kernel 3.5 and device-tree support...

Yes, looks like 'coherent_pool=1M' worked! Thanks. My RAID1 is rebuilding right now, seems to be all a-ok.

Is there a way to put this 'coherent_pool' into the DTS file? Somewhere other than the bootargs, as I override this in U-boot because of my RAID...



Edited 1 time(s). Last edit at 08/15/2012 12:36AM by rpress.

Kurlon thanks i would apreciate that im still new to all this submitting patches and stuff so dont know what and where to send them i subscribed to a mailing list a few days ago and tried to send in a mail to get help when i was makeing the dts file but as far as i can tell it did not get posted or anything so thanks.

I will try that "coherent_pool=1M" fix now and see if it works for me.

Edit:
Just tried it and it looks like it works for me aswel.

I have another question though does anyone else have trouble with their raid array mounting very slow on any kernel after 3.3.7? ever since then my raid has been slow to mount and anytime i use a program like git or that use's java the system crash's anyone have a similar issue?
i see this:

  md0: unknown partition table
Success: assembled all arrays.

then it continues and hangs here:

Mounting local filesystems...

then continues to a successful boot.



Edited 2 time(s). Last edit at 08/15/2012 04:26AM by Buttzy10169.

I also see the "unknown partition table" error, but I figured because I had my RAID set up as two RAID1 arrays under a regular partition table, vs having one RAID1 array and then two partitions under that, it was normal. I have it set this way so I can boot off the first RAID1 array, and it all works awesome, even allowing booting from either drive when the array is degraded.

I don't have any unusual problems or delays...

This new kernel seems faster somehow than 3.4, but it's probably placebo effect. :D

even on this latest 3.6 kernel im seeing the delay right when mounting my raid. The last kerenl i could use without issue was the 3.3.7 kernel and since then there have been hangs at the mounting of the raid and if i use git like for example:

git clone http://git.chromium.org/webm/libvpx.git

the nas just stops working until i unplug it and plug it back in.

rpress have you put your iconnect buttons into the dts file? and if you have are you haveing any trouble with kenerl panics? and if not could you post your button configs from your dts file?

I am using the mainline 3.6-rc2 kernel, and my DTS file is unchanged. I haven't had any kernel panics on either the iConnect or the GFN.

My iConnect is now working again after I fixed U-Boot. It seems pretty stable, so here is the U-Boot (and patch), built 3.6-rc2 kernel (and .config), and DTB file that I am using for those that want to give it a shot. You'll need to set up the U-Boot scripts to load the DTB file and execute bootm with the loaded blob address.
iconnect_FDT_uboot_kernel_3.6-rc2.tar.gz

The kernel above also works on GFN (with bootargs 'coherent_pool=1M') and the U-Boot patch also adds GFN support.

I have my ix2-200 up and running on 3.6-rc1 but when i move to 3.6-rc2 or linux next the buttons dont work and are causing a panic.

How are you guys building your kernels because im getting very odd and mixed results when building them for example:

a 3.3.7 kernel built a while ago works perfect

but the 3.3.7 kernel i rebuilt yesterday out of curiosity crash's / stops whenever i run git clone?

Hi All,

I've been lurking on this thread, trying to get my iConnect booting kernel 3.5++ into Debian testing on a usb connected ext2 /boot partition. This used to work under older kernels, with appropriate hackage to the old rd88f6281-setup.c board file. However, the DTB stuff now has apparently thrown a few curveballs...

I pm-ed with rpress a few weeks back, and he pointed me towards his u-boot and iconnect.dtb containing tarball. (Thanks, BTW!!!)

At the moment, I've still got my mtd partitions pretty much the same as from the stock firmware, and I kind of like being able to use the original as a rescue floppy type system. In that mtd config, there is a large-ish ubi partition allocated for /boot, and it is fairly empty. I've actually copied over the iconnect.dtb file into that partition, because I think that will be significantly less volatile than everything else in my normal Debian /boot. The version of u-boot from rpress appears to have support for ubi filesystems, as long as the mtd partitions are set up correctly.

My question has to do with getting the mtd partitions set up in u-boot. Apparently, there is some missing environment for u-boot to be able to recognize the mtd partitions, and that environment was never in the original firmware's config. (I also note that the new board-iconnect.c file appears to specify all of flash as a single partition. Will this cause a conflict with u-boot?)

