| Message ID | tencent_3E8054E385AF0DCBE94DC17FAD12060E6707@qq.com |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [1/2] configs/milkv_duo: new defconfig | expand |
Hi Hanyuan, Thank you for your patch. Please find below a quick, incomplete review. On 20/11/2023 16:20, Hanyuan Zhao via buildroot wrote: > Milk-V Duo is an ultra-compact embedded development platform based on the CV1800B chip. > It can run Linux and RTOS, providing a reliable, low-cost, and high-performance platform > for professionals, industrial ODMs, AIoT enthusiasts, DIY hobbyists, and creators. > > Co-authored-by: Emil S <emil.soltys@outlook.com> > Co-authored-by: GP Orcullo <kinsamanka@gmail.com> > Signed-off-by: Hanyuan Zhao <hanyuan-z@qq.com> > --- > board/milkv/duo/README.md | 64 +++++ > board/milkv/duo/genimage.cfg | 30 ++ > board/milkv/duo/overlay/etc/dnsmasq.conf | 4 + > board/milkv/duo/overlay/etc/init.d/S05runonce | 32 +++ > board/milkv/duo/overlay/etc/init.d/S10hwinit | 44 +++ > board/milkv/duo/overlay/etc/init.d/S15swap | 25 ++ > board/milkv/duo/overlay/etc/init.d/S99user | 27 ++ > board/milkv/duo/overlay/etc/milkv-duo.conf | 56 ++++ > board/milkv/duo/overlay/etc/profile | 36 +++ > .../duo/overlay/etc/runonce.d/01make-swap | 6 + > .../overlay/etc/runonce.d/05store-eth0-mac | 12 + > .../duo/overlay/etc/runonce.d/ran/.gitkeep | 0 > .../milkv/duo/overlay/opt/milkv-duo/blink.sh | 25 ++ > .../duo/overlay/opt/milkv-duo/duo-init.sh | 8 + > .../duo/overlay/opt/milkv-duo/run_usb.sh | 265 ++++++++++++++++++ > .../milkv/duo/overlay/opt/milkv-duo/uhubon.sh | 63 +++++ > .../duo/overlay/opt/milkv-duo/usb-host.sh | 3 + > .../overlay/opt/milkv-duo/usb-mass-storage.sh | 10 + > .../duo/overlay/opt/milkv-duo/usb-rndis.sh | 14 + > ...ensbi-porting-adapt-to-CV180X-CV181X.patch | 44 +++ > ...csr-zifencei-and-kill-compile-errors.patch | 59 ++++ > ...ize-the-value-of-a-variable-in-stack.patch | 26 ++ > board/milkv/duo/post-image.sh | 52 ++++ Buildroot defconfigs should be _minimal_, i.e. just enough to boot the device and perhaps get a (wired) network connection. Looking at all those files, clearly these defconfigs are _not_ minimal. I think pretty much everything in the overlay can be removed. > configs/milkv_duo_musl_riscv64_64mb_defconfig | 94 +++++++ > configs/milkv_duo_musl_riscv64_defconfig | 93 ++++++ > 25 files changed, 1092 insertions(+) > create mode 100644 board/milkv/duo/README.md > create mode 100644 board/milkv/duo/genimage.cfg > create mode 100644 board/milkv/duo/overlay/etc/dnsmasq.conf > create mode 100755 board/milkv/duo/overlay/etc/init.d/S05runonce > create mode 100755 board/milkv/duo/overlay/etc/init.d/S10hwinit > create mode 100755 board/milkv/duo/overlay/etc/init.d/S15swap > create mode 100755 board/milkv/duo/overlay/etc/init.d/S99user > create mode 100644 board/milkv/duo/overlay/etc/milkv-duo.conf > create mode 100644 board/milkv/duo/overlay/etc/profile > create mode 100755 board/milkv/duo/overlay/etc/runonce.d/01make-swap > create mode 100755 board/milkv/duo/overlay/etc/runonce.d/05store-eth0-mac > create mode 100644 board/milkv/duo/overlay/etc/runonce.d/ran/.gitkeep > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/blink.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/run_usb.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-host.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-mass-storage.sh > create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-rndis.sh > create mode 100644 board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch > create mode 100644 board/milkv/duo/patches/opensbi/0002-fix-set-zicsr-zifencei-and-kill-compile-errors.patch > create mode 100644 board/milkv/duo/patches/opensbi/0003-fix-initialize-the-value-of-a-variable-in-stack.patch > create mode 100755 board/milkv/duo/post-image.sh > create mode 100644 configs/milkv_duo_musl_riscv64_64mb_defconfig > create mode 100644 configs/milkv_duo_musl_riscv64_defconfig > > diff --git a/board/milkv/duo/README.md b/board/milkv/duo/README.md > new file mode 100644 > index 0000000000..f302d24de9 > --- /dev/null > +++ b/board/milkv/duo/README.md The readme should be called readme.txt (and not be in markdown). > @@ -0,0 +1,64 @@ > + > +## Milk-V Duo Introduction > + > +Milk-V Duo is an ultra-compact embedded development platform based on the rv64 CV1800B chip. > + > +It can run Linux and RTOS, providing a reliable, cost-effective, and high-performance platform for professionals, industrial ODMs, AIoT enthusiasts, DIY enthusiasts, and creators. Please wrap text at 80 columns. > + > +Features: > + > +- [x] The RISC-V C906 processor with clock speeds of 1GHz and 700MHz. > +- [x] CSI-2 (MIPI Serial Camera Interface) > +- [x] Ethernet PHY > +- [x] USB RNDIS & USB HOST & USB Mass Storage > +- [x] Multiple peripherals > + - [x] - Up to 3x I2C > + - [x] - Up to 5x UART > + - [x] - Up to 1x SDIO1 > + - [x] - Up to 1x SPI > + - [x] - Up to 2x ADC > + - [x] - Up to 7x PWM > + - [x] - Up to 1x RUN > + - [x] - Up to 1x JTAG > + > +## How to build > + > +```shell > +$ make milkv_duo_musl_riscv64_defconfig > +$ make > +``` > + > +By default, a portion of the RAM is allocated to ION, which is the memory used when running algorithms with the camera. If you are not using the camera, please try: > + > +```shell > +$ make milkv_duo_musl_riscv64_64mb_defconfig > +$ make > +``` > + > +To have all the 64mb free memory. > + > + > +To configure the kernel or u-boot or others, please try: > + > +```shell > +$ make linux-menuconfig # act like kernel's make menuconfig > +$ make linux-rebuild > +``` > + > +## Usage of the sdcard.img > + > +The `sdcard.img` is in `output/images`, you can directly write the image into your TF card, which could automatically boot and load the USB as rndis function by default. Wow, I had to consult wikipedia to find out about a TF card! I never heard a microSD card being called a TF card before. I think the same goes for most people, so please use microSD or uSD or just SD card. > + > +## Usage of the rootfs > + > +Find the `rootfs.tar.xz` in `output/images`, mount the `mmcblk0p2` part to a directory. Then: > + > +```shell > +$ cd path/to/mmcblk0p2 > +$ rm -rf * > +$ sync > + > +$ tar -xf rootfs.tar.xz --directory=path/to/mmcblk0p2 I didn't understand this... Is the idea to avoid rewriting the entire SD card? Doesn't seem very useful to me, the SD card image is pretty small so it should flash fast, especially with `dd conv=sparse`. > +$ cd path/to/mmcblk0p2 > +$ sync > +``` > diff --git a/board/milkv/duo/genimage.cfg b/board/milkv/duo/genimage.cfg > new file mode 100644 > index 0000000000..ed19e4fd3d > --- /dev/null > +++ b/board/milkv/duo/genimage.cfg > @@ -0,0 +1,30 @@ > +image boot.vfat { > + vfat { > + files = { > + "fip.bin", > + "boot.sd", > + } > + } > + size = 64M 64M? Really? Shouldn't a few MB be sufficient? > +} > + > +image sdcard.img { > + hdimage { > + } > + > + partition boot { > + partition-type = 0xC > + bootable = "true" > + image = "boot.vfat" > + } > + > + partition rootfs { > + partition-type = 0x83 > + image = "rootfs.ext4" > + } > + > + partition empty { > + partition-type = 0x82 > + size = 256M > + } Drop this partition, there is no need for it (see also below). > +} > \ No newline at end of file Make sure there's a newline at the end of each file. [snip] > diff --git a/board/milkv/duo/overlay/etc/init.d/S10hwinit b/board/milkv/duo/overlay/etc/init.d/S10hwinit > new file mode 100755 > index 0000000000..ef7dc0f8c4 > --- /dev/null > +++ b/board/milkv/duo/overlay/etc/init.d/S10hwinit > @@ -0,0 +1,44 @@ > +#!/bin/sh > + > +# > +# Initialize hardware > +# > +export SYSTEMPATH=/opt/milkv-duo/ We don't use /opt in upstream buildroot, /opt is for local stuff. > +CONFIG=/etc/milkv-duo.conf > +source ${CONFIG} > + > +case "$1" in > + start) > + > + if [ -f $SYSTEMPATH/duo-init.sh ]; then > + . $SYSTEMPATH/duo-init.sh & > + fi > + > + if [ ! -z $ETH0_MAC_ADDR ]; then > + ifconfig eth0 hw ether ${ETH0_MAC_ADDR} Nice trick to save the random MAC address that is generated on first boot, but we don't include such things in the minimal defconfig, especially if it's not board specific. > + fi > + > + if [ "$DUO_USB_FUNC" == "RNDIS" ]; then > + if [ -f $SYSTEMPATH/usb-rndis.sh ]; then > + . $SYSTEMPATH/usb-rndis.sh & > + fi > + elif [ "$DUO_USB_FUNC" == "HOST" ]; then > + if [ -f $SYSTEMPATH/usb-host.sh ]; then > + . $SYSTEMPATH/usb-host.sh & > + fi > + elif [ "$DUO_USB_FUNC" == "MASS-STORAGE" ]; then > + if [ -f $SYSTEMPATH/usb-mass-storage.sh ]; then > + . $SYSTEMPATH/usb-mass-storage.sh & > + fi This USB stuff is also not board-specific, is just generic USB gadget configuration. It might make sense to create a Buildroot package with those scripts though. > + fi > + ;; > + stop) > + ;; > + restart|reload) > + ;; > + *) > + echo "Usage: $0 {start|stop|restart}" > + exit 1 > +esac > + > +exit $? > diff --git a/board/milkv/duo/overlay/etc/init.d/S15swap b/board/milkv/duo/overlay/etc/init.d/S15swap > new file mode 100755 > index 0000000000..6b5ef76326 > --- /dev/null > +++ b/board/milkv/duo/overlay/etc/init.d/S15swap > @@ -0,0 +1,25 @@ > +#!/bin/sh > + > +# > +# Activate swap, depending on option > +# > +CONFIG=/etc/milkv-duo.conf > +source ${CONFIG} > + > +case "$1" in > + start) > + > + if [ "$ENABLE_SWAP" == "1" ]; then > + swapon /dev/mmcblk0p3 I would be very surprised if swap is actually useful, and for sure we don't want it in a defconfig. Also, it's probably better to manage swap in fstab. > + fi > + ;; > + stop) > + ;; > + restart|reload) > + ;; > + *) > + echo "Usage: $0 {start|stop|restart}" > + exit 1 > +esac > + > +exit $? > diff --git a/board/milkv/duo/overlay/etc/init.d/S99user b/board/milkv/duo/overlay/etc/init.d/S99user > new file mode 100755 > index 0000000000..3abd073738 > --- /dev/null > +++ b/board/milkv/duo/overlay/etc/init.d/S99user > @@ -0,0 +1,27 @@ > +#!/bin/sh > +${CVI_SHOPTS} > + > +# > +# Start firmware > +# > +export SYSTEMPATH=/opt/milkv-duo/ > + > +case "$1" in > + start) > + > + echo "Starting app..." > + > + if [ -f $SYSTEMPATH/blink.sh ]; then > + . $SYSTEMPATH/blink.sh & This "app" also doesn't seem very board-specific (except for the specific GPIO that corresponds to a LED). Also, it's a completely useless app. Blinking a LED in Linux is doen with the LED Timer Trigger. [snip] > diff --git a/board/milkv/duo/overlay/opt/milkv-duo/blink.sh b/board/milkv/duo/overlay/opt/milkv-duo/blink.sh > new file mode 100755 > index 0000000000..3d14977cc9 > --- /dev/null > +++ b/board/milkv/duo/overlay/opt/milkv-duo/blink.sh > @@ -0,0 +1,25 @@ > +#!/bin/sh > + > +CONFIG=/etc/milkv-duo.conf > +source ${CONFIG} > + > +if test -d ${LED_PATH}; then > + echo "GPIO${LED_GPIO} already exported" > +else > + ${DUO_WRITE} ${LED_GPIO} >/sys/class/gpio/export If the GPIO is wired to a LED, it should be specified as such in the device tree (and bound to the leds_gpio driver). > +fi > + > +${DUO_WRITE} out >${LED_PATH}/direction > + > +while true; do > + source ${CONFIG} > + if [ "$ENABLE_BLINK" == "1" ]; then > + ${DUO_WRITE} 0 >${LED_PATH}/value > + sleep 0.5 > + ${DUO_WRITE} 1 >${LED_PATH}/value > + sleep 0.5 > + else > + ${DUO_WRITE} 0 >${LED_PATH}/value > + sleep 2 > + fi > +done > diff --git a/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh b/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh > new file mode 100755 > index 0000000000..298e43bc8d > --- /dev/null > +++ b/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh > @@ -0,0 +1,8 @@ > +#!