| Age | Commit message (Collapse) | Author |
|---|
|
The device tree clock structure for the mipi_dsi is
unnecessarily redundant.
The default clock parent of IMX8MM_CLK_DSI_PHY_REF is
already IMX8MM_CLK_24M, so there is no need to set the
parent-child relationship between them. The default clock
rates for IMX8MM_SYS_PLL1_266M and IMX8MM_CLK_24M are
already defined to be 266MHz and 24MHz respectively,
so there is no need to define those clock rates.
On i.MX8M[MNP] the samsung,pll-clock-frequency is not
necessary, because the driver will read it from sclk_mipi
which is also already set to 24MHz making it also
redundant.
Signed-off-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Frieder Schrempf <frieder.schrempf@kontron.de>
Tested-by: Frieder Schrempf <frieder.schrempf@kontron.de> # Kontron BL i.MX8MM
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Add the 128 KiB on-chip SRAM at address 0x900000.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
This adds the device tree support for the MIPI-DSI block. The block can
be used as encoder for the parallel signals coming from the lcdif block.
Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Roland Hieber <rhi@pengutronix.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Fix some errors in the Marvell MV88E6xxx switch descriptions:
- switch0@0 is not OK, should be ethernet-switch@0
- ports should be ethernet-ports
- port should be ethernet-port
- phy should be ethernet-phy
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The DTS code coding style expects exactly one space before and after '='
sign.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
i.MX8DXL's ddr pmu has port/channel filter capabilities, but it still is
compatible with "fsl,imx8-ddr-pmu". This will change the compatible.
Signed-off-by: Xu Yang <xu.yang_2@nxp.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
'dr_mode' is part of the USB DWC3 core, not the glue layer. Remove the
property from glue layer. Fixes the dtbs_check warning:
arch/arm64/boot/dts/freescale/imx8mp-tqma8mpql-mba8mpxl.dtb: usb@32f10108:
'dr_mode' does not match any of the regexes: '^usb@[0-9a-f]+$',
'pinctrl-[0-9]+'
from schema $id: http://devicetree.org/schemas/usb/fsl,imx8mp-dwc3.yaml#
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The 'shared-interrupt' property is not documented nor used anywhere.
Remove it.
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The 'shared-interrupt' property is not documented nor used anywhere.
Remove it.
This fixes the following schema warning:
imx93-11x11-evk.dtb: dma-controller@42000000: Unevaluated properties are not allowed ('shared-interrupt' was unexpected)
from schema $id: http://devicetree.org/schemas/dma/fsl,edma.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Per gpio-sbu-mux.yaml, the compatible entry is incomplete.
The imx8qxp-mek board uses a CBDTU02043, so complete the gpio-sbu-mux
compatible accordingly.
This fixes the following schema warning:
imx8qxp-mek.dtb: gpio-sbu-mux: compatible:0: 'gpio-sbu-mux' is not one of ['onnn,fsusb43l10x', 'pericom,pi3usb102']
from schema $id: http://devicetree.org/schemas/usb/gpio-sbu-mux.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The property 'phy-connection-type' can be used to describe the interface
type between the Ethernet device and the Ethernet PHY device.
However, snps,dwmac.yaml gives the following warning:
imx8mp-debix-model-a.dtb: ethernet@30bf0000: 'phy-mode' is a required property
from schema $id: http://devicetree.org/schemas/net/snps,dwmac.yaml#
To avoid the warning, switch to the more commonly used, 'phy-mode'
property instead.
Signed-off-by: Fabio Estevam <festevam@denx.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Per i2c-mux-pca954x.yaml, the I2C subnodes should follow the
'i2c@' format.
Change it to fix the following schema warning:
imx8mm-nitrogen-r2.dtb: i2c-mux@70: Unevaluated properties are not allowed ('i2c3@0' was unexpected)
from schema $id: http://devicetree.org/schemas/i2c/i2c-mux-pca954x.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Per snps,dwmac.yaml, the interrupt-names entries should be in the
following order: "macirq", "eth_wake_irq";
Change it to fix the following schema warnings.
imx8dxl-evk.dtb: ethernet@5b050000: interrupt-names:0: 'macirq' was expected
from schema $id: http://devicetree.org/schemas/net/snps,dwmac.yaml#
imx8dxl-evk.dtb: ethernet@5b050000: interrupt-names:1: 'macirq' is not one of ['eth_wake_irq', 'eth_lpi']
from schema $id: http://devicetree.org/schemas/net/snps,dwmac.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Per nxp,pcf8575.yaml, #gpio-cells should be 2.