The linux boottime commandline args for the old (my current) partition layout look like:

mtdparts=orion_nand:0xc0000@0x0(uboot),0x20000@0xa0000(env),0x300000@0x100000(zImage),0x300000@0x540000(initrd),0x1f400000@0x980000(boot)

Does anyone have any hints as to how to translate that into both the u-boot parameters and the board-iconnect.c mtd_partition arguments (if needed)? I find the u-boot "manpage" for mtdparts to be more than a little confusing, and an example (perhaps from another board) would be greatly appreciated. I hope the kernel board-file documentation is a little better too...

Thanks in advance for any help you might be able to provide!

Cheers,
Frank Horowitz

@davygravy,

Do you have any plan to upgrade uBoot to support Device-Tree? It seems Kurlon has done a lot of works to get device tree for some Kirwood devices into kernel 3.6.x. It would be nice if we can start migrating there to get quite a bit of kernel improvements.

I'm going to test kernel 3.6.x to see if we can still boot the old Dockstar, Pogo E02/V4/Mobile and GoFlexHome uBoot.

Looking at kernel 3.6.x mainline, we have device tree board files iConnect, and Goflex Net. It seems that to use the current uBoot with 3.6.x, we (this Debian forum users) will need to pack dt board file into uImage binary to boot these 2 boxes.

Thank,
bodhi



Edited 1 time(s). Last edit at 03/16/2013 02:41AM by bodhi.

Nope. It does not seems to work booting old mach-types while the Device Tree is enable in the kernel.

I've built and tested kernel 3.6.11 with patches to remove device tree setup. Still not sure about using a mixed of device tree and old mach-types, it did not work for me.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Kurlon Wrote:
-------------------------------------------------------
> I was just about to start a new thread about the
> impeding DTopcalypse, but then remembered this
> one.
>
> So, short story: I have my GoFlex Net booting 3.5
> via Device Tree. It's including in linux-next now
> and should be showing up formally in 3.6.
> Kirkwood DT support has gotten much better, pretty
> much everything can be defined via GPIO now, about
> the only holdout is ethernet but that's being
> worked on as we speak on the linux-arm list,
> guessing that'll land in 3.7.
>
> For those wanting to try, build a DT enabled
> kernel and make sure to enable the option to
> append a DTB to the end of the zImage. This means
> your kernel will only support that one device
> whose .dtb you append. The build process ends up
> being:
>
> make menuconfig
> make
> make kirkwood-goflexnet.dtb
> cat arch/arm/boot/kirkwood-goflexnet.dtb >>
> arch/arm/boot/zImage
> make uImage
> cp arch/arm/boot/uImage /boot/uImage
> reboot
>
> For the future, the ideal setup would be a newer
> u-boot (2012.04 era?) built with DT support
> enabled. This should allow loading of the .dtb
> separate from the kernel, ala:
>
> ext2load ide 0:1 0x80000 /boot/zImage
> ext2load ide 0:1 some other addr
> /boot/kirkwood-goflexnet.dtb
> bootm 0x80000 - dtb addr
>
> Yeah, zImage, newer u-boots are supposed to be
> happy with a zImage instead of a uImage. :D I
> don't know what the normal dtb load addr is yet
> for kirkwoods but I'm going to ping those playing
> with it on the linux-arm list. If we can get
> u-boot playing along, ARCH for example will be
> able to supply a linux-kirkwood-dtb kernel package
> that supplies a zImage and all the .dtbs, each
> device can use an u-boot env setting to tell it
> which one to load a boot just like we use a
> specific arc number now to signal the kernel as to
> which board it's booting on.
>
> So, if you're up for beating on u-boot davy, I'm
> game to beat on kernel support for all the various
> boards.

Anybody has success in booting a mainline Debian kernel 3.5 to 3.8 using Kurlon instruction? my attempt so far has not been successful. I tried to boot the IConnect using Open-Wrt uBoot, and kernel 3.6.11 compiled with no patch.

It would be nice if we can use Davy's uBoot with DT support. Because currently, to use a mainline kernel we need to append the dtb file to zImage, as descibed here. So a different uImage is needed for each device.



Edited 1 time(s). Last edit at 04/12/2013 01:01PM by bodhi.

I should also mention that I've created a GoFlex Home Dt board file. But do not have a way to test it yet until I can boot either the iConnect or the GoFlex Net using this method.