/bin/sh > + > +# Set Pin-29(PWR_SEQ2) to GPIO > +devmem 0x0502707c 32 0x111 > +devmem 0x03001068 32 0x3 > + > +# Set Pin-19(SD0_PWR_EN) to GPIO > +devmem 0x0300101c 32 0x3 I have no idea what this is supposed to do, but shouldn't it have a proper Linux driver instead devmem? [snip] > diff --git a/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh b/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh > new file mode 100755 > index 0000000000..8a06949e3f > --- /dev/null > +++ b/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh This script _is_ board specific, so it does make sense to have it in the overlay! However, it should be in /usr/bin, not in /opt. This script should definitely be mentioned in the readme. Normally, host/device mode is selected automatically with the ID pin, so if this is not the case on this board it's probably better to explain it. > @@ -0,0 +1,63 @@ > +CONFIG=/etc/milkv-duo.conf > +source ${CONFIG} > + > +hub_on() { > + echo "turn on usb hub" > + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBPORT_EN ]; then > + ${DUO_WRITE} $GPIO_HUBPORT_EN >/sys/class/gpio/export > + fi > + > + if [ ! -d $SYS_GPIO/gpio$GPIO_ROLESEL ]; then > + ${DUO_WRITE} $GPIO_ROLESEL >/sys/class/gpio/export > + fi > + > + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBRST ]; then > + ${DUO_WRITE} $GPIO_HUBRST >/sys/class/gpio/export > + fi > + > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/direction > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_ROLESEL/direction > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBRST/direction Please don't use the deprecated sysfs access for GPIOs, instead use gpioset and friends from libgpiod. > + > + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/value > + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_ROLESEL/value > + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_HUBRST/value > +} > + > +hub_off() { > + echo "turn off usb hub" > + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBPORT_EN ]; then > + ${DUO_WRITE} $GPIO_HUBPORT_EN >/sys/class/gpio/export > + fi > + > + if [ ! -d $SYS_GPIO/gpio$GPIO_ROLESEL ]; then > + ${DUO_WRITE} $GPIO_ROLESEL >/sys/class/gpio/export > + fi > + > + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBRST ]; then > + ${DUO_WRITE} $GPIO_HUBRST >/sys/class/gpio/export > + fi > + > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/direction > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_ROLESEL/direction > + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBRST/direction > + > + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/value > + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_ROLESEL/value > + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_HUBRST/value > +} > + > +case "$1" in > + host) > + insmod /mnt/system/ko/dwc2.ko This is weird... Modules should be in /lib/modules. Also, the module should have been loaded automatically. > + ${DUO_WRITE} host > /proc/cviusb/otg_role > + ;; > + device) > + ${DUO_WRITE} device > /proc/cviusb/otg_role > + ;; > + *) > + echo "Usage: $0 host" > + echo "Usage: $0 device" > + exit 1 > +esac > +exit $? [snip] > diff --git a/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch b/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch > new file mode 100644 > index 0000000000..cd784948be > --- /dev/null > +++ b/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch > @@ -0,0 +1,44 @@ > +From 2e8190643f1abdfcf6c8b5981b6d6863dea78a08 Mon Sep 17 00:00:00 2001 > +From: "sam.xiang" <sam.xiang@sophgo.com> > +Date: Tue, 21 Feb 2023 20:43:18 +0800 > +Subject: [PATCH 1/3] [opensbi] porting: adapt to CV180X / CV181X There should not be a 1/3 in the subject line. Run 'make check-package' to detect such issues. > + > +Change-Id: Id7b3c64d203eb2c9af6c66f195bf0d8a05f0164c > +Signed-off-by: Hanyuan Zhao <hanyuan-z@qq.com> There's also an Upstream: link missing. > +--- > + lib/utils/serial/fdt_serial_uart8250.c | 1 + > + platform/generic/config.mk | 3 ++- > + 2 files changed, 3 insertions(+), 1 deletion(-) > + mode change 100644 => 100755 lib/utils/serial/fdt_serial_uart8250.c [snip] > diff --git a/board/milkv/duo/post-image.sh b/board/milkv/duo/post-image.sh > new file mode 100755 > index 0000000000..bebd76a4ad > --- /dev/null > +++ b/board/milkv/duo/post-image.sh > @@ -0,0 +1,52 @@ > +#!/bin/sh > + > +########################################################### > +# File: post-image.sh > +# Author: GP Orcullo <kinsamanka@gmail.com> > +# Description: this sh will make the fip.bin and boot.sd, > +# then pack everything to an image file. > +########################################################### > +if grep -Eq "^BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL=y$" ${BR2_CONFIG}; then > + if grep -Eq "^BR2_PACKAGE_MILKV_DUO_SMALLCORE_FREERTOS=y$" ${BR2_CONFIG}; then Both conditions are set in both defconfigs, so there probably isn't a need to check for them. > + ${BINARIES_DIR}/fiptool.py genfip ${BINARIES_DIR}/fip.bin \ Please use tab for indentation in shell scripts. It's not yet done consistently, but we're trying to go in that direction. > + --MONITOR_RUNADDR=0x80000000 \ > + --CHIP_CONF=${BINARIES_DIR}/chip_conf.bin \ > + --NOR_INFO=FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF \ > + --NAND_INFO=00000000 \ > + --BL2=${BINARIES_DIR}/bl2.bin \ > + --BLCP_IMG_RUNADDR=0x05200200 \ > + --BLCP_PARAM_LOADADDR=0 \ > + --DDR_PARAM=${BINARIES_DIR}/ddr_param.bin \ > + --MONITOR=${BINARIES_DIR}/fw_dynamic.bin \ > + --LOADER_2ND=${BINARIES_DIR}/u-boot.bin \ > + --BLCP=${BINARIES_DIR}/empty.bin \ > + --BLCP_2ND=${BINARIES_DIR}/cvirtos.bin \ Actually, I think it's better to check for the existence of this file to decide whether or not to include the smallcore image. > + --BLCP_2ND_RUNADDR=0x83f40000 \ > + > ${BINARIES_DIR}/fip.log 2>&1 > + echo "[Duo Post-Image fiptool.py] FreeRTOS integrated" > + else > + ${BINARIES_DIR}/fiptool.py genfip ${BINARIES_DIR}/fip.bin \ > + --MONITOR_RUNADDR=0x80000000 \ > + --CHIP_CONF=${BINARIES_DIR}/chip_conf.bin \ > + --NOR_INFO=FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF \ > + --NAND_INFO=00000000 \ > + --BL2=${BINARIES_DIR}/bl2.bin \ > + --BLCP_IMG_RUNADDR=0x05200200 \ > + --BLCP_PARAM_LOADADDR=0 \ > + --DDR_PARAM=${BINARIES_DIR}/ddr_param.bin \ > + --MONITOR=${BINARIES_DIR}/fw_dynamic.bin \ > + --LOADER_2ND=${BINARIES_DIR}/u-boot.bin \ > + > ${BINARIES_DIR}/fip.log 2>&1 > + echo "[Duo Post-Image fiptool.py] No FreeRTOS integrated" > + fi > + > + cp ${BINARIES_DIR}/u-boot.dtb ${BINARIES_DIR}/cv1800b_milkv_duo_sd.dtb > + lzma -fk ${BINARIES_DIR}/Image > + mkimage -f ${BINARIES_DIR}/multi.its ${BINARIES_DIR}/boot.sd > + echo "[Duo Post-Image] > boot.sd generated!" > + support/scripts/genimage.sh -c $(pwd)/board/milkv/duo/genimage.cfg > + gzip -fk ${BINARIES_DIR}/sdcard.img > + echo "[Duo Post-Image] > sdcard.img generated!" > +else > + echo "[Duo Post-Image] Not requested to generate the boot files and sdcard.img" > +fi > \ No newline at end of file > diff --git a/configs/milkv_duo_musl_riscv64_64mb_defconfig b/configs/milkv_duo_musl_riscv64_64mb_defconfig > new file mode 100644 > index 0000000000..e97677d891 > --- /dev/null > +++ b/configs/milkv_duo_musl_riscv64_64mb_defconfig > @@ -0,0 +1,94 @@ > +BR2_riscv=y > +BR2_riscv_g=y > +BR2_RISCV_ISA_RVC=y Please put comments in the defconfig, like in e.g. configs/canaan_kd233_defconfig > +BR2_TOOLCHAIN_BUILDROOT_MUSL=y Does it really have to be musl? Our defconfigs normally use uClibc. > +BR2_KERNEL_HEADERS_5_10=y Please use the default BR2_KERNEL_HEADERS_AS_KERNEL. > +BR2_GCC_VERSION_13_X=y Please use the default GCC version; if that is not appropriate, add a comment why. > +BR2_TOOLCHAIN_BUILDROOT_CXX=y > +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" Wow, this definitely needs some explanation! Oh, hang on, it's using that non-standard T-Head CPU, that explains it... And it also explains why you need GCC 13. I'm surprised though that it's sufficient to put it in LDFLAGS, don't you need it for compilation as well? I actually think we probably should have a riscv subarchitecture option for the T-Head, since there will be multiple boards using that SoC (including the BeagleV-Ahead). > +BR2_GLOBAL_PATCH_DIR="board/milkv/duo/patches" > +BR2_TARGET_GENERIC_HOSTNAME="milkv-duo" > +BR2_TARGET_GENERIC_ISSUE="Welcome to Milk-V" > +BR2_TARGET_GENERIC_ROOT_PASSWD="milkv" > +BR2_ROOTFS_OVERLAY="board/milkv/duo/overlay" > +BR2_ROOTFS_POST_IMAGE_SCRIPT="board/milkv/duo/post-image.sh" > +BR2_LINUX_KERNEL=y > +BR2_LINUX_KERNEL_CUSTOM_GIT=y > +BR2_LINUX_KERNEL_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-linux.git" > +BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION="duo-linux-5.10.4" > +BR2_LINUX_KERNEL_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" > +BR2_LINUX_KERNEL_LZMA=y > +BR2_PACKAGE_BZIP2=y > +BR2_PACKAGE_GZIP=y > +BR2_PACKAGE_LZIP=y > +BR2_PACKAGE_UNZIP=y > +BR2_PACKAGE_ZIP=y > +BR2_PACKAGE_GDB=y > +BR2_PACKAGE_GDB_DEBUGGER=y > +BR2_PACKAGE_SPIDEV_TEST=y > +BR2_PACKAGE_STRACE=y > +BR2_PACKAGE_BINUTILS=y > +BR2_PACKAGE_GREP=y > +BR2_PACKAGE_MAKE=y > +BR2_PACKAGE_PATCH=y > +BR2_PACKAGE_TREE=y > +BR2_PACKAGE_DOSFSTOOLS=y > +BR2_PACKAGE_E2FSPROGS=y > +BR2_PACKAGE_E2FSPROGS_RESIZE2FS=y None of the packages above should be needed for a minimal bootable system. > +BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL=y > +BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL_64MB=y > +BR2_PACKAGE_MILKV_DUO_SMALLCORE_FREERTOS=y These are only added in patch 2/2, so the two patches should be swapped - the other one should come first. (But the other one should actually be split into three as well, one per package.) Although this firmware is probably not strictly needed for booting, I do think it's a good idea to include it, so you have a complete system. The readme.txt should point to some documentation of that FreeRTOS image however. > +BR2_PACKAGE_MILKV_DUO_PINMUX=y > +BR2_PACKAGE_EVTEST=y > +BR2_PACKAGE_PYTHON3=y > +BR2_PACKAGE_PYTHON_EVDEV=y > +BR2_PACKAGE_PYTHON_LXML=y > +BR2_PACKAGE_PYTHON_MODBUS_TK=y > +BR2_PACKAGE_PYTHON_PILLOW=y > +BR2_PACKAGE_PYTHON_PIP=y > +BR2_PACKAGE_PYTHON_SMBUS_CFFI=y > +BR2_PACKAGE_PYTHON_SPIDEV=y > +BR2_PACKAGE_FREETYPE=y > +BR2_PACKAGE_JSON_C=y > +BR2_PACKAGE_DHCPCD=y > +BR2_PACKAGE_DNSMASQ=y > +BR2_PACKAGE_DROPBEAR=y > +BR2_PACKAGE_LRZSZ=y > +BR2_PACKAGE_NTP=y > +BR2_PACKAGE_NTP_NTPDATE=y > +BR2_PACKAGE_NTP_NTPTIME=y > +BR2_PACKAGE_WPA_SUPPLICANT=y > +BR2_PACKAGE_WPA_SUPPLICANT_CLI=y > +BR2_PACKAGE_WPA_SUPPLICANT_WPA_CLIENT_SO=y > +BR2_PACKAGE_NEOFETCH=y > +BR2_PACKAGE_TMUX=y > +BR2_PACKAGE_WHICH=y > +BR2_PACKAGE_HTOP=y > +BR2_PACKAGE_KMOD=y > +BR2_PACKAGE_NANO=y > +BR2_TARGET_ROOTFS_EXT2=y > +BR2_TARGET_ROOTFS_EXT2_4=y > +BR2_TARGET_ROOTFS_EXT2_SIZE="200M" If you don't include all those packages, the default 60M should be sufficient. > +BR2_TARGET_ROOTFS_TAR_XZ=y Normally there should not be a reason to have the tarball at all. > +BR2_TARGET_OPENSBI=y > +BR2_TARGET_OPENSBI_CUSTOM_GIT=y > +BR2_TARGET_OPENSBI_CUSTOM_REPO_URL="https://github.com/T-head-Semi/opensbi.git" > +BR2_TARGET_OPENSBI_CUSTOM_REPO_VERSION="4e77060e0512ad981eee55d5a2501f6d88a41fd9" > +BR2_TARGET_OPENSBI_PLAT="generic" > +BR2_TARGET_OPENSBI_UBOOT_PAYLOAD=y > +BR2_TARGET_OPENSBI_FW_FDT_PATH=y > +BR2_TARGET_UBOOT=y > +BR2_TARGET_UBOOT_BUILD_SYSTEM_KCONFIG=y > +BR2_TARGET_UBOOT_CUSTOM_GIT=y > +BR2_TARGET_UBOOT_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-u-boot.git" > +BR2_TARGET_UBOOT_CUSTOM_REPO_VERSION="v2021.10_64mb" > +BR2_TARGET_UBOOT_BOARD_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" > +BR2_TARGET_UBOOT_NEEDS_DTC=y > +BR2_TARGET_UBOOT_NEEDS_OPENSBI=y > +BR2_PACKAGE_HOST_CMAKE=y There normally shouldn't be any reason to set this option. > +BR2_PACKAGE_HOST_DOSFSTOOLS=y > +BR2_PACKAGE_HOST_GENIMAGE=y > +BR2_PACKAGE_HOST_LZMA_ALONE=y > +BR2_PACKAGE_HOST_MTOOLS=y > +BR2_PACKAGE_HOST_UBOOT_TOOLS=y > +BR2_PACKAGE_HOST_UBOOT_TOOLS_FIT_SUPPORT=y > diff --git a/configs/milkv_duo_musl_riscv64_defconfig b/configs/milkv_duo_musl_riscv64_defconfig > new file mode 100644 > index 0000000000..bd79f138f6 > --- /dev/null > +++ b/configs/milkv_duo_musl_riscv64_defconfig I'm not sure if we want both defconfigs. I think it's sufficient to put in the readme.txt that if you want to use ION/camera, that U-Boot should switch to the other defconfig. Regards, Arnout [snip]
