https://gitlab.com/gitlab-org/gitlab/-/issues/19746

This reverts commit 1f5ddefa.

The cma=512M@3G setting is only necessary for LS1028A and breaks on imx
when an old u-boot version only enables 2 GB of RAM.

Johannes Schauer Marin Rodrigues authored 2 years ago and minute committed 1 year ago

When connecting a FTDI FT232RL usb to ttl uart adapter to the reform
with TX, RX and GND but not connecting that adapter to another machine
via usb, then the adapter will send garbage signals to the reform during
boot. These will interrupt autoboot and prevent the reform from booting
until either the RX/TX/GND are disconnected or the adapter is connected
via USB to another machine.

To prevent this from happening, do not interrupt autoboot on any key
press but require the space key to be pressed.

.gitlab-ci.yml: re-use build.sh to avoid maintaining the list of necessary environment variables in two places

Makefile: Use relative paths for debugging symbols.

See merge request !18

Vagrant Cascadian authored 2 years ago and Johannes Schauer Marin Rodrigues committed 1 year ago

The KBUILD_CFLAGS and KBUILD_AFLAGS variables are adjusted to use
-ffile-prefix-map and --debug-prefix-map, respectively, to use
relative paths for occurrences of __FILE__ and debug paths.

This enables reproducible builds regardless of the absolute path to
the build directory:

  https://reproducible-builds.org/docs/build-path/



Series-to: u-boot

Signed-off-by: Vagrant Cascadian <vagrant@reproducible-builds.org>

.gitlab-ci.yml: set SOURCE_DATE_EPOCH to the timestamp of the last git commit...

See merge request !17

.gitlab-ci.yml: set SOURCE_DATE_EPOCH to the timestamp of the last git commit for reproducible builds

I didn't author this patch, just committing it. See below.

Co-authored-by: Patrick Wildt <patrick@blueri.se>

board/boundary/nitrogen8m_som/nitrogen8m_som.c: remove printf("set_env_vars()")

See merge request !14

Mainscreen turn on

See merge request !12

This seems to cause data abort panics on "mmc rescan"
command.

Also remove the pinmuxing options on the GP_EMMC_RESET
and USDHC2_RST signals. We keep it how spl left it.

Sigrid Haflínudóttir authored 2 years ago and cinap_lenrek committed 2 years ago

In dual mode (with lcdif and hdmi both used) Linux would end up
with built-in display flashing randomly at the top and otherwise
black screen. Instead of disabling the clocks, shut down the bridge
(SN65DSI86) exactly by spec and leave the clocks be.

Also increase the delay on init to make sure the bridge is up as
sometimes it did not respond in time, causing the display not work
in U-Boot.

The LCDIF needs to be shut down in a particular way.
Disable display clocks, reset and power down mipi dsi.
Disable pwm and edp bridge.
Also stop USB in any case.

Linux kernel mode setting seems to fail to
properly reset the mipi core, causing the
screen to be shifted to the right.

As a work around, we reset the lcdif and
mipi core in a board_quiesce_devices()
function.

GPIO3_20 is the enable for the SN65DSI86,
so power cycling the sdcard would have the
screen turn off.

Modifier keys seem to be broken in boot protocol
of the reform keyboard. But it sends scancodes
for the modifier keys, so just emulate the modifier
key state bits.

Co-authored-by: q66 <daniel@octaforge.org>

Update lpddr4_timing and blobs from upstream

See merge request !7

disable the saveenv command

See merge request !8

set fdtfile environment variable

See merge request !10