In the upstream Linux kernel, the Mediatek Genio 350-EVK board has a
devicetree named "mt8365-evk.dts" and not "mt8365-genio-350-evk.dts",
so fix the name value in include so the fdtfile in default env is
configured with the correct name.

Signed-off-by: Louis-Alexis Eyraud <louisalexis.eyraud@collabora.com>

Signed-off-by: Julien Massot <julien.massot@collabora.com>

Signed-off-by: Julien Massot <julien.massot@collabora.com>

Signed-off-by: Julien Massot <julien.massot@collabora.com>

Signed-off-by: Julien Massot <julien.massot@collabora.com>

Julien Massot authored 2 months ago and Ariel D'Alessandro committed 2 months ago

The Radxa NIO 12L use an UFS.

The UFS is hardware partionned in 3 LUNs
- LUN0 with BL2 (4MiB)
- LUN1 for u-boot env (4MiB)
- LUN2 for the others software

Julien Massot authored 2 months ago and Ariel D'Alessandro committed 2 months ago

The lk platform is sometimes different than the drmam_plat.

Julien Massot authored 2 months ago and Ariel D'Alessandro committed 2 months ago

The upstream u-boot doesn't have support for ufs mtk_image

Julien Massot authored 2 months ago and Ariel D'Alessandro committed 2 months ago

Update the semantic, the FIP offset is not really related
to the MMC or UFS. It's just an offset in the BL2 storage.

imported from
https://git@github.com/radxa/u-boot.git
sha1: 560dd037fe48cf6bacbdd2679480de9d13ee8ef9

* Kernel device tree paths contain an extra dtb extension. Remove it.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

This project has been cloned from lava mediatek-boot [0]. Update
references and links.

[0] https://gitlab.collabora.com/lava/mediatek-boot/



Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

This commit extend the following fixes to all Genio EVK boards:

* Fix ramdisk image overlapping
* Match upstream kernel device tree blob names

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Genio 350 EVK board often timeouts on a tftp boot sequence reporting:

```
The remote end did not respond in time.
```

Set the default timeout value to 20s, which is enough for these cases.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

This commit picks upstream patch series refactoring the USB ethernet
driver to allow manually binding the gadget driver.

As mentioned in commit ed62305b ("u-boot: GENIO: configs: 510/700/1200
EVK: Disable USB Ethernet Gadget") USB Ethernet Gadget is causing
enumaration errors on the host side.

To workaround this issue, upstream [0] approach is followed, i.e.:

* drop usb_ether_init() from board files, as it is deprecated and
  causing the enumeration errors.
* use bind command to enable usb ethernet when required.

However, while binding the driver, a NULL deref bug was hit. See the
following error log testing on Genio 700 EVK:

```
=> bind /soc/usb@11200000/ssusb@11200000 usb_ether
=> dhcp
ethernet@11021000 Waiting for PHY auto negotiation to complete......... TIMEOUT !
phy_startup() failed: -110FAILED: -110using mtu3-gadget, OUT ep1out IN ep1in STATUS ep2in
MAC de:ad:be:ef:00:01
HOST MAC de:ad:be:ef:00:00
RNDIS ready
dev_get_priv: null device
"Synchronous Abort" handler, esr 0x96000004
elr: 000000004c060200 lr : 000000004c05f9a0 (reloc)
elr: 00000000fff50200 lr : 00000000fff4f9a0
x0 : eafffffeeafffffe x1 : 0000000000000000
x2 : 00000000ff9eedb0 x3 : 000000000000007b
x4 : 00000000ff9eee70 x5 : 00000000ff9f34a0
x6 : 000000000000007b x7 : 00000000ff9e9240
x8 : 0000000000000010 x9 : 0000000000000008
x10: fffffffffffffff0 x11: 0000000000000000
x12: 0000000000000002 x13: 00000000ff9f96c0
x14: 0000000000000000 x15: 0000000000000200
x16: 00000000fff35bac x17: 0000000000000000
x18: 00000000ff9eedb0 x19: 0000000000000000
x20: 0000000000000bb8 x21: 000000000002773b
x22: 0000000100001000 x23: 000000010000b000
x24: 000000010000b000 x25: 00000000ff9fdf50
x26: 0000000000000000 x27: 0000000000000000
x28: 0000000000000000 x29: 00000000ff9e9230

Code: a90153f3 aa0103f3 f90013f5 f9400820 (b9500000)
```

[0] https://patchwork.ozlabs.org/project/uboot/patch/20240726083102.380719-1-admin@hifiphile.com/



Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Disable USB Ethernet Gadget on GENIO 510/700/1200 EVK boards as this is
messing up the USB host side. This boards uses ethernet to boot, so the
USB gadget is not required.

Due to backported upstream commit 718f1d414eb ("usb: gadget: ether:
Handle gadget driver registration in probe and remove"), the USB
ethernet gadget is initialized early, causing enumaration errors on the
host side.