Hello Arnout, Thanks for your patience and careful review! Let me explain something you requested. > Buildroot defconfigs should be _minimal_, i.e. just enough to boot the device and perhaps get a (wired) network connection. Looking at all those files, clearly these defconfigs are _not_ minimal. I think pretty much everything in the overlay can be removed. > Nice trick to save the random MAC address that is generated on first boot, but we don't include such things in the minimal defconfig, especially if it's not board specific. > This USB stuff is also not board-specific, is just generic USB gadget configuration. It might make sense to create a Buildroot package with those scripts though. I think the digest of the changes is requesting us to remove the unnecessary things such as the packages we use in the defconfig and files in the overlay. I am sorry for my previous misunderstanding of your minimal philosophy. The reason I did them is that I want to let the system we build could be nearly the same as the official image. And the features are mainly about the USB RNDIS or other multiplex functions. Now I plan to remove the whole overlay folder and move our files which give the features into a new package (e.g. named BR2_PACKAGE_MILKV_DUO_USB_TOOLS). Could you please give me a review on that? > I didn't understand this... Is the idea to avoid rewriting the entire SD card? Doesn't seem very useful to me, the SD card image is pretty small so it should flash fast, especially with `dd conv=sparse`. OK. I will remove it. > 64M? Really? Shouldn't a few MB be sufficient? Yes, certainly sufficient. I will change. > Drop this partition, there is no need for it (see also below). OK. I will refactor the whole file. > There's also an Upstream: link missing. Is the Upstream a must? I have found many patches in buildroot do not have this. > Both conditions are set in both defconfigs, so there probably isn't a need to check for them. I think we still need to check the conditions. Because the user might change them. If not checked, the error will occur and stop the whole make process. > Please use tab for indentation in shell scripts. It's not yet done consistently, but we're trying to go in that direction. OK. I am always happy to follow your format. > Actually, I think it's better to check for the existence of this file to decide whether or not to include the smallcore image. To me, maybe we can ignore this? If the user request the smallcore, and actually the smallcore image does not exist, then the fiptool.py will emit an error. If we just include the smallcore as long as the image exists, I think it might not be aligned with the user’s expect. > Does it really have to be musl? Our defconfigs normally use uClibc. I also think there’s no need for musl. However, the official sdk is using musl. So I have to use musl to keep the consistency. >> +BR2_GCC_VERSION_13_X=y > > Please use the default GCC version; if that is not appropriate, add a comment why. > >> +BR2_TOOLCHAIN_BUILDROOT_CXX=y >> +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" > > Wow, this definitely needs some explanation! > > Oh, hang on, it's using that non-standard T-Head CPU, that explains it... And it also explains why you need GCC 13. I'm surprised though that it's sufficient to put it in LDFLAGS, don't you need it for compilation as well? > > I actually think we probably should have a riscv subarchitecture option for the T-Head, since there will be multiple boards using that SoC (including the BeagleV-Ahead). Yes. You have already found our reasons of using gcc-13. However, here I do make a mistake. I plan to copy the -march into the BR2_TARGET_OPTIMIZATION, which will then be passed to TARGET_CFLAGS and TARGET_CXXFLAGS. I will comment these on the defconfig. > Although this firmware is probably not strictly needed for booting, I do think it's a good idea to include it, so you have a complete system. The readme.txt should point to some documentation of that FreeRTOS image however. Nonono, this firmware is strictly needed for booting! Without this, we could not make the whole SD card image! The boot steps of the chip cv1800b is complex. The firmware produces fip.bin, which stores the fsbl, opensb,freertosi and the u-boot, and boot.sd, which is the kernel in actual. We include this firmware, to build the first stage boot loader, while involving the other binaries to give the two files. If the user is not saying Y to the firmware packages, the buildroot would only produce the rootfs.tar. The whole SD image will be produced if and only if the firmware packages are selected. > If you don't include all those packages, the default 60M should be sufficient. OK. I will strictly imitate other board’s config. Make sure most things are default. > There normally shouldn't be any reason to set this option. Does this option mean the HOST CMAKE? > I'm not sure if we want both defconfigs. I think it's sufficient to put in the readme.txt that if you want to use ION/camera, that U-Boot should switch to the other defconfig. We add two defconfigs, imitating the acmesystems_acqua_a5_256mb_defconfig. And if the user want to use ION/camera, not only the u-boot but also the firmware need to be configured. So to the convenience, we add two configs. Thanks again for your nice review and great patience! Hope this email explains. Best regards, Hanyuan
On 22/11/2023 04:47, hanyuan wrote: > Hello Arnout, > > Thanks for your patience and careful review! Let me explain something you requested. > >> Buildroot defconfigs should be _minimal_, i.e. just enough to boot the device and perhaps get a (wired) network connection. Looking at all those files, clearly these defconfigs are _not_ minimal. I think pretty much everything in the overlay can be removed. >> Nice trick to save the random MAC address that is generated on first boot, but we don't include such things in the minimal defconfig, especially if it's not board specific. >> This USB stuff is also not board-specific, is just generic USB gadget configuration. It might make sense to create a Buildroot package with those scripts though. > > I think the digest of the changes is requesting us to remove the unnecessary things such as the packages we use in the defconfig and files in the overlay. I am sorry for my previous misunderstanding of your minimal philosophy. The reason I did them is that I want to let the system we build could be nearly the same as the official image. And the features are mainly about the USB RNDIS or other multiplex functions. > > Now I plan to remove the whole overlay folder and move our files which give the features into a new package (e.g. named BR2_PACKAGE_MILKV_DUO_USB_TOOLS). Could you please give me a review on that? As I wrote somewhere in my review (but you didn't include it below), I think the script that toggles the USB hub host/gadget mode is still relevant, because that is something board specific. The rest of the overlay indeed can be removed, I think. About BR2_PACKAGE_MILKV_DUO_USB_TOOLS: as far as I could see, those scripts are not at all board-specific, they simply configure the generic gadget functions. So the package name doesn't need to have milkv-duo in it. Perhaps initscripts-usb-gadget ? I'm not entirely sure if it is appropriate to be part of Buildroot anyway, it would be more appropriate in its own repository. But then again, if we have it in Buildroot itself, we can maintain it independently. Note that this initscripts-usb-gadget package will probably take a few more iterations before it gets accepted as well. So probably best to submit it separately from the reset of this series. Here are a few hints already that can help to make sure the initial submission is as good as possible. - Configuration should be in /etc/default. - Helper scripts should be in /usr/bin. - Use <PKG>_LINUX_CONFIG_FIXUPS to enable the required configs in the kernel. - The package should probably depend on !BR2_INIT_SYSTEMD. - Make sure to follow the init script skeleton (but I think you already did that). > >> I didn't understand this... Is the idea to avoid rewriting the entire SD card? Doesn't seem very useful to me, the SD card image is pretty small so it should flash fast, especially with `dd conv=sparse`. > > OK. I will remove it. > >> 64M? Really? Shouldn't a few MB be sufficient? > > Yes, certainly sufficient. I will change. > >> Drop this partition, there is no need for it (see also below). > > OK. I will refactor the whole file. > >> There's also an Upstream: link missing. > > Is the Upstream a must? I have found many patches in buildroot do not have this. The Upstream: tag is a must, check-package complains if it's not there. We require it exactly to motivate people to send their patches upstream. We want it upstream for two reasons: we don't want to maintain patches, and often upstream is better positioned to review the patches. The latter certainly applies to the OpenSBI patches. >> Both conditions are set in both defconfigs, so there probably isn't a need to check for them. > > I think we still need to check the conditions. Because the user might change them. If not checked, the error will occur and stop the whole make process. Well, if the user changes them, they probably shouldn't be using this post-image script either, right? > >> Please use tab for indentation in shell scripts. It's not yet done consistently, but we're trying to go in that direction. > > OK. I am always happy to follow your format. > >> Actually, I think it's better to check for the existence of this file to decide whether or not to include the smallcore image. > > To me, maybe we can ignore this? If the user request the smallcore, and actually the smallcore image does not exist, then the fiptool.py will emit an error. If we just include the smallcore as long as the image exists, I think it might not be aligned with the user’s expect. The point is that the user may have its own alternative firmware for the smallcore image and just install that in BINARIES_DIR. That said, the post-{build,image} scripts of defconfigs just need to work for the defconfig, there's no need to make them too generic for custom use cases. So just checking the config options is fine. > >> Does it really have to be musl? Our defconfigs normally use uClibc. > > I also think there’s no need for musl. However, the official sdk is using musl. So I have to use musl to keep the consistency. Yeah, our defconfigs use uClibc unless there's a good reason not to. > > >>> +BR2_GCC_VERSION_13_X=y >> >> Please use the default GCC version; if that is not appropriate, add a comment why. >> >>> +BR2_TOOLCHAIN_BUILDROOT_CXX=y >>> +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" >> >> Wow, this definitely needs some explanation! >> >> Oh, hang on, it's using that non-standard T-Head CPU, that explains it... And it also explains why you need GCC 13. I'm surprised though that it's sufficient to put it in LDFLAGS, don't you need it for compilation as well? >> >> I actually think we probably should have a riscv subarchitecture option for the T-Head, since there will be multiple boards using that SoC (including the BeagleV-Ahead). > > Yes. You have already found our reasons of using gcc-13. However, here I do make a mistake. I plan to copy the -march into the BR2_TARGET_OPTIMIZATION, which will then be passed to TARGET_CFLAGS and TARGET_CXXFLAGS. I will comment these on the defconfig. Indeed, I thought it should be in BR2_TARGET_OPTIMIZATION. We have that for example in configs/snps_archs38_hsdk_defconfig. However, as I mentioned: it's probably better to create a RISC-V subarchitecture for T-Head. Something like: choice prompt "Target Architecture Variant" default BR2_riscv_g config BR2_riscv_g bool "General purpose (G)" [...] config BR2_riscv_thead bool "T-Head <whatever it is>" select BR2_ARCH_NEEDS_GCC_AT_LEAST_13 select [...] help T-Head <whatever it is> SoCs use an instruction set definition for the <...