3.11-rc1+ multiplatform kernel+dtb pandaboard issues

3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Kyle McMartin]

Hey,

We've been trying to get booting with a detached DTB working in Fedora
rawhide for a while, with no luck, the kernel keeps hanging after
"Starting kernel ..." with no output.

I've suspected there might be a config issue, as omap2_defconfig
works alright, as does booting a uImage with an appended
omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
(or even just zImage without initrd + dtb.)

I guess for 3.11 we could just use a legacy uImage in the interim until
the OMAP dtb stuff is a bit more baked, but we're willing to try to
debug the issue.

Our .config is available at [1] in case it's something obvious that
I'm just missing that anyone can hint at, but I've futzed a lot of
the options from omap2_defconfig in with little luck.

Thanks in advance!

regards, Kyle 

1: http://kyle.fedorapeople.org/config-3.11.0-0.rc6.git0.1.fc20.armv7hl

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Kyle McMartin]

On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> I've suspected there might be a config issue, as omap2_defconfig
> works alright, as does booting a uImage with an appended
> omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> (or even just zImage without initrd + dtb.)
> 

after some further investigation, something odd is going on... turns
out things have been working (at least until initrd time, but that's
another story) fine but none of the bootargs have made it into the
kernel, so console= was unset and no messages came out.

not sure if this is appropriate, but explicitly putting the "default"
serial consoles into the dtb, at least helps me get along debugging
issues...

---
From: Kyle McMartin <kyle@redhat.com>

Specify default bootargs consoles for the Beaglebone and Pandaboards
in their respective device tree files.

Signed-off-by: Kyle McMartin <kyle@redhat.com>

--- a/arch/arm/boot/dts/am335x-bone.dts
+++ b/arch/arm/boot/dts/am335x-bone.dts
@@ -19,6 +19,10 @@
 		};
 	};
 
+	chosen {
+		bootargs = "console=ttyO0,115200n8";
+	};
+
 	memory {
 		device_type = "memory";
 		reg = <0x80000000 0x10000000>; /* 256 MB */
diff --git a/arch/arm/boot/dts/omap4-panda-common.dtsi b/arch/arm/boot/dts/omap4-panda-common.dtsi
index faa95b5..b424d00 100644
--- a/arch/arm/boot/dts/omap4-panda-common.dtsi
+++ b/arch/arm/boot/dts/omap4-panda-common.dtsi
@@ -16,6 +16,10 @@
 		reg = <0x80000000 0x40000000>; /* 1 GB */
 	};
 
+	chosen {
+		bootargs = "console=ttyO2,115200";
+	};
+
 	leds: leds {
 		compatible = "gpio-leds";
 		pinctrl-names = "default";

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Sebastian Reichel]

[-- Attachment #1: Type: text/plain, Size: 789 bytes --]

Hi,

On Wed, Aug 21, 2013 at 03:41:43PM -0400, Kyle McMartin wrote:
> On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> > I've suspected there might be a config issue, as omap2_defconfig
> > works alright, as does booting a uImage with an appended
> > omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> > (or even just zImage without initrd + dtb.)
> > 
> 
> after some further investigation, something odd is going on... turns
> out things have been working (at least until initrd time, but that's
> another story)

The initrd problem may be the same error I got, when I tried to boot
an OMAP DT kernel with an initrd file (I had no problems with kernel
arguments):

http://www.spinics.net/lists/linux-omap/msg95504.html

-- Sebastian

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Tony Lindgren]

* Kyle McMartin <kmcmarti@redhat.com> [130821 12:48]:
> On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> > I've suspected there might be a config issue, as omap2_defconfig
> > works alright, as does booting a uImage with an appended
> > omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> > (or even just zImage without initrd + dtb.)
> > 
> 
> after some further investigation, something odd is going on... turns
> out things have been working (at least until initrd time, but that's
> another story) fine but none of the bootargs have made it into the
> kernel, so console= was unset and no messages came out.

Hmm I've seen that with extra quotes around the u-boot bootargs line.
But if things work OK with omap2plus_defconfig, then it sounds that
something gets trashed with your .config options. Care to post your
.config file somewhere?