This has been reported upstream in [0]:

```
From the host's point of view, a device is ready to be enumerated.
However, since dm_usb_gadget_handle_interrupts() is only called when
ethernet function is started, at this stage USB events are not managed,
host's enumeration attempts will fail and resulting error like:
    usb 1-1: new high-speed USB device number 50 using xhci_hcd
    usb 1-1: device descriptor read/64, error -110
    usb 1-1: device descriptor read/64, error -110
    usb usb1-port1: attempt power cycle
```

However, patch fix wasn't accepted as upstream u-boot handles USB
ethernet gadget in a different way now, requiring manual binding the
gadget driver. From [0] thread:

```
[...] usb_ether_init() should not be used anymore.

Instead, to enable usb ethernet, we should manually bind the UDC driver
to the usb_ether gadget.
```

[0] https://patchwork.ozlabs.org/project/uboot/patch/20240726083102.380719-1-admin@hifiphile.com/



Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

As previously done for genio-700-evk board, this extends fixes for the
genio-510-evk target.

* Match kernel device tree path names.
* Current kernel size is too big, resulting in overlapping areas:

    ERROR: RD image overlaps OS image (OS=0x45000000..0x47ef0000)

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Enable MediaTek watchdog timer as it comes enabled from BL2 TFA.
Otherwise, this causes a reset after a certain amount of time.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Currently, `booti` command doesn't seem to accept the u-boot header for
ramdisks, and LAVA can't currently tftp ramdisks without it.

Enable CONFIG_LEGACY_IMAGE_FORMAT for RAMDISK support.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

The Genio 350 EVK board reports the `dhcp` command constantly causing a
Synchronous Abort, like the following:

```
=> dhcp
using mtu3-gadget, OUT ep1out IN ep1in STATUS ep2in
MAC de:ad:be:ef:00:01
HOST MAC de:ad:be:ef:00:00
RNDIS ready
high speed config #2: 2 mA, Ethernet Gadget, using RNDIS
USB RNDIS network up!
BOOTP broadcast 1
DHCP client bound to address 192.168.202.106 (3 ms)
"Synchronous Abort" handler, esr 0x96000061
elr: 000000004c023958 lr : 000000004c01a200 (reloc)
elr: 00000000fff12958 lr : 00000000fff09200
x0 : 00000000ff9f68b0 x1 : 0000000000000020
x2 : 00000000fffb3540 x3 : 0000000000000001
x4 : 0000000000000030 x5 : 00000000ff9e9594
x6 : 0000000000000002 x7 : 00000000ff9e9540
x8 : 00000000fffffff0 x9 : 0000000000000008
x10: 000000000000000a x11: 0000000000000006
x12: 000000000001869f x13: 00000000ff9e8eb0
x14: 000000000000002e x15: 0000000000000000
x16: 00000000fff813e4 x17: 0000000000000000
x18: 00000000ff9eddc0 x19: 0000000000000020
x20: 00000000fffb3500 x21: 00000000fffa3705
x22: 0000000000000003 x23: 0000000000000200
x24: 000000000000000e x25: 00000000ff9fa110
x26: 0000000000000000 x27: 0000000000000000
x28: 0000000000000000 x29: 00000000ff9e9430

Code: 540001c0 a9408c01 f9400c02 927ef421 (f9000c62)
Resetting CPU ...
```

From a quick research, the abort happens in `drivers/usb/gadget/ether.c`
rx_submit() tries to dereference a freed dev struct pointer.

This backports upstream commit 718f1d414e ("usb: gadget: ether: Handle
gadget driver registration in probe and remove"), which apparently fixes
the sequence of des/registration.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

These jobs don't have any dependencies, let's run them as soon as
possible independently.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Lightweight runners are kept on standby while for all other runners the
VMs are provisioned on demand. So lightweight runners can start faster,
more suitable for these short-running jobs.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Improves build speed a little bit.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

This allows jobs to be run as soon as the required artifacts are
available.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Genio 510 EVK board compilation is picking the wrong libbasea binary.
Fix this by setting the right one for this SoC.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

* Kernel device tree is mediatek/mt8390-genio-700-evk.dtb.
* Current kernel size is too big, resulting in overlapping areas:

    ERROR: RD image overlaps OS image (OS=0x45000000..0x47ef0000)

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Currently, Genio 510 EVK keeps going into bootrom mode rather than
starting u-boot. The lk.bin artifact differs from the one shipped in
Mediatek pre-built images, which works just fine. Let's match the same
version.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Avoid using latest, as it might change at some point.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Having a different directory hierarchy for the final artifacts can be
confusing. Let's remove it and have the same one across all targets.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

We don't need the single binaries, so let's not keep them to avoid
confusion.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

This extends support to store and load FIP image from mmc boot partition
to all Genio 350/510/700/1200 EVK platforms: mt8188/8195/8365

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

We now have multiple jobs of the same kind, let's group them in
different stages.

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>

Signed-off-by: Ariel D'Alessandro <ariel.dalessandro@collabora.com>