> extension that was not yet accepted at the time the chip was designed. Therefore, it needs a special compiler variant. config BR2_riscv_custom bool "Custom architecture" select BR2_RISCV_ISA_RVI endchoice and in arch/arch.mk.riscv: ifeq ($(BR2_riscv_thead),y) GCC_TARGET_ARCH = rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync endif The above change should obviously be a separate patch as well. >> Although this firmware is probably not strictly needed for booting, I do think it's a good idea to include it, so you have a complete system. The readme.txt should point to some documentation of that FreeRTOS image however. > > Nonono, this firmware is strictly needed for booting! Without this, we could not make the whole SD card image! The boot steps of the chip cv1800b is complex. The firmware produces fip.bin, which stores the fsbl, opensb,freertosi and the u-boot, and boot.sd, which is the kernel in actual. We include this firmware, to build the first stage boot loader, while involving the other binaries to give the two files. If the user is not saying Y to the firmware packages, the buildroot would only produce the rootfs.tar. The whole SD image will be produced if and only if the firmware packages are selected. OK. > >> If you don't include all those packages, the default 60M should be sufficient. > > OK. I will strictly imitate other board’s config. Make sure most things are default. > >> There normally shouldn't be any reason to set this option. > > Does this option mean the HOST CMAKE? Yes, but host-cmake is built automatically when it's needed, so you normally don't need to enable the config option. It's only needed if you want to build an SDK which includes cmake, which is not the case in a defconfig. >> I'm not sure if we want both defconfigs. I think it's sufficient to put in the readme.txt that if you want to use ION/camera, that U-Boot should switch to the other defconfig. > > We add two defconfigs, imitating the acmesystems_acqua_a5_256mb_defconfig. And if the user want to use ION/camera, not only the u-boot but also the firmware need to be configured. So to the convenience, we add two configs. Ah yes that's a good reason. Regards, Arnout > > > Thanks again for your nice review and great patience! Hope this email explains. > > Best regards, > Hanyuan >
Hello Arnout, > 2023年11月23日 04:19，Arnout Vandecappelle <arnout@mind.be> 写道： > > > On 22/11/2023 04:47, hanyuan wrote: >> Hello Arnout, >> Thanks for your patience and careful review! Let me explain something you requested. >>> Buildroot defconfigs should be _minimal_, i.e. just enough to boot the device and perhaps get a (wired) network connection. Looking at all those files, clearly these defconfigs are _not_ minimal. I think pretty much everything in the overlay can be removed. >>> Nice trick to save the random MAC address that is generated on first boot, but we don't include such things in the minimal defconfig, especially if it's not board specific. >>> This USB stuff is also not board-specific, is just generic USB gadget configuration. It might make sense to create a Buildroot package with those scripts though. >> I think the digest of the changes is requesting us to remove the unnecessary things such as the packages we use in the defconfig and files in the overlay. I am sorry for my previous misunderstanding of your minimal philosophy. The reason I did them is that I want to let the system we build could be nearly the same as the official image. And the features are mainly about the USB RNDIS or other multiplex functions. >> Now I plan to remove the whole overlay folder and move our files which give the features into a new package (e.g. named BR2_PACKAGE_MILKV_DUO_USB_TOOLS). Could you please give me a review on that? > > As I wrote somewhere in my review (but you didn't include it below), I think the script that toggles the USB hub host/gadget mode is still relevant, because that is something board specific. The rest of the overlay indeed can be removed, I think. > > About BR2_PACKAGE_MILKV_DUO_USB_TOOLS: as far as I could see, those scripts are not at all board-specific, they simply configure the generic gadget functions. So the package name doesn't need to have milkv-duo in it. Perhaps > initscripts-usb-gadget ? > > I'm not entirely sure if it is appropriate to be part of Buildroot anyway, it would be more appropriate in its own repository. But then again, if we have it in Buildroot itself, we can maintain it independently. In fact, all the scripts concerning USB in the previous overlay are all board specific, to some extent. Please take a look at the following lines: ${DUO_WRITE} host > /proc/cviusb/otg_role CVI_DIR=/tmp/usb CVI_GADGET=$CVI_DIR/usb_gadget/cvitek # Set the USB configuration mkdir $CVI_GADGET/configs/c.1 mkdir $CVI_GADGET/configs/c.1/strings/0x409 ${DUO_WRITE} "config1">$CVI_GADGET/configs/c.1/strings/0x409/configuration We could have two strategies. Firstly, if we think they are relevant and want to keep something in the overlay, let’s compress the files into one single big script. And do I have other alternatives of /usr/bin? I don’t want to move the USB initial scripts into the user’s PATH. Secondly, we import the package BR2_PACKAGE_MILKV_DUO_USB_TOOLS. The USB functions are not must for a minimal system to run. However, if we use these shell scripts, by default the USB rndis will be up, allowing us to login through ssh by a usb cable. If not, we have to use the serial port. What do you think, overlay or package? > > Note that this initscripts-usb-gadget package will probably take a few more iterations before it gets accepted as well. So probably best to submit it separately from the reset of this series. Here are a few hints already that can help to make sure the initial submission is as good as possible. > > - Configuration should be in /etc/default. > - Helper scripts should be in /usr/bin. > - Use <PKG>_LINUX_CONFIG_FIXUPS to enable the required configs in the kernel. > - The package should probably depend on !BR2_INIT_SYSTEMD. > - Make sure to follow the init script skeleton (but I think you already did that). If we do use the package to provide the USB function, I highly hope that it could be named milkv-duo, which will be convenient because we could directly move our code into it. And we will not have the need to make it general. > >>> I didn't understand this... Is the idea to avoid rewriting the entire SD card? Doesn't seem very useful to me, the SD card image is pretty small so it should flash fast, especially with `dd conv=sparse`. >> OK. I will remove it. >>> 64M? Really? Shouldn't a few MB be sufficient? >> Yes, certainly sufficient. I will change. >>> Drop this partition, there is no need for it (see also below). >> OK. I will refactor the whole file. >>> There's also an Upstream: link missing. >> Is the Upstream a must? I have found many patches in buildroot do not have this. > > The Upstream: tag is a must, check-package complains if it's not there. We require it exactly to motivate people to send their patches upstream. We want it upstream for two reasons: we don't want to maintain patches, and often upstream is better positioned to review the patches. The latter certainly applies to the OpenSBI patches. Could I use a customed repo to avoid use patches? The upstream is still a big work in progress. I could not give an upstream tag currently. We hope to let our users have a taste of buildroot quickly. > > >>> Both conditions are set in both defconfigs, so there probably isn't a need to check for them. >> I think we still need to check the conditions. Because the user might change them. If not checked, the error will occur and stop the whole make process. > > Well, if the user changes them, they probably shouldn't be using this post-image script either, right? Oh yes, you are right. Thanks for pointing it out! > > >>> Please use tab for indentation in shell scripts. It's not yet done consistently, but we're trying to go in that direction. >> OK. I am always happy to follow your format. >>> Actually, I think it's better to check for the existence of this file to decide whether or not to include the smallcore image. >> To me, maybe we can ignore this? If the user request the smallcore, and actually the smallcore image does not exist, then the fiptool.py will emit an error. If we just include the smallcore as long as the image exists, I think it might not be aligned with the user’s expect. > > The point is that the user may have its own alternative firmware for the smallcore image and just install that in BINARIES_DIR. > > That said, the post-{build,image} scripts of defconfigs just need to work for the defconfig, there's no need to make them too generic for custom use cases. So just checking the config options is fine. Got that. Thanks! > > >>> Does it really have to be musl? Our defconfigs normally use uClibc. >> I also think there’s no need for musl. However, the official sdk is using musl. So I have to use musl to keep the consistency. > > Yeah, our defconfigs use uClibc unless there's a good reason not to. > >>>> +BR2_GCC_VERSION_13_X=y >>> >>> Please use the default GCC version; if that is not appropriate, add a comment why. >>> >>>> +BR2_TOOLCHAIN_BUILDROOT_CXX=y >>>> +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" >>> >>> Wow, this definitely needs some explanation! >>> >>> Oh, hang on, it's using that non-standard T-Head CPU, that explains it... And it also explains why you need GCC 13. I'm surprised though that it's sufficient to put it in LDFLAGS, don't you need it for compilation as well? >>> >>> I actually think we probably should have a riscv subarchitecture option for the T-Head, since there will be multiple boards using that SoC (including the BeagleV-Ahead). >> Yes. You have already found our reasons of using gcc-13. However, here I do make a mistake. I plan to copy the -march into the BR2_TARGET_OPTIMIZATION, which will then be passed to TARGET_CFLAGS and TARGET_CXXFLAGS. I will comment these on the defconfig. > > Indeed, I thought it should be in BR2_TARGET_OPTIMIZATION. We have that for example in configs/snps_archs38_hsdk_defconfig. > > However, as I mentioned: it's probably better to create a RISC-V subarchitecture for T-Head. Something like: > > choice > prompt "Target Architecture Variant" > default BR2_riscv_g > > config BR2_riscv_g > bool "General purpose (G)" > [...] > > config BR2_riscv_thead > bool "T-Head <whatever it is>" > select BR2_ARCH_NEEDS_GCC_AT_LEAST_13 > select [...] > help > T-Head <whatever it is> SoCs use an instruction set definition for the > <...> extension that was not yet accepted at the time the chip was > designed. Therefore, it needs a special compiler variant. > > config BR2_riscv_custom > bool "Custom architecture" > select BR2_RISCV_ISA_RVI > > endchoice > > > and in arch/arch.mk.riscv: > > ifeq ($(BR2_riscv_thead),y) > GCC_TARGET_ARCH = rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync > endif > > > The above change should obviously be a separate patch as well. This is totally beyond my control, haha, willing to contribute in the future. > > >>> Although this firmware is probably not strictly needed for booting, I do think it's a good idea to include it, so you have a complete system. The readme.txt should point to some documentation of that FreeRTOS image however. >> Nonono, this firmware is strictly needed for booting! Without this, we could not make the whole SD card image! The boot steps of the chip cv1800b is complex. The firmware produces fip.bin, which stores the fsbl, opensb,freertosi and the u-boot, and boot.sd, which is the kernel in actual. We include this firmware, to build the first stage boot loader, while involving the other binaries to give the two files. If the user is not saying Y to the firmware packages, the buildroot would only produce the rootfs.tar. The whole SD image will be produced if and only if the firmware packages are selected. > > OK. > >>> If you don't include all those packages, the default 60M should be sufficient. >> OK. I will strictly imitate other board’s config. Make sure most things are default. >>> There normally shouldn't be any reason to set this option. >> Does this option mean the HOST CMAKE? > > Yes, but host-cmake is built automatically when it's needed, so you normally don't need to enable the config option. It's only needed if you want to build an SDK which includes cmake, which is not the case in a defconfig. Got it. Sorry for my misunderstanding. Thanks again for your kindly help and careful review! Best regards, Hanyuan
