I'm searching for one of the old BB images that allowed to write the EEPROM
from u-boot. One of those that you could use to access the u-boot even if
the EEPROM was empty.
You don't have to go old for that. Robert continues to build updated
u-boot images that ignore the EEPROM such that you can boot and
program it. Realize there is a test-point on the board if you want to
un-write-protect the EEPROM.
The image-builder specifies the image to get here:
I'm not sure where the sources are! Robert, please jump in here as it
isn't obvious, especially since the image build script doesn't have an
option for rebuilding the bootloader (it seems it should).
I'm searching for one of the old BB images that allowed to write the EEPROM
from u-boot. One of those that you could use to access the u-boot even if
the EEPROM was empty.
You don't have to go old for that. Robert continues to build updated
u-boot images that ignore the EEPROM such that you can boot and
program it. Realize there is a test-point on the board if you want to
un-write-protect the EEPROM.
I'm not sure where the sources are! Robert, please jump in here as it
isn't obvious, especially since the image build script doesn't have an
option for rebuilding the bootloader (it seems it should).
Well, i really don't have a script to build "one" bootloader.
I kinda ended up in a situation where i'm tracking 15 boards in u-boot
mainline. opps!
As u-boot releases happen, i push those patches to " GitHub - eewiki/u-boot-patches " so we can hard link to
them. The patches in Bootloader-Builder tend to churn a lot.
Oh, I'm just ignoring the fact u-boot has a "am335x-boneblack" config,
as "am335x-evm" supports both beaglebone's... Easier to share a
microSD between the white/black.
Robert?!?
I believe we should push the am335x-boneblack.h file into mainline and
annotate it clearly to be used only in cases of a board without an
EEPROM.
I'm searching for one of the old BB images that allowed to write the EEPROM
from u-boot. One of those that you could use to access the u-boot even if
the EEPROM was empty.
You don't have to go old for that. Robert continues to build updated
u-boot images that ignore the EEPROM such that you can boot and
program it. Realize there is a test-point on the board if you want to
un-write-protect the EEPROM.
I'm not sure where the sources are! Robert, please jump in here as it
isn't obvious, especially since the image build script doesn't have an
option for rebuilding the bootloader (it seems it should).
Well, i really don't have a script to build "one" bootloader.
I kinda ended up in a situation where i'm tracking 15 boards in u-boot
mainline. opps!
As u-boot releases happen, i push those patches to " GitHub - eewiki/u-boot-patches " so we can hard link to
them. The patches in Bootloader-Builder tend to churn a lot.
Oh, I'm just ignoring the fact u-boot has a "am335x-boneblack" config,
as "am335x-evm" supports both beaglebone's... Easier to share a
microSD between the white/black.
OK, so where you use it in setup-card.sh is simply a hack to give a
different URL to fetch it?
I believe we should push the am335x-boneblack.h file into mainline and
annotate it clearly to be used only in cases of a board without an
EEPROM.
The
Seems like you hesitated here. Perhaps we could #include
<am335x-evm.h> and then overwrite some settings and add a clear note
that this is for ignoring that EEPROM? The reasons are that lots of
people look for boneblack in the u-boot sources (without success),
lots of people try to build clones without EEPROMs and struggle to
figure out how, and out-of-tree patches are always hard to locate with
confidence they are used in the build of the software at hand.
Thanks for pointing it out. I flashed the image and put the TP2 to GND (using the BeagleBone white).
Tried it and got an infinite loop with the following:
U-Boot SPL 2014.04-00015-gb4422bd (Apr 22 2014 - 13:24:29) Incorrect magic number (0xffffffff) in EEPROM Could not get board ID. Incorrect magic number (0xffffffff) in EEPROM Could not get board ID. Unknown board: assuming BeagleBone Black.Incorrect magic number (0xffffffff) in EEPROM Could not get board ID.
U-Boot SPL 2014.04-00015-gb4422bd (Apr 22 2014 - 13:24:29) Incorrect magic number (0xffffffff) in EEPROM Could not get board ID. Incorrect magic number (0xffffffff) in EEPROM Could not get board ID. Unknown board: assuming BeagleBone Black.Incorrect magic number (0xffffffff) in EEPROM Could not get board ID.
Oh, I'm just ignoring the fact u-boot has a "am335x-boneblack" config,
as "am335x-evm" supports both beaglebone's... Easier to share a
microSD between the white/black.
OK, so where you use it in setup-card.sh is simply a hack to give a
different URL to fetch it?
Yeah, i use the "--uboot boneblack_flasher" flag in my setup-card.sh
creator script to grab the "blank eeprom" u-boot, vs the normal
am335x_evm build.
I believe we should push the am335x-boneblack.h file into mainline and
annotate it clearly to be used only in cases of a board without an
EEPROM.
The
Seems like you hesitated here. Perhaps we could #include
<am335x-evm.h> and then overwrite some settings and add a clear note
that this is for ignoring that EEPROM? The reasons are that lots of
people look for boneblack in the u-boot sources (without success),
lots of people try to build clones without EEPROMs and struggle to
figure out how, and out-of-tree patches are always hard to locate with
confidence they are used in the build of the software at hand.
Well, when i set this up originally, we didn't have much for a clone
market. So we definitely need to look at this again. I just didn't
want people running this image on clones that did have the 1Ghz or the
black's ldo setup. I need to think, what's still safe for a blank
bbw..
I tried booting with different images, specially with one that in the past probed to do the job (refer to this thread): Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.05-beaglebone-2012.11.22.img.xz.
The odd thing is that the board doesn’t boot (the PWR LED doesn’t even light up) when this image is used. If new images are used PWR LED switches on but i get messages as:
-Boot SPL 2013.07-dirty (Sep 01 2013 - 12:08:18) Incorrect magic number (0xffffffff) in EEPROM Could not get board ID. Unknown board, cannot configure pinmux.### ERROR ### Please RESET the board ###
Not being able to access u-boot to flash the EEPROM. Has anyone seen such a behaviour (some images make the board work and some not)?
@Gerald, do you thing this could be hardware related?
This is a new board developed to support BeaglePilot (a simplification of what i’ve been doing with Erle) thereby i need to flash the EEPROM myself.
After removing some components that were creating conflicts with SYS_BOOT pins the behavior changed:
Out of 5 boards we manufactured we’ve been able to program the EEPROM of only one using the image i mentioned before (Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.05-beaglebone-2012.11.22.img.xz). This is what leads me to thing to a manufacturing issue.
The other 4 boards, apparently look good (tty interface appears when connected through miniUSB, no microSD card on) but once I introduce a microSD card with Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.05-beaglebone-2012.11.22.img.xz image and the miniUSB is connected, the PWR LED turns on for about a second and then it switches off. No new tty interface appears under /dev/.
This behavior doesn’t happen with newer images but neither i can use them to flash the EEPROM.