4.1 TI Delta: https://github.com/RobertCNelson/ti-linux-kernel/compare/2803cc758722bcfe8fdd2c3ee813bd5e390354b2...1f0ed8c5eb8c05d613458b8cc3c35ab28485a611



Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

When compiling a kernel without CONFIG_MULTI_CACHE enabled the
dma access functions end up not being exported. Fix it.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>

Add the node into the am33xx.dtsi file for the SGX GPU
that is found in some variants of the SoC.

Signed-off-by: Darren Etheridge <detheridge@ti.com>

Use pdata_quirks to provide platform data to the sgx driver.
The data that is provided includes:

1) Function pointers for the driver to use to reset the h/w block.
2) The reset name that matches with what is used in hwmod.

Signed-off-by: Darren Etheridge <detheridge@ti.com>

Add a vsync callback registration API that is identical to what was done
for da8xx-fb.c.

Need to find if there is a better way using the DRM infrastructure from
kernel space.  Either that or change the userspace window manager stuff in
the gpu libraries to make use of the DRM provided syncronization
mechanisms.

Signed-off-by: Darren Etheridge <detheridge@ti.com>

Increase the size of the buffer that is created in the fbdev emulation
helpers.  And fill in the var structure with the amount that was allocated.

Signed-off-by: Darren Etheridge <detheridge@ti.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

I have encountered the same issue(s) on A6A boards.

I couldn't find a patch,  so I wrote this patch to update the device tree
in the davinci_mdio driver in the 3.15.1 tree, it seems to correct it. I
would welcome any input on a different approach.

https://groups.google.com/d/msg/beagleboard/9mctrG26Mc8/SRlnumt0LoMJ



v4.1-rcX: added hack around CONFIG_OF_OVERLAY
v4.2-rc3+: added if (of_machine_is_compatible("ti,am335x-bone")) so we do
not break dual ethernet am335x devices

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

The cpsw device runs at 100/10mbits. Deferring a read for 96k
wipes out the dcache thrice over. It is better to take more interrupts
as they come, at 100mbit.

IoT devices such as the beaglebone black come with a default PHY
that can only do 100Mbit, and a fairly large tx ring. BQL moderates
the effect of that considerably when saturating the network at that
speed.

Tested on a beaglebone black.

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Nathaniel Lewis <linux.robotdude@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Remove CONFIG_EXPERIMENTAL dependency on CONFIG_GPIO_OF_HELPER as CONFIG_EXPERIMENTAL has been removed with 3.9

Signed-off-by: Tobias Müller <Tobias_Mueller@twam.info>

Signed-off-by: Adrian Remonda <adrianremonda@gmail.com>
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

BeagleBone DTS : Enable run-time pinmux for HDMI Add cape-universal-hdmi pin
info to am335x-bone-common-universal.dtsi Edit hdmi dtsi include files to use
new mode= setting to set HDMI mode at startup, leaving pinmux configurable at
runtime.

Signed-off-by: Charles Steinkuehler <charles@steinkuehler.net>
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

bone-pinmux-helper: Add support for "mode" device-tree parameter The new mode parameter is used to set the initial pinmux mode to something other than "default" or NULL, which is what happens currently. This allows enabling SoC hardware via device-tree which requires specific pinmux settings to function on boot, but still leaves the pinmux register under control of the bone-pinmux- helper driver meaning the pinmux setting can be changed at run time via user-mode access to sysfs.

Signed-off-by: Charles Steinkuehler <charles@steinkuehler.net>

This adds gpio and pinmux helpers to the majority of available expansion header pins
based on the cape-universal work from Charles Steinkuehler making them userspace
configurable. This is not a substitute for Capemgr as it doesn't perform the
configuration based on cape detection, nor does it enable dynamic configuration of
all types of peripherals that could be on a cape. It does, however, enable many
developers to rapidly experiment with a lesser degree of complexity.

Derived from:
https://github.com/cdsteinkuehler/beaglebone-universal-io/blob/52461b52ef3203e648399c16c7e160c848a04b5c$



Signed-off-by: Jason Kridner <jdk@ti.com>
Cc: Charles Steinkuehler <charles@steinkuehler.net>
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Jason Kridner <jdk@ti.com>

Signed-off-by: Charles Steinkuehler <charles@steinkuehler.net>

A gpio OF helper driver that allows configuration to be done via
DT.

devm_kfree warned out... why? no idea.

Pinctrl already supports multiple states. Just make them visible.

This patch was derived from 2 commits, but I removed the non-pinmux-helper portions
and added the helper to the bb.org_defconfig:

capemgr: Capemgr makefiles and Kconfig fragments.

Introduce a cape loader using DT overlays and dynamic
DT objects.

Makefile and Kconfig fragments.

Signed-off-by: Pantelis Antoniou <panto@antoniou-consulting.com>

Conflicts:
	arch/arm/mach-omap2/Kconfig
	drivers/misc/Kconfig
	drivers/misc/Makefile

And:

Pinmux helper driver.

    That's just a hack to get a pinmux helper driver working.

    Define in the DT

    	helper {
    		compatible = "bone-pinmux-helper";
    		pinctrl-names = "default";
    		pinctrl-0 = <&helper_pins>;
    		status = "okay";
    	};

Add support for the WL1835MOD cape.
This cape conflicts with the eMMC and HDMI on board the BeagleBone Black.

This change requires that the board be booted from the SD card slot by
holding the user/boot button down at power on and reset.

Signed-off-by: Eyal Reizer <eyalr@ti.com>

updated dtb to work with cape manager

Rewritten using includes, v3.16.1

Signed-off-by: Dave Lambert <dave@lambsys.com>
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Added pin mux definitions for collision and carrier sense errors coming
from the ethernet phy. Without these two signals the MAC ends up ignoring
the errors detected by the PHY which end up causing a lower throughput
problem when the phy is in half duplex mode such as connected through a hub.

Fixes: 2ba35493

 (ARM: dts: am335x-bone*: add DT for BeagleBone Black)
Cc: Koen Kooi <koen@dominion.thruhere.net>
Cc: Tom Rini <trini@ti.com>
Cc: Kevin Hilman <khilman@linaro.org>
Cc: <stable@vger.kernel.org> # v3.13+
Signed-off-by: Schuyler Patton <spatton@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>

https://groups.google.com/d/msg/beagleboard/634Xm1m3XA8/B028x1FzAyAJ



Reported-by: Mark A. Yoder <mark.a.yoder@gmail.com>
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

Signed-off-by: Robert Nelson <robertcnelson@gmail.com>

This probably misplaced (in drivers/misc) patch allows use
of device tree overlays on the two kinds of probeable busses
that count nowadays, PCI & USB.

It does so by dynamically creating device nodes for the busses &
devices that are probed and according to user-configuration applying
an overlay when they appear.

It is still a WIP but it's coming along nicely.

Issues: Only PCI works for now, the generated bindings are not
correct according to the openfirmware spec.

Signed-off-by: Pantelis Antoniou <pantelis.antoniou@konsulko.com>