On 23/11/2023 02:22, hanyuan wrote: > Hello Arnout, > >> 2023年11月23日 04:19，Arnout Vandecappelle <arnout@mind.be> 写道： >> >> >> On 22/11/2023 04:47, hanyuan wrote: >>> Hello Arnout, >>> Thanks for your patience and careful review! Let me explain something you >>> requested. >>>> Buildroot defconfigs should be _minimal_, i.e. just enough to boot the >>>> device and perhaps get a (wired) network connection. Looking at all those >>>> files, clearly these defconfigs are _not_ minimal. I think pretty much >>>> everything in the overlay can be removed. >>>> Nice trick to save the random MAC address that is generated on first boot, >>>> but we don't include such things in the minimal defconfig, especially if >>>> it's not board specific. >>>> This USB stuff is also not board-specific, is just generic USB gadget >>>> configuration. It might make sense to create a Buildroot package with those >>>> scripts though. >>> I think the digest of the changes is requesting us to remove the unnecessary >>> things such as the packages we use in the defconfig and files in the overlay. >>> I am sorry for my previous misunderstanding of your minimal philosophy. The >>> reason I did them is that I want to let the system we build could be nearly >>> the same as the official image. And the features are mainly about the USB >>> RNDIS or other multiplex functions. >>> Now I plan to remove the whole overlay folder and move our files which give >>> the features into a new package (e.g. named BR2_PACKAGE_MILKV_DUO_USB_TOOLS). >>> Could you please give me a review on that? >> >> As I wrote somewhere in my review (but you didn't include it below), I think >> the script that toggles the USB hub host/gadget mode is still relevant, >> because that is something board specific. The rest of the overlay indeed can >> be removed, I think. >> >> About BR2_PACKAGE_MILKV_DUO_USB_TOOLS: as far as I could see, those scripts >> are not at all board-specific, they simply configure the generic gadget >> functions. So the package name doesn't need to have milkv-duo in it. Perhaps >> initscripts-usb-gadget ? >> >> I'm not entirely sure if it is appropriate to be part of Buildroot anyway, it >> would be more appropriate in its own repository. But then again, if we have it >> in Buildroot itself, we can maintain it independently. > > In fact, all the scripts concerning USB in the previous overlay are all board > specific, to some extent. Please take a look at the following lines: Okay, I didn't look at it too carefully it seems. > > ${DUO_WRITE} host > /proc/cviusb/otg_role > CVI_DIR=/tmp/usb > CVI_GADGET=$CVI_DIR/usb_gadget/cvitek > # Set the USB configuration > mkdir $CVI_GADGET/configs/c.1 > mkdir $CVI_GADGET/configs/c.1/strings/0x409 > ${DUO_WRITE} "config1">$CVI_GADGET/configs/c.1/strings/0x409/configuration > > We could have two strategies. Firstly, if we think they are relevant and want to > keep something in the overlay, let’s compress the files into one single big > script. And do I have other alternatives of /usr/bin? I don’t want to move the > USB initial scripts into the user’s PATH. Secondly, we import the package > BR2_PACKAGE_MILKV_DUO_USB_TOOLS. The USB functions are not must for a minimal > system to run. However, if we use these shell scripts, by default the USB rndis > will be up, allowing us to login through ssh by a usb cable. If not, we have to > use the serial port. What do you think, overlay or package? For sure, the first step is a milkv_duo_defconfig without any of this stuff, and add the USB things in a follow-up patch. Expect a _lot_ of review comments on those, and possibly finally a rejection. >> Note that this initscripts-usb-gadget package will probably take a few more >> iterations before it gets accepted as well. So probably best to submit it >> separately from the reset of this series. Here are a few hints already that >> can help to make sure the initial submission is as good as possible. >> >> - Configuration should be in /etc/default. >> - Helper scripts should be in /usr/bin. >> - Use <PKG>_LINUX_CONFIG_FIXUPS to enable the required configs in the kernel. >> - The package should probably depend on !BR2_INIT_SYSTEMD. >> - Make sure to follow the init script skeleton (but I think you already did that). > > If we do use the package to provide the USB function, I highly hope that it > could be named milkv-duo, which will be convenient because we could directly > move our code into it. And we will not have the need to make it general. Yeah, if it really is board-specific, the milkv-duo naming is OK. [snip] >>>> There's also an Upstream: link missing. >>> Is the Upstream a must? I have found many patches in buildroot do not have this. >> >> The Upstream: tag is a must, check-package complains if it's not there. We >> require it exactly to motivate people to send their patches upstream. We want >> it upstream for two reasons: we don't want to maintain patches, and often >> upstream is better positioned to review the patches. The latter certainly >> applies to the OpenSBI patches. > > Could I use a customed repo to avoid use patches? The upstream is still a big > work in progress. I could not give an upstream tag currently. We hope to let our > users have a taste of buildroot quickly. Yes, a vendor fork repository is acceptable. [snip] >> ifeq ($(BR2_riscv_thead),y) >> GCC_TARGET_ARCH = >> rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync >> endif >> >> >> The above change should obviously be a separate patch as well. > > This is totally beyond my control, haha, willing to contribute in the future. In the mean time I found out that the T-Head CPU used in the BeagleV A-Head is in fact a different one, with different incompatibilities, so my idea of unifying them is not valid. Regards, Arnout [snip]
Hello Arnout, Glad to receive your review! These days we modified our patches and resubmitted them. Now there are two patches marked as v2 available, which almost meet your requirements. Could you please take a look on that? I am grateful for your patience and kind assistance. Best regards, Hanyuan
diff --git a/board/milkv/duo/README.md b/board/milkv/duo/README.md new file mode 100644 index 0000000000..f302d24de9 --- /dev/null +++ b/board/milkv/duo/README.md @@ -0,0 +1,64 @@ + +## Milk-V Duo Introduction + +Milk-V Duo is an ultra-compact embedded development platform based on the rv64 CV1800B chip. + +It can run Linux and RTOS, providing a reliable, cost-effective, and high-performance platform for professionals, industrial ODMs, AIoT enthusiasts, DIY enthusiasts, and creators. + +Features: + +- [x] The RISC-V C906 processor with clock speeds of 1GHz and 700MHz. +- [x] CSI-2 (MIPI Serial Camera Interface) +- [x] Ethernet PHY +- [x] USB RNDIS & USB HOST & USB Mass Storage +- [x] Multiple peripherals + - [x] - Up to 3x I2C + - [x] - Up to 5x UART + - [x] - Up to 1x SDIO1 + - [x] - Up to 1x SPI + - [x] - Up to 2x ADC + - [x] - Up to 7x PWM + - [x] - Up to 1x RUN + - [x] - Up to 1x JTAG + +## How to build + +```shell +$ make milkv_duo_musl_riscv64_defconfig +$ make +``` + +By default, a portion of the RAM is allocated to ION, which is the memory used when running algorithms with the camera. If you are not using the camera, please try: + +```shell +$ make milkv_duo_musl_riscv64_64mb_defconfig +$ make +``` + +To have all the 64mb free memory. + + +To configure the kernel or u-boot or others, please try: + +```shell +$ make linux-menuconfig # act like kernel's make menuconfig +$ make linux-rebuild +``` + +## Usage of the sdcard.img + +The `sdcard.img` is in `output/images`, you can directly write the image into your TF card, which could automatically boot and load the USB as rndis function by default. + +## Usage of the rootfs + +Find the `rootfs.tar.xz` in `output/images`, mount the `mmcblk0p2` part to a directory. Then: + +```shell +$ cd path/to/mmcblk0p2 +$ rm -rf * +$ sync + +$ tar -xf rootfs.tar.xz --directory=path/to/mmcblk0p2 +$ cd path/to/mmcblk0p2 +$ sync +``` diff --git a/board/milkv/duo/genimage.cfg b/board/milkv/duo/genimage.cfg new file mode 100644 index 0000000000..ed19e4fd3d --- /dev/null +++ b/board/milkv/duo/genimage.cfg @@ -0,0 +1,30 @@ +image boot.vfat { + vfat { + files = { + "fip.bin", + "boot.sd", + } + } + size = 64M +} + +image sdcard.img { + hdimage { + } + + partition boot { + partition-type = 0xC + bootable = "true" + image = "boot.vfat" + } + + partition rootfs { + partition-type = 0x83 + image = "rootfs.ext4" + } + + partition empty { + partition-type = 0x82 + size = 256M + } +} \ No newline at end of file diff --git a/board/milkv/duo/overlay/etc/dnsmasq.conf b/board/milkv/duo/overlay/etc/dnsmasq.conf new file mode 100644 index 0000000000..fcdb620ffa --- /dev/null +++ b/board/milkv/duo/overlay/etc/dnsmasq.conf @@ -0,0 +1,4 @@ +interface=usb0 +dhcp-range=192.168.42.2,192.168.42.242,1h +dhcp-option=3 +dhcp-option=6 diff --git a/board/milkv/duo/overlay/etc/init.d/S05runonce b/board/milkv/duo/overlay/etc/init.d/S05runonce new file mode 100755 index 0000000000..fe91bab998 --- /dev/null +++ b/board/milkv/duo/overlay/etc/init.d/S05runonce @@ -0,0 +1,32 @@ +#!/bin/sh + +# +# Execute scripts, that are intended to run only once +# +case "$1" in + start) + + for file in /etc/runonce.d/* + do + if [ ! -f "$file" ] + then + continue + fi + + if "$file" >> /tmp/runonce.log 2>&1 + then + mv "$file" "/etc/runonce.d/ran/" + sync + fi + done + ;; + stop) + ;; + restart|reload) + ;; + *) + echo "Usage: $0 {start|stop|restart}" + exit 1 +esac + +exit $? diff --git a/board/milkv/duo/overlay/etc/init.d/S10hwinit b/board/milkv/duo/overlay/etc/init.d/S10hwinit new file mode 100755 index 0000000000..ef7dc0f8c4 --- /dev/null +++ b/board/milkv/duo/overlay/etc/init.d/S10hwinit @@ -0,0 +1,44 @@ +#!