Le 14/11/2013 14:10, Alexander Graf a écrit :
>
> Am 14.11.2013 um 05:03 schrieb Guillaume Gardet <[email protected]>:
>
>> Le 12/11/2013 13:49, Alexander Graf a écrit :
>>> Cool, thanks a lot for thd patch! See a few comments inline.
>>>
>>> Am 11.11.2013 um 05:18 schrieb <[email protected]>:
>>>
>>>> home:Guillaume_G:branches:openSUSE:13.1:Ports/JeOS ->
>>>> openSUSE:13.1:Ports/JeOS
>>>>
>>>>
>>>> https://build.opensuse.org/request/show/206130
>>>>
>>>> Description: * Various fixes:
>>>> - rename loadfdt (already used differently upstream) in should_load_fdt
>>>> - rename usefdt in should_use_fdt
>>>> - Undef $bootpart before using it in for loop otherwise var is not updated
>>>> (bug or feature?)
>>>>
>>>> ++++++ uboot-image-arndale-setup
>>>> --- uboot-image-arndale-setup
>>>> +++ uboot-image-arndale-setup
>>>> @@ -64,7 +64,7 @@
>>>> # mkinitrd to point to the kernel and initrd in order
>>>> # to support kernel updates
>>>> # ----
>>>> - kernel=uImage
>>>> + kernel=zImage
>>> Don't we need uImages at all anymore? That change is not documented in the
>>> description btw :).
>> I wanted to default to zImage as most images are zImage now. Did I drop
>> uImage support with this modification?
> Well, I don't see sny target setting this to uImage. But maybe I just missed
> it :).
>
> I think it makes sense to go with zImage brw, but I'm not 100% sure all out
> u-boot versions are capable of it.
I reset it to uImage ATM. We should not break more our images now. We could
work on that on Factory later. ;)
>
>>>> initrd=initrd
>>>>
>>>> # gather kernel command line
>>>> @@ -100,7 +100,7 @@
>>>> case $flavor in
>>>> highbank)
>>>> bootdevs=scsi
>>>> - usefdt=1
>>>> + should_use_fdt=1
>>>> setdev=1
>>>> # Highbank's u-boot already prepends 0x to the file size
>>>> sizeprefix=
>>>> @@ -114,30 +114,30 @@
>>>> boottype=bootm
>>>> kerneladdr=0x48000000
>>>> ramdiskaddr=0x43100000
>>>> - loadfdt=1
>>>> - fdt=$flavor.bin
>>>> + should_load_fdt=1
>>>> + fdtfile=$flavor.bin
>>>> fdt_addr=0x43000000
>>>> ;;
>>>>
>>>> loco)
>>>> - usefdt=1
>>>> - loadfdt=1
>>>> + should_use_fdt=1
>>>> + should_load_fdt=1
>>>> units="0 1"
>>>> - fdt=imx53-qsb.dtb
>>>> + fdtfile=imx53-qsb.dtb
>>>> ;;
>>>> arndale)
>>>> boottype=bootm
>>>> - usefdt=1
>>>> - loadfdt=1
>>>> - fdt=exynos5250-arndale.dtb
>>>> + should_use_fdt=1
>>>> + should_load_fdt=1
>>>> + fdtfile=exynos5250-arndale.dtb
>>>> fdt_addr=0x44000000
>>>> ;;
>>>> chromebook)
>>>> bootdevs="mmc usb"
>>>> bootparts='${unit}:2'
>>>> - usefdt=1
>>>> - loadfdt=1
>>>> - fdt=exynos5250-snow.dtb
>>>> + should_use_fdt=1
>>>> + should_load_fdt=1
>>>> + fdtfile=exynos5250-snow.dtb
>>>> ;;
>>>> esac
>>>>
>>>> @@ -150,9 +150,9 @@
>>>> fdt_high=0xffffffff
>>>>
>>>> # copy bash variables into boot script
>>>> -for variable in kernel initrd fdt flavor target bootargs bootdevs
>>>> bootparts \
>>>> - initrd_high fdt_high boottype kerneladdr ramdiskaddr
>>>> usefdt \
>>>> - loadfdt fdt_addr setdev units sizeprefix; do
>>>> +for variable in kernel initrd fdtfile flavor target bootargs bootdevs
>>>> bootparts \
>>>> + initrd_high fdt_high boottype kerneladdr ramdiskaddr
>>>> should_use_fdt \
>>>> + should_load_fdt fdt_addr setdev units sizeprefix; do
>>>> value=$(eval "echo $(echo \$$variable)")
>>>> # only set variables that contain data, leave the others alone
>>>> if [ "$value" ]; then
>>>> @@ -165,18 +165,18 @@
>>>> printenv ramdiskaddr|| setenv ramdiskaddr ${ramdisk_addr_r}
>>>> setenv load_kernel 'ext2load ${bootdev} ${bootpart} ${kerneladdr}
>>>> ${kernel}'
>>>> setenv load_initrd 'ext2load ${bootdev} ${bootpart} ${ramdiskaddr}
>>>> ${initrd}; setenv rd_filesize ${sizeprefix}${filesize}'
>>>> - if itest 1${loadfdt} == 11; then
>>>> - setenv load_fdt 'ext2load ${bootdev} ${bootpart} ${fdt_addr} ${fdt}'
>>>> + if itest 1${should_load_fdt} == 11; then
>>> This makes sense
>>>
>>>> + setenv should_load_fdt 'ext2load ${bootdev} ${bootpart} ${fdt_addr}
>>>> ${fdtfile}'
>>> Here the should is a misnomer, sorry. It should rather be 'do_load_fdt' ir
>>> something similar, as the contents of that variable are going to be
>>> executed later. We should give it a name that makes its function obvious.
>>> Sorry for misguiding you on it earlier.
>> My fault. We could use loadfdt as upstream does.
> Works for me :)
SR #206890
Guillaume
--
To unsubscribe, e-mail: [email protected]
To contact the owner, e-mail: [email protected]
