* imported Rockchip project patches from libreelec
* disable duplicate/unapplicable patches and rework others
* add patches for tinkerboard, miqi and xt-q8l-v10 to enable hevc, hevc_mmu and rga nodes
- starting from Home Assistant OS (HASSOS) setup/patchset, which is stable with pure-mainline
- *huge thanks* to Stefan Agner (@agners) from Home Assistant for his work and help with this
- CONFIG_DRM_MESON=y fixes the order of shutdown callbacks, allowing it to reboot without patches
- this increases the kernel size a bit, due to some dependencies like DRM=y
- remove the shutdown revert patch, allowing other meson64 boards to mainline-reboot
- add fdt/no-map patch fixing the dreaded "SError Interrupt on CPU4" panics that plagued the N2
- move the N2 to mainline u-boot on current (rebased, already done)
- add SPI-enabling DTB variants, which allow writing to SPI NOR flash (and usage instructions in comments)
- add UBOOT_TARGET_MAP and write_uboot_platform_mtd to current, for nand-sata-install to SPI/USB goodness
- remove the UHS-enabling patch, to avoid 1.8v vs 3.3v mess. Make sure to use only A1-rated SD cards as recommended.
- change default cpufreq governor for 24h+ stability
- other patch cleanups:
- move deprecated/non-applying patch to need-check
- remove the HardKernel ODROID-specific reboot driver and DTS patches (odroid-reboot.c) which are now useless.
- remove documentation-only patches (we have enough patches in meson64 without those)
- remove duplicated SAR ADC patches
- remove multiple different and conflicting patches for DTS and emulated-PWM GPIO fan
- now using simple on/off GPIO fan at 30 celsius, unrelated to passive cooling
* remove 0-rever-rk3328.dtsi-.patch
* split rk3328-dtsi-mmc-vdec-usb3-tweaks.patch into different specific patches (sdmmc-ext, mmc-reset, power domains, usb3 resets)
* split rk3328-audio-and-renegade-supplies.patch into specific roc-cc audio and supplies patch and general rk3328 spdif patch
* add "dtsi" infix to rk3328 patches that deal with dtsi files
* add back mali gpu operating points patch
* add workaround against default MAC Address (userland utility bdaddr does not work to set MAC Address as expected)
* enable bits for xt-q8l-v10 board (ap6330/bcm4330 part), switching AP6330 bluetooth to btbcm driver using serdev
facilities to avoid userspace machinery with deprecated hciattach tool, added bits to enable driver support in kernel.
Removed ap6330 bluetooth service from bsp packages, added initramfs-tools hook to include bcm4330b1.hcd firmware in initramfs
* Backporting patch to 5.10 kernel makes sense. Lets do it.
Co-authored-by: iamdrq <iamdrq@qq.com>
Co-authored-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Bumping EDGE kernel to 5.14.y
Meson64:
- removing Odroid reboot shutdown patch since its probably not needed anymore
Rockchip64:
- removing Rockpi S. No interest to maintain this any further
- removing PBP suspend. Doesn’t align. Need inspection if some other way was mainstreamed
- temporally removing Orangepi R1
* Re-adding rockpis, pbp suspend, HFLPS170 wifi and cleanup
* Removing deprecated patch, fixing ap6256 wifi
* Re-enable Opi R1 plus, untest
* Add and fix Radxa Zero
* Add support for Radxa Zero
u-boot: new patch based on v2021.04
kernel: use patches sent to kernel list by chewitt
Tested with minimal build and cinnamon desktop:
- SD/eMMC, ok
- USB3, works in USB 2 mode
- HDMI, ok
- WiFi/BT, ok
Signed-off-by: Jack Ma <jack@radxa.com>
* Add build targets
* Cleanup and set min speed to 666000
* Change to WIP target
Support status undetermined
Co-authored-by: Jack Ma <jack@radxa.com>
- fix/dup bootscript, since new u-boot fixed n2_plus to n2-plus variant
- g12b: if n2 and edge, enable new u-boot; include non-SD version as SPI binary; write_uboot_platform_mtd
- Kernel DTS patch to add SPI-flash-enabled DTBs (slower eMMC), produces -spi .dtbs for n2 and n2-plus
- This could also be an overlay, but I dunno overlays.
