This adds kernel support support for Sagemcom F@st 2704 wireless ADSL
router.
It's a BCM6328-based 802.11n wireless router with USB port and ADSL2+
modem equipped with 64 MiB RAM and 8 MiB flash.
Signed-off-by: Marcin Jurkowski <marcin1j@gmail.com>
Patchwork: http://patchwork.openwrt.org/patch/4266/
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39671
Add kernel support for the BCM963269BHR board, USB and Ethernet switch,
LEDs and buttons are working.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39363
Allow using the HSSPI controller since it works just fine, as well as
the integrated switch core, since that one works too.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39362
Update a few devices with their actual port numbers, especially those
with two ports.
Should allow properly using the second port on the Neufboxen and HW553
and HW556.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39325
Allow setting the number of available usb host ports for boards with
additional sanity checks to allow using the second port on devices
where it is available.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39324
USB will fail to work if compiling with BCM6318 only if BCM63XX_HAS_EHCI
and BCM63XX_HAS_OHCI are not selected.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39322
Return the same clock for usbh and usbd on BCM6318 to prevent the clock
getting disabled despite being supposed to be enabled, e.g. by the
following sequence:
clk_enable(usbh); <- usbh->usage = 1, enabled
clk_enable(usbd); <- usbd->usage = 1, enabled (no effect)
clk_disable(usbd); <- usbd->usage = 0, -> disabled!
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39321
Changeset r39302 updated a board section and added twice .has_ehci0,
this is not necessary, so fix it.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39304
BCM6318 has a special initialization sequence which involves touching
PLL control registers in the USBH_PRIV register space, as well as
toggling a bit the Simulation control register.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39302
Enable the use of the USB gadget device at the board level so we can use
bcm63xx_udc on this board.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39292
Also fixing a build warning due to a misplaced parenthesis in the
previous submission.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39291
Very similar to the BCM96318REF board except for the name and the
SDRAM/DDR configuration which is totally transparent for us.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39286
Add basic support for the BCM96318REF board which is comprised of:
- HS-SPI flash support
- Ethernet switch support
- USB slave device support
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39283
Provide clock toggling for the USB host and slave controller as well as
for the Ethernet Switch. The bcm63xx_enetsw driver is changed not to use
request_mem_region() as the Roboswitch register overlaps with the
Roboswitch IUDMA channels, hence making the driver return -EBUSY. A
future fix might be to use variable size RSET_ENETSW sizes depending on
the chip we are targetting.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39282
While adding support for BCM6345 Ethernet, some changes in the macros
ENETDMA{C,S} where introduced which now make the bcm63xx USB gadget
driver fail to build. Fix this.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39279
A missing condition check in dev-hsspi.c prevented the controller from
being registered on 6318 where it is fully functional.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39278
varid might both unused and unitialized when building for e.g: bcm6318,
fix these two warnings turned into errors.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 39277
BCM63XX's ohci controller does not need the frame number shift, but that
was guarded with a PPC symbol. Fix this by making the no-fix generally
available.
This is a regression from the previous conversion to ohci platform driver.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39274
CFE seems to leave the SPI flash mapping in an invalid state after
loading the kernel on some reference boards, so fix it up on boot.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39273
Add inital support for BCM6318, but keep it disabled for now until
most things are supported.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39272
Add initial support for the BCM63268 family of SoCs, but keep it
disabled for now as most things don't work yet.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39271
Some SoCs have variants which are mostly the same, but use a different
chip id (or not). Add code for detecting them and handling them as
their standard counterparts.
This adds support for e.g. BCM6369.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39269
Remove the non runtime detect code since its effectiveness is dubious
and almost never used.
Also update affinity patches to work on top of it.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39268
Update bmips cleanup patches with upstream submission and backport a few
bmips fixes.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39267
Fixes generic SPI flash accesses on BCM6358/BCM6368. BCM6338 still needs
additional fixes for M25P80, so it remains broken for now.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39266
Fixes e.g. SPI flash access on BCM6358/BCM6368. Transfers < 256 bytes
are not affected.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39265
Update the HSSPI driver with the upstream submitted one that has a
workaround for the auto cs down issue.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 39264
- Some fixes for the incorrect GPIO assignment of leds on Huawei HW556.
- Addition of unassigned (LAN) leds which can be found on the back of the device.
Signed-off-by: Angga Reza Fardana <angga@norture.com>
Patchwork: http://patchwork.openwrt.org/patch/3744/
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 37102
In order to support both normal images and initramfs, ensure that each
target sets KERNELNAME properly so that the generic kernel building code
can copy the corresponding files over $(KDIR) with the appropriate
extension. Update the various paths to the kernel and wrapper images
from $(LINUX_DIR)/arch/$(ARCH)/boot/$(foo) to $(KDIR)/$(foo).
