There have been several attempts to set the dma-names property on the
SoC level (in rk356x.dtsi), but that appears to cause problems when set
on channels without flow control.
Quoting part of a previous attempt for clarification:
> Nah, enabling it for bluetooth is fine because you have flow control.
> My issues have been on channels without flow control. Without DMA it
> simply drops messages or the channel hangs until you close and reopen
> it. With DMA, when an overflow locks up the channel it is usually
> unavailable until the board is rebooted.
Setting it on the board level for the bluetooth connection was deemed
safe, so do so for the Quartz64 Model B.
This fixes the following error/warning:
of_dma_request_slave_channel: dma-names property of node
'/serial@fe650000' missing or empty
dw-apb-uart fe650000.serial: failed to request DMA
Signed-off-by: Diederik de Haas <didi.debian@cknow.org>
Link: https://libera.irclog.whitequark.org/armlinux/2024-02-29
Link: https://lore.kernel.org/linux-rockchip/18284546.sWSEgdgrri@diego/
Reviewed-by: Dragan Simic <dsimic@manjaro.org>
Link: https://lore.kernel.org/r/20240628120130.24076-1-didi.debian@cknow.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The rtl82xx DT bindings do not require ethernet-phy-ieee802.3-c22
as the fallback compatible string. There are fewer users of the
Realtek PHY compatible string with fallback compatible string than
there are users without fallback compatible string, so drop the
fallback compatible string from the few remaining users:
$ git grep -ho ethernet-phy-id001c....... | sort | uniq -c
1 ethernet-phy-id001c.c816",
2 ethernet-phy-id001c.c915",
2 ethernet-phy-id001c.c915";
5 ethernet-phy-id001c.c916",
13 ethernet-phy-id001c.c916";
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202406290316.YvZdvLxu-lkp@intel.com/
Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20240630034910.173552-2-marex@denx.de
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The iommu@fe043e00 on RK356x SoC shares the VOP power domain, but the
power-domains property was not provided when the node has been added.
The consequence is that an attempt to reload the rockchipdrm module will
freeze the entire system. That is because on probe time,
pm_runtime_get_suppliers() gets called for vop@fe040000, which blocks
when pm_runtime_get_sync() is being invoked for iommu@fe043e00.
Fix the issue by adding the missing property.
Fixes: 9d6c6d978f ("arm64: dts: rockchip: rk356x: Add VOP2 nodes")
Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Link: https://lore.kernel.org/r/20240702-rk356x-fix-vop-mmu-v1-1-a66d1a0c45ea@collabora.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The commit 587b4ee24f ("arm64: dts: rockchip: add core dtsi file for
RK3399Pro SoCs") describes the RK3399Pro's PCI Express interface as the way
built-in NPU communicates with the rest of the SoC. All available evidence
shows this not to be accurate, as described in detail below. Moreover, the
rk3399pro.dtsi isn't used anywhere, so let's delete it.
The publicly available schematics of the Radxa Rock Pi N10 carrier board [1]
and the Vamrs VMARC RK3399Pro SoM, [2] which put together form the currently
single supported RK3399Pro-based board, clearly show that the PCI Express x4
interface of this SoC is fully functional and actually not used by the SoC
to communicate with the built-in NPU. In more detail, the VMARC SoM exports
the SoC's PCI Express interface at its board-to-board connector, and the Rock
Pi N10 routes it to an M.2 M-key slot with four PCI Express lanes.
Both the Rockchip RK3399Pro datasheet, version 1.1, [3] and the Rockchip
RK3399Pro technical reference manual (TRM), first part of the version 1.0, [4]
don't describe that the SoC's PCI Express interface is reserved for the NPU.
Instead, the RK3399Pro TRM describes that the NPU uses AHB and AXI interfaces
as the host interface (HIF). The RK3399Pro datasheet clearly describes that
the PCI Express x4 interface is available for general-purpose use, just like
it's the case with the standard Rockchip RK3399 SoC, [5] albeit with a bit
shorter feature list provided in the RK3399Pro datasheet.
Even the publicly available reference RK3399Pro schematic from Rockchip [6]
shows the availability of a standard PCI Express slot with four lanes, which
would be pretty much impossible if the PCI Express interface was reserved
for the communication with the built-in NPU.