Change it to fix the following schema warning:
imx8mm-emcon-avari.dtb: gpio@3a: #gpio-cells:0:0: 2 was expected
from schema $id: http://devicetree.org/schemas/gpio/nxp,pcf8575.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Per fsl-lpuart.yaml, when the 'dmas' property is used 'dma-names' should
also be present.
Pass the lpuart 'dma-names' property to fix the following schema
warnings:
imx8dxl-evk.dtb: serial@5a060000: dma-names:0: 'rx' was expected
from schema $id: http://devicetree.org/schemas/serial/fsl-lpuart.yaml#
imx8dxl-evk.dtb: serial@5a060000: dma-names:1: 'tx' was expected
from schema $id: http://devicetree.org/schemas/serial/fsl-lpuart.yaml#
imx8dxl-evk.dtb: serial@5a060000: Unevaluated properties are not allowed ('dma-names' was unexpected)
from schema $id: http://devicetree.org/schemas/serial/fsl-lpuart.yaml#
Signed-off-by: Fabio Estevam <festevam@denx.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Add Skov i.MX8MP based climate controller.
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The following warning is shown when probing device:
pca953x 1-0020: supply vcc not found, using dummy regulator
Define a new fixed 3.3v regulator for carrier board peripherals,
enabled by mosfet switch Q2 after the SOM_3V3 supply rises (no software
control).
Add this new regulator as vcc supply to the PCA9534 to silence the warning.
Signed-off-by: Hugo Villeneuve <hvilleneuve@dimonoff.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The RVE gateway board is based on a Variscite VAR-SOM-NANO,
with a NXP MX8MN nano CPU.
Signed-off-by: Hugo Villeneuve <hvilleneuve@dimonoff.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Provide the 1.8 and 3.3 volt regulators that are utilised on the Debix
SOM BMB-08 base board.
Facilitate this by also supplying the pin control used to enable the
regulators on the second MIPI CSI port.
Reviewed-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Newer variants of Ixora boards require a power-up delay when powering up
the CAN transceiver of up to 1ms.
Signed-off-by: Andrejs Cainikovs <andrejs.cainikovs@toradex.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
When the EEPROM is probed, we have this warning:
at24 0-0052: supply vcc not found, using dummy regulator
Add fixed 3.3v regulator to silence the warning.
Signed-off-by: Hugo Villeneuve <hvilleneuve@dimonoff.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Several schema warnings are seen when running:
make dtbs_check DT_SCHEMA_FILES=pci-bus.yaml
Fix them.
Signed-off-by: Fabio Estevam <festevam@denx.de>
Acked-by: Tim Harvey <tharvey@gateworks.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Even if the 'dsp' node is disabled the memory intended to be used by the
DSP is reserved. This limits the memory range suitable for CMA allocation.
Thus disable the dsp_reserved node. DSP users need to enable it in parallel
to the 'dsp' node.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The NPU is based on the Vivante GC8000 and its power-domain
is controlled my pgc_mlmix. Since the power-domain uses
some of these clocks, setup the clock parent and rates
inside the power-domain, and add the NPU node.
The data sheet states the CLK_ML_AHB should be 300MHz for
nominal, but 800MHz clock will divide down to 266 instead.
Boards which operate in over-drive mode should update the
clocks on their boards accordingly. When the driver loads,
the NPU numerates as:
etnaviv-gpu 38500000.npu: model: GC8000, revision: 8002
Signed-off-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Map the 'RUN' LED present on the Debix-SOM as a heartbeat.
Reviewed-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
This adds an overlay for the supported LVDS display AUO G133HAN01.
Configure the video PLL frequency to exactly match typical pixel clock of
141.200 MHz.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Remove unnecessary #address-cells and #size-cells from tpm node.
Fixes: 531936b218d8 ("arm64: dts: imx8mp-venice-gw74xx: update to revB PCB")
Signed-off-by: Tim Harvey <tharvey@gateworks.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The DTS code coding style expects exactly one space before and after '='
sign.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Add mmc aliases to ensure a consistent mmc device naming across the
Toradex SoM family, with this commit mmc0 is the on-module eMMC
boot device and the not available mmc interfaces are removed.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Add mmc aliases to ensure a consistent mmc device naming across the
Toradex SoM family, with this commit mmc0 is the on-module eMMC
boot device and the not available mmc interfaces are removed.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Add mmc aliases to ensure a consistent mmc device naming across the
Toradex SoM family, with this commit mmc0 is the on-module eMMC
boot device and the not available mmc interfaces are removed.
