unable to boot new odroidxu4-v4.6

[rhkean]

I've ordered USB-UART module so that I can view the boot console, but it hasn't arrived yet, so I was wondering if anyone else is having this problem with the current branch
thanks,

steps followed:

1. cd /usr/src
2. git clone --depth 1 https://github.com/tobetter/linux.git -b odroidxu4-v4.6 linux-4.6.0
3. cd linux-4.6.0
4. make distclean; make mrproper
5. make odroidxu4-defconfig
6. make -j4
7. make modules_install
8. cp arch/arm/boot/zImage /boot/zImage-make kernelrelease``
9. cp arch/arm/boot/dts/exynos5422-odroidxu4.dtb /boot/odroidxu4.dtb-make kernelrelease``
10. update-initramfs -c -k make kernelrelease``
11. mkimage -A arm -O linux -T ramdisk -C none -a 0 -e 0 -n uInitrd -d /boot/initrd.img-make kernelrelease /boot/uInitrd-make kernelrelease``
12. edit /boot/boot.ini, set bootcmd line to:
    setenv bootcmd "fatload mmc 0:1 0x40008000 zImage-4.6.0+; fatload mmc 0:1 0x42000000 uInitrd-4.6.0+; fatload mmc 0:1 0x44000000 odroidxu4.dtb-4.6.0+; bootz 0x40008000 0x42000000 0x44000000"

[uDude]

I have the 4.6 kernel booting but only the 4 A7 cores show up. I don't know
if the problem is with the device tree or not.
On Jun 23, 2016 3:39 PM, "Robert Kean" [email protected] wrote:

I've ordered USB-UART module so that I can view the boot console, but it
hasn't arrived yet, so I was wondering if anyone else is having this
problem with the current branch
thanks,

steps followed:

1. cd /usr/src
2. git clone --depth 1 https://github.com/tobetter/linux.git -b
   odroidxu4-v4.6 linux-4.6.0
3. cd linux-4.6.0
4. make distclean; make mrproper
5. make odroidxu4-defconfig
6. make -j4
7. make modules_install
8. cp arch/arm/boot/zImage /boot/zImage-make kernelrelease
9. cp arch/arm/boot/dts/exynos5422-odroidxu4.dtb
   /boot/odroidxu4.dtb-make kernelrelease
10. update-initramfs -c -k make kernelrelease
11. mkimage -A arm -O linux -T ramdisk -C none -a 0 -e 0 -n uInitrd -d
    /boot/initrd.img-make kernelrelease /boot/uInitrd-make kernelrelease
12. edit /boot/boot.ini, set bootcmd line to:
    setenv bootcmd "fatload mmc 0:1 0x40008000 zImage-4.6.0+; fatload mmc
    0:1 0x42000000 uInitrd-4.6.0+; fatload mmc 0:1 0x44000000
    odroidxu4.dtb-4.6.0+; bootz 0x40008000 0x42000000 0x44000000"

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#6, or mute the thread
https://github.com/notifications/unsubscribe/AIiMeAgDyAySIryUIygK-EAOZJ5ahJkXks5qOv0tgaJpZM4I9RFG
.

[rhkean]

I can't get it to boot. Did you modify any of the code in this branch or are you using a different repo?

[jobenvil]

@uDude the problems are probably following config kernel features: CONFIG_BIG_LITTLE=yand CONFIG_BL_SWITCHER=y. You should disable them both. By other hand you should enable the two uppers (this worked anycase in compiling 4.7.0-rc-4 from @ioft = @umiddelb repositories, new_mater branch.

the Multi Core Schedulers Support and SMT Scheduler Support
@rhkean do you have a/boot directory in DOS partition? If you worked inboot.iniwith devices (i.e:mmcblk0p2) your should try with UUID, because I observed the weird behaviour that some 4.x kernel series read in mmcblk1p2 instead of mmcblk0p2. Here the mentioned line where you should use UUID: setenv bootrootfs "console=tty1 console=ttySAC2,115200n8 root=UUID=e139ce78-9841-40fe-8823-96a304a09859 rootfstype=ext4 rootwait rw fsck.repair=yes"