Based on the RK3399Pro datasheet [3] and the board schematics, [2][6] the
built-in NPU actually exports NPU_PCIE as a separate PCI Express x2 interface
that's partially pinmuxed with the NPU's separate USB 3.0 interface, which is
described further in the next paragraph. However, the NPU's separate PCI
Express x2 interface is left undocumented in the publicly available RK3399Pro
documentation, in which it's clearly described as reserved for internal use
and not intended for the communication with the NPU. Finally, the evidently
independent nature of the separate NPU_PCIE x2 interface makes ignoring it
safe when it comes to determining the nature and the availability of the
RK3399Pro's main PCI Express x4 interface.
The actual application-level communication with the built-in NPU, including
powering it up and down and uploading the NPU firmware, is performed through
the separate USB 2.0 and USB 3.0 interfaces exported by the NPU, [7] which
the VMARC SoM [2] and the reference board design from Rockchip [6] route to
the SoC's standard USB 2.0 and USB 3.0 interfaces, to make the NPU accessible
to software running on the SoC's ARM cores.
[1] https://dl.radxa.com/rockpin10/docs/hw/rockpi_n10_sch_v1.1_20190909.pdf
[2] https://dl.radxa.com/rockpin10/docs/hw/VMARC_RK3399Pro_sch_V1.1_20190619.pdf
[3] https://www.rockchip.fr/RK3399Pro%20datasheet%20V1.1.pdf
[4] https://www.rockchip.fr/Rockchip%20RK3399Pro%20TRM%20V1.0%20Part1.pdf
[5] https://www.rockchip.fr/RK3399%20datasheet%20V1.8.pdf
[6] https://opensource.rock-chips.com/images/e/e4/RK_EVB_RK3399PRO_LP3S178P332SD8_V11_20181113_RZF.pdf
[7] https://wiki.radxa.com/RockpiN10/dev/NPU-booting
Signed-off-by: Dragan Simic <dsimic@manjaro.org>
Link: https://lore.kernel.org/r/4449f7d4eead787308300e2d1d37b88c9d1446b2.1717308862.git.dsimic@manjaro.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The 'mic-in-differential' DT property supported by the RK809/RK817 audio
codec driver is actually valid if prefixed with 'rockchip,':
DTC_CHK arch/arm64/boot/dts/rockchip/rk3568-rock-3a.dtb
rk3568-rock-3a.dtb: pmic@20: codec: 'mic-in-differential' does not match any of the regexes: 'pinctrl-[0-9]+'
from schema $id: http://devicetree.org/schemas/mfd/rockchip,rk809.yaml#
However, the board doesn't make use of differential signaling, hence
drop the incorrect property and the now unnecessary 'codec' node.
Fixes: 22a442e658 ("arm64: dts: rockchip: add basic dts for the radxa rock3 model a")
Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Link: https://lore.kernel.org/r/20240622-rk809-fixes-v2-3-c0db420d3639@collabora.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add rock5b overlays for PCIe endpoint mode support.
If using the rock5b as an endpoint against a normal PC, only the
rk3588-rock-5b-pcie-ep.dtbo needs to be applied.
If using two rock5b:s, with one board as EP and the other board as RC,
rk3588-rock-5b-pcie-ep.dtbo and rk3588-rock-5b-pcie-srns.dtbo has to
be applied to the respective boards.
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Link: https://lore.kernel.org/r/20240607-rockchip-pcie-ep-v1-v5-13-0a042d6b0049@kernel.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The VOP on RK3328 needs to run at a higher rate in order to produce a
proper 3840x2160 signal.
Change to use 300MHz for VIO clk and 400MHz for VOP clk, same rates used
by vendor 4.4 kernel.
Fixes: 52e02d377a ("arm64: dts: rockchip: add core dtsi file for RK3328 SoCs")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20240615170417.3134517-2-jonas@kwiboo.se
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
RK3588j uses a different set of OPPs for its GPU, both in terms of
allowed frequencies and in terms of voltages.
Move the GPU OPPs table into per-variant .dtsi files to accommodate
for this difference.
The table for RK3588j is adapted from Rockchip downstream sources [1],
while RK3588 one is moved verbatim into the per-variant .dtsi file.
The values provided for RK3588 in the downstream sources match those
in the original commit.
[1] 604cec4004/arch/arm64/boot/dts/rockchip/rk3588s.dtsi
Fixes: 6fca4edb93 ("arm64: dts: rockchip: Add rk3588 GPU node")
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-8-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
RK3588j is the 'industrial' variant of RK3588, and it uses a different
set of OPPs both in terms of allowed frequencies and in terms of
applicable voltages at each frequency setpoint.
