[U-Boot] u-boot master on a pandaboard won't boot
Hello, I've recently been playing around with u-boot on a pandaboard Rev A1. The rest of my environment is an MLO built from x-loader git hash c4289f1bee035dea41536c5ba5e1bc36c7d493c4, and booting an ubuntu 11.04 kernel 2.6.38-1208-omap4. Up until tag v2011.06, u-boot works flawlessly (as far as I've tested) on this combination. However, if I take any u-boot release after that, including v2011.09 and u-boot git head, u-boot hangs very early on:
Texas Instruments X-Loader 1.5.1 (Nov 2 2011 - 08:55:25) Reading boot sector Loading u-boot.bin from mmc
And the status LEDs blink rapidly.
I bisected the problem down to commit bcae72116257201d7288cb8c525a29aea4875b95; prior to that commit, I can successfully[1] boot u-boot on the pandaboard, but after that I cannot.
I'm going to poke at it a bit more, but does anyone have any ideas of what is going on here?
--- Chris Lalancette
[1] Well, I can at least get beyond "Loading u-boot.bin from mmc" and boot the kernel and userland. For some reason with this setup I can't get ethernet to work, but one problem at a time.
On Wed, Nov 2, 2011 at 6:03 AM, Lalancette, Christopher clalancette@irobot.com wrote:
Hello, I've recently been playing around with u-boot on a pandaboard Rev A1. The rest of my environment is an MLO built from x-loader git hash c4289f1bee035dea41536c5ba5e1bc36c7d493c4, and booting an ubuntu 11.04 kernel 2.6.38-1208-omap4. Up until tag v2011.06, u-boot works flawlessly (as far as I've tested) on this combination. However, if I take any u-boot release after that, including v2011.09 and u-boot git head, u-boot hangs very early on:
Texas Instruments X-Loader 1.5.1 (Nov 2 2011 - 08:55:25) Reading boot sector Loading u-boot.bin from mmc
If you use SPL instead (so the 'MLO' that U-Boot also builds) does everything work again?
If you use SPL instead (so the 'MLO' that U-Boot also builds) does everything work again?
Oh, I didn't even notice that u-boot now builds an MLO. Thanks for the heads up.
Unfortunately, that doesn't help; when I copy both the new MLO and the new u-boot.bin into place (again from git head), I don't get any output from my serial console at all. I presume that means it is hanging up somewhere in the MLO.
--- Chris Lalancette
On Wed, Nov 2, 2011 at 3:12 PM, Lalancette, Christopher clalancette@irobot.com wrote:
If you use SPL instead (so the 'MLO' that U-Boot also builds) does everything work again?
Oh, I didn't even notice that u-boot now builds an MLO. Thanks for the heads up.
Unfortunately, that doesn't help; when I copy both the new MLO and the new u-boot.bin into place (again from git head), I don't get any output from my serial console at all. I presume that means it is hanging up somewhere in the MLO.
MLO loads u-boot.img by default not u-boot.bin. That may be your problem.
--john
Hi Christopher,
On Wednesday 02 November 2011 06:33 PM, Lalancette, Christopher wrote:
Hello, I've recently been playing around with u-boot on a pandaboard Rev A1. The rest of my environment is an MLO built from x-loader git hash c4289f1bee035dea41536c5ba5e1bc36c7d493c4, and booting an ubuntu 11.04 kernel 2.6.38-1208-omap4. Up until tag v2011.06, u-boot works flawlessly (as far as I've tested) on this combination. However, if I take any u-boot release after that, including v2011.09 and u-boot git head, u-boot hangs very early on:
Texas Instruments X-Loader 1.5.1 (Nov 2 2011 - 08:55:25) Reading boot sector Loading u-boot.bin from mmc
And the status LEDs blink rapidly.
I bisected the problem down to commit bcae72116257201d7288cb8c525a29aea4875b95; prior to that commit, I can successfully[1] boot u-boot on the pandaboard, but after that I cannot.
This commit changes the load address of u-boot and according to the new relocation scheme u-boot should be loaded at where it is built for. x-loader is assuming the load address of TI Internal u-boot. You have two options:
1. Use SPL - when you build mainline u-boot you get an MLO as by-product. Use it instead of x-loader. 2. Change the u-boot load address in x-loader to 0x80100000 (CFG_LOADADDR in include/configs/omap4430panda.h)
BTW, I am considering unifying the two load addresses.
br, Aneesh
I'm going to poke at it a bit more, but does anyone have any ideas of what is going on here?
Chris Lalancette
[1] Well, I can at least get beyond "Loading u-boot.bin from mmc" and boot the kernel and userland. For some reason with this setup I can't get ethernet to work, but one problem at a time.
participants (4)
-
Aneesh V -
John Rigby -
Lalancette, Christopher -
Tom Rini