[rhkean]

my bootrootfs looks like this:
setenv bootrootfs "console=tty1 root=UUID=e139ce78-9841-40fe-8823-96a304a09859 rootwait ro fsck.repair=yes"

my partition layout like this:

root@odroid1:~# parted /dev/mmcblk0 p
Model: SD 00000 (sd/mmc)
Disk /dev/mmcblk0: 31.4GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Disk Flags: 

Number  Start   End     Size    Type     File system  Flags
 1      1049kB  75.5MB  74.4MB  primary  fat16        lba
 2      75.5MB  31.4GB  31.3GB  primary  ext4

and my uuid's:

root@odroid1:~# ls -Fl /dev/disk/by-uuid/ | grep mmcblk
lrwxrwxrwx 1 root root 15 Jun 23 22:44 96C3-9298 -> ../../mmcblk0p1
lrwxrwxrwx 1 root root 15 Jun 23 22:44 e139ce78-9841-40fe-8823-96a304a09859 -> ../../mmcblk0p2

[jobenvil]

@rhkean Try to copy the uInitrd-4.6+, zImage-4.6+ and odroidxu4*.dtb.4.6+ to /media/boot, where is placed your boot.ini, because this is the DOS partition

[rhkean]

Yes all of the files are going to the correct location. Part1 is mounted
on /boot. Part2 is my root fs.

I'm building and booting the @hardkernel 4.2.0-rc1 kernel without an issue.
On Jun 24, 2016 11:51 AM, "José Benlloch" [email protected] wrote:

@rhkean https://github.com/rhkean Try to copy the uInitrd-4.6+,
zImage-4.6+ and odroidxu4*.dtb.4.6+ to /media/boot?, where is placed your
boot.ini, because this is the DOS partition

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#6 (comment), or mute
the thread
https://github.com/notifications/unsubscribe/AOZAhOSZRlZoQzxHcsPrDDkF1nEno95Qks5qO_0VgaJpZM4I9RFG
.

[jobenvil]

oops, clear. This morning I got the serial UART. I developed a feeling without UART cable trying to boot last weeks. Can you tell me if the blue led lights continuously and not giving you the heartbreak?
wait...paste bin follows.. here, maybe helps you

[umiddelb]

Part1 is mounted on /boot.

Hm, Part1 is supposed to be mounted as /media/boot (if you follow the ODROID way)

[rhkean]

@umiddelb
wasn't aware of the /media/boot part.... I've never had a problem with my initrd and /boot... in fact, in shouldn't matter since the mount points are determined after the kernel loads the uInitrd and performs a switch_root

@jobenvil
I am pretty sure that it loads the kernel, but doesn't get to the heartbeat. Sorry, I'm not at home at the moment, but I'm pretty sure it was on-on-pause

[rhkean]

wait...paste bin follows.. here, maybe helps you

@jobenvil , is that your boot.ini?

nevermind, I just followed the link... I'll review and get back to you

[jobenvil]

it is, why, it's so ugly? ;-)

[rhkean]

and, where are you pulling a working 4.7 kernel from?

[jobenvil]

@uDude the problems are probably following config kernel features: CONFIG_BIG_LITTLE=yand CONFIG_BL_SWITCHER=y. You should disable them both. By other hand you should enable the two uppers (this worked anycase in compiling 4.7.0-rc-4 from @ioft = @umiddelb repositories, new_mater branch.
This thread, at the beginning. I have running this one from new_master.

root@hiperborea ~ # uname -a
Linux hiperborea 4.7.0-rc4-g50f5c3f-dirty #2 SMP Thu Jun 23 23:11:41 CEST 2016 armv7l armv7l armv7l GNU/Linux

[rhkean]

I did confirm by scanning the hardkernel U-Boot repo that the mount point of the boot partition after the kernel is loaded is not relevant (or even required) during the U-Boot Stage.

very brief explanation:
After U-Boot is loaded into memory from the 1st 1MB of the mmcblk0 device, it then scans for a boot.ini file by first looking for [FAT Part 1]/boot.ini, then [ext4 Part 1]/boot/boot.ini, then [ext4 Part2]/boot/boot.ini (note, I think this technically this means we can use a single partition layout if we put all of our boot stuff in the the /boot directory of an ext4 partition):

        "cfgload - boot.ini textfile load from FAT32/ext4\n",
        "<interface(only support mmc 0:1/0:2)>\n"
        "       - boot.ini file load from FAT32/ext4 on 'interface'\n"
        "Will attempt boot from: \n"
        "       fat mmc 0:1 /boot.ini\n"
        "       ext4 0:1 /boot/boot.ini\n"
        "       ext4 0:2 /boot/boot.ini\n"
        "on the above sequence\n"

Once a boot.ini file is found it is parsed and this statement:
fatload mmc 0:1 0x40008000 zImage
says load from fat partition on mmc device 0, partition 1 a file called zImage into memory starting at address 0x40008000.... again no mount point dependency.