- SPI+USB3 boot works with "pendrive" storage, untested with cabled/high power USB->SATA or USB->NVMe
Signed-off-by: Ricardo Pardini <ricardo@pardini.net>
* [ meson64 ] nanopi K2 ethernet fix
Fix NanoPi K2 Ethernet issues
Do some cleanup of the Meson64 folder while at it
* [ Meson64 ] revert accidental delete of C2 audio patch
* [ RK3328 ] boost GPU voltages to avoid starving Memory controller
The RK3328 has the dynamic memory controller and GPU on the same power rail.
Both are able to be dynamically clocked, both have opp tables.
- DMC min voltage: 1.075V
- GPU min voltage: 0.950V
On boards with only the GPU enabled, they memory controller can be under-voltaged.
On boards attempting to use both, the disagreeing opps result in the GPU failing to init
This patch applies a mV range opp to the GPU and fixed opps to the DMC on rk3328-roc-cc
to allow everything to get alongin all possible devfreq situations.
* [ rockchip64 ] clean up disabled patches
* [ rock64 ] add mali supply and remove unused 5V supply
- New Mali opps will keep from starving ram controller
- extra supply was causing dmesg errors.
As it shares enable with another supply and both were marked "always on"
I removed the second. Neither are referenced by any ports/devices.
- add mtd-utils to PACKAGE_LIST_BOARD.
- this should be enough to wipe Petitboot;
- more investigative work and u-boot building is needed for SPI+SATA boot
Signed-off-by: Ricardo Pardini <ricardo@pardini.net>
* Upgrading sunxi and sunxi64 EDGE to 5.13
Both variants tested for building
* Bump imx6 to 5.13.y
Tested
* Bump Jetson nano kernel to 5.13.y
* Move mvebu and mvebu64 edge to 5.13.y
* add DFI driver to provide hardware-based memory controller load
* adapted DRAM Memory Controller driver from rk3328, add necessary headers
* provide device tree overlays to enable DDR3 clock scaling
* adapted rk322x-box to allow DDR speed bin selection to user
* add ipb900 board gpio configuration
* add rtl8723cs driver to linux-current kernel
* use ddrbin v1.10 set to 330 Mhz at boot
* provide overlays for ddr3 at 330, 528, 660 and 800 mhz
* split emmc overlays to enable options by user choice
* modified rk322x-config to support dmc, emmc overlays
* remove optee trust os in favor of rockchip proprietary blob
for dmc functionality
* [ Rockchip64 ] Add Tinker Board 2
Using Mainline ATF/U-boot/Kernel
What works:
- USB (NOT TYPE C)
- PCIe
- LED's
- Ethernet
- HDMI
TODO:
- Type-C (fusb302 stuff)
- Type-C DP
- Regulators for big cores and GPU (sitting at default values now, no DVFS)
- Formware/driver for RTL8822CE
U-boot spits out some trash right after handing off control from SPL to U-Boot,
but doesn't impact boot. Needs looked at.
* [ rockchip64 ] remove unrelated changes to 369ab2decdc5a3c49973d8eeb783c95558dcc41a
* [ rockchip64 ] fix merge mess from symlinks step 1
* [ rockchip64 ] fix merge mess from symlinks step 2
* [ rockchip64 ] fix merge mess from symlinks step 3
* [ rockchip64 ] add asus driver for FAN53200 regulator
* Rename tinkerboard.conf to tinkerboard.wip
set as WIP for merge to trunk. Not ready for images just yet.
* revert rename to wip
wrong tinkerboard
* Rename tinkerboard-2.conf to tinkerboard-2.wip
prepare for merge to trunk. Not ready for images
* Tinker Board 2 GPU and Type-C
Support USB3 over type-C connector. display Port is still non-functional, framework in device tree however
Enable GPU
* Tinkerboard 2 DTS tweaks
Added this round:
Type-C DP now works
thermal sensors enabled (oops)
* Tinker Board 2 update config
-Adds wifi support for the RTL8822
* remove obsolete and broken patch for pine64so u-boot
This prevents the image from being built.
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
* sunxi: fix pine64so phy-mode and add yt8511 support
Pine A64-LTS board should have RGMII-TXID mode instead of RGMII-ID (the
Realtek PHY chip's RXID is sometimes broken).
In addition, new Pine A64-LTS v2 board will use YT8511 instead of
RTL8211E, which still works under RGMII-TXID mode.