/bin/sh + +# +# Initialize hardware +# +export SYSTEMPATH=/opt/milkv-duo/ +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +case "$1" in + start) + + if [ -f $SYSTEMPATH/duo-init.sh ]; then + . $SYSTEMPATH/duo-init.sh & + fi + + if [ ! -z $ETH0_MAC_ADDR ]; then + ifconfig eth0 hw ether ${ETH0_MAC_ADDR} + fi + + if [ "$DUO_USB_FUNC" == "RNDIS" ]; then + if [ -f $SYSTEMPATH/usb-rndis.sh ]; then + . $SYSTEMPATH/usb-rndis.sh & + fi + elif [ "$DUO_USB_FUNC" == "HOST" ]; then + if [ -f $SYSTEMPATH/usb-host.sh ]; then + . $SYSTEMPATH/usb-host.sh & + fi + elif [ "$DUO_USB_FUNC" == "MASS-STORAGE" ]; then + if [ -f $SYSTEMPATH/usb-mass-storage.sh ]; then + . $SYSTEMPATH/usb-mass-storage.sh & + fi + fi + ;; + stop) + ;; + restart|reload) + ;; + *) + echo "Usage: $0 {start|stop|restart}" + exit 1 +esac + +exit $? diff --git a/board/milkv/duo/overlay/etc/init.d/S15swap b/board/milkv/duo/overlay/etc/init.d/S15swap new file mode 100755 index 0000000000..6b5ef76326 --- /dev/null +++ b/board/milkv/duo/overlay/etc/init.d/S15swap @@ -0,0 +1,25 @@ +#!/bin/sh + +# +# Activate swap, depending on option +# +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +case "$1" in + start) + + if [ "$ENABLE_SWAP" == "1" ]; then + swapon /dev/mmcblk0p3 + fi + ;; + stop) + ;; + restart|reload) + ;; + *) + echo "Usage: $0 {start|stop|restart}" + exit 1 +esac + +exit $? diff --git a/board/milkv/duo/overlay/etc/init.d/S99user b/board/milkv/duo/overlay/etc/init.d/S99user new file mode 100755 index 0000000000..3abd073738 --- /dev/null +++ b/board/milkv/duo/overlay/etc/init.d/S99user @@ -0,0 +1,27 @@ +#!/bin/sh +${CVI_SHOPTS} + +# +# Start firmware +# +export SYSTEMPATH=/opt/milkv-duo/ + +case "$1" in + start) + + echo "Starting app..." + + if [ -f $SYSTEMPATH/blink.sh ]; then + . $SYSTEMPATH/blink.sh & + fi + ;; + stop) + ;; + restart|reload) + ;; + *) + echo "Usage: $0 {start|stop|restart}" + exit 1 +esac + +exit $? diff --git a/board/milkv/duo/overlay/etc/milkv-duo.conf b/board/milkv/duo/overlay/etc/milkv-duo.conf new file mode 100644 index 0000000000..c074254b55 --- /dev/null +++ b/board/milkv/duo/overlay/etc/milkv-duo.conf @@ -0,0 +1,56 @@ +# set 0 to disable blinking +ENABLE_BLINK=1 + +# set 1 to enable swap +ENABLE_SWAP=0 + +# command to write data into gpio or other devices +DUO_WRITE="echo -n" + +# LED gpio information +LED_GPIO=440 +LED_PATH=/sys/class/gpio/gpio${LED_GPIO} + +# USB Function: RNDIS, HOST, MASS-STORAGE +DUO_USB_FUNC=RNDIS +# Please ensure that the filesystem of the specified +# path is compatible with the operating system intended +# to be inserted, as failure to do so may result in +# damage to the filesystem +DUO_USB_MASS_STORAGE_ROOT=/dev/mmcblk0p1 + +# USB Configurations +GPIO_HUBPORT_EN=449 +GPIO_ROLESEL=450 +GPIO_HUBRST=451 +SYS_GPIO=/sys/class/gpio + +CLASS=acm +VID=0x3346 +PID=0x1003 +MSC_PID=0x1008 +RNDIS_PID=0x1009 +UVC_PID=0x100A +UAC_PID=0x100B +ADB_VID=0x18D1 +ADB_PID=0x4EE0 +ADB_PID_M1=0x4EE2 +ADB_PID_M2=0x4EE4 +MANUFACTURER="Cvitek" +PRODUCT="USB Com Port" +PRODUCT_RNDIS="RNDIS" +PRODUCT_UVC="UVC" +PRODUCT_UAC="UAC" +PRODUCT_ADB="ADB" +ADBD_PATH=/usr/bin/ +SERIAL="0123456789" +MSC_FILE=$3 +CVI_DIR=/tmp/usb +CVI_GADGET=$CVI_DIR/usb_gadget/cvitek +CVI_FUNC=$CVI_GADGET/functions +FUNC_NUM=0 +MAX_EP_NUM=4 +TMP_NUM=0 +INTF_NUM=0 +EP_IN=0 +EP_OUT=0 diff --git a/board/milkv/duo/overlay/etc/profile b/board/milkv/duo/overlay/etc/profile new file mode 100644 index 0000000000..f6f3784024 --- /dev/null +++ b/board/milkv/duo/overlay/etc/profile @@ -0,0 +1,36 @@ +export PATH="/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin" + +if [ "$PS1" ]; then + if [ "`id -u`" -eq 0 ]; then + export PS1='# ' + else + export PS1='$ ' + fi +fi + +export PAGER='/bin/more ' +export EDITOR='/bin/vi' + +# Source configuration files from /etc/profile.d +for i in /etc/profile.d/*.sh ; do + if [ -r "$i" ]; then + . $i + fi + unset i +done + +export HOSTNAME="$(hostname)" +export OLDPWD=/root + +if [ '$USER' == 'root' ]; then + export PS1='[\u@\h]\w\# ' +else + export PS1='[\u@\h]\w\$ ' +fi + +alias ll='ls -alF' +alias la='ls -A' +alias l='ls -CF' + +export TERM=xterm-256color +export TERMINFO=/usr/share/terminfo diff --git a/board/milkv/duo/overlay/etc/runonce.d/01make-swap b/board/milkv/duo/overlay/etc/runonce.d/01make-swap new file mode 100755 index 0000000000..64c361109f --- /dev/null +++ b/board/milkv/duo/overlay/etc/runonce.d/01make-swap @@ -0,0 +1,6 @@ +#!/bin/sh + +# +# Set up swap area +# +mkswap /dev/mmcblk0p3 diff --git a/board/milkv/duo/overlay/etc/runonce.d/05store-eth0-mac b/board/milkv/duo/overlay/etc/runonce.d/05store-eth0-mac new file mode 100755 index 0000000000..1c56640cd2 --- /dev/null +++ b/board/milkv/duo/overlay/etc/runonce.d/05store-eth0-mac @@ -0,0 +1,12 @@ +#!/bin/sh + +# +# Store first boot eth0 MAC address to make it persistent +# +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +if [ -z $ETH0_MAC_ADDR ]; then + echo -e "\n# stored at first boot" >> ${CONFIG} + echo "ETH0_MAC_ADDR=$(cat /sys/class/net/eth0/address)" >> ${CONFIG} +fi diff --git a/board/milkv/duo/overlay/etc/runonce.d/ran/.gitkeep b/board/milkv/duo/overlay/etc/runonce.d/ran/.gitkeep new file mode 100644 index 0000000000..e69de29bb2 diff --git a/board/milkv/duo/overlay/opt/milkv-duo/blink.sh b/board/milkv/duo/overlay/opt/milkv-duo/blink.sh new file mode 100755 index 0000000000..3d14977cc9 --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/blink.sh @@ -0,0 +1,25 @@ +#!/bin/sh + +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +if test -d ${LED_PATH}; then + echo "GPIO${LED_GPIO} already exported" +else + ${DUO_WRITE} ${LED_GPIO} >/sys/class/gpio/export +fi + +${DUO_WRITE} out >${LED_PATH}/direction + +while true; do + source ${CONFIG} + if [ "$ENABLE_BLINK" == "1" ]; then + ${DUO_WRITE} 0 >${LED_PATH}/value + sleep 0.5 + ${DUO_WRITE} 1 >${LED_PATH}/value + sleep 0.5 + else + ${DUO_WRITE} 0 >${LED_PATH}/value + sleep 2 + fi +done diff --git a/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh b/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh new file mode 100755 index 0000000000..298e43bc8d --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh @@ -0,0 +1,8 @@ +#!/bin/sh + +# Set Pin-29(PWR_SEQ2) to GPIO +devmem 0x0502707c 32 0x111 +devmem 0x03001068 32 0x3 + +# Set Pin-19(SD0_PWR_EN) to GPIO +devmem 0x0300101c 32 0x3 diff --git a/board/milkv/duo/overlay/opt/milkv-duo/run_usb.sh b/board/milkv/duo/overlay/opt/milkv-duo/run_usb.sh new file mode 100755 index 0000000000..911dd2883a --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/run_usb.sh @@ -0,0 +1,265 @@ +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +case "$2" in + acm) + CLASS=acm + ;; + msc) + CLASS=mass_storage + PID=$MSC_PID + ;; + cvg) + CLASS=cvg + ;; + rndis) + CLASS=rndis + PID=$RNDIS_PID + PRODUCT=$PRODUCT_RNDIS + ;; + uvc) + CLASS=uvc + PID=$UVC_PID + PRODUCT=$PRODUCT_UVC + ;; + uac1) + CLASS=uac1 + PID=$UAC_PID + PRODUCT=$PRODUCT_UAC + ;; + adb) + CLASS=ffs.adb + VID=$ADB_VID + PID=$ADB_PID + PRODUCT=$PRODUCT_ADB + ;; + *) + if [ "$1" = "probe" ] ; then + echo "Usage: $0 probe {acm|msc|cvg|rndis|uvc|uac1|adb}" + exit 1 + fi +esac + +calc_func() { + FUNC_NUM=$(ls $CVI_GADGET/functions -l | grep ^d | wc -l) + echo "$FUNC_NUM file(s)" +} + +res_check() { + TMP_NUM=$(find $CVI_GADGET/functions/ -name "acm*" | wc -l) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + TMP_NUM=$(($TMP_NUM * 2)) + EP_IN=$(($EP_IN+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name "mass_storage*" | wc -l) + EP_IN=$(($EP_IN+$TMP_NUM)) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name "cvg*" | wc -l) + EP_IN=$(($EP_IN+$TMP_NUM)) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name "rndis*" | wc -l) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + TMP_NUM=$(($TMP_NUM * 2)) + EP_IN=$(($EP_IN+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name "uvc*" | wc -l) + TMP_NUM=$(($TMP_NUM * 2)) + EP_IN=$(($EP_IN+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name "uac1*" | wc -l) + TMP_NUM=$(($TMP_NUM * 2)) + EP_IN=$(($EP_IN+$TMP_NUM)) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + TMP_NUM=$(find $CVI_GADGET/functions/ -name ffs.adb | wc -l) + EP_IN=$(($EP_IN+$TMP_NUM)) + EP_OUT=$(($EP_OUT+$TMP_NUM)) + INTF_NUM=$(($INTF_NUM+$TMP_NUM)) + + if [ "$CLASS" = "acm" ] ; then + EP_IN=$(($EP_IN+2)) + EP_OUT=$(($EP_OUT+1)) + fi + if [ "$CLASS" = "mass_storage" ] ; then + EP_IN=$(($EP_IN+1)) + EP_OUT=$(($EP_OUT+1)) + fi + if [ "$CLASS" = "cvg" ] ; then + EP_IN=$(($EP_IN+1)) + EP_OUT=$(($EP_OUT+1)) + fi + if [ "$CLASS" = "rndis" ] ; then + EP_IN=$(($EP_IN+2)) + EP_OUT=$(($EP_OUT+1)) + fi + if [ "$CLASS" = "uvc" ] ; then + EP_IN=$(($EP_IN+2)) + fi + if [ "$CLASS" = "uac1" ] ; then + EP_IN=$(($EP_IN+1)) + EP_OUT=$(($EP_OUT+1)) + fi + if [ "$CLASS" = "ffs.adb" ] ; then + EP_IN=$(($EP_IN+1)) + EP_OUT=$(($EP_OUT+1)) + fi + echo "$EP_IN in ep" + echo "$EP_OUT out ep" + if [ $EP_IN -gt $MAX_EP_NUM ]; then + echo "reach maximum resource" + exit 1 + fi + if [ $EP_OUT -gt $MAX_EP_NUM ]; then + echo "reach maximum resource" + exit 1 + fi +} + +probe() { + if [ ! -d $CVI_DIR ]; then + mkdir $CVI_DIR + fi + if [ ! -d $CVI_DIR/usb_gadget ]; then + # Enale USB ConfigFS + mount none $CVI_DIR -t configfs + # Create gadget dev + mkdir $CVI_GADGET + # Set the VID and PID + ${DUO_WRITE} $VID >$CVI_GADGET/idVendor + ${DUO_WRITE} $PID >$CVI_GADGET/idProduct + # Set the product information string + mkdir $CVI_GADGET/strings/0x409 + ${DUO_WRITE} $MANUFACTURER>$CVI_GADGET/strings/0x409/manufacturer + ${DUO_WRITE} $PRODUCT>$CVI_GADGET/strings/0x409/product + ${DUO_WRITE} $SERIAL>$CVI_GADGET/strings/0x409/serialnumber + # Set the USB configuration + mkdir $CVI_GADGET/configs/c.1 + mkdir $CVI_GADGET/configs/c.1/strings/0x409 + ${DUO_WRITE} "config1">$CVI_GADGET/configs/c.1/strings/0x409/configuration + # Set the MaxPower of USB descriptor + ${DUO_WRITE} 120 >$CVI_GADGET/configs/c.1/MaxPower + fi + # get current functions number + calc_func + # assign the class code for composite device + if [ ! $FUNC_NUM -eq 0 ]; then + ${DUO_WRITE} 0xEF >$CVI_GADGET/bDeviceClass + ${DUO_WRITE} 0x02 >$CVI_GADGET/bDeviceSubClass + ${DUO_WRITE} 0x01 >$CVI_GADGET/bDeviceProtocol + fi + # resource check + res_check + # create the desired function + if [ "$CLASS" = "ffs.adb" ] ; then + # adb shall be the last function to probe. Override the pid/vid + ${DUO_WRITE} $VID >$CVI_GADGET/idVendor + ${DUO_WRITE} $PID >$CVI_GADGET/idProduct + # choose pid for different function number + if [ $INTF_NUM -eq 1 ]; then + ${DUO_WRITE} $ADB_PID_M1 >$CVI_GADGET/idProduct + fi + if [ $INTF_NUM -eq 2 ]; then + ${DUO_WRITE} $ADB_PID_M2 >$CVI_GADGET/idProduct + fi + mkdir $CVI_GADGET/functions/$CLASS + else + mkdir $CVI_GADGET/functions/$CLASS.usb$FUNC_NUM + fi + if [ "$CLASS" = "mass_storage" ] ; then + ${DUO_WRITE} $MSC_FILE >$CVI_GADGET/functions/$CLASS.usb$FUNC_NUM/lun.0/file + fi + if [ "$CLASS" = "rndis" ] ; then + #OS STRING + ${DUO_WRITE} 1 >$CVI_GADGET/os_desc/use + ${DUO_WRITE} 0xcd >$CVI_GADGET/os_desc/b_vendor_code + ${DUO_WRITE} MSFT100 >$CVI_GADGET/os_desc/qw_sign + #COMPATIBLE ID + ${DUO_WRITE} RNDIS >$CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/compatible_id + #MAKE c.1 THE ONE ASSOCIATED WITH OS DESCRIPTORS + ln -s $CVI_GADGET/configs/c.1 $CVI_GADGET/os_desc + #MAKE "Icons" EXTENDED PROPERTY + mkdir $CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Icons + ${DUO_WRITE} 2 >$CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Icons/type + ${DUO_WRITE} "%SystemRoot%\\system32\\shell32.dll,-233" >$CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Icons/data + #MAKE "Label" EXTENDED PROPERTY + mkdir $CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Label + ${DUO_WRITE} 1 >$CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Label/type + ${DUO_WRITE} "XYZ Device" >$CVI_FUNC/rndis.usb$FUNC_NUM/os_desc/interface.rndis/Label/data + fi + +} + +start() { + # link this function to the configuration + calc_func + if [ $FUNC_NUM -eq 0 ]; then + echo "Functions Empty!" + exit 1 + fi + if [ -d $CVI_GADGET/functions/ffs.adb ]; then + FUNC_NUM=$(($FUNC_NUM-1)) + fi + for i in `seq 0 $(($FUNC_NUM-1))`; + do + find $CVI_GADGET/functions/ -name "*.usb$i" | xargs -I % ln -s % $CVI_GADGET/configs/c.1 + done + if [ -d $CVI_GADGET/functions/ffs.adb ]; then + ln -s $CVI_GADGET/functions/ffs.adb $CVI_GADGET/configs/c.1 + mkdir /dev/usb-ffs/adb -p + mount -t functionfs adb /dev/usb-ffs/adb + if [ -f $ADBD_PATH/adbd ]; then + $ADBD_PATH/adbd & + fi + else + # Start the gadget driver + UDC=`ls /sys/class/udc/ | awk '{print $1}'` + ${DUO_WRITE} ${UDC} >$CVI_GADGET/UDC + fi +} + +stop() { + if [ -d $CVI_GADGET/configs/c.1/ffs.adb ]; then + pkill adbd + rm $CVI_GADGET/configs/c.1/ffs.adb + else + ${DUO_WRITE} "" >$CVI_GADGET/UDC + fi + find $CVI_GADGET/configs/ -name "*.usb*" | xargs rm -f + rmdir $CVI_GADGET/configs/c.