Date: Mon, 22 Jul 2013 13:35:47 -0700 From: Justin Hibbits <jrh29@alumni.cwru.edu> To: Konstantin Belousov <kostikbel@gmail.com> Cc: FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: contigmalloc won't allocate more than one page Message-ID: <CAHSQbTD9en3P3WeCVK22jfaw08VuLD_nrD8GyJ7-qDeMSOs9ZQ@mail.gmail.com> In-Reply-To: <20130722045425.GF5991@kib.kiev.ua> References: <CAHSQbTCT2e0UKSP34u05WSjCwdhQ_4z9GjxVxdN5kXXO35cxFQ@mail.gmail.com> <20130722045425.GF5991@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Jul 21, 2013 at 9:54 PM, Konstantin Belousov <kostikbel@gmail.com>wrote: > On Sun, Jul 21, 2013 at 06:51:22PM -0700, Justin Hibbits wrote: > > I've been spending the day trying to get ofwdump to work correctly on > > ppc64, and found that for at least one property it needs a full page for > > the property value. So, to fix this I checked the maximum ioctl sizes, > and > > with a buffer it totals 5 pages, so I tried to get > ofw_real_bounce_alloc() > > to allocate 5 pages instead of one. WIth this, it hangs when allocating > > the memory for the buffer. So, I tried just two pages, and got the same > > result. > Could you test on other architectures, esp. on x86i, please ? > Unfortunately for this test, all I have physical access to are PowerPC. It's possible that contigmalloc() can allocate more memory, but possibly later in the boot cycle, as this code runs very early on. > What version do you use, head ? > Yes, running head as of a few days ago. - Justin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAHSQbTD9en3P3WeCVK22jfaw08VuLD_nrD8GyJ7-qDeMSOs9ZQ>