On Mon, Apr 09, 2012 at 08:24:51PM -0500, Omar Ramirez Luna wrote:
== Omar Ramirez <omarrmz> ==
=== Highlights ===
- Working on device tree for mailbox:
Find a kernel that works with DT for pandaboard (OMAP4):
- k3.4-rc1 doesn't recognize DT blob, k3.3 DT panics at boot.
- DT has to be appended to zImage.
Find a DT uboot:
- secretlab.ca uboot doesn't have usb support for panda, having
problems with fatload and mmc.
- linaro uboot has problems recognizing dt blob in memory (bootm
${kern} - ${dtb}; doesn't work properly).
This is a bit unexpected -- is there really nobody currently booting Panda using a DT?
On Tue, Apr 10, 2012 at 5:17 PM, Christian Robottom Reis kiko@linaro.org wrote:
On Mon, Apr 09, 2012 at 08:24:51PM -0500, Omar Ramirez Luna wrote:
== Omar Ramirez <omarrmz> ==
=== Highlights ===
- Working on device tree for mailbox:
Find a kernel that works with DT for pandaboard (OMAP4): - k3.4-rc1 doesn't recognize DT blob, k3.3 DT panics at boot. - DT has to be appended to zImage.
Find a DT uboot: - secretlab.ca uboot doesn't have usb support for panda, having problems with fatload and mmc. - linaro uboot has problems recognizing dt blob in memory (bootm ${kern} - ${dtb}; doesn't work properly).
This is a bit unexpected -- is there really nobody currently booting Panda using a DT?
The routine tests we do on each monthly release test for device tree and this test has passed for panda for months now. That is the LT kernel that may include some DT patches so maybe that explains this?
On 04/11/2012 07:41 AM, Somebody in the thread at some point said:
On Tue, Apr 10, 2012 at 5:17 PM, Christian Robottom Reis kiko@linaro.org wrote:
On Mon, Apr 09, 2012 at 08:24:51PM -0500, Omar Ramirez Luna wrote:
== Omar Ramirez <omarrmz> ==
=== Highlights ===
- Working on device tree for mailbox:
Find a kernel that works with DT for pandaboard (OMAP4):
- k3.4-rc1 doesn't recognize DT blob, k3.3 DT panics at boot.
- DT has to be appended to zImage.
Find a DT uboot:
- secretlab.ca uboot doesn't have usb support for panda, having
problems with fatload and mmc.
- linaro uboot has problems recognizing dt blob in memory (bootm
${kern} - ${dtb}; doesn't work properly).
This is a bit unexpected -- is there really nobody currently booting Panda using a DT?
The routine tests we do on each monthly release test for device tree and this test has passed for panda for months now. That is the LT kernel that may include some DT patches so maybe that explains this?
Right, LT kernels work fine with DT boot, we have done DT boot on OMAP4 since last Autumn. We still need to add at least this for a long time
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=commi...
I don't know if that's all that's required.
Last week we tried v3.3 mainline and it dies in some problem about UART.
Omar, you should look here
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=summa...
use tilt-tracking and omap_5430evm_defconfig
That's workable on 4430, 4460 (and Omap5) with DT boot. We have a race bug about mmc probe outstanding but we can't reproduce it at the moment. Otherwise for 4460 it should be running on all cylinders.
-Andy
On 10 April 2012 17:01, Andy Green andy.green@linaro.org wrote:
On 04/11/2012 07:41 AM, Somebody in the thread at some point said:
On Tue, Apr 10, 2012 at 5:17 PM, Christian Robottom Reis kiko@linaro.org wrote:
On Mon, Apr 09, 2012 at 08:24:51PM -0500, Omar Ramirez Luna wrote:
== Omar Ramirez <omarrmz> ==
=== Highlights ===
- Working on device tree for mailbox:
Find a kernel that works with DT for pandaboard (OMAP4): - k3.4-rc1 doesn't recognize DT blob, k3.3 DT panics at boot. - DT has to be appended to zImage.
