RMI4 F1A supports the simple capacitive buttons function, it's used for
example on embedded devices such as smartphones for capacitive Android
back and recents buttons.
This finally allowed me to get past the 2 ath10k_info()'s in qmi.c!
However, this doesn't fix everything; it seems you *have* to run
diag-router from https://github.com/andersson/diag to calm the firmware
down and stop constantly crashing; need to look into a way to get Wi-Fi
working later without needing this debug crap to be running.
This adds support for accessing SRAM. That includes requesting and releasing
access to SRAM by setting some registers and handling the mem-avail IRQ,
configuring access, and reading from/writing to it.
Pre-gen3 fuel gauges require reading from SRAM to get voltage, current
and temperature data. Getting capacity is identical to gen3.
Tested on Xiaomi Mi Note 2 with PMI8996.
The OnePlus 5 and 5T both have a haptics engine connected to PMI8998.
Signed-off-by: Jami Kettunen <jami.kettunen@somainline.org>
Signed-off-by: Caleb Connolly <caleb@connolly.tech>
Enable the pmi8998 SPMI haptics driver and configure the play rate for the
hardware.
Signed-off-by: Joel Selvaraj <jo@jsfamily.in>
Signed-off-by: Caleb Connolly <caleb@connolly.tech>
Add support for the haptics found in pmi8998 and related PMICs.
Based on the ff-memless interface. Currently this driver provides
a partial implementation of hardware features.
This driver only supports LRAs (Linear Resonant Actuators) in the "buffer"
mode with a single wave pattern.
Signed-off-by: Caleb Connolly <caleb@connolly.tech>
The nt35950 IC from Novatek is a Driver IC used to drive MIPI-DSI panels,
with Static RAM for content retention in command mode and also supports
video mode with VESA Frame Buffer Compression or Display Stream Compression
on single, or dual dsi port(s).
This DDIC is also capable of upscaling an input image to the panel's native
resolution, for example it can upscale a 1920x1080 input to 3840x2160 with
either bilinear interpolation or pixel duplication.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Implement the support for SAW v2.3, used in at least MSM8976, MSM8956
and APQ variants and, while at it, also add the configuration for the
MSM8976's little (a53) and big (a72) clusters cache power management.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Introduce a driver for the Qualcomm interconnect busses found in
the MSM8976/8956 SoCs.
The topology consists of several NoCs that are controlled by a
remote processor that collects the aggregated bandwidth for each
master-slave pairs.
According to the documentation any combination of LEDs can be used,
which is specified in the DT using qcom,enabled-strings as u32 array.
The numbers specified in this array represent indices of the led strings
that are to be enabled and disabled.
The value is appropriately used to enable string modules, but completely
disregarded in the set_brightness path which only iterates over "string
indices" 0<=i<num_strings.
This means that in order to have a custom set of strings enabled in the
DT, qcom,num-strings needs to be at least equal to the highest string
index plus one, even if that doesn't reflect the actual number of
strings used on the system.
In turn, gaps in the strings list need to be filled with duplicate
indices in order to prevent enabling other strings not listed in the
array (those are specified by the default for wled3 and 4, but not
overwritten when the array is read from the DT).
Take for example a situation where only string 1 is desired to be turned
on. Specifying an array of <1> will only overwrite the first entry of
enabled_strings, while the other numbers stick to their default. Unless
qcom,num-strings is adjusted to 1 too, other (unspecified) string
modules get enabled too.
However, this causes set_brightness to only write a brightness value to
the string at index 0.
This patch aims to simplify the case for custom string combinations by
using enabled_strings in set_brightness, as well as updating num_strings
with the length of the array when it is set.
It is assumed here that setting an array of enabled strings
TODO BETTER WORDS rules out
the use of autodetection, or at least only apply autodetection on the
specified LED modules.
TODO: THIS LATTER PART IS CURRENTLY NOT ACTUALLY THE CASE... WHAT DO WE
WANT?
Furthermore, in order to ensure the autodetection keeps working, the
default {0, 1, 2, 3} enabled_strings value has been adopted from wled3.
This is identical to iterating the indices from 0<=i<num_strings where
num_strings defaults to 4.
Keep in mind that wled4 did not have a sensible default, which would
cause only string 0 to be enabled in wled4_setup with autodetection
disabled, even when num_strings is 4.
TODO TODO TODO
SECOND FINALLY: There is another error. The default of qcom,num_strings
in case it's unset is 4. The cfg.num_strings = cfg.num_strings + 1;
causes writes to an unexisting or overlapping register in set_brightness
and other functions relying on num_strings.
TODO TODO TODO CLEAN UP WORD CONVENTION. Are we talking about strings,
modules, string modules, LED modules??? What's the unit here?
[mainline todo]: num-strings is not optional atm.
Signed-off-by: Marijn Suijten <marijn.suijten@somainline.org>
of_property_read_u32_array takes the number of elements to read as last
argument. This does not always need to be 4 (sizeof(u32)) but should
instead be the size of the array in DT as read just above with
of_property_count_elems_of_size.
To not make such an error go unnoticed again the driver now bails
accordingly when of_property_read_u32_array returns an error.
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Signed-off-by: Marijn Suijten <marijn.suijten@somainline.org>
Add MSM8998 and SDM660's mss-pil compatibles to switch the default
iommu domain type to IDENTITY, as similarly required by SDM845 and
others.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Similarly to MSM8996 DragonBoard 820c, MSM8998 and SDM630 are equipped
with Adreno 5xx series, which doesn't have separate pagetables support
at the moment of writing.
Skip the TTBR1 quirk for these two SoCs as to get Adreno in a usable
state.
Add generic slimbus configuration to MSM8998 and keep it disabled.
It is expected that this will be enabled in board-specific device
trees when supported, along with the right audio codec.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Add MSM8998 sound support: this driver supports basic functionality
and headphone jack.
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
