Some (older) CFEs are loaded at 0x80401000 and ramdisks are loaded at
0x80010000, which means that ramdisk size limit is 0x3F1000 (almost 4M).
Therefore, current ramdisks (~4MB) are overwritting CFE in these devices,
which results in a crash.
This commit changes the address where ramdisks are loaded to 0x80a00000,
which is the same address where kernel is loaded when booting from the flash.
Therefore, lzma-loader will now be loaded at 0x80a00000, but it will still
decompress the kernel at 0x80010000.
Tested with huawei,hg556a-b, which has its CFE loaded at 0x80401000.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Apparently, Sercomm allows loading a BCM WFI image via CFE, but this image
destroys "serial" and "protect" nand partitions, which is wrong.
It will also set both bootflags to the same value, which causes booting
issues with cferam (cferom will alternatively boot from cferam1 or cferam2
each time the device is rebooted).
Now that OEM Sercomm images are supported it's time to remove this hacky
cfe.bin image support.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
BCM6368 and newer devices are compatible with any lzma compression parameters.
Add a new legacy device definition and use it on BCM6358 and older devices.
Compressed kernel size is reduced by ~1.35%.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Allows to keep a backup firmware in case active firmware is corrupted.
Also fix hsspi address warning.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
This images can be flashed from the official firmware, as opposed to CFE
images, which can only be flashed from CFE and require opening the case.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Support Sercomm firmware partition split.
WFI partition must be defined after bootflag partitions in order for the
parser to properly find bootflag1 and bootflag2 partitions.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
ADB P.DG A4001N A-000-1A1-AX a.k.a. Telecom Italia ADSL2+ Wi-Fi N (AGPWI)
has the same PCB as the OpenWrt's ADB P.DG A4001N1 with LEDs connected
to different GPIO PINs in active low configuration.
OpenWrt's ADB P.DG A4001N image is made for the ADB P.DG A4001N A-000-1A1-AE.
It has different LEDs configuration and flash size/layout
w.r.t the ADB P.DG A4001N A-000-1A1-AX.
Hardware:
* Board ID: 96328avng
* SoC: Broadcom BCM6328
* RAM DDR2-800: 32 Mbyte - winbond W9725G6KB-25
* Serial flash: 16 Mbyte - MXIC MX25L 12845EMI-10G
* Ethernet: 4x Ethernet 10/100 baseT
* Wifi 2.4GHz: Broadcom Corporation BCM43224/5 Wireless Network Adapter (rev 01)
* LEDs: 2x Power, 2x ADSL, 2x Internet, 2x Wi-Fi, 2x Service
* Buttons: 1x Reset, 1x WPS (named WiFi/LED)
* UART: 1x TTL 115200n8, TX NC RX, on J5 connector (short R192 and R193)
NC GND NC
Installation via CFE:
* Stock CFE has to be overwriten with one for 96328avng boards that can upload
.bin images with no signature check (cfe-A4001N-V0000_96328avng.bin)
* connect a serial port to the board
* Stop the boot process after power on by pressing enter
* set static IP 192.168.1.2 and subnet mask 255.255.255.0
* navigate to http://192.168.1.1/
* upload the OpenWrt image file
Signed-off-by: Daniele Castro <danielecastro@hotmail.it>
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
This extracts the model part of the board name and uses it for the
LED string identifiers in 01_leds. As this makes statements more
generic, it will allow to merge more cases in the future.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
While "ok" is recognized in DT parsing, only "okay" is actually
mentioned as valid value. Replace it accordingly.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Introduce support for generating JFFS2 CFE partition tags.
This is used in NAND devices in order to verify the integrity of the JFFS2
partition.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
There are older devices which require overriding the RGMII ports, so this
shouldn't be limited and forced to BCM63268.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
NAND is used as extra storage on this device.
Signed-off-by: Daniel Gonzalez Cabanelas <dgcbueu@gmail.com>
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
lzma-loader uart output wasn't working on BCM3380/BCM6362 because these
SoCs have the same processor ID.
Let's use CHIP_ID for establishing the UART base address.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Until now only HW modded SPI flash version was supported.
BCM6328 with 64M RAM and 128M NAND.
More info: https://openwrt.org/toh/sercomm/ad1018
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
BCM63169 with 128M RAM, 128M NAND and BCM53125 switch.
Switch is connected by HSSPI to CS5.
More info: https://openwrt.org/toh/comtrend/vg8050
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
BCM6362 with 64M RAM, 32M NAND and BCM53125 switch.
Switch is connected by MMAP, which is currently unsupported (no VLANs).
More info: https://openwrt.org/toh/netgear/dgnd3700v2
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
The bcm6345-periph-intc driver only targets a single CPU at a time, even
if the notional affinity is wider. Let's inform the core code about this.