@davygravy,

Could you please help bodhi by posting your patch for uboot?

-syong

I've also tried with Davy's uBoot on a GoFlex Net.

U-Boot 2011.12 (Apr 18 2012 - 23:08:20)
Seagate GoFlexNet

I've attached here the config file I used to build this DT kernel.

syong Wrote:
-------------------------------------------------------
> @davygravy,
>
> Could you please help bodhi by posting your patch
> for uboot?

I think we do need the patch. Eventhough I found that davy's pogo e02 board file is in the mainline u-boot, the envs address is at a different address so it won't work right away as drop-in for the current set up.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

I've posted this here before, have you tried it? The patch inside here is for Iomega iConnect and GoFlex Net with device-tree:

https://docs.google.com/open?id=0B8FzJlhz-BzBbl9IRWZ6a1J5cTQ

And here is the device-tree U-Boot binary for GoFlex Net:

https://docs.google.com/open?id=0B8FzJlhz-BzBdXdtdmoxZjhqTk0

But before you flash it, be sure to test it in the old U-Boot by loading it into RAM and then doing a "go" to the new U-Boot. Otherwise if you flash it and it doesn't work you will need JTAG to unbrick your unit.

For device-tree booting you'll need to edit the scripts to load the DTB file, and then add another parameter to the bootm command with the DTB address in RAM.

Thanks rpress,

I've read your post, already downloaded your tarball :) when I started to look at the kirkwood code for the latest uBoot (u-boot-2013.01), I realized that the board files for iconnect, goflex net, dockstar ... were already in there! and they all have env addr set to 0x600000. So I was a little unsure and wanted to gather more info.

I will try and post back.

Thanks again,
- bodhi

@rpress,

It does not seem I can run it from RAM.

U-Boot 2012.04.01 (Nov 20 2012 - 06:47:34) Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0 [PRIME]
Hit any key to stop autoboot:  0 

iconnect => usb start
(Re)start USB...
USB:   Register 10011 NbrPorts 1
USB EHCI 1.00
scanning bus for devices... 3 USB Device(s) found
       scanning bus for storage devices... 1 Storage Device(s) found
iconnect => mw 0x600000 0 1
iconnect => ext2load usb 0:1 0x600000 u-boot-iconnect-rpress.mtd0
Loading file "u-boot-iconnect-rpress.mtd0" from usb device 0:1 (usbda1)
524288 bytes read
iconnect => go x600200
## Starting application at 0x00000000 ...

It stuck here. I recalled the problem about the uBoot env addr. But looking at your patch, you did set it to 0xC0000! any idea why it loaded but did not go?

bodhi Wrote:
-------------------------------------------------------
> iconnect => go x600200
> ## Starting application at 0x00000000 ...
> [/code]
>
> It stuck here. I recalled the problem about the
> uBoot env addr. But looking at your patch, you did
> set it to 0xC0000! any idea why it loaded but did
> not go?

Typo? Shouldn't it be "go 0x600200" and then you will see "## Starting application at 0x00600200".

Sorry, that was a typo! but just tried it again and got same result:

U-Boot 2012.04.01 (Nov 20 2012 - 06:47:34) Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0 [PRIME]
Hit any key to stop autoboot:  0 
iconnect => usb start
(Re)start USB...
USB:   Register 10011 NbrPorts 1
USB EHCI 1.00
scanning bus for devices... 3 USB Device(s) found
       scanning bus for storage devices... 1 Storage Device(s) found
iconnect => mw 0x600000 0 1
iconnect => ext2load usb 0:1 0x600000 u-boot-iconnect-rpress.mtd0
Loading file "u-boot-iconnect-rpress.mtd0" from usb device 0:1 (usbda1)
524288 bytes read
iconnect => go 0x600200
## Starting application at 0x00600200 ...

Thanks,

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Why would you load to 0x00600000 and then jump to 0x00600200 to kick off the chain loaded uboot? (uboot newb here, sorry if this is obvious/common knowledge)

Kurlon,

I thought that's how we boot uBoot image with kirkwood header in the first 0x200 bytes. I believe these uBoot images all built that way. But it strange because I'm running the old iConnect Open-WRT uBoot, and load uBoot image that people have tried running in RAM successfully before (I've also tried Davy's uBoot for Pogo E02, same hang after starting).
Weird!