Signed-off-by: Hiago De Franco <hiago.franco@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
The PICO-IMX7D SoM has the usual power supply configuration using
output sw1a of PF3000 PMIC, which was defined in downstream derivative
of linux-imx (see link) in the sources for "Android Things" devkit.
It is required to support CPU frequency scaling.
Map the respective "cpu-supply" nodes of each core to sw1a of the PMIC.
Enabling them causes cpufreq-dt, and imx-thermal drivers to probe
successfully, and CPU frequency scaling to function.
Link: https://android.googlesource.com/platform/hardware/bsp/kernel/nxp/imx-v4.1/+/o-iot-preview-5/arch/arm/boot/dts/imx7d-pico.dtsi#849
Cc: Fabio Estevam <festevam@gmail.com>
Cc: Shawn Guo <shawnguo@kernel.org>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Lech Perczak <lech.perczak@gmail.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
Use preferred device_get_match_data() instead of of_match_device() to
get the driver match data. With this, adjust the includes to explicitly
include the correct headers.
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
|
|
It's designed for hardwiring Bluetooth devices to it.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Link: https://lore.kernel.org/r/20231122144208.21114-2-zajec5@gmail.com
Signed-off-by: Florian Fainelli <florian.fainelli@broadcom.com>
|
|
Combined dtb builds are only useful for making sure that the overlay
applies cleanly on the base dtb.
So we move all such combined blobs under a `dtb- +=` section that is
only built when CONFIG_OF_ALL_DTBS is enabled.
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-9-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
RPi v2 Camera (IMX219) is an 8MP camera that can be used with SK-AM62A
through the 22-pin CSI-RX connector.
Same overlay can be used across SK-AM62* boards that have a 15/22-pin
FFC connector, so we name it with the k3-am62x- prefix.
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-8-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
Enable I2C-2 as it is used to control CSI based sensors. Also enable
IO-EXP-2 as it controls the mux between different CSI-2 connectors.
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-7-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
Three different OV5640 modules are supported using the 15-pin FFC
connector on SK-AM62:
- Digilent PCam 5C
- ALINX AN5641
- TEVI-OV5640-*-RPI
The Digilent and ALINX modules supply a 12Mhz XCLK to the sensor, while
the TEVI module supplies a 24Mhz XCLK, thus requiring a separate
overlay.
These overlays can be used on other boards of the SK-AM62* family that
have a 15/22-pin FFC connector, so we name the overlays with the prefix
k3-am62x-.
Tested-by: Martyn Welch <martyn.welch@collabora.com>
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-6-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
CSI cameras are controlled using I2C, on SK-AM62 and derivative boards
this is routed to I2C-2, so enable that bus.
Specific sensor connected to this bus will be described in the DT
overlay for each sensor.
Tested-by: Martyn Welch <martyn.welch@collabora.com>
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-5-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
Three different OV5640 modules are supported using the FFC connector on
BeaglePlay:
- Digilent PCam 5C
- ALINX AN5641
- TEVI-OV5640-*-RPI
The Digilent and ALINX modules supply a 12Mhz XCLK to the sensor, while
the TEVI module supplies a 24Mhz XCLK, thus requiring a separate
overlay.
Reviewed-by: Andrew Davis <afd@ti.com>
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-4-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
Add nodes for Cadence DPHY, CSI2RX and TI's pixel-grabbing wrapper.
AM62A uses a dedicated BCDMA instance for CSI-RX traffic, so enable
that as well.
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-3-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
The CSI2RX subsystem can be used to capture video frames from CSI-2
cameras. Add nodes for the CSI core, SHIM layer, and the DPHY.
Tested-by: Martyn Welch <martyn.welch@collabora.com>
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-2-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
AM62 and other K3 based SoCs use Cadence DPHY and CSI-RX bridge
drivers, along with a DMA wrapper CSI IP for the camera pipeline.
Enable the same to get camera functionality on AM62x-SK, BeaglePlay and
AM62Ax SK among other platforms.
Tested-by: Martyn Welch <martyn.welch@collabora.com>
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Link: https://lore.kernel.org/r/20231201-csi_dts-v3-1-9f06f31080fe@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
The AM652 is basically a AM654 but with 2 cores instead of 4. Add a
DTSI file for AM652 matching AM654 except this core difference.