This patch gets rid of the kernel message:
"genirq: irq_chip bcm6345-periph-intc did not update eff. affinity mask
of irq 52"
Signed-off-by: Daniel Gonzalez Cabanelas <dgcbueu@gmail.com>
In file included from ./arch/mips/include/asm/io.h:34,
from ./arch/mips/include/asm/mmiowb.h:5,
from ./include/linux/spinlock.h:60,
from ./include/linux/irq.h:14,
from drivers/irqchip/irq-bcm6345-ext.c:10:
drivers/irqchip/irq-bcm6345-ext.c: In function 'bcm6345_ext_intc_of_init':
./arch/mips/include/asm/mach-bcm63xx/ioremap.h:48:9: warning: 'base' may be used uninitialized in this function [-Wmaybe-uninitialized]
return is_bcm63xx_internal_registers((unsigned long)addr);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/irqchip/irq-bcm6345-ext.c:255:16: note: 'base' was declared here
void __iomem *base;
^~~~
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
drivers/irqchip/irq-bcm6345-periph.c: In function 'bcm6345_periph_irq_handle':
drivers/irqchip/irq-bcm6345-periph.c:55:21: warning: 'block' may be used uninitialized in this function [-Wmaybe-uninitialized]
struct intc_block *block;
^~~~~
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
While "ok" is recognized in DT parsing, only "okay" is actually
mentioned as valid value. Replace it accordingly.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This commit removes the target-specific diag.sh script. This way, the
generic one is used for the target, which uses DT-aliases to specify the
LEDs used.
Though generic diag.sh allows to use different LEDs to indicate different
states, this patch just moves the old assignment and does not try to
"improve" the assignment by using additional colors.
However, individual proposals to do so are welcome.
For the few cases where status_led2 was used in old diag.sh, only the
primary LED was migrated.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Seems stable after 6 days of testing on some of my devices.
Let's switch to 5.4 in order to get more feedback.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Device with 4 MiB flash and 32 MiB RAM won't be able to run OpenWrt in
a sufficient manner without tweaks, so don't build images for them by
default.
This includes all BCM6338, BCM6345 and BCM6348 "generic" devices,
as there are no supported devices of these with more than that.
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
Since stable kernel for this target is still 4.14, kernel 4.19
has never been used much (and actually was broken for some devices).
So, since we bump testing kernel to 5.4, there is no real need to
keep 4.19 and have an additional version to care about.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This uses 5.4 as testing kernel. Since 4.19 has not seen broad
testing yet, just keep 4.14 as stable kernel until 5.4 is ready.
Tested on Comtrend AR-5387un (Thanks to @Noltari).
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
In kernel patch 92ce7e83b4e5 ("driver_find_device: Unify the match
function with class_find_device()") the arguments of functions used
with driver_find_device are adjusted. Do the same for our local
user.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The helper syscon_regmap_lookup_by_pdevname has been removed in kernel patch
29d14b668d2f ("mfd: Remove unused helper syscon_regmap_lookup_by_pdevname")
due to lack of users.
However, we use this function in our local pinctrl drivers for BCM6358 and
BCM6368. Thus, we have to add it locally.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
In kernel 5.4 -Werror=implicit-fallthrough is treated as error:
arch/mips/bcm63xx/cpu.c: In function 'detect_cpu_clock':
arch/mips/bcm63xx/cpu.c:158:2: error: this statement may fall through [-Werror=implicit-fallthrough=]
{
^
arch/mips/bcm63xx/cpu.c:179:2: note: here
case BCM6328_CPU_ID:
^~~~
cc1: all warnings being treated as errors
This can be fixed by adding "Fall through" as a comment where intended,
and has to be fixed by returning a proper default otherwise.
In case of the default clock frequency for BCM6318 we fixed this by
returning the default value taken from BCM6328 and BCM6362.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
This patch follows the other patches that added the watchdog
core to various (armvirt, malta, ath79, ...) targets that
have been hit by the following build error:
Package kmod-hwmon-sch5627 is missing dependencies for the following libraries:
watchdog.ko
In theory, we could have just added the CONFIG_WATCHDOG_CORE=y
to the Kconfig variable of kmod-hwmon-sch5627's package definition.
This would have forced the watchdog core to be builtin and less
architectures would need to be updated. But we might as well follow
through here.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Refresh patches to make them apply to kernel 5.4.
The removed patches have been merged upstream.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Copy config and patches to kernel 5.4.
make kernel_oldconfig has been run on 4.19 beforehand.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Place DEVICE_VARS assignments at the top of the file or above Device/Default
to make them easier to find.
For ramips, remove redundant values already present in parent file.
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
[do not touch ar71xx, extend commit message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
There is no such role as target maintainer anymore, one should always
send corresponding changes for the review and anyone from the commiters
is allowed to merge them or eventually use the hand break and NACK them.
Lets make it clear, that it is solely a community doing the maintenance
tasks.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Acked-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Piotr Dymacz <pepe2k@gmail.com>
This change makes the names of Broadcom targets consistent by using
the common notation based on SoC/CPU ID (which is used internally
anyway), bcmXXXX instead of brcmXXXX.
This is even used for target TITLE in make menuconfig already,
only the short target name used brcm so far.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