Add the OPPs that apply to RK3588j (and apparently RK3588m too) to
enable dynamic CPU frequency scaling.
OPP values are derived from Rockchip downstream sources [1] by taking
only those OPPs which have the highest frequency for a given voltage
level and dropping the rest (if they are included, the kernel complains
at boot time about them being inefficient)
[1] 604cec4004/arch/arm64/boot/dts/rockchip/rk3588s.dtsi
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-7-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
By default the CPUs on RK3588 start up in a conservative performance
mode. Add frequency and voltage mappings to the device tree to enable
dynamic scaling via cpufreq.
OPP values are adapted from Radxa's downstream kernel for Rock 5B [1],
stripping them down to the minimum frequency and voltage combinations
as expected by the generic upstream cpufreq-dt driver, and also dropping
those OPPs that don't differ in voltage but only in frequency (keeping
the top frequency OPP in each case).
Note that this patch ignores voltage scaling for the CPU memory
interface which the downstream kernel does through a custom cpufreq
driver, and which is why the downstream version has two sets of voltage
values for each OPP (the second one being meant for the memory
interface supply regulator). This is done instead via regulator
coupling between CPU and memory interface supplies on affected boards.
This has been tested on Rock 5B with u-boot 2023.11 compiled from
Collabora's integration tree [2] with binary bl31 and appears to be
stable both under active cooling and passive cooling (with throttling)
[1] https://github.com/radxa/kernel/blob/stable-5.10-rock5/arch/arm64/boot/dts/rockchip/rk3588s.dtsi
[2] https://gitlab.collabora.com/hardware-enablement/rockchip-3588/u-boot
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-6-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
RK3588 chips allow for their CPU cores to be powered by a different
supply vs. their corresponding memory interfaces, and two of the
boards currently upstream do that (EVB1 and QuartzPro64).
The voltage of the memory interface though has to match that of the
CPU cores that use it, which downstream kernels achieve by the means
of a custom cpufreq driver which adjusts both at the same time.
It seems that regulator coupling is a more appropriate generic
interface for it, so this patch introduces coupling to affected
device trees to ensure that memory interface voltage is also updated
whenever cpufreq switches between CPU OPPs.
Note that other boards, such as Radxa Rock 5B, define both the CPU
and memory interface regulators as aliases to the same DT node, so
this doesn't apply there.
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-5-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
align with other Radxa products.
- mmc0 is eMMC
- mmc1 is microSD
for ZERO 3E, there is no eMMC, but aliases should start at 0, so mmc0
is microSD as exception.
Fixes: 1a5c8d307c ("arm64: dts: rockchip: Add Radxa ZERO 3W/3E")
Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Changes in v3:
- fix syntax error in rk3566-radxa-zero-3e.dts
Changes in v2:
- microSD is mmc0 instead of mmc1 for ZERO 3E
Link: https://lore.kernel.org/r/20240620224435.2752-1-naoki@radxa.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
This commit adds SFC node for Radxa ROCK 5A.
since sdhci and sfc on RK3588s share pins(i.e. exclusive), it cannot
be enabled both nodes at the same time. so status = "okay" is omitted
here.
you may be able to enable sfc (and disable sdhci) by fdt overlay.
SPI NOR flash chip may vary, so use safe(lowest) spi-max-frequency.
Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Link: https://lore.kernel.org/r/20240623023329.1044-2-naoki@radxa.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
This enables the on-chip thermal monitoring sensor (TSADC) on all
RK3588(s) boards that don't have it enabled yet. It provides temperature
monitoring for the SoC and emergency thermal shutdowns, and is thus
important to have in place before CPU DVFS is enabled, as high CPU
operating performance points can overheat the chip quickly in the
absence of thermal management.
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-2-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
This includes the necessary device tree data to allow thermal
monitoring on RK3588(s) using the on-chip TSADC device, along with
trip points for automatic thermal management.
Each of the CPU clusters (one for the little cores and two for
the big cores) get a passive cooling trip point at 85C, which
will trigger DVFS throttling of the respective cluster upon
reaching a high temperature condition.
All zones also have a critical trip point at 115C, which will
trigger a reset.
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240617-rk-dts-additions-v5-1-c1f5f3267f1e@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Rename the Rockchip RK3588 SoC dtsi files and, consequently, adjust their
contents appropriately, to prepare them for the ability to specify different
CPU and GPU OPPs for each of the supported RK3588 SoC variants.
