Rockchip SoCs used to have a random number generator as part of their
crypto device, and support for it has to be added to the corresponding
driver.
Newer Rockchip SoCs like the RK3568 have an independent True Random
Number Generator device. Import pending patchset which adds a driver for
it, include it in Kconfig and enable it in the device tree.
Doing so significantly reduces the time needed to boot devices based on
those SoCs, from about 27 seconds until Ethernet is up and running to
less than 13 seconds with a minimal snapshot image.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add "linux64-loongarch64-openwrt" into openssl configurations to enable
building on loongarch64 machines.
Signed-off-by: Weijie Gao <hackpascal@gmail.com>
* Allow kmod-acpi-video to be built for loongarch64:
The x86-specific CONFIG_ACPI_WMI will be split from default
kmod-acpi-video as a board-specific addition.
* Allow kmod-drm-amdgpu to be built for loongarch64:
Also add loongarch64-specific configs and modules.
Signed-off-by: Weijie Gao <hackpascal@gmail.com>
Add target for Loongson LoongArch64-based boards.
LoongArch is a new RISC ISA developed by Loongson. It's a bit like
MIPS or RISC-V. LoongArch includes both 32-bit and 64-bit versions
(LoongArch32/LoongArch64).
Loongson 3A5000 and 3A6000 are the two existing CPUs of LoongArch64
and is used for PC products. It's BIOS supports ACPI and UEFI-only
boot. These CPUs supports SMP and SMT.
At present only LoongArch64 is supported by linux kernel.
Toolchain requirement:
binutils >= 2.40
gcc >= 13.1
For details, please check the following links:
https://lwn.net/Articles/861951/https://loongson.github.io/LoongArch-Documentation/README-EN.html
Signed-off-by: Weijie Gao <hackpascal@gmail.com>
GCC has changed musl dynamic linker name from
ld-musl-loongarch-lp64d.so.1 to ld-musl-loongarch64.so.1 recently [1].
This means there are two dynamic linker names will be used across different
ersions of GCC. But musl 1.2.5 only supports the new name while the GCC
we're currently using uses the old name.
To maintain compatibility with all versions of GCC, the musl is then patched
to generate two symbolic links to libc.so with both old and new names.
[1] https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=8bccee51f0deac64b79cd9ad75df599422f4c8ff
Signed-off-by: Weijie Gao <hackpascal@gmail.com>
GCC has changed musl dynamic linker name from
ld-musl-loongarch-lp64d.so.1 to ld-musl-loongarch64.so.1 recently [1].
Meanwhile musl 1.2.5 only supports the new name. So it's better to follow
the new name.
[1] https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=8bccee51f0deac64b79cd9ad75df599422f4c8ff
Signed-off-by: Weijie Gao <hackpascal@gmail.com>
This reverts commit 3fe239fcf8.
Now that we switched to Linux 6.6 this is no longer needed, and resulted
in a left-over file because it's removal was not included in the commit
removing all the other files intended for Linux 6.1.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Remove unnecessary 'if' macros for previous kernel versions.
After removing kernel 6.1 the kernel is always >= 6.6 so the conditions
are unnecessary.
Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
[removed some more and also no longer include version.h]
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Remove unnecessary 'if' macros for previous kernel versions.
After removing kernel 6.1 the kernel is always >= 6.6 so the conditions
are unnecessary.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Import pending patch to fix the cmdline parsing of the "blkdevparts="
parameter which has been broken somewhen between Linux 6.1 and Linux 6.6.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Commit d82c5884c6 ("treewide: make use of new toolchain define")
changed $(TOOLCHAIN_DIR)/include to the new variable
$(TOOLCHAIN_INC_DIRS) that now can contain multiple entry.
Because of this only the first include in $(TOOLCHAIN_INC_DIRS) was
actually included with -isystem, making the other producing warning with
ignored inputs.
Fix this by parsing each entry in $(TOOLCHAIN_INC_DIRS) and adding the
-isystem prefix to correctly include them in the BPF_KERNEL_INCLUDE.
Fixes: d82c5884c6 ("treewide: make use of new toolchain define")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
To enable verbose log for xdp-tools compilation, we check for "c" in
the OPENWRT_VERBOSE, but verbose.mk supports only "w" and "s" for V=1
and V=99.
Fix the wrong matching and correctly enable verbose output matching for
"s".
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Now that 6.6 is the default, remove the 6.1 config and the