Regards,

Tony 

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Kyle McMartin]

On Wed, Aug 21, 2013 at 11:34:50PM -0700, Tony Lindgren wrote:
> * Kyle McMartin <kmcmarti@redhat.com> [130821 12:48]:
> > On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> > > I've suspected there might be a config issue, as omap2_defconfig
> > > works alright, as does booting a uImage with an appended
> > > omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> > > (or even just zImage without initrd + dtb.)
> > > 
> > 
> > after some further investigation, something odd is going on... turns
> > out things have been working (at least until initrd time, but that's
> > another story) fine but none of the bootargs have made it into the
> > kernel, so console= was unset and no messages came out.
> 
> Hmm I've seen that with extra quotes around the u-boot bootargs line.
> But if things work OK with omap2plus_defconfig, then it sounds that
> something gets trashed with your .config options. Care to post your
> .config file somewhere?
> 

Sure, CONFIG_CMDLINE is "" for us...

http://kyle.fedorapeople.org/config-3.11.0-0.rc6.git0.1.fc20.armv7hl

--Kyle

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Kyle McMartin]

On Thu, Aug 22, 2013 at 02:25:59AM +0200, Sebastian Reichel wrote:
> Hi,
> 
> On Wed, Aug 21, 2013 at 03:41:43PM -0400, Kyle McMartin wrote:
> > On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> > > I've suspected there might be a config issue, as omap2_defconfig
> > > works alright, as does booting a uImage with an appended
> > > omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> > > (or even just zImage without initrd + dtb.)
> > > 
> > 
> > after some further investigation, something odd is going on... turns
> > out things have been working (at least until initrd time, but that's
> > another story)
> 
> The initrd problem may be the same error I got, when I tried to boot
> an OMAP DT kernel with an initrd file (I had no problems with kernel
> arguments):
> 

Definitely looks similar... I don't even get that far though. Spooky.
Something definitely seems wonky.

--Kyle

Re: 3.11-rc1+ multiplatform kernel+dtb pandaboard issues

[Tony Lindgren]

* Kyle McMartin <kmcmarti@redhat.com> [130822 07:08]:
> On Wed, Aug 21, 2013 at 11:34:50PM -0700, Tony Lindgren wrote:
> > * Kyle McMartin <kmcmarti@redhat.com> [130821 12:48]:
> > > On Wed, Aug 21, 2013 at 11:43:21AM -0400, Kyle McMartin wrote:
> > > > I've suspected there might be a config issue, as omap2_defconfig
> > > > works alright, as does booting a uImage with an appended
> > > > omap4-panda.dtb, but there's no love with a zImage + initrd + dtb
> > > > (or even just zImage without initrd + dtb.)
> > > > 
> > > 
> > > after some further investigation, something odd is going on... turns
> > > out things have been working (at least until initrd time, but that's
> > > another story) fine but none of the bootargs have made it into the
> > > kernel, so console= was unset and no messages came out.
> > 
> > Hmm I've seen that with extra quotes around the u-boot bootargs line.
> > But if things work OK with omap2plus_defconfig, then it sounds that
> > something gets trashed with your .config options. Care to post your
> > .config file somewhere?
> > 
> 
> Sure, CONFIG_CMDLINE is "" for us...
> 
> http://kyle.fedorapeople.org/config-3.11.0-0.rc6.git0.1.fc20.armv7hl

Tried this on my 4430 SDP with the current tip of linux-omap tree
at -rc6, had to disable FTRACE and ARCH_ZYNQ to compile.

I can confirm that earlyprintk works, but no output from the
serial console. Looking at the console output, looks like serial-omap
probe is deferred because of no pinctrl driver. So having:

CONFIG_PINCTRL_SINGLE=y

fixes the issue at least for me booting to the point where it tries
to mount root. It seems to be set to module in your .config.

Note that there's still that one bug where earlyprintk does not properly
work with serial-omap because of the idling of the device.. But that's a
separate issue :)

Regards,

Tony