Backport YT8511 driver and phy-mode change from linux-next.
Signed-off-by: Icenowy Zheng <icenowy@aosc.io>
Co-authored-by: Igor Pečovnik <igorpecovnik@users.noreply.github.com>
* Move Jetson patches to right location, add upstream patch
Also add upstream patch for sunxi
* Add upstream patches for Odroidxu4 current
* Add meson64 upstream patches
* Remove unnecessary patch
- this patch works when compiling and installing manually.
- the dtb with the changes is taken from the kernel and not from u-boot, therefore this patch does not work.
* Add enable-audio-codec-bpim2ultra-kernel.patch
- Add patch to enable audio codec in bpi-m2-ultra with h3 compatible
- Built and tested on Armbian Buster with Kernel 5.11.x (edge)
* Bpi-M2-Ultra enable audio on branches
- This finishes enabling the audio codec in the other kernels (5.4 - 5.10 - 5.11)
- Remove patch from u-boot folder (my bug)
- Compiled and tested in their respective kernel with debian buster
* Bpi-M2-Ultra enable audio kernel 5.12
- add patch to kernel 5.12
* Improve sunxi fix-a64-timejump.patch
Current patch still throws BUG:
WARNING: CPU: 2 PID: 31 at drivers/clocksource/arm_arch_timer.c:364 sun50i_a64_read_cntpct_el0+0x2c/0x38
Cause: Timer values are required to be exactly the same, while it may increase in between reads. Seems to happen especially during boot
Timer values are like: x2 : 0000000015014561 - x0 : 0000000015014567
Solution: Normalize timer values and return normalized value.
* Add patch for 5.12.y
Note. Small changes on upstream solution:
GENMASK(9, 0 ... -> GENMASK(8, 0 ...
Co-authored-by: Igor Pecovnik <igor.pecovnik@gmail.com>
* Bump imx6 edge to 5.12.y
* Bump odroid xu4 edge to 5.12.y
* Bump Jetson nano edge to 5.12.y
* WIP: Bump sunxi to 5.12.y
- create a copy of patches and make a new link
- wireless patches needs adjutement to 5.12
- some patches are still failing
* Adjust wireless patches to build on 5.12.y
* Remove template wireless patch
* Move Espressobin edge to 5.12.y
* Bump mvebu to 5.12
- patch 92-mvebu-gpio-remove-hardcoded-timer-assignment.patch was disabled @heisath
- adjusted wireless driver for 8723ds
* Adjust sunxi / sunxi64 edge to compile
* Fix kernel config for Jetson nano edge
* bump meson64 to 5.12.y
* bump rockchip edge to 5.12.y
* Bump rockchip64 to 5.12.y
* Bump rk322x edge to 5.12.y
@paolosabatino
* Cleanup im6, re-add 5.10.y patches since one is missing in 5.12.y
* Update odroidxu4 edge links
* rk322x: moved rk322x-edge to kernel 5.12, fix overlay compilation, retouched included kernel modules
* Cleanup: remove deprecated config
Co-authored-by: Paolo Sabatino <paolo.sabatino@gmail.com>
* Remove unnecessary patch
- this patch works when compiling and installing manually.
- the dtb with the changes is taken from the kernel and not from u-boot, therefore this patch does not work.
* Add enable-sound-pinecube-kernel.patch
- Add patch to enable audio codec in pinecube with v3s compatible
- Built and tested on Armbian Buster with Kernel 5.11.x (edge)
A long story short:
- patch 'meson64_add_shutdown_function_to_meson_drv.patch' (adds meson_drv_shutdown) added on 04/03/2021
- same patch renamed to '.patch_disabled' on 07/03/2021, since it broke what it was supposed to fix
- kernel 5.10.24 upstream merged similar version of patch, which indeed fixed some khadas stuff, but broke odroid, at least HC4
- now I add this patch to revert upstream.
- also remove the .patch_disabled to avoid confusion
- to say the truth, the HC4 is not reboot-proof, it still hangs sometimes. so this is not the end of it.
- ps: without this revert, HC4 hangs 100% of the time.
- ps2: did not test on edge 5.11 yet, but it looks like the same
Signed-off-by: Ricardo Pardini <ricardo@pardini.net>
There are already emmc nodes in rk3288-tinker-s.dts, adding these nodes
to rk3288-tinker.dtsi will cause emmc busy loop on original Tinkerboard.
