roman Kurakin wrote: > >> Kris Kennaway wrote: >> >>> On Mon, Sep 13, 2004 at 12:56:51PM +0400, Roman Kurakin wrote: >>> >>>> Hi, >>>> >>>> I have the following idea/problem. I want to build world in memory. >>>> I've put in to my system 1G of memory and tried to put on it sources. >>>> But all times I've tried to do that system went to panic. I've used md >>>> with malloc type. >>>> >>>> Any ideas? >>>> >>> >>> >>> Please provide more information. It's likely you're trying to create >>> a md that is too big, though. >>> >> It is about 800Mb. But system panics while I try to extract source >> tree. It is not very recent >> current. The intent of my question was to get some hints how to set >> up such build area, I >> belive it should help to make compilation much faster than using hard >> drive. My home system >> is not ready for panic investigation now. I'll update it to current >> and set console to it and will >> try again. >> >> >> PS. Sorry to all for my last posts to current that are so weak in >> "technical" details, the debug >> technique I used before was without remote debugging since most of >> problems were very >> near my code and it was enough for me anlo three things: my mind, >> printed sources, and printf. >> So now I am in stage of adaptation to/setting up remote debugging. >> >> rik > > > What kind of panic was it? Try using swap backing instead of malloc. kmem_map too small. > Malloc will consume wired memory and will generally make life difficult > for other parts of the kernel that need memory later. Swap-backed pages > will stay in RAM until the VM needs to push them out due to memory > pressure. Ok. I'll try to try this this night. Thanks. rik > Scott > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41475B46.8030002>