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 04/16/2013 04:55PM by bodhi.

Ok, I just tested it and it works at address 0x800000 instead. Address 0x600000 does freeze up, I guess it's not valid or something else is there.

U-Boot 2012.04.01-08086-gfe65963-dirty (Aug 16 2012 - 22:40:59)
Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0
88E1116 Initialized on egiga0
Hit any key to stop autoboot:  0 
u-boot>> usb start
(Re)start USB...
USB:   Register 10011 NbrPorts 1
USB EHCI 1.00
scanning bus for devices... 3 USB Device(s) found
       scanning bus for storage devices... 1 Storage Device(s) found
u-boot>> fatload usb 0:1 0x800000 u-boot-iconnect.mtd0
reading u-boot-iconnect.mtd0

524288 bytes read
u-boot>> go 0x800200
## Starting application at 0x00800200 ...


U-Boot 2012.04.01-08086-gfe65963-dirty (Aug 16 2012 - 22:40:59)
Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0
88E1116 Initialized on egiga0
Hit any key to stop autoboot:  0 
u-boot>>

Here is my environment by the way:

arcNumber=3089
baudrate=115200
bootcmd=run bootcmd_usb; usb stop; run bootcmd
bootcmd_pogo=fsload uboot-original-mtd0.kwb; go 0x800200
bootcmd_usb=run usb_init; run usb_load_uimage; run set_bootargs_usb; run usb_boot;
bootdelay=3
console=ttyS0,115200
ethact=egiga0
ethaddr=00:D0:B8:0C:FC:38
filesize=3169D0
led_error=orange blinking
led_exit=green off
led_init=green blinking
mainlineLinux=yes
mtdids=nand0=orion_nand
mtdparts=mtdparts=orion_nand:1M(u-boot),4M(uImage),32M(rootfs),-(data)
partition=nand0,2
set_bootargs_usb=setenv bootargs console=$console root=$usb_root rootdelay=$usb_rootdelay rootfstype=$usb_rootfstype $mtdparts coherent_pool=$usb_coherent_pool
stderr=serial
stdin=serial
stdout=serial
usb_boot=ext2load usb 0:1 0x1100000 /boot/uInitrd; ext2load usb 0:1 0x1000000 /boot/devicetree.dtb; bootm 0x800000 0x1100000 0x1000000
usb_coherent_pool=2M
usb_device=0:1
usb_init=usb start
usb_load_uimage=mw 0x800000 0 1; ext2load usb 0:1 0x800000 /boot/uImage
usb_root=/dev/sda1
usb_rootdelay=3
usb_rootfstype=ext3

No dice for me at 0x800000 either!

U-Boot 2012.04.01 (Nov 20 2012 - 06:47:34) Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0 [PRIME]
Hit any key to stop autoboot:  0 

iconnect => usb start
(Re)start USB...
USB:   Register 10011 NbrPorts 1
USB EHCI 1.00
scanning bus for devices... 3 USB Device(s) found
       scanning bus for storage devices... 1 Storage Device(s) found
iconnect => mw 0x800000 0 1
iconnect => ext2load usb 0:1 0x800000 u-boot-iconnect-rpress.mtd0              
Loading file "u-boot-iconnect-rpress.mtd0" from usb device 0:1 (usbda1)
524288 bytes read
iconnect => go 0x800200
## Starting application at 0x00800200 ...

Davy seems to indicate that if uBoot version you're loading into RAM is too different than the one running then it won't work either:
http://forum.doozan.com/read.php?3,1593

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Yeah I guess I can see it not working with some versions. After all, U-Boot doesn't expect to be run from anything other than a full reset condition.

Well if it makes you feel better I just flashed that version again and my device still works:

U-Boot 2012.04.01-08086-gfe65963-dirty (Aug 16 2012 - 23:17:18)
Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0
88E1116 Initialized on egiga0
Hit any key to stop autoboot:  0 
u-boot>> usb reset
(Re)start USB...
USB:   Register 10011 NbrPorts 1
USB EHCI 1.00
scanning bus for devices... 3 USB Device(s) found
       scanning bus for storage devices... 1 Storage Device(s) found
u-boot>>  fatload usb 0 0x800000 u-boot-iconnect.mtd0
reading u-boot-iconnect.mtd0

524288 bytes read
u-boot>> nand erase 0 0xC0000

NAND erase: device 0 offset 0x0, size 0xc0000
Erasing at 0xa0000 -- 100% complete.
OK
u-boot>> nand write 0x800000 0 0xC0000

NAND write: device 0 offset 0x0, size 0xc0000
 786432 bytes written: OK
u-boot>> reset
resetting ...