As already discussed, [1][2][3][4] some of the RK3588 SoC variants require
different OPPs, and it makes more sense to have the OPPs already defined when
a board dts(i) file includes one of the SoC variant dtsi files (rk3588.dtsi,
rk3588j.dtsi or rk3588s.dtsi), rather than requiring the board dts(i) file
to also include a separate rk3588*-opp.dtsi file. The choice of the SoC
variant is already made by the inclusion of the SoC dtsi file into the board
dts(i) file, and it doesn't make much sense to, effectively, allow the board
dts(i) file to include and use an incompatible set of OPPs for the already
selected RK3588 SoC variant.
The new naming scheme for the RK3588 SoC dtsi files uses "-base" and "-extra"
suffixes to denote the DT data shared between all RK5588 SoC variants, and
the DT data shared between the unrestricted SoC variants, respectively.
For example, the DT data for the RK3588 includes both rk3588-base.dtsi and
rk3588-extra.dtsi, because it's an unrestricted SoC variant, while the DT
data for the RK3588S variant includes rk3588-base.dtsi only, because it's
a restricted SoC variant, feature- and interface-wise. This achieves a more
logical naming of the RK3588 SoC dtsi files, which reflects the way DT data
for the SoC variants is built by "stacking" the SoC variant features made
available through the "-base" and "-extra" SoC dtsi files. Additionally,
the SoC variant dtsi files (rk3588.dtsi, rk3588j.dtsi and rk3588s.dtsi) are
no longer parents to any other SoC variant dtsi files, which should help with
making the new "stacking" approach cleaner and easier to follow.
The RK3588 pinctrl dtsi files are also renamed in the same way, for the sake
of consistency. This also keeps the "-base" and "-extra" groups of the dtsi
files together when looked at in a directory listing, which is helpful.
The per-SoC-variant OPPs should go directly into the SoC dtsi files, if no
more than one SoC variant uses those OPPs, or be put into a separate "-opp"
dtsi file that's shared between and included from two or more SoC variant
dtsi files. An example for the former is the non-shared OPP data that should
go directly into the RK3588J SoC variant dtsi file (i.e. rk3588j.dtsi), and
an example for the latter is the shared OPP data that should be put into
rk3588-opp.dtsi and be included from the RK3588 and RK3588S SoC variant dtsi
files (i.e. rk3588.dtsi and rk3588s.dtsi, respectively). Consequently, if
the OPPs for the RK3588 and RK3588S SoC variants are ever made different,
the shared rk3588-opp.dtsi file should be deleted and the new OPPs should
be put directly into rk3588.dtsi and rk3588s.dtsi. [4]
No functional changes are introduced, which was validated by decompiling and
comparing all affected dtb files before and after these changes.
As a side note, due to the nature of introduced changes, this commit is best
viewed using the --break-rewrites option for git-log(1).
[1] https://lore.kernel.org/linux-rockchip/646a33e0-5c1b-471c-8183-2c0df40ea51a@cherry.de/
[2] https://lore.kernel.org/linux-rockchip/CABjd4Yxi=+3gkNnH3BysUzzYsji-=-yROtzEc8jM_g0roKB0-w@mail.gmail.com/
[3] https://lore.kernel.org/linux-rockchip/035a274be262528012173d463e25b55f@manjaro.org/
[4] https://lore.kernel.org/linux-rockchip/673dcf47596e7bc8ba065034e339bb1bbf9cdcb0.1716948159.git.dsimic@manjaro.org/T/#u
Signed-off-by: Dragan Simic <dsimic@manjaro.org>
Link: https://lore.kernel.org/r/9ffedc0e2ca7f167d9d795b2a8f43cb9f56a653b.1717923308.git.dsimic@manjaro.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The CM3588 NAS by FriendlyElec pairs the CM3588 compute module, based on
the Rockchip RK3588 SoC, with the CM3588 NAS Kit carrier board.
To reflect the hardware setup, add device tree sources for the SoM and
the NAS daughter board as separate files.