Signed-off-by: Jack Chen <redchenjs@live.com>
Adds Xilinx Bootgen tools to build environment (alongside sunxi-tools, etc ..)
Adds new SoC family Zynq (7000 series, not MP/UltraScale)
Adds new board Avnet MicroZed
* add device tree overlays to handle boards with power regulation issues,
* fix misconfiguration issues in overlays
* add missing wireguard module for rk322x-current and -edge
* Re-enable network checksum TX offload on rockchip64 family
* Adjust gmac PBL setting for rk3399 legacy to fix network issues with MTU 1500
* Adjust gmac PBL setting for rk3328 in rockchip64-legacy to fix network issues with MTU 1500
Co-authored-by: Piotr Szczepanik <piter75@gmail.com>
- Enabled nfc on rk322x-dev and disable on rk322x-current (need further development)
- Tidied up rk322x-current device tree
- enabled nfc rockchip driver enabled in rk322x-dev kernel config
- Enabled EHCI controller in u-boot (added patch for inno-phy, device tree and config bits), better device detection for dwc2 usb otg port
- Removed SPL_FIT_GENERATOR from u-boot configuration, fixed .its file to use binman
- fixed rk322x its file (now includes dtb), reverted u-boot to v2020.10 and changed dev_* into log_debug() calls
* Adjust the RK808 buck step to improve stability
* Adjust vdd_log and enable vdd_center init voltage
For some reason, regulator-init-microvolt property under PMIC does not applied. Set the voltage on board file.
* Explicitly set vdd_log voltage for NanoPi M4V2 - possibly fix stability issues
* Regulator tweaks for NanoPi M4V2
* Add vdd_log to kernel and switch vdd_center back to 900mV
* Switch vdd_center to 0.95
* Make rk80x bucks voltage steps shorter to make the NanoPi M4V2 stable
* Tweak u-boot config for NanoPi M4V2
* Made the rk808 voltage steps configurable and configured 50mV for NanoPi M4V2
* Bump rockchip DEV 32b to 5.11.y
* Move mvebu64 DEV to 5.11.y
* Move Rockchip64 DEV to 5.11.y
Disabling general-rkvdec_nv15.patch - need more attention
* Add Meson DEV / Odroid C1
* Move OdroidXU4 DEV to tobetter's 5.11.y branch
* Bump sunxi DEV to 5.11.y
* Bump mvebu DEV to 5.11.y
* mvebu: Patch kernel to fix SATA write performance regression
Add an upstream patch that fixes SATA write speed regression that
appeared in kernel 5.9.
Signed-off-by: Maxim Mikityanskiy <maxtram95@gmail.com>
* Create 0001-Revert-block-simplify-set_init_blocksize-to-regain-l.patch
Co-authored-by: Igor Pečovnik <igorpecovnik@users.noreply.github.com>
- for manual/auto control of pwm fan (in the ODROID HC4 for example)
- original is at 8a63de9e2a
- discussion https://patchwork.kernel.org/project/linux-hwmon/patch/20201125163242.GA1264232@paju/
- I'd be much happier with a cooling map in the DTB, but this makes my fans spin correctly via fancontrol
- this affects all of meson64, but only for PWMs that are already exposed to hwmon
Signed-off-by: Ricardo Pardini <ricardo@pardini.net>
* Enable RTC (hym8563) for Station P1 in mainline (renaming DT in process)
* Switch Station P1 to rockchip64
* Fix linking for Station P1 device tree in legacy
* rockchip64: Added XHCI HCD USB TRB ENT quirk
On some xHCI controllers (e.g. Rockchip RK3399/RK3328/RK1808), they need
to enable the ENT flag in the TRB data structure to force xHC to
prefetch the next TRB of a TD.
Enable the quirk on RK3399 through device tree node properties.
Ported from Rockchip Linux 4.19
* add to dev branch
Polished and fixed xt-q8lv-10 rockchip-legacy dtb
Reworked act8846 reboot patch for rockchip-legacy (applies, but not yet working though)
Renamed kernel updated patches in rockchip-legacy directory adding "04-" prefix to make them uniform with existing
This patch (most likely taken from OpenWrt) only applies to ARMADA-XP
hardware, none of which is supported by Armbian.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