This removes the need to remove the extra cores from AM654 manually
in DT files for boards that use the AM652 variant. Do that for the
IOT2050 boards here.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Bryan Brattlof <bb@ti.com>
Link: https://lore.kernel.org/r/20231205162358.23904-1-afd@ti.com
Signed-off-by: Nishanth Menon <nm@ti.com>
|
|
io_uring sets up the io worker kernel thread via a syscall out of an
user space prrocess. This process might have used FPU and since
copy_thread() didn't clear FPU states for kernel threads a BUG()
is triggered for using FPU inside kernel. Move code around
to always clear FPU state for user and kernel threads.
Cc: stable@vger.kernel.org
Reported-by: Aurelien Jarno <aurel32@debian.org>
Closes: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1055021
Suggested-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
|
|
There are many types of revsered memory passed from firmware
that should be reserved in memblock, and UMA memory passed
from firmware that should be added to system memory for system
to use.
Also for memblock there is no need to align those space into page,
which actually cause problems.
Handle them properly to prevent memory corruption on some systems.
Cc: stable@vger.kernel.org
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
|
|
There are some Loongson64 systems come with broken coherent DMA
support, firmware will set a bit in boot_param and pass nocoherentio
in cmdline.
However nonconherent support was missed out when spin off Loongson-2EF
form Loongson64, and that boot_param change never made itself into
upstream.
Support DMA noncoherent properly to get those systems working.
Cc: stable@vger.kernel.org
Fixes: 71e2f4dd5a65 ("MIPS: Fork loongson2ef from loongson64")
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
|
|
vgabios is passed from firmware to kernel on Loongson64 systems.
Sane firmware will keep this pointer in reserved memory space
passed from the firmware but insane firmware keeps it in low
memory before kernel entry that is not reserved.
Previously kernel won't try to allocate memory from low memory
before kernel entry on boot, but after converting to memblock
it will do that.
Fix by resversing those memory on early boot.
Cc: stable@vger.kernel.org
Fixes: a94e4f24ec83 ("MIPS: init: Drop boot_mem_map")
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
|
|
rcutree_report_cpu_starting() must be called before
clockevents_register_device() to avoid the following lockdep splat triggered by
calling list_add() when CONFIG_PROVE_RCU_LIST=y:
WARNING: suspicious RCU usage
...
-----------------------------
kernel/locking/lockdep.c:3680 RCU-list traversed in non-reader section!!
other info that might help us debug this:
RCU used illegally from offline CPU!
rcu_scheduler_active = 1, debug_locks = 1
no locks held by swapper/1/0.
...
Call Trace:
[<ffffffff8012a434>] show_stack+0x64/0x158
[<ffffffff80a93d98>] dump_stack_lvl+0x90/0xc4
[<ffffffff801c9e9c>] __lock_acquire+0x1404/0x2940
[<ffffffff801cbf3c>] lock_acquire+0x14c/0x448
[<ffffffff80aa4260>] _raw_spin_lock_irqsave+0x50/0x88
[<ffffffff8021e0c8>] clockevents_register_device+0x60/0x1e8
[<ffffffff80130ff0>] r4k_clockevent_init+0x220/0x3a0
[<ffffffff801339d0>] start_secondary+0x50/0x3b8
raw_smp_processor_id() is required in order to avoid calling into lockdep
before RCU has declared the CPU to be watched for readers.
See also commit 29368e093921 ("x86/smpboot: Move rcu_cpu_starting() earlier"),
commit de5d9dae150c ("s390/smp: move rcu_cpu_starting() earlier") and commit
99f070b62322 ("powerpc/smp: Call rcu_cpu_starting() earlier").
Signed-off-by: Stefan Wiehler <stefan.wiehler@nokia.com>
Reviewed-by: Huacai Chen <chenhuacai@loongson.cn>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
|
|
In current code, init_irq_stacks() will call cpu_to_node().
The cpu_to_node() depends on percpu "numa_node" which is initialized in:
arch_call_rest_init() --> rest_init() -- kernel_init()
--> kernel_init_freeable() --> smp_prepare_cpus()
But init_irq_stacks() is called in init_IRQ() which is before
arch_call_rest_init().
So in init_irq_stacks(), the cpu_to_node() does not work, it
always return 0. In NUMA, it makes the node 1 cpu accesses the IRQ stack which
is in the node 0.
This patch fixes it by:
1.) export the early_cpu_to_node(), and use it in the init_irq_stacks().
2.) change init_irq_stacks() to __init function.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Huang Shijie <shijie@os.amperecomputing.com>
Link: https://lore.kernel.org/r/20231124031513.81548-1-shijie@os.amperecomputing.com
Signed-off-by: Will Deacon <will@kernel.org>
|
