Date: Wed, 14 May 2014 14:42:34 +0300 From: "Sulev-Madis Silber (ketas)" <madis555@hot.ee> To: "Sulev-Madis Silber (ketas)" <madis555@hot.ee> Cc: freebsd-arm <freebsd-arm@freebsd.org>, Ian Lepore <ian@freebsd.org> Subject: Re: Patch to make BBB properly boot from eMMC every time Message-ID: <537356AA.9090704@hot.ee> In-Reply-To: <5373048C.7030009@hot.ee> References: <5371E1F3.6080002@hot.ee> <1399987808.56626.2.camel@revolution.hippie.lan> <CADH-AwGftVJCi9NrAfQs6CYNifM04DZu9rtVF4k6Cd6irFk7fg@mail.gmail.com> <1399991632.56626.7.camel@revolution.hippie.lan> <CADH-AwHs2bNXyh-ADuMy8NgkdaakKi4hx1qz8bjB4heicFJy3w@mail.gmail.com> <5372789E.8020103@hot.ee> <CADH-AwE56LmekRdKfbZYAbvrjaH9pFy0RY5L%2BY_ne0YfwDtOcA@mail.gmail.com> <5373048C.7030009@hot.ee>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2014-05-14 08:52, Sulev-Madis Silber (ketas) wrote: > On 2014-05-13 23:33, Winston Smith wrote: >> On Tue, May 13, 2014 at 3:55 PM, Sulev-Madis Silber (ketas) >> <madis555@hot.ee> wrote: >>> That's bit different issue. >> >> I can't help but think the hang here is related, particularly to the >> subtle timing issues that are plaguing the 4 vs 8 bit detection. >> Seems like there's just something not right in the eMMC >> detection/startup? Maybe the [faster] 1GHZ CPU is further throwing >> off the timing. >> >>> I would actually want to have 1GHz WITH eMMC... now it's impossible. >>> First that, then maybe variable freq later (if possible). I certainly >>> have usage for such speeds somewhere. >> >> Why is it impossible? It seems to work, just not reliably! >> > > So you have uboot that could run BBB 1GHz while booting from eMMC? > Indeed, 2014.04 has appeared in crochet... while it should appear in > ports as *-devel, meh. I haven't tried that yet. > > >> Seems silly to have the 1GHZ BBB running at only 550Mhz. >> >>> But right now BBB is so stable. I would put snapshot of CURRENT to >>> production right now. I now also have taken all debugging options out. I >>> didn't realize how SLOW things are with WITNESS & others on. >>> Maybe when I get more than one copy of same HW, I could run some of them >>> with WITNESS on too (would it actually give me some useful data I could >>> report back?) >> >> I'm in the same boat. However, Ian is in the process of backporting >> the CURRENT arm updates to STABLE. If I can get 1GHZ working with >> 10-STABLE, then I'll be very happy. >> >> WITNESS is an easy one, create a config like this: >> >> ident MYCONF >> >> include BEAGLEBONE >> nooptions WITNESS >> >> >> And build with it (set KERNCONF if you're using crotchet-freebsd) and >> you'll get a WITNESS-free kernel. >> > > No, I'm not using crochet... > > http://ketas.si.pri.ee/bbb/ > > There are most of things I use to customize tree for BBB. > I got it booting from "eMMC". I had to find microsd for that. Only one that was unused was 64MB which was pretty enough to hold 7M /boot & etc/fstab... even 5M /rescue! And I have another partition there which holds same data + one FAT part for future use. So, now I have this silly boot path: ----------------------------------------------------------- eMMC, FAT : small uboot 2014.04 || \/ eMMC, FAT : regular uboot 2014.04 || \/ eMMC, FAT : uboot loader || \/ SD, UFS : loader files, kernel || \/ eMMC, UFS : rootfs ----------------------------------------------------------- Surely ubldr could be fixed for that. I tried to find cause but it's likely somewhere deep. Also, uboot env is absolutely filled with content that's only required for booting Linux kernel. It was even hard to understand what was wrong there. Actually the error was same with previous 1GHz uboot (2014.01). Only I didn't notice that loader actually runs, only it fails to find eMMC.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?537356AA.9090704>