Neat stuff...

[tobetter]

Wow, how the conversion is being here without me? Please keep in mind that v4.6 is not test yet and committing to release new Debian/Ubuntu binary - only headless version. Hope to be released by end of this month. Would be glad to have your suggestion or patches to fix v4.6 branch.

[moonlinux]

The issue with cloud shell image is you need to update the boot.ini to set UUID.
boot.ini is present in /boot ie: /dev/mmcblk0

setenv bootrootfs "root=UUID=669a4d12-4c0d-4b61-a5fc-9de45699973b rootwait ro fsck.repair=yes"

Also you need to append the /etc/fstab

odroid@odroid-server:~$ cat /etc/fstab
UUID=1F12-97C4 /boot vfat rw,relatime,fmask=0022,dmask=0022,codepage=437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro 0 0
UUID=669a4d12-4c0d-4b61-a5fc-9de45699973b / ext4 defaults 0 0

Note: Their is space limitation in /dev/mmcblk0
So you need to delete few of the unnecessary files or keep backup in other partition before you upgrade using apt-get.

To install latest kernel image deb

$ sudo apt-get update
$ sudo apt-get upgrade
$ sudo apt-get install linux-image-4.6.0-odroidxu4

[rhkean]

OK, I'm at a loss.... neither @tobetter 's 4.2 nor 4.6 branches will boot on my XU4, but the @hardkernel 4.2.0-rc1 will boot (following same instructions as OP above in all 3 cases)... maybe I just need to walk away for a day or so and come back to this...

[moonlinux]

I also face the same issue earlier, as their was a slight changes in MMC driver side which made it mandatory to use UUID into boot.ini and /etc/fstab

So it depend on what is the UUID of your microSD or eMMC card you need to append these file to boot into filesystem.

odroid@odroid-server:~$ sudo blkid
/dev/mmcblk1p1: SEC_TYPE="msdos" LABEL="BOOT" UUID="1F12-97C4" TYPE="vfat"
/dev/mmcblk1p2: LABEL="rootfs" UUID="669a4d12-4c0d-4b61-a5fc-9de45699973b" TYPE="ext4"
/dev/sda1: UUID="A646DDB646DD8787" TYPE="ntfs"
/dev/sda2: UUID="0CA0C949A0C93A4A" TYPE="ntfs"
/dev/sda3: LABEL="odroiddev" UUID="30eed8e4-85ce-4357-b648-6de45c3e56ec" TYPE="ext4"
odroid@odroid-server:~$

After you boot into these you can update using apt-get.

[rhkean]

@moonlinux
I'm not clear on what you refer to as you have now made reference to using apt-get twice. apt-get for what? We are discussing compiling and booting from source.

In reference to the UUID, I am already using a UUID in the boot.ini for the rootfs:

setenv bootrootfs "console=tty1 root=UUID=e139ce78-9841-40fe-8823-96a304a09859 rootwait ro fsck.repair=yes"

[moonlinux]

Opps: I am referring the cloud shell image for Odroid XU4 image for server.
http://dn.odroid.com/5422/ODROID-XU4/release/

Here is the install process I follow.
Build the kernel.

