* add xpadneo driver for Xbox One wireless gamepad
* xpadneo: add kernel patches
should prevent generic hid module to claim the gamepad
* xpadneo: add additional settings to modprobe conf
* Add event service package and a systemd service for OGA
This is a service that listens for certain keys on certain devices and
injects commands to Retroarch.
* Update Retroarch and add command support
This adds a new function hook that can be defined for devices that have
slightly different cmdline within the same configuration.
Remove vcchost regulator since it is not used and just makes the rumble
go crazy during kernel boot.
This reenables Wifi (MT7601U) that was deleted by a previous commit.
Also enables the secondary SD card reader by fixing the voltages.
Finally re-enable low power OPPs that were disabled. This brings power
savings on idle mode and low demanding emus (in particular voltages are
lower for those points, reducing power consumption significantly)
* Fix OGA builds
No need for the u-boot patch, update the INI files instead.
Added the new userspace mali G31 blob and update the build script (we
pull headers still from Rockchip repo).
Add the RTL8821 patch to fix linux build with newer GCCs too.
This makes OGAs bootable again (tested it on OGA Super)
* Adding aarch64 support
This is essentially an OGA clone with a bigger (640x480) non-rotated
screen. Requires a joypad kernel patch (a copy of OGA's + small bits) and a new
DTB.
Changed a bit the way splash screens work, now you can specify proper
resolutions (instead of just vertical res) so that it works well for
cases like rotated screens.
Tweaked Retroarch to ship both drivers (DRM + GO2 DRM) and the GO2
driver to inhibit itself in case the console has a non-rotated display.
This should work for any console as long as the resolution is reported
correctly. The regular DRM driver should have 1 frame less latency!
What's not working: Console is still rotated (minor issue) and kernel
panics on poweroff :) Debugging it, but I'll need to solder the UART for
that.
This console requires a separate DTS and a USB controller config.
U-Boot cannot reliably distinguish the two consoles since the HW ID
for it is the same as OGA v1.1 (looking into the schematics did not
yeld any GPIO differences that could be used for that).
I plan to dig a bit more into u-boot, perhaps enabling USB and
looking for the USB controller could be a way to differentiate them.
In the meantime this just adds another OGA flavor, which adds little
overhead to the build system.
Another TODO is rumble support, mapped to PWM1, which is not trivial
to support in RetroArch.
* fix linux u-boot build under host gcc 10
Debian 11 has gcc 10, thus Lakka build failed.
port upstream patch to fix build issue is Debian 11.
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e33a814e772cdc36436c8c188d8c42d019fda639
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* melonds depends on mesa on x86_64
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* python is python2
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* Gamegirl: fix kernel build
vmlinux link to ko, thus fail to build.
config related driver to ko.
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* Gamegirl: set PKG_TOOLCHAIN to manual
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* Gamegirl: build GL
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* Gamegirl: fix bootload build
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* fix chailove build issue and enable it
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
* build_all: remove some cache to make sure rebuild pass
when the previous build stops in the middle, must remove more to pass build.
Signed-off-by: Zhang Ning <832666+zhangn1985@users.noreply.github.com>