*/strings/0x409/ + tmp_dirs=$(find $CVI_GADGET/os_desc/c.* -type d) + if [ -n tmp_dirs ]; then + echo "remove os_desc!" + rm -rf $CVI_GADGET/os_desc/c.*/ + find $CVI_GADGET/functions/ -name Icons | xargs rmdir + find $CVI_GADGET/functions/ -name Label | xargs rmdir + fi + rmdir $CVI_GADGET/configs/c.*/ + rmdir $CVI_GADGET/functions/* + rmdir $CVI_GADGET/strings/0x409/ + rmdir $CVI_GADGET + umount $CVI_DIR + rmdir $CVI_DIR +} + +case "$1" in + start) + start + ;; + stop) + stop + ;; + probe) + probe + ;; + UDC) + ls /sys/class/udc/ >$CVI_GADGET/UDC + ;; + *) + echo "Usage: $0 probe {acm|msc|cvg|uvc|uac1} {file (msc)}" + echo "Usage: $0 start" + echo "Usage: $0 stop" + exit 1 +esac +exit $? diff --git a/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh b/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh new file mode 100755 index 0000000000..8a06949e3f --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh @@ -0,0 +1,63 @@ +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +hub_on() { + echo "turn on usb hub" + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBPORT_EN ]; then + ${DUO_WRITE} $GPIO_HUBPORT_EN >/sys/class/gpio/export + fi + + if [ ! -d $SYS_GPIO/gpio$GPIO_ROLESEL ]; then + ${DUO_WRITE} $GPIO_ROLESEL >/sys/class/gpio/export + fi + + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBRST ]; then + ${DUO_WRITE} $GPIO_HUBRST >/sys/class/gpio/export + fi + + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/direction + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_ROLESEL/direction + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBRST/direction + + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/value + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_ROLESEL/value + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_HUBRST/value +} + +hub_off() { + echo "turn off usb hub" + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBPORT_EN ]; then + ${DUO_WRITE} $GPIO_HUBPORT_EN >/sys/class/gpio/export + fi + + if [ ! -d $SYS_GPIO/gpio$GPIO_ROLESEL ]; then + ${DUO_WRITE} $GPIO_ROLESEL >/sys/class/gpio/export + fi + + if [ ! -d $SYS_GPIO/gpio$GPIO_HUBRST ]; then + ${DUO_WRITE} $GPIO_HUBRST >/sys/class/gpio/export + fi + + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/direction + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_ROLESEL/direction + ${DUO_WRITE} "out" >/sys/class/gpio/gpio$GPIO_HUBRST/direction + + ${DUO_WRITE} 0 >/sys/class/gpio/gpio$GPIO_HUBPORT_EN/value + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_ROLESEL/value + ${DUO_WRITE} 1 >/sys/class/gpio/gpio$GPIO_HUBRST/value +} + +case "$1" in + host) + insmod /mnt/system/ko/dwc2.ko + ${DUO_WRITE} host > /proc/cviusb/otg_role + ;; + device) + ${DUO_WRITE} device > /proc/cviusb/otg_role + ;; + *) + echo "Usage: $0 host" + echo "Usage: $0 device" + exit 1 +esac +exit $? diff --git a/board/milkv/duo/overlay/opt/milkv-duo/usb-host.sh b/board/milkv/duo/overlay/opt/milkv-duo/usb-host.sh new file mode 100755 index 0000000000..4072ecda46 --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/usb-host.sh @@ -0,0 +1,3 @@ +#!/bin/sh + +echo host > /proc/cviusb/otg_role >> /tmp/usb.log 2>&1 diff --git a/board/milkv/duo/overlay/opt/milkv-duo/usb-mass-storage.sh b/board/milkv/duo/overlay/opt/milkv-duo/usb-mass-storage.sh new file mode 100755 index 0000000000..571e59acdb --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/usb-mass-storage.sh @@ -0,0 +1,10 @@ +#!/bin/sh + +CONFIG=/etc/milkv-duo.conf +source ${CONFIG} + +/opt/milkv-duo/uhubon.sh device >> /tmp/usb.log 2>&1 +/opt/milkv-duo/run_usb.sh probe msc ${DUO_USB_MASS_STORAGE_ROOT} >> /tmp/usb.log 2>&1 +/opt/milkv-duo/run_usb.sh start >> /tmp/usb.log 2>&1 + +[ $? = 0 ] && echo "Mass Storage started successfully" || echo "Fail to start Mass Storage" diff --git a/board/milkv/duo/overlay/opt/milkv-duo/usb-rndis.sh b/board/milkv/duo/overlay/opt/milkv-duo/usb-rndis.sh new file mode 100755 index 0000000000..787e044b90 --- /dev/null +++ b/board/milkv/duo/overlay/opt/milkv-duo/usb-rndis.sh @@ -0,0 +1,14 @@ +#!/bin/sh + +/opt/milkv-duo/uhubon.sh device >> /tmp/rndis.log 2>&1 +/opt/milkv-duo/run_usb.sh probe rndis >> /tmp/rndis.log 2>&1 +/opt/milkv-duo/run_usb.sh start rndis >> /tmp/rndis.log 2>&1 + +sleep 0.5 +ifconfig usb0 192.168.42.1 + +count=`ps | grep dnsmasq | grep -v grep | wc -l` +if [ ${count} -lt 1 ] ;then + echo "/etc/init.d/S80dnsmasq start" >> /tmp/rndis.log 2>&1 + /etc/init.d/S80dnsmasq start >> /tmp/rndis.log 2>&1 +fi diff --git a/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch b/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch new file mode 100644 index 0000000000..cd784948be --- /dev/null +++ b/board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch @@ -0,0 +1,44 @@ +From 2e8190643f1abdfcf6c8b5981b6d6863dea78a08 Mon Sep 17 00:00:00 2001 +From: "sam.xiang" <sam.xiang@sophgo.com> +Date: Tue, 21 Feb 2023 20:43:18 +0800 +Subject: [PATCH 1/3] [opensbi] porting: adapt to CV180X / CV181X + +Change-Id: Id7b3c64d203eb2c9af6c66f195bf0d8a05f0164c +Signed-off-by: Hanyuan Zhao <hanyuan-z@qq.com> +--- + lib/utils/serial/fdt_serial_uart8250.c | 1 + + platform/generic/config.mk | 3 ++- + 2 files changed, 3 insertions(+), 1 deletion(-) + mode change 100644 => 100755 lib/utils/serial/fdt_serial_uart8250.c + +diff --git a/lib/utils/serial/fdt_serial_uart8250.c b/lib/utils/serial/fdt_serial_uart8250.c +old mode 100644 +new mode 100755 +index 918193ae9..c72eeeb89 +--- a/lib/utils/serial/fdt_serial_uart8250.c ++++ b/lib/utils/serial/fdt_serial_uart8250.c +@@ -26,6 +26,7 @@ static int serial_uart8250_init(void *fdt, int nodeoff, + } + + static const struct fdt_match serial_uart8250_match[] = { ++ { .compatible = "snps,dw-apb-uart" }, + { .compatible = "ns16550" }, + { .compatible = "ns16550a" }, + { }, +diff --git a/platform/generic/config.mk b/platform/generic/config.mk +index 8151974f2..6e09f3332 100644 +--- a/platform/generic/config.mk ++++ b/platform/generic/config.mk +@@ -28,7 +28,8 @@ else + # This needs to be 2MB aligned for 64-bit system + FW_JUMP_ADDR=$(shell printf "0x%X" $$(($(FW_TEXT_START) + 0x200000))) + endif +-FW_JUMP_FDT_ADDR=$(shell printf "0x%X" $$(($(FW_TEXT_START) + 0x2200000))) ++#set FDT_ADDR 0xB0000000 ++FW_JUMP_FDT_ADDR=$(shell printf "0x%X" $$(($(FW_TEXT_START) + 0x30000000))) + FW_PAYLOAD=y + ifeq ($(PLATFORM_RISCV_XLEN), 32) + # This needs to be 4MB aligned for 32-bit system +-- +2.34.1 + diff --git a/board/milkv/duo/patches/opensbi/0002-fix-set-zicsr-zifencei-and-kill-compile-errors.patch b/board/milkv/duo/patches/opensbi/0002-fix-set-zicsr-zifencei-and-kill-compile-errors.patch new file mode 100644 index 0000000000..37b009dc48 --- /dev/null +++ b/board/milkv/duo/patches/opensbi/0002-fix-set-zicsr-zifencei-and-kill-compile-errors.patch @@ -0,0 +1,59 @@ +From 6b0054b83422b0b964b2ecc5496e136987c1ac16 Mon Sep 17 00:00:00 2001 +From: Hanyuan Zhao <hanyuan-z@qq.com> +Date: Mon, 16 Oct 2023 21:44:19 +0800 +Subject: [PATCH 2/3] fix: set zicsr zifencei and kill compile errors + +Signed-off-by: Hanyuan Zhao <hanyuan-z@qq.com> +--- + Makefile | 10 ++++------ + 1 file changed, 4 insertions(+), 6 deletions(-) + +diff --git a/Makefile b/Makefile +index eeffe6bad..fa3b18ddc 100644 +--- a/Makefile ++++ b/Makefile +@@ -157,7 +157,7 @@ ifndef PLATFORM_RISCV_ABI + endif + ifndef PLATFORM_RISCV_ISA + ifneq ($(PLATFORM_RISCV_TOOLCHAIN_DEFAULT), 1) +- PLATFORM_RISCV_ISA = rv$(PLATFORM_RISCV_XLEN)imafdc ++ PLATFORM_RISCV_ISA = rv$(PLATFORM_RISCV_XLEN)imafdc_zicsr_zifencei + else + PLATFORM_RISCV_ISA = $(OPENSBI_CC_ISA) + endif +@@ -203,10 +203,9 @@ GENFLAGS += $(libsbiutils-genflags-y) + GENFLAGS += $(platform-genflags-y) + GENFLAGS += $(firmware-genflags-y) + +-CFLAGS = -g -Wall -Werror -ffreestanding -nostdlib -fno-strict-aliasing -O2 ++CFLAGS = -g -Wall -Werror -ffreestanding -nostdlib -fno-stack-protector -fno-strict-aliasing -O2 + CFLAGS += -fno-omit-frame-pointer -fno-optimize-sibling-calls + CFLAGS += -mno-save-restore -mstrict-align +-CFLAGS += -mabi=$(PLATFORM_RISCV_ABI) -march=$(PLATFORM_RISCV_ISA) + CFLAGS += -mcmodel=$(PLATFORM_RISCV_CODE_MODEL) + CFLAGS += $(GENFLAGS) + CFLAGS += $(platform-cflags-y) +@@ -217,10 +216,9 @@ CPPFLAGS += $(GENFLAGS) + CPPFLAGS += $(platform-cppflags-y) + CPPFLAGS += $(firmware-cppflags-y) + +-ASFLAGS = -g -Wall -nostdlib ++ASFLAGS = -g -Wall -nostdlib -fno-stack-protector -D__ASSEMBLY__ + ASFLAGS += -fno-omit-frame-pointer -fno-optimize-sibling-calls + ASFLAGS += -mno-save-restore -mstrict-align +-ASFLAGS += -mabi=$(PLATFORM_RISCV_ABI) -march=$(PLATFORM_RISCV_ISA) + ASFLAGS += -mcmodel=$(PLATFORM_RISCV_CODE_MODEL) + ASFLAGS += $(GENFLAGS) + ASFLAGS += $(platform-asflags-y) +@@ -236,7 +234,7 @@ MERGEFLAGS += -r + MERGEFLAGS += -b elf$(PLATFORM_RISCV_XLEN)-littleriscv + MERGEFLAGS += -m elf$(PLATFORM_RISCV_XLEN)lriscv + +-DTSCPPFLAGS = $(CPPFLAGS) -nostdinc -nostdlib -fno-builtin -D__DTS__ -x assembler-with-cpp ++DTSCPPFLAGS = $(CPPFLAGS) -nostdinc -nostdlib -fno-stack-protector -fno-builtin -D__DTS__ -x assembler-with-cpp + + # Setup functions for compilation + define dynamic_flags +-- +2.34.1 + diff --git a/board/milkv/duo/patches/opensbi/0003-fix-initialize-the-value-of-a-variable-in-stack.patch b/board/milkv/duo/patches/opensbi/0003-fix-initialize-the-value-of-a-variable-in-stack.patch new file mode 100644 index 0000000000..0b1f5e7302 --- /dev/null +++ b/board/milkv/duo/patches/opensbi/0003-fix-initialize-the-value-of-a-variable-in-stack.patch @@ -0,0 +1,26 @@ +From 1d96d3e9f82e81991152aafb44fa5f0ec493a9d2 Mon Sep 17 00:00:00 2001 +From: Hanyuan Zhao <zhaohy22@mails.tsinghua.edu.cn> +Date: Mon, 16 Oct 2023 21:48:21 +0800 +Subject: [PATCH 3/3] fix: initialize the value of a variable in stack + +Signed-off-by: Hanyuan Zhao <zhaohy22@mails.tsinghua.edu.cn> +--- + lib/utils/reset/fdt_reset_thead.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/lib/utils/reset/fdt_reset_thead.c b/lib/utils/reset/fdt_reset_thead.c +index 95f8c369c..805067859 100644 +--- a/lib/utils/reset/fdt_reset_thead.c ++++ b/lib/utils/reset/fdt_reset_thead.c +@@ -62,7 +62,7 @@ static int thead_reset_init(void *fdt, int nodeoff, + void *p; + const fdt64_t *val; + const fdt32_t *val_w; +- int len, i, cnt; ++ int len, i, cnt=0; + u32 t, tmp = 0; + + /* Prepare clone csrs */ +-- +2.34.1 + diff --git a/board/milkv/duo/post-image.sh b/board/milkv/duo/post-image.sh new file mode 100755 index 0000000000..bebd76a4ad --- /dev/null +++ b/board/milkv/duo/post-image.sh @@ -0,0 +1,52 @@ +#!