Hardware features:
- Rockchip RK3588 SoC
- 4GB/8GB/16GB LPDDR4x RAM
- 64GB eMMC
- MicroSD card slot
- 1x RTL8125B 2.5G Ethernet
- 4x M.2 M-Key with PCIe 3.0 x1 (via bifurcation) for NVMe SSDs
- 2x USB 3.0 (USB 3.1 Gen1) Type-A, 1x USB 2.0 Type-A
- 1x USB 3.0 Type-C with DP AltMode support
- 2x HDMI 2.1 out, 1x HDMI in
- MIPI-CSI Connector, MIPI-DSI Connector
- 40-pin GPIO header
- 4 buttons: power, reset, recovery, MASK, user button
- 3.5mm Headphone out, 2.0mm PH-2A Mic in
- 5V Fan connector, PWM beeper, IR receiver, RTC battery connector
PCIe bifurcation is used to handle all four M.2 sockets at PCIe 3.0 x1
speed. Data lane mapping in the DT is done like described in commit
f8020dfb31 ("phy: rockchip-snps-pcie3: fix bifurcation on rk3588").
This device tree includes support for eMMC, SD card, ethernet, all USB2
and USB3 ports, all four M.2 slots, GPU, beeper, IR, RTC, UART debugging
as well as the buttons and LEDs.
The GPIOs are labeled according to the schematics.
Reviewed-by: Space Meyer <git@the-space.agency>
Signed-off-by: Sebastian Kropatsch <seb-dev@mail.de>
Link: https://lore.kernel.org/r/20240616215354.40999-3-seb-dev@mail.de
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
By default the BT WAKE signal inside the M.2 key E connector on Radxa
Rock 5B is driven low, which results in the Bluetooth function being
disabled even if the inserted M.2 card supports it. Expose this signal
as an RFKILL device so that it can be enabled by the userspace.
Tested with an Intel AX210 card, which connects a Bluetooth device over
the USB 2.0 bus.
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Link: https://lore.kernel.org/r/20240517122509.4626-1-alchark@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Update WiFi SDIO and BT UART related props to better reflect details
about the optional onboard RTL8723DS WiFi/BT module.
Also correct the compatible used for bluetooth to match the WiFi/BT
module used on the board.
Fixes: bc3753aed8 ("arm64: dts: rockchip: rock-pi-s add more peripherals")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20240521211029.1236094-14-jonas@kwiboo.se
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
UAR0 CTS/RTS is not wired to any pin and is not used for the default
serial console use of UART0 on ROCK Pi S.
Override the SoC defined pinctrl props to limit configuration of the
two xfer pins wired to one of the GPIO pin headers.
Fixes: 2e04c25b13 ("arm64: dts: rockchip: add ROCK Pi S DTS support")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20240521211029.1236094-6-jonas@kwiboo.se
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Add cap-mmc-highspeed to allow use of high speed MMC mode using an eMMC
to uSD board. Use disable-wp to signal that no physical write-protect
line is present. Also add vcc_io used for card and IO line power as
vmmc-supply.
Fixes: 2e04c25b13 ("arm64: dts: rockchip: add ROCK Pi S DTS support")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20240521211029.1236094-5-jonas@kwiboo.se
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
The Radxa ZERO 3W/3E is an ultra-small, high-performance single board
computer based on the Rockchip RK3566, with a compact form factor and
rich interfaces.
The ZERO 3W and ZERO 3E are basically the same size and model, but
differ only in storage and network interfaces.
- eMMC (3W)
- SD-card (both)
- Ethernet (3E)
- WiFi/BT (3W)
Add initial support for eMMC, SD-card, Ethernet, HDMI and USB.
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20240521202810.1225636-3-jonas@kwiboo.se
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
percpu.h depends on smp.h, but doesn't include it directly because of
circular header dependency issues; percpu.h is needed in a bunch of low
level headers.
This fixes a randconfig build error on mips:
include/linux/alloc_tag.h: In function '__alloc_tag_ref_set':
include/asm-generic/percpu.h:31:40: error: implicit declaration of function 'raw_smp_processor_id' [-Werror=implicit-function-declaration]
Reported-by: kernel test robot <lkp@intel.com>
Fixes: 24e44cc22a ("mm: percpu: enable per-cpu allocation tagging")
Closes: https://lore.kernel.org/oe-kbuild-all/202405210052.DIrMXJNz-lkp@intel.com/
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull perf tool fix from Arnaldo Carvalho de Melo:
"Revert a patch causing a regression.
This made a simple 'perf record -e cycles:pp make -j199' stop working
on the Ampere ARM64 system Linus uses to test ARM64 kernels".
* tag 'perf-tools-fixes-for-v6.10-1-2024-05-26' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools:
Revert "perf parse-events: Prefer sysfs/JSON hardware events over legacy"
