Age | Commit message (Collapse) | Author | Files | Lines |
|
According to a downstream commit I found in the Khadas vendor kernel,
the GPU on G12b is wired up for ACE-lite, so (now that Panfrost knows
how to handle this properly) we should describe it as such. Otherwise
the mismatch leads to all manner of fun with mismatched attributes and
inadvertently snooping stale data from caches, which would account for
at least some of the brokenness observed on this platform.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Tested-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/765446e529e50b304af63432da7836c4d31eb8d4.1600780574.git.robin.murphy@arm.com
|
|
The G12B thermal nodes should be in the meson-g12b.dtsi file.
Fixes: 195f140318a9 ("arm64: dts: meson: g12b: add cooling properties")
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Link: https://lore.kernel.org/r/20200512093916.19676-2-narmstrong@baylibre.com
|
|
Add missing #colling-cells field for G12B SoC
Add cooling-map for passive and hot trip point
Tested-by: Christian Hewitt <christianshewitt@gmail.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Amit Kucheria <amit.kucheria@linaro.org>
Signed-off-by: Guillaume La Roque <glaroque@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
Meson G12B SoCs (S922X and A311D) are a big-little design where not all CPUs
are equal; the A53s cores are weaker than the A72s.
Include capacity-dmips-mhz properties to tell the OS there is a difference
in processing capacity. The dmips values are based on similar submissions for
other A53/A72 SoCs: HiSilicon 3660 [1] and Rockchip RK3399 [2].
This change is particularly beneficial for use-cases like retro gaming where
emulators often run on a single core. The OS now chooses an A72 core instead
of an A53 core.
[1] https://lore.kernel.org/patchwork/patch/862742/
[2] https://patchwork.kernel.org/patch/10836577/
Signed-off-by: Frank Hartung <supervisedthinking@gmail.com>
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
The power domain declared in the g12a and g12b dtsi are the same.
Move the declaration of these power domains in the g12 common dtsi.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
While the sm1 is very close to the g12a/b family, somethings apply
differently on the g12a/b and not the sm1. This introduce a new layer
of dtsi for part which apply to the g12a and g12b but not the sm1.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
Replace the VPU-centric power domain controller by the generic system-wide
Everything-Else power domain controller and setup the right power-domains
properties on the VPU, Ethernet & USB nodes.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
[khilman: minor subject edit: add dts]
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
Meson g12b ships with a low-speed (S922X) and high-speed (A311D) variant
so remove cpu_opp_table nodes in meson-g12b.dtsi and create two new dtsi
that can be included in device-specific dts files. Opp points were taken
from the vendor BSP kernel.
Also make meson-g12b-odroid-n2.dts include the new meson-g12b-s922x.dtsi.
Signed-off-by: Christian Hewitt <christianshewitt@gmail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
Add the OPP table taken from the HardKernel Odroid-N2 DTS.
The Amlogic G12B SoC seems to available in 2 types :
- low-speed: Cortex-A73 Cluster up to 1,704GHz
- high-speed: Cortex-A73 Cluster up to 2.208GHz
The Cortex-A73 Cluster can be clocked up to 1,896GHz for both types.
The Vendor Amlogic A311D OPP table are slighly different, with lower
voltages than the HardKernel S922X tables but seems to be high-speed type.
This adds the conservative OPP table with the S922X higher voltages
and the maximum low-speed OPP frequency.
The values were tested to be stable on an HardKernel Odroid-N2 board
running the arm64 cpuburn at [1] and cycling between all the possible
cpufreq translations for both clusters and checking the final frequency
using the clock-measurer, script at [2].
[1] https://github.com/ssvb/cpuburn-arm/blob/master/cpuburn-a53.S
[2] https://gist.github.com/superna9999/d4de964dbc0f84b7d527e1df2ddea25f
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
To simplify the representation of differences betweem the G12A and G12B
SoCs, move the common nodes into a meson-g12-common.dtsi file and
express the CPU nodes and differences in meson-g12a.dtsi and meson-g12b.dtsi.
This separation will help for DVFS and future Amlogic SM1 Family support.
The sd_emmc_a quirk is added in the g12a/g12b since since it's already
known the sd_emmc_a controller is fixed in the next SM1 SoC family.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Kevin Hilman <khilman@baylibre.com>
Tested-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|
|
This patch adds basic support for :
- Amlogic G12B, which is very similar to G12A
- The HardKernel Odroid-N2 based on the S922X SoC
The Amlogic G12B SoC is very similar with the G12A SoC, sharing
most of the features and architecture, but with these differences :
- The first CPU cluster only has 2xCortex-A53 instead of 4
- G12B has a second cluster of 4xCortex-A73
- Both cluster can achieve 2GHz instead of 1,8GHz for G12A
- CPU Clock architecture is difference, thus needing a different
compatible to handle this slight difference
- Supports a MIPI CSI input
- Embeds a Mali-G52 instead of a Mali-G31, but integration is the same
Actual support is done in the same way as for the GXM support, including
the G12A dtsi and redefining the CPU clusters.
Unlike GXM, the first cluster is different, thus needing to remove
the last 2 cpu nodes of the first cluster.
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Tested-by: Anand Moon <linux.amoon@gmail.com>
[khilman: add vin-supply for vcc_v5 as suggested by Anand Moon]
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
|