/bin/sh + +########################################################### +# File: post-image.sh +# Author: GP Orcullo <kinsamanka@gmail.com> +# Description: this sh will make the fip.bin and boot.sd, +# then pack everything to an image file. +########################################################### +if grep -Eq "^BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL=y$" ${BR2_CONFIG}; then + if grep -Eq "^BR2_PACKAGE_MILKV_DUO_SMALLCORE_FREERTOS=y$" ${BR2_CONFIG}; then + ${BINARIES_DIR}/fiptool.py genfip ${BINARIES_DIR}/fip.bin \ + --MONITOR_RUNADDR=0x80000000 \ + --CHIP_CONF=${BINARIES_DIR}/chip_conf.bin \ + --NOR_INFO=FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF \ + --NAND_INFO=00000000 \ + --BL2=${BINARIES_DIR}/bl2.bin \ + --BLCP_IMG_RUNADDR=0x05200200 \ + --BLCP_PARAM_LOADADDR=0 \ + --DDR_PARAM=${BINARIES_DIR}/ddr_param.bin \ + --MONITOR=${BINARIES_DIR}/fw_dynamic.bin \ + --LOADER_2ND=${BINARIES_DIR}/u-boot.bin \ + --BLCP=${BINARIES_DIR}/empty.bin \ + --BLCP_2ND=${BINARIES_DIR}/cvirtos.bin \ + --BLCP_2ND_RUNADDR=0x83f40000 \ + > ${BINARIES_DIR}/fip.log 2>&1 + echo "[Duo Post-Image fiptool.py] FreeRTOS integrated" + else + ${BINARIES_DIR}/fiptool.py genfip ${BINARIES_DIR}/fip.bin \ + --MONITOR_RUNADDR=0x80000000 \ + --CHIP_CONF=${BINARIES_DIR}/chip_conf.bin \ + --NOR_INFO=FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF \ + --NAND_INFO=00000000 \ + --BL2=${BINARIES_DIR}/bl2.bin \ + --BLCP_IMG_RUNADDR=0x05200200 \ + --BLCP_PARAM_LOADADDR=0 \ + --DDR_PARAM=${BINARIES_DIR}/ddr_param.bin \ + --MONITOR=${BINARIES_DIR}/fw_dynamic.bin \ + --LOADER_2ND=${BINARIES_DIR}/u-boot.bin \ + > ${BINARIES_DIR}/fip.log 2>&1 + echo "[Duo Post-Image fiptool.py] No FreeRTOS integrated" + fi + + cp ${BINARIES_DIR}/u-boot.dtb ${BINARIES_DIR}/cv1800b_milkv_duo_sd.dtb + lzma -fk ${BINARIES_DIR}/Image + mkimage -f ${BINARIES_DIR}/multi.its ${BINARIES_DIR}/boot.sd + echo "[Duo Post-Image] > boot.sd generated!" + support/scripts/genimage.sh -c $(pwd)/board/milkv/duo/genimage.cfg + gzip -fk ${BINARIES_DIR}/sdcard.img + echo "[Duo Post-Image] > sdcard.img generated!" +else + echo "[Duo Post-Image] Not requested to generate the boot files and sdcard.img" +fi \ No newline at end of file diff --git a/configs/milkv_duo_musl_riscv64_64mb_defconfig b/configs/milkv_duo_musl_riscv64_64mb_defconfig new file mode 100644 index 0000000000..e97677d891 --- /dev/null +++ b/configs/milkv_duo_musl_riscv64_64mb_defconfig @@ -0,0 +1,94 @@ +BR2_riscv=y +BR2_riscv_g=y +BR2_RISCV_ISA_RVC=y +BR2_TOOLCHAIN_BUILDROOT_MUSL=y +BR2_KERNEL_HEADERS_5_10=y +BR2_GCC_VERSION_13_X=y +BR2_TOOLCHAIN_BUILDROOT_CXX=y +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" +BR2_GLOBAL_PATCH_DIR="board/milkv/duo/patches" +BR2_TARGET_GENERIC_HOSTNAME="milkv-duo" +BR2_TARGET_GENERIC_ISSUE="Welcome to Milk-V" +BR2_TARGET_GENERIC_ROOT_PASSWD="milkv" +BR2_ROOTFS_OVERLAY="board/milkv/duo/overlay" +BR2_ROOTFS_POST_IMAGE_SCRIPT="board/milkv/duo/post-image.sh" +BR2_LINUX_KERNEL=y +BR2_LINUX_KERNEL_CUSTOM_GIT=y +BR2_LINUX_KERNEL_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-linux.git" +BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION="duo-linux-5.10.4" +BR2_LINUX_KERNEL_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" +BR2_LINUX_KERNEL_LZMA=y +BR2_PACKAGE_BZIP2=y +BR2_PACKAGE_GZIP=y +BR2_PACKAGE_LZIP=y +BR2_PACKAGE_UNZIP=y +BR2_PACKAGE_ZIP=y +BR2_PACKAGE_GDB=y +BR2_PACKAGE_GDB_DEBUGGER=y +BR2_PACKAGE_SPIDEV_TEST=y +BR2_PACKAGE_STRACE=y +BR2_PACKAGE_BINUTILS=y +BR2_PACKAGE_GREP=y +BR2_PACKAGE_MAKE=y +BR2_PACKAGE_PATCH=y +BR2_PACKAGE_TREE=y +BR2_PACKAGE_DOSFSTOOLS=y +BR2_PACKAGE_E2FSPROGS=y +BR2_PACKAGE_E2FSPROGS_RESIZE2FS=y +BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL=y +BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL_64MB=y +BR2_PACKAGE_MILKV_DUO_SMALLCORE_FREERTOS=y +BR2_PACKAGE_MILKV_DUO_PINMUX=y +BR2_PACKAGE_EVTEST=y +BR2_PACKAGE_PYTHON3=y +BR2_PACKAGE_PYTHON_EVDEV=y +BR2_PACKAGE_PYTHON_LXML=y +BR2_PACKAGE_PYTHON_MODBUS_TK=y +BR2_PACKAGE_PYTHON_PILLOW=y +BR2_PACKAGE_PYTHON_PIP=y +BR2_PACKAGE_PYTHON_SMBUS_CFFI=y +BR2_PACKAGE_PYTHON_SPIDEV=y +BR2_PACKAGE_FREETYPE=y +BR2_PACKAGE_JSON_C=y +BR2_PACKAGE_DHCPCD=y +BR2_PACKAGE_DNSMASQ=y +BR2_PACKAGE_DROPBEAR=y +BR2_PACKAGE_LRZSZ=y +BR2_PACKAGE_NTP=y +BR2_PACKAGE_NTP_NTPDATE=y +BR2_PACKAGE_NTP_NTPTIME=y +BR2_PACKAGE_WPA_SUPPLICANT=y +BR2_PACKAGE_WPA_SUPPLICANT_CLI=y +BR2_PACKAGE_WPA_SUPPLICANT_WPA_CLIENT_SO=y +BR2_PACKAGE_NEOFETCH=y +BR2_PACKAGE_TMUX=y +BR2_PACKAGE_WHICH=y +BR2_PACKAGE_HTOP=y +BR2_PACKAGE_KMOD=y +BR2_PACKAGE_NANO=y +BR2_TARGET_ROOTFS_EXT2=y +BR2_TARGET_ROOTFS_EXT2_4=y +BR2_TARGET_ROOTFS_EXT2_SIZE="200M" +BR2_TARGET_ROOTFS_TAR_XZ=y +BR2_TARGET_OPENSBI=y +BR2_TARGET_OPENSBI_CUSTOM_GIT=y +BR2_TARGET_OPENSBI_CUSTOM_REPO_URL="https://github.com/T-head-Semi/opensbi.git" +BR2_TARGET_OPENSBI_CUSTOM_REPO_VERSION="4e77060e0512ad981eee55d5a2501f6d88a41fd9" +BR2_TARGET_OPENSBI_PLAT="generic" +BR2_TARGET_OPENSBI_UBOOT_PAYLOAD=y +BR2_TARGET_OPENSBI_FW_FDT_PATH=y +BR2_TARGET_UBOOT=y +BR2_TARGET_UBOOT_BUILD_SYSTEM_KCONFIG=y +BR2_TARGET_UBOOT_CUSTOM_GIT=y +BR2_TARGET_UBOOT_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-u-boot.git" +BR2_TARGET_UBOOT_CUSTOM_REPO_VERSION="v2021.10_64mb" +BR2_TARGET_UBOOT_BOARD_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" +BR2_TARGET_UBOOT_NEEDS_DTC=y +BR2_TARGET_UBOOT_NEEDS_OPENSBI=y +BR2_PACKAGE_HOST_CMAKE=y +BR2_PACKAGE_HOST_DOSFSTOOLS=y +BR2_PACKAGE_HOST_GENIMAGE=y +BR2_PACKAGE_HOST_LZMA_ALONE=y +BR2_PACKAGE_HOST_MTOOLS=y +BR2_PACKAGE_HOST_UBOOT_TOOLS=y +BR2_PACKAGE_HOST_UBOOT_TOOLS_FIT_SUPPORT=y diff --git a/configs/milkv_duo_musl_riscv64_defconfig b/configs/milkv_duo_musl_riscv64_defconfig new file mode 100644 index 0000000000..bd79f138f6 --- /dev/null +++ b/configs/milkv_duo_musl_riscv64_defconfig @@ -0,0 +1,93 @@ +BR2_riscv=y +BR2_riscv_g=y +BR2_RISCV_ISA_RVC=y +BR2_TOOLCHAIN_BUILDROOT_MUSL=y +BR2_KERNEL_HEADERS_5_10=y +BR2_GCC_VERSION_13_X=y +BR2_TOOLCHAIN_BUILDROOT_CXX=y +BR2_TARGET_LDFLAGS="-march=rv64imafdc_xtheadba_xtheadbb_xtheadbs_xtheadcmo_xtheadcondmov_xtheadfmemidx_xtheadfmv_xtheadint_xtheadmac_xtheadmemidx_xtheadmempair_xtheadsync" +BR2_GLOBAL_PATCH_DIR="board/milkv/duo/patches" +BR2_TARGET_GENERIC_HOSTNAME="milkv-duo" +BR2_TARGET_GENERIC_ISSUE="Welcome to Milk-V" +BR2_TARGET_GENERIC_ROOT_PASSWD="milkv" +BR2_ROOTFS_OVERLAY="board/milkv/duo/overlay" +BR2_ROOTFS_POST_IMAGE_SCRIPT="board/milkv/duo/post-image.sh" +BR2_LINUX_KERNEL=y +BR2_LINUX_KERNEL_CUSTOM_GIT=y +BR2_LINUX_KERNEL_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-linux.git" +BR2_LINUX_KERNEL_CUSTOM_REPO_VERSION="duo-linux-5.10.4" +BR2_LINUX_KERNEL_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" +BR2_LINUX_KERNEL_LZMA=y +BR2_PACKAGE_BZIP2=y +BR2_PACKAGE_GZIP=y +BR2_PACKAGE_LZIP=y +BR2_PACKAGE_UNZIP=y +BR2_PACKAGE_ZIP=y +BR2_PACKAGE_GDB=y +BR2_PACKAGE_GDB_DEBUGGER=y +BR2_PACKAGE_SPIDEV_TEST=y +BR2_PACKAGE_STRACE=y +BR2_PACKAGE_BINUTILS=y +BR2_PACKAGE_GREP=y +BR2_PACKAGE_MAKE=y +BR2_PACKAGE_PATCH=y +BR2_PACKAGE_TREE=y +BR2_PACKAGE_DOSFSTOOLS=y +BR2_PACKAGE_E2FSPROGS=y +BR2_PACKAGE_E2FSPROGS_RESIZE2FS=y +BR2_PACKAGE_MILKV_DUO_FIRMWARE_FSBL=y +BR2_PACKAGE_MILKV_DUO_SMALLCORE_FREERTOS=y +BR2_PACKAGE_MILKV_DUO_PINMUX=y +BR2_PACKAGE_EVTEST=y +BR2_PACKAGE_PYTHON3=y +BR2_PACKAGE_PYTHON_EVDEV=y +BR2_PACKAGE_PYTHON_LXML=y +BR2_PACKAGE_PYTHON_MODBUS_TK=y +BR2_PACKAGE_PYTHON_PILLOW=y +BR2_PACKAGE_PYTHON_PIP=y +BR2_PACKAGE_PYTHON_SMBUS_CFFI=y +BR2_PACKAGE_PYTHON_SPIDEV=y +BR2_PACKAGE_FREETYPE=y +BR2_PACKAGE_JSON_C=y +BR2_PACKAGE_DHCPCD=y +BR2_PACKAGE_DNSMASQ=y +BR2_PACKAGE_DROPBEAR=y +BR2_PACKAGE_LRZSZ=y +BR2_PACKAGE_NTP=y +BR2_PACKAGE_NTP_NTPDATE=y +BR2_PACKAGE_NTP_NTPTIME=y +BR2_PACKAGE_WPA_SUPPLICANT=y +BR2_PACKAGE_WPA_SUPPLICANT_CLI=y +BR2_PACKAGE_WPA_SUPPLICANT_WPA_CLIENT_SO=y +BR2_PACKAGE_NEOFETCH=y +BR2_PACKAGE_TMUX=y +BR2_PACKAGE_WHICH=y +BR2_PACKAGE_HTOP=y +BR2_PACKAGE_KMOD=y +BR2_PACKAGE_NANO=y +BR2_TARGET_ROOTFS_EXT2=y +BR2_TARGET_ROOTFS_EXT2_4=y +BR2_TARGET_ROOTFS_EXT2_SIZE="200M" +BR2_TARGET_ROOTFS_TAR_XZ=y +BR2_TARGET_OPENSBI=y +BR2_TARGET_OPENSBI_CUSTOM_GIT=y +BR2_TARGET_OPENSBI_CUSTOM_REPO_URL="https://github.com/T-head-Semi/opensbi.git" +BR2_TARGET_OPENSBI_CUSTOM_REPO_VERSION="4e77060e0512ad981eee55d5a2501f6d88a41fd9" +BR2_TARGET_OPENSBI_PLAT="generic" +BR2_TARGET_OPENSBI_UBOOT_PAYLOAD=y +BR2_TARGET_OPENSBI_FW_FDT_PATH=y +BR2_TARGET_UBOOT=y +BR2_TARGET_UBOOT_BUILD_SYSTEM_KCONFIG=y +BR2_TARGET_UBOOT_CUSTOM_GIT=y +BR2_TARGET_UBOOT_CUSTOM_REPO_URL="https://github.com/milkv-duo/milkv-duo-u-boot.git" +BR2_TARGET_UBOOT_CUSTOM_REPO_VERSION="v2021.10_ion" +BR2_TARGET_UBOOT_BOARD_DEFCONFIG="cvitek_cv1800b_milkv_duo_sd" +BR2_TARGET_UBOOT_NEEDS_DTC=y +BR2_TARGET_UBOOT_NEEDS_OPENSBI=y +BR2_PACKAGE_HOST_CMAKE=y +BR2_PACKAGE_HOST_DOSFSTOOLS=y +BR2_PACKAGE_HOST_GENIMAGE=y +BR2_PACKAGE_HOST_LZMA_ALONE=y +BR2_PACKAGE_HOST_MTOOLS=y +BR2_PACKAGE_HOST_UBOOT_TOOLS=y +BR2_PACKAGE_HOST_UBOOT_TOOLS_FIT_SUPPORT=y
Milk-V Duo is an ultra-compact embedded development platform based on the CV1800B chip. It can run Linux and RTOS, providing a reliable, low-cost, and high-performance platform for professionals, industrial ODMs, AIoT enthusiasts, DIY hobbyists, and creators. Co-authored-by: Emil S <emil.soltys@outlook.com> Co-authored-by: GP Orcullo <kinsamanka@gmail.com> Signed-off-by: Hanyuan Zhao <hanyuan-z@qq.com> --- board/milkv/duo/README.md | 64 +++++ board/milkv/duo/genimage.cfg | 30 ++ board/milkv/duo/overlay/etc/dnsmasq.conf | 4 + board/milkv/duo/overlay/etc/init.d/S05runonce | 32 +++ board/milkv/duo/overlay/etc/init.d/S10hwinit | 44 +++ board/milkv/duo/overlay/etc/init.d/S15swap | 25 ++ board/milkv/duo/overlay/etc/init.d/S99user | 27 ++ board/milkv/duo/overlay/etc/milkv-duo.conf | 56 ++++ board/milkv/duo/overlay/etc/profile | 36 +++ .../duo/overlay/etc/runonce.d/01make-swap | 6 + .../overlay/etc/runonce.d/05store-eth0-mac | 12 + .../duo/overlay/etc/runonce.d/ran/.gitkeep | 0 .../milkv/duo/overlay/opt/milkv-duo/blink.sh | 25 ++ .../duo/overlay/opt/milkv-duo/duo-init.sh | 8 + .../duo/overlay/opt/milkv-duo/run_usb.sh | 265 ++++++++++++++++++ .../milkv/duo/overlay/opt/milkv-duo/uhubon.sh | 63 +++++ .../duo/overlay/opt/milkv-duo/usb-host.sh | 3 + .../overlay/opt/milkv-duo/usb-mass-storage.sh | 10 + .../duo/overlay/opt/milkv-duo/usb-rndis.sh | 14 + ...ensbi-porting-adapt-to-CV180X-CV181X.patch | 44 +++ ...csr-zifencei-and-kill-compile-errors.patch | 59 ++++ ...ize-the-value-of-a-variable-in-stack.patch | 26 ++ board/milkv/duo/post-image.sh | 52 ++++ configs/milkv_duo_musl_riscv64_64mb_defconfig | 94 +++++++ configs/milkv_duo_musl_riscv64_defconfig | 93 ++++++ 25 files changed, 1092 insertions(+) create mode 100644 board/milkv/duo/README.md create mode 100644 board/milkv/duo/genimage.cfg create mode 100644 board/milkv/duo/overlay/etc/dnsmasq.conf create mode 100755 board/milkv/duo/overlay/etc/init.d/S05runonce create mode 100755 board/milkv/duo/overlay/etc/init.d/S10hwinit create mode 100755 board/milkv/duo/overlay/etc/init.d/S15swap create mode 100755 board/milkv/duo/overlay/etc/init.d/S99user create mode 100644 board/milkv/duo/overlay/etc/milkv-duo.conf create mode 100644 board/milkv/duo/overlay/etc/profile create mode 100755 board/milkv/duo/overlay/etc/runonce.d/01make-swap create mode 100755 board/milkv/duo/overlay/etc/runonce.d/05store-eth0-mac create mode 100644 board/milkv/duo/overlay/etc/runonce.d/ran/.gitkeep create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/blink.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/duo-init.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/run_usb.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/uhubon.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-host.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-mass-storage.sh create mode 100755 board/milkv/duo/overlay/opt/milkv-duo/usb-rndis.sh create mode 100644 board/milkv/duo/patches/opensbi/0001-opensbi-porting-adapt-to-CV180X-CV181X.patch create mode 100644 board/milkv/duo/patches/opensbi/0002-fix-set-zicsr-zifencei-and-kill-compile-errors.patch create mode 100644 board/milkv/duo/patches/opensbi/0003-fix-initialize-the-value-of-a-variable-in-stack.patch create mode 100755 board/milkv/duo/post-image.sh create mode 100644 configs/milkv_duo_musl_riscv64_64mb_defconfig create mode 100644 configs/milkv_duo_musl_riscv64_defconfig