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 37049
We currently used 16 DMA Ethernet channels, but the hardware only
supports up to 8 (with 2 per enet and the remaining for USB slave).
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 36979
Cable modem devices currently poke at some random location in RAM and
may end up having an invalid MAC address which is not suitable. Provide
a default one which is sane.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 36957
Cable Modem devices cannot load an ELF file directly so we need to
provide them with a HCS image even when targetting initramfs.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 36954
$(KDIR)/vmlinux is already a binary image, there is no need to create
another one.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 36949
Check bootloader argument in mtd.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
[jogo@openwrt.org: split into two patches, one for detection,
one letting bcm63xxpart use it.]
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 36659
Don't build the CRAMFS file system module in the kernel, and save some bytes.
No bcm63xx board needs this obsolete file system, neither those ones using
the Redboot bootloader (liveboxes). Thus, there is no need to build it, since it has
been replaced long time ago by squashfs.
Signed-off-by: Daniel Gonzalez <dgcbueu@gmail.com>
SVN-Revision: 36616
And while at it, also include a define check for BCM6328, thanks to
Gabor for spotting this.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 36506
Use a different length for calculating the nvram checksum on hauwei
boards. Fixes boot on HW553 and HW556.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 35533
The first preinit script 03_do_brcm63xx.sh does the right job of
exporting variables for the other subsequent preinit scripts to use
them.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
SVN-Revision: 35504
Actually this board is the very same as the rta770bw,
where the additional 'b' within the name just
indicates 'Annex B'. The ADSL Modem itself is able
to handle both, Annex A as well as Annex B -
the loaded firmware makes the only difference
SVN-Revision: 35500
Switch to using the B53 switch driver where appropriate.
Use a non-tagged default config where it isn't confirmed to be
working.
Remove the switch configuration in preinit as the switch will now
be configured by the driver to use port isolation to prevent leakage
between lan and wan.
While at it, also provide a switch config for Neufbox 6.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 35307
The SPI driver is needed for supporting to boot from SPI flash on BCM6358
and BCM6368, and for SPI connected switches.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 35304
Allow bcm63xxpart to receive a caldata offset if calibration data is contained in flash.
Tested-by: Álvaro Fernández Rojas <noltari@gmail.com>
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
SVN-Revision: 35010
This patch adds support for defining rt2x00 eeprom data.
This eeprom data can be extracted from flash or loaded directly from a firmware file.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
SVN-Revision: 34322
Some boards need specifying LED pin in order to get it working.
For example, in board HW556 (Huawei HG556a) "led_pin" must be "2".
By default led_pin is "0", so dsl_274xb_rev_f is changed to "-1".
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
SVN-Revision: 34321
* Accept transfers without bits_per_word set.
* Work around the inability of the hardware of keeping CS asserted.
Signed-off-by: Jonas Gorski <jogo@openwrt.org>
SVN-Revision: 34320
This reverts commit r34230, as it is causing build failures:
make[3] -C package/mtd compile
imagetag.c:35:32: fatal error: linux/bcm963xx_tag.h: No such file or directory
compilation terminated.
make[4]: *** [imagetag.o] Error 1
Any patches modifying exported kernel headers need to reside in generic
as toolchain/kernel-headers only applies generic patches.
SVN-Revision: 34255
Fixes the following error:
drivers/usb/host/ohci-bcm63xx.c: In function 'ohci_bcm63xx_start':
drivers/usb/host/ohci-bcm63xx.c:33:3: error: implicit declaration of function 'err' [-Werror=implicit-function-declaration]
Reported-by: Álvaro Fernández Rojas <noltari@gmail.com>
SVN-Revision: 34138
Some ath9k caldatas need to be "endian checked".
This patch allows selecting wether to enable endian check or not.
Endian check is disabled by default.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
SVN-Revision: 34105
Export bcm963xx_tag.h so mtd can use it without using kernel includes.
Signed-off-by: Catalin Patulea <cat@vv.carleton.ca>
[jonas.gorski@gmail.com: fixup bcm63xx patches, completely remove header
from original location, refresh patches, use a more matching patch
number, port to 3.6]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
SVN-Revision: 34103
Jonas Gorski
Florian Fainelli
Gabor Juhos
Hauke Mehrtens
Felix Fietkau
John Crispin
Mirko Vogt
Jo-Philipp Wich
Imre Kaloz