Find a DT uboot: - secretlab.ca uboot doesn't have usb support for panda, having problems with fatload and mmc. - linaro uboot has problems recognizing dt blob in memory (bootm ${kern} - ${dtb}; doesn't work properly).
This is a bit unexpected -- is there really nobody currently booting Panda using a DT?
The routine tests we do on each monthly release test for device tree and this test has passed for panda for months now. That is the LT kernel that may include some DT patches so maybe that explains this?
Right, LT kernels work fine with DT boot, we have done DT boot on OMAP4 since last Autumn. We still need to add at least this for a long time
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=commi...
This is from Dec and still not in mainline. Grant, Arnd, Rajendra, do you know the reason why?
I don't know if that's all that's required.
Last week we tried v3.3 mainline and it dies in some problem about UART.
Have you tried 3.4-rc kernels?
Omar, you should look here
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=summa...
use tilt-tracking and omap_5430evm_defconfig
That's workable on 4430, 4460 (and Omap5) with DT boot. We have a race bug about mmc probe outstanding but we can't reproduce it at the moment. Otherwise for 4460 it should be running on all cylinders.
Can someone on the LT, working with Omar, take some time to dig into what all is needed to boot mainline with DT support? I'm a little surprised too that it doesn't just work out of the box. :/
~Deepak
On 04/11/2012 08:29 AM, Somebody in the thread at some point said:
On 10 April 2012 17:01, Andy Green andy.green@linaro.org wrote:
On 04/11/2012 07:41 AM, Somebody in the thread at some point said:
On Tue, Apr 10, 2012 at 5:17 PM, Christian Robottom Reis kiko@linaro.org wrote:
On Mon, Apr 09, 2012 at 08:24:51PM -0500, Omar Ramirez Luna wrote:
== Omar Ramirez <omarrmz> ==
=== Highlights ===
- Working on device tree for mailbox:
Find a kernel that works with DT for pandaboard (OMAP4):
- k3.4-rc1 doesn't recognize DT blob, k3.3 DT panics at boot.
- DT has to be appended to zImage.
Find a DT uboot:
- secretlab.ca uboot doesn't have usb support for panda, having
problems with fatload and mmc.
- linaro uboot has problems recognizing dt blob in memory (bootm
${kern} - ${dtb}; doesn't work properly).
This is a bit unexpected -- is there really nobody currently booting Panda using a DT?
The routine tests we do on each monthly release test for device tree and this test has passed for panda for months now. That is the LT kernel that may include some DT patches so maybe that explains this?
Right, LT kernels work fine with DT boot, we have done DT boot on OMAP4 since last Autumn. We still need to add at least this for a long time
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=commi...
This is from Dec and still not in mainline. Grant, Arnd, Rajendra, do you know the reason why?
Last time I looked, at v3.3 since that's where we're stuck right now, dts for Panda is still really minimal. But all over the kernel, there's dt stuff gone in that's not getting tested even with DT boot path. If it didn't happen inbetweentimes it might be worth trying at the same time to generate a "maximum dts" that exercises the new goodies.
I don't know if that's all that's required.
Last week we tried v3.3 mainline and it dies in some problem about UART.
Have you tried 3.4-rc kernels?
No. Normally I would have rebased immediately but as I explained in another post, we are camped at v3.3 until our new contributors make it to the 3.4-rc1 rendezvous point. Since just the -core and -pm topics are already >650 patches, and it's > 1400 total, making a throwaway test rebase when there's other things still to do is too expensive.
Omar, you should look here
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=summa...
use tilt-tracking and omap_5430evm_defconfig
That's workable on 4430, 4460 (and Omap5) with DT boot. We have a race bug about mmc probe outstanding but we can't reproduce it at the moment. Otherwise for 4460 it should be running on all cylinders.