#Build the kernel using odroidxu3_defconfig
#make mrproper
make exynos_defconfig
make -j8 dtbs zImage modules LOCALVERSION=-xu4ml
make modules_install LOCALVERSION=-xu4ml

Install the kernel

kernelversion=`cat ./include/config/kernel.release`
cp -v .config /boot/config-$kernelversion
echo "kernel version" $kernelversion
cp -v arch/arm/boot/dts/exynos5422-odroidxu3.dtb /media/boot/exynos5422-odroidxu3-$kernelversion.dtb
cp -v arch/arm/boot/dts/exynos5422-odroidxu3.dtb /media/boot/exynos5422-odroidxu3.dtb
cp -v arch/arm/boot/dts/exynos5422-odroidxu4.dtb /media/boot/exynos5422-odroidxu4-$kernelversion.dtb
cp -v arch/arm/boot/dts/exynos5422-odroidxu4.dtb /media/boot/exynos5422-odroidxu4.dtb
cp -v arch/arm/boot/zImage /media/boot/zImage-$kernelversion
cp -v arch/arm/boot/zImage /media/boot/vmlinuz-$kernelversion
cp -v arch/arm/boot/zImage /media/boot/zImage
update-initramfs -c -k $kernelversion
mkimage -A arm -O linux -T ramdisk -C gzip -a 0x0 -e 0x0 -n uInitrd -d /boot/initrd.img-$kernelversion uInitrd-$kernelversion
cp -v  uInitrd-$kernelversion /boot/uInitrd-$kernelversion
cp -v  uInitrd-$kernelversion /boot/uInitrd
cp -v .config /media/boot/config-$kernelversion
sync

It seem that in bootcmd incorrect in boot.ini
Below is simple version, only thing you need to update the version

ODROIDXU-UBOOT-CONFIG

# U-Boot Parameters (DO NOT MODIFY)
setenv version 4.6.0-88
#setenv version 4.6.0-87
setenv zimage vmlinuz-${version}
setenv uinitrd uInitrd-${version}
setenv fdtbin exynos5422-odroidxu4-${version}.dtb

setenv zimage_addr_r 0x40008000
setenv initrd_addr_r 0x42000000
setenv fdtbin_addr_r 0x44000000

setenv initrd_high "0xffffffff"
setenv fdt_high "0xffffffff"

# Default boot argument
setenv bootrootfs "root=UUID=669a4d12-4c0d-4b61-a5fc-9de45699973b rootwait ro fsck.repair=yes"
setenv console "console=ttySAC2,115200n8"

setenv bootargs "${bootrootfs} ${console}"

# boot commands
fatload mmc 0:1 ${zimage_addr_r} ${zimage}
fatload mmc 0:1 ${initrd_addr_r} ${uinitrd}
fatload mmc 0:1 ${fdtbin_addr_r} ${fdtbin}
bootz ${zimage_addr_r} ${initrd_addr_r} ${fdtbin_addr_r}"

[umiddelb]

wasn't aware of the /media/boot part.... I've never had a problem with my initrd and /boot... in fact, in shouldn't matter since the mount points are determined after the kernel loads the uInitrd and performs a switch_root
I did confirm by scanning the hardkernel U-Boot repo that the mount point of the boot partition after the kernel is loaded is not relevant (or even required) during the U-Boot Stage.

You're right, the mount point becomes to be relevant when you install/upgrade the official HK kernel package via apt.

[rhkean]

Got my tobetter-4.2.0 kernel to boot using the config from my hardkernel-4.2.0-rc1 kernel.

zcat /proc/config.gz > .config
make oldconfig

Now, to figure out what's so different between the 2 odroidxu4_defconfig files.

As for my tobetter-4.6.0 kernel... still no luck. I tried the same thing as above and took the default values on the new stuff. The board is definitely loading the kernel (according to the blue led), but not loading the uInitrd.

[jobenvil]

Now, to figure out what's so different between the 2 odroidxu4_defconfig files.

it's "somehow" easy, generate both .config and then git diff .config1 .config2 this way you can see what is added or deleted. I did this with the .config's for exynos_defconfig and odroidxu4_defconfig, after doing make exynos_defconfig and make odroidxu4_defconfig

[uDude]

@rhkean I'll submit a small patch -- I don't think it will fix your issue, you really need that console cable to arrive.

odroid@od2:/repos/tobetter/linux$ uname -a
Linux od2 4.6.0-g3bb2b88-dirty #1 SMP PREEMPT Thu Jun 16 09:31:11 MDT 2016 armv7l armv7l armv7l GNU/Linux
odroid@od2:/repos/tobetter/linux$ git diff
diff --git a/drivers/phy/phy-exynos5-usbdrd.c b/drivers/phy/phy-exynos5-usbdrd.c
index 70ef7e3..d5e1be4 100644
--- a/drivers/phy/phy-exynos5-usbdrd.c
+++ b/drivers/phy/phy-exynos5-usbdrd.c
@@ -629,7 +629,7 @@ static int crport_ctrl_write(struct exynos5_usbdrd_phy *phy_drd,