U-Boot 2012.04.01-08086-gfe65963-dirty (Aug 16 2012 - 23:17:18)
Iomega iConnect Wireless

SoC:   Kirkwood 88F6281_A0
DRAM:  256 MiB
WARNING: Caches not enabled
NAND:  512 MiB
In:    serial
Out:   serial
Err:   serial
Net:   egiga0
88E1116 Initialized on egiga0
Hit any key to stop autoboot:  0 
u-boot>>

@rpress,

Thanks for the uBoot patch. I'm going to build with your patch and see if it will make any difference. If it does not, I'll go ahead and flash your uBoot image to the iConnect mtd0. It should run OK with non-DT kernel as before?

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 04/16/2013 09:12PM by bodhi.

bodhi Wrote:
-------------------------------------------------------
> @rpress,
>
> Thanks for the uBoot patch. I'm going to build
> with your patch and see if it will make any
> difference. If it does not, I'll go ahead and
> flash your uBoot image to the iConnect mtd0. It
> should run OK with non-DT kernel as before?

Yes, I just tested with a non-DT kernel 2.6.38 and it booted just fine.

@rpress,

I'm having problem with the patch. Here is the build log.

Thanks,
- bodhi

- patching:

root@tldDebian:/usr/src/u-boot-2012.04.01# patch -p1 < ../builds-u-boot-2012.04.01/u-boot-marvell.patch 
patching file board/Iomega/iconnect/Makefile
patching file board/Iomega/iconnect/config.mk
patching file board/Iomega/iconnect/iconnect.c
patching file board/Iomega/iconnect/iconnect.h
patching file board/Iomega/iconnect/kwbimage.cfg
patching file board/Seagate/goflexnet/Makefile
patching file board/Seagate/goflexnet/goflexnet.c
patching file board/Seagate/goflexnet/goflexnet.h
patching file board/Seagate/goflexnet/kwbimage.cfg
patching file include/configs/goflexnet.h
patching file include/configs/iconnect.h
patching file boards.cfg
Hunk #1 FAILED at 158.
1 out of 1 hunk FAILED -- saving rejects to file boards.cfg.rej
patching file include/ide.h

- So I manually changed boards.cfg to include goflexnet and iConnect.

root@tldDebian:/usr/src/u-boot-2012.04.01# cat boards.cfg | egrep -i '(iconnect|goflex)'
iconnect                     arm         arm926ejs   -                   Iomega         kirkwood
goflexnet                    arm         arm926ejs   -                   Seagate        kirkwood

- Error during build (I've included the tail of the log here)

root@tldDebian:/usr/src/u-boot-2012.04.01# make u-boot.kwb |& tee build-uboot-2012.04.01-goflexnet.log
...

gcc  -g  -Os   -fno-common -ffixed-r8 -msoft-float  -D__KERNEL__ -DCONFIG_SYS_TEXT_BASE=0x00600000 -I/usr/src/u-boot-2012.04.01/include -fno-builtin -ffreestanding -nostdinc -isystem /usr/lib/gcc/arm-linux-gnueabi/4.6/include -pipe  -DCONFIG_ARM -D__ARM__ -marm -mabi=aapcs-linux -mno-thumb-interwork -march=armv5te -Wall -Wstrict-prototypes -fno-stack-protector -Wno-format-nonliteral -Wno-format-security -fstack-usage     -o goflexnet.o goflexnet.c -c
goflexnet.c: In function ‘board_early_init_f’:
goflexnet.c:103:2: error: too many arguments to function ‘kirkwood_mpp_conf’
/usr/src/u-boot-2012.04.01/include/asm/arch/mpp.h:315:6: note: declared here
make[1]: *** [goflexnet.o] Error 1
make[1]: Leaving directory `/usr/src/u-boot-2012.04.01/board/Seagate/goflexnet'
make: *** [board/Seagate/goflexnet/libgoflexnet.o] Error 2

Rebuilt once more, still seeing this error.



Edited 3 time(s). Last edit at 04/17/2013 01:05AM by bodhi.