Can someone on the LT, working with Omar, take some time to dig into what all is needed to boot mainline with DT support? I'm a little surprised too that it doesn't just work out of the box. :/
Jassi already took a look last week at v3.3 DT status when he generated a minimal config for mainline LAVA tests for Andrey.
Jassi maybe you can give pure mainline 3.4-rc1 or HEAD a go and identify what needs sorting for DT boot path.
-Andy
On 11 April 2012 06:10, Andy Green andy.green@linaro.org wrote:
Can someone on the LT, working with Omar, take some time to dig into what all is needed to boot mainline with DT support? I'm a little surprised too that it doesn't just work out of the box. :/
Jassi already took a look last week at v3.3 DT status when he generated a minimal config for mainline LAVA tests for Andrey.
Jassi maybe you can give pure mainline 3.4-rc1 or HEAD a go and identify what needs sorting for DT boot path.
Sorry for the not-so-prompt reply. I was dating another issue.
Yes, it's the patch "e7f11e8c3", except the redundant (for the purpose) omap4_panda_map_io function.
Regards, -Jassi
On 10 April 2012 19:29, Deepak Saxena dsaxena@linaro.org wrote:
Last week we tried v3.3 mainline and it dies in some problem about UART.
Have you tried 3.4-rc kernels?
I tried 3.4-rc1 and failed with nfs and mmc, it does work with cramfs which I didn't try until now, according to Benoit Cousson there is a fix which should hit 3.3 stable and is now on queue for 3.4
http://www.spinics.net/lists/linux-serial/msg05942.html
Can someone on the LT, working with Omar, take some time to dig into what all is needed to boot mainline with DT support? I'm a little surprised too that it doesn't just work out of the box. :/
It would be helpful to sort out if I'm missing something, given that right now only cramfs works and my dtb has to be appended to generate a bootable uImage. Anyway, I'll be trying LT kernel to see if /dev/nfs or /dev/mmc are working there + dtb passed as a blob instead of appended to uImage.
Regards,
Omar
On 11 April 2012 21:17, Omar Ramirez Luna omar.luna@linaro.org wrote:
Can someone on the LT, working with Omar, take some time to dig into what all is needed to boot mainline with DT support? I'm a little surprised too that it doesn't just work out of the box. :/
It would be helpful to sort out if I'm missing something, given that right now only cramfs works and my dtb has to be appended to generate a bootable uImage. Anyway, I'll be trying LT kernel to see if /dev/nfs or /dev/mmc are working there + dtb passed as a blob instead of appended to uImage.
Omar, give it a try.... compile uImage and omap4-panda.dtb. Have the u-boot load them and run by
# setenv initrd_high "0xffffffff" ; setenv fdt_high "0xffffffff" # mmc rescan 0; fatload mmc 0:1 0x80200000 uImage; fatload mmc 0:1 0x81600000 uInitrd; fatload mmc 0:1 0x815f0000 omap4-panda.dtb # setenv bootargs "console=tty0 console=ttyO2,115200n8 root=/dev/mmcblk0p2 rootwait ro earlyprintk fixrtc nocompcache vram=48M omapfb.vram=0:24M mem=456M@0x80000000 mem=512M@0xA0000000" # bootm 0x80200000 0x81600000 0x815f0000
hth.
On 10 April 2012 19:01, Andy Green andy.green@linaro.org wrote:
Omar, you should look here
http://git.linaro.org/gitweb?p=landing-teams/working/ti/kernel.git%3Ba=summa...
use tilt-tracking and omap_5430evm_defconfig
That's workable on 4430, 4460 (and Omap5) with DT boot. We have a race bug about mmc probe outstanding but we can't reproduce it at the moment. Otherwise for 4460 it should be running on all cylinders.
Ok, thanks. I'll try it for testing purposes, however my patches need to be based on linux-omap hopefully there won't be any dependency.
Regards,
Omar
-
Andy Green -
Christian Robottom Reis -
Deepak Saxena -
Jassi Brar -
John Rigby -
Omar Ramirez Luna