• SuperSpeed requirements on Exynos5420 and Exynos5800 systems,
• which have 28nm USB 3.0 DRD PHY.
  */
  -static void exynos5420_usbdrd_phy_calibrate(struct exynos5_usbdrd_phy *phy_drd)
  +static int exynos5420_usbdrd_phy_calibrate(struct exynos5_usbdrd_phy *phy_drd)
  {
  unsigned int temp;
  int ret = 0;

-===================
also, I'll snag my config fiel and upload it... NOTE: I did have issues with the base network interface eth0 (I don't use wifi). Network manager kept dropping eth0 and I took a lazy approach and added the following to /etc/rc.local which fixes thie issue (but the default interface is no lonter eth0):

/sbin/rmmod r8152 /sbin/modprobe r8152 /sbin/dhclient

[jobenvil]

Here my serial output, maybe helps you. I will recommend to start with some already tested kernel branch, which it works at first shot, without any further modification, until you get the serial cable. I can provide .config, boot.ini and /etc/fstab (or alreday you have from the first pastebin)

[uDude]

@rhkean I hope this dan interfae works for me today. I had a couple of issues on the 4.6 kernel.
FIrst note unmae -a
droid@od2:~/repos/tobetter/linux$ uname -a Linux od2 4.6.0-g3bb2b88-dirty #1 SMP PREEMPT Thu Jun 16 09:31:11 MDT 2016 armv7l armv7l armv7l GNU/Linux

Next Note that I only see 4 cores in /proc/cpuino (the A7s not the A15s).

I had a problem with the network interface not coming up. Network manager was disabling it after it started. So I lazily put the following lines in /etc/rc.local

/sbin/rmmod r8152
/sbin/modprobe r8152
/sbin/dhclient

This gave me an interface (not named eth0). I will keep looking at how to get all 8 cores visible in cpuinfo.

Oh yes, in building the kernel, I had an error that I patched -- I'll issue a pull request today.

odroid@od2:~/repos/tobetter/linux$ git diff
diff --git a/drivers/phy/phy-exynos5-usbdrd.c b/drivers/phy/phy-exynos5-usbdrd.c
index 70ef7e3..d5e1be4 100644
--- a/drivers/phy/phy-exynos5-usbdrd.c
+++ b/drivers/phy/phy-exynos5-usbdrd.c
@@ -629,7 +629,7 @@ static int crport_ctrl_write(struct exynos5_usbdrd_phy *phy_drd,
* SuperSpeed requirements on Exynos5420 and Exynos5800 systems,
* which have 28nm USB 3.0 DRD PHY.
*/
-static void exynos5420_usbdrd_phy_calibrate(struct exynos5_usbdrd_phy *phy_drd)
+static int exynos5420_usbdrd_phy_calibrate(struct exynos5_usbdrd_phy *phy_drd)
{
unsigned int temp;
int ret = 0;

uDude

[jobenvil]

I will keep looking at how to get all 8 cores visible in cpuinfo.

@uDude I/we had this wronng 4 cores (LITTLE) behaviour as well. The odroid forum talks about it. I solved it disabling CONFIG_BIG_LITTLE=y and CONFIG_BL_SWITCHER=y. See beginning of this GitHub issue.

[uDude]

@jobenvil thx. I was just making this change -- it is nice to know that this will do this trick.

Much appreciated.

[rhkean]

@uDude
rhkean I'll submit a small patch -- I don't think it will fix your issue, you really need that console cable to arrive.
So true.... 😆 (it's that darn 1.8V limitation, otherwise I'd just wire it directly to one of the uart's on a spare RPi and use minicom. I did that with my PineA64 boards and it worked perfectly)

I doubt it's the usbdrd Warning, though... I get that on the 4.2 kernels, also. (that's not to say it shouldn't be patch, though. it most definitely should)

@jobenvil
I will recommend to start with some already tested kernel branch, which it works at first shot,

I attempted to build that 4.7 kernel and it panic'd my machine during the make -j4. I have to try again without the -j4 so that I can determine what caused the crash. BTW, the @ioft repository got patched up to -rc5 last night.