From mboxrd@z Thu Jan 1 00:00:00 1970 From: erik quanstrom Date: Sun, 12 Sep 2010 12:48:22 -0400 To: 9fans@9fans.net Message-ID: <6dc83bc76090e982543b4f10231530ec@brasstown.quanstro.net> In-Reply-To: References: <20100911163536.7348cd25@gmail.com> <243456116470afd150abf281a3f0f371@plug.quanstro.net> MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit Subject: Re: [9fans] 9vx mk install chokes on gs Topicbox-Message-UUID: 55580cee-ead6-11e9-9d60-3106f5b1d025 On Sun Sep 12 11:45:31 EDT 2010, rsc@swtch.com wrote: > On Sun, Sep 12, 2010 at 12:59 AM, erik quanstrom wrote: > >> -#define      USTKTOP         (0x4000000)             /* byte just beyond user stack */ > >> +#define      USTKTOP         (0x8000000)             /* byte just beyond user stack */ > > > > shouldn't you add a 0 to that?  what's wrong with giving a process 2gb > > of address space?  fundamental 9vx limits? > > there might not be 2gb of contiguous address space to have. > this is running inside a unix process. > > another reason for the low size was so that it was easier > to keep multiple processes mapped at the same time, > to reduce context switch latency. that makes sense. unfortunately, this means that any process that uses significant memory on plan 9 needs to be re-checked for 9vx. even 100mb is tiny. - erik