Re: Nanobsd for RPI4

Reply: Guido Falsi : "Re: Nanobsd for RPI4"
In reply to: Guido Falsi : "Re: Nanobsd for RPI4"
Go to: [ bottom of page ] [ top of archives ] [ this month ]
From: Andrew Hankinson <andrew.hankinson_at_gmail.com>
Date: Tue, 16 Jan 2024 16:48:55 UTC
Thank you! Using this I've gotten a bit further. 

Would you mind sharing what you set for the NANO_SLICE_CODE_SIZE value? I think this (so far) is the only undefined variable. I tried guessing at it, but I think I chose poorly...

Thanks,
-Andrew

> On 16 Jan 2024, at 12:49, Guido Falsi <mad@madpilot.net> wrote:
> 
> On 16/01/24 11:34, Andrew Hankinson wrote:
>>> On 16 Jan 2024, at 10:46, Guido Falsi <mad@madpilot.net> wrote:
>>> 
>>> On 16/01/24 10:30, Andrew Hankinson wrote:
>>>>> On 16 Jan 2024, at 10:03, Guido Falsi <mad@madpilot.net> wrote:
>>>>> 
>>>>> On 16/01/24 09:52, Andrew Hankinson wrote:
>>>>>> Hello,
>>>>>> I saw a recent posting to this list about nanobsd for RPI3:
>>>>>> Re: Nanobsd builds for rpi3 out of embedded blow up during build <https://lists.freebsd.org/archives/freebsd-arm/2024-January/003509html>
>>>>>> lists.freebsd.org <https://lists.freebsd.org/archives/freebsd-arm/2024-January/003509.html>
>>>>>> favicon.ico <https://lists.freebsd.org/archives/freebsd-arm/2024-January/003509.html>
>>>>>> <https://lists.freebsdorg/archives/freebsd-arm/2024-January/003509.html>
>>>>>> I just wanted to chime in to say I am seeing the exact same thing for a RPI4 build with both 14.0 release and 15.0 current. I asked about it on the forums, but there didn’t seem to be any solution.
>>>>>> Has anyone here managed to get a nanobsd build working?
>>>>> 
>>>>> I am using nanobsd with RPi3. Never tried RPi4. I did some customization to the nanobsd scripts and configuration to get it working though.
>>>>> 
>>>>> Last time I rebuilt my image was in mid december. I'm going to build a new image in the next pair of weeks.
>>>>> 
>>>>> Is the problem limited to the boot code?
>>>> I haven't been able to get past the boot code bit, but up until this point it works OK. I've simply copied the RPI3.cfg file and changed a few variables.
>>>> My build stops with the following logs:
>>>> # cat /usr/embedded/rpi4/_.cust.dos_boot_part
>>>> + dos_boot_part
>>>> + local 'd=/usr/local/share/u-boot/u-boot-rpi4'
>>>> + local 'f=/usr/embedded/rpi4/_.fat'
>>>> + rm -rf /usr/embedded/rpi4/_.fat
>>>> + mkdir /usr/embedded/rpi4/_.fat
>>>> + chdir /usr/embedded/rpi4/_.fat
>>>> + cp /usr/local/share/u-boot/u-boot-rpi4/README /usr/local/share/u-boot/u-boot-rpi4/metadata /usr/local/share/u-boot/u-boot-rpi4/u-boot.bin .
>>>> + touch uEnv.txt
>>>> + cp '/usr/embedded/rpi4/_.w/boot/dtb/*.dtb' .
>>>> cp: /usr/embedded/rpi4/_.w/boot/dtb/*.dtb: No such file or directory
>>>> This seems to be a problem with the dos_boot_part() function in /usr/src/tools/tools/nanobsd/embedded/common.
>>>> Your code seems to bypass this and uses the rpi-firmware package dtbo files.
>>> 
>>> I was seeing problems simlar to what you are describing, so I wrote my own function.
>>> 
>>>> So I guess my questions are:
>>>> 1. What changed that the .dtb files are no longer created in the regular build?
>>>> 2. Should the dos_boot_part() function be changed to take the files from the rpi-firmware pkg instead?
>>>> Not being familiar with the nanobsd building process, I'm not really sure what is the best way to go.
>>> 
>>> I wrote this code some time ago and I don't remember the details, anyway my code does take the files from the pkg.
>>> 
>>> I took idea from the release scripts for aarch64 from:
>>> 
>>> src/release/release.sh
>>> src/release/arm64/aarch64.conf
>>> src/release/tools/arm.subr
>>> 
>>> which have more current code.
>> Would you be willing to share your full nanobsd config file? Or is that the only custom part you needed to get it working?
> 
> I have some more custom parts, but most are unrelated to RPi3.
> 
> Actually mine is a configuration I used with a PCEngine APU2 board, I converted to RPi3.
> 
> The only other relevant configuration is this part:
> 
> calculate_partitioning ( ) (
> )
> 
> create_code_slice ( ) (
>    pprint 2 "build code slice"
>    pprint 3 "log: ${NANO_OBJ}/_.cs"
> 
>    (
>    IMG=${NANO_DISKIMGDIR}/_.disk.image
>    MNT=${NANO_OBJ}/_.mnt
>    mkdir -p ${MNT}
>    CODE_SIZE=${NANO_SLICE_CODE_SIZE}
> 
>    if [ "${NANO_MD_BACKING}" = "swap" ] ; then
>        MD=`mdconfig -a -t swap -s ${CODE_SIZE} -x ${NANO_SECTS} \
>            -y ${NANO_HEADS}`
>    else
>        echo "Creating md backing file..."
>        rm -f ${IMG}
>        dd if=/dev/zero of=${IMG} seek=${CODE_SIZE} count=0
>        MD=`mdconfig -a -t vnode -f ${IMG} -x ${NANO_SECTS} \
>            -y ${NANO_HEADS}`
>    fi
> 
>    trap "echo 'Running exit trap code' ; df -i ${MNT} ; umount ${MNT} || true ; mdconfig -d -u $MD" 1 2 15 EXIT
> 
>    gpart create -s bsd ${MD}
>    gpart add -t freebsd-ufs -a 64k -b 64k ${MD}
>    #gpart bootcode -b ${NANO_WORLDDIR}/boot/boot ${MD}
> 
>    # Create first image
>    populate_slice /dev/${MD}${NANO_PARTITION_ROOT} ${NANO_WORLDDIR} ${MNT} "${NANO_ROOT}"
>    mount /dev/${MD}${NANO_PARTITION_ROOT} ${MNT}
>    echo "Generating mtree..."
>    ( cd ${MNT} && mtree -c ) > ${NANO_OBJ}/_.mtree
>    ( cd ${MNT} && du -k ) > ${NANO_OBJ}/_.du
>    nano_umount ${MNT}
> 
>    if [ "${NANO_MD_BACKING}" = "swap" ] ; then
>        echo "Writing out _.disk.image..."
>        dd conv=sparse if=/dev/${MD} of=${NANO_DISKIMGDIR}/_.disk.image bs=64k
>    fi
>    mdconfig -d -u $MD
> 
>    trap - 1 2 15 EXIT
> 
>    ) > ${NANO_OBJ}/_.cs 2>&1
> )
> 
> # Custom disk image command, from common embedded image
> create_diskimage ( ) (
>    pprint 2 "build diskimage"
>    pprint 3 "log: ${NANO_LOG}/_.di"
> 
>    (
> 
>    IMG=${NANO_DISKIMGDIR}/${NANO_IMGNAME}
>    MNT=${NANO_OBJ}/_.mnt
>    mkdir -p ${MNT}
> 
>    if [ "${NANO_MD_BACKING}" = "swap" ] ; then
>        MD=`mdconfig -a -t swap -s ${NANO_MEDIASIZE} -x ${NANO_SECTS} \
>            -y ${NANO_HEADS}`
>    else
>        echo "Creating md backing file..."
>        rm -f ${IMG}
>        dd if=/dev/zero of=${IMG} seek=${NANO_MEDIASIZE} count=0
>        MD=`mdconfig -a -t vnode -f ${IMG} -x ${NANO_SECTS} \
>            -y ${NANO_HEADS}`
>    fi
> 
>    trap "echo 'Running exit trap code' ; df -i ${MNT} ; nano_umount ${MNT} || true ; mdconfig -d -u $MD" 1 2 15 EXIT
> 
>    gpart create -s MBR ${MD}
>    gpart add -t '!12' -a 512k -s ${NANO_SLICE_EFI_SIZE} ${MD}
>    gpart set -a active -i 1 ${MD}
>    newfs_msdos -L efi -F 16 /dev/${MD}${NANO_SLICE_EFI}
>    gpart add -t freebsd -s ${NANO_SLICE_CFG_SIZE} ${MD}
>    gpart add -t freebsd -s ${NANO_SLICE_CODE_SIZE} ${MD}
>    gpart add -t freebsd -s ${NANO_SLICE_CODE_SIZE} ${MD}
> 
>    mount_msdosfs /dev/${MD}${NANO_SLICE_EFI} ${MNT}
>    populate_boot_part ${MNT}
>    umount ${MNT}
> 
>    echo "Writing code image..."
>    dd conv=sparse if=${NANO_DISKIMGDIR}/_.disk.image of=/dev/${MD}${NANO_SLICE_ROOT} bs=64k
> 
>    # Create Config slice
>    populate_cfg_slice /dev/${MD}${NANO_SLICE_CFG} "${NANO_CFGDIR}" ${MNT} "${NANO_SLICE_CFG}"
> 
>    if [ "${NANO_MD_BACKING}" = "swap" ] ; then
>        if [ ${NANO_IMAGE_MBRONLY} ]; then
>            echo "Writing out _.disk.mbr..."
>            dd if=/dev/${MD} of=${NANO_DISKIMGDIR}/_.disk.mbr bs=512 count=1
>        else
>            echo "Writing out ${NANO_IMGNAME}..."
>            dd if=/dev/${MD} of=${IMG} bs=64k
>        fi
> 
>        echo "Writing out ${NANO_IMGNAME}..."
>        dd conv=sparse if=/dev/${MD} of=${IMG} bs=64k
>    fi
> 
>    mdconfig -d -u ${MD}
> 
>    trap - 1 2 15 EXIT
> 
>    ) > ${NANO_LOG}/_.di 2>&1
> )
> 
>>> 
>>> -- 
>>> Guido Falsi <mad@madpilot.net>
>>> 
> 
> -- 
> Guido Falsi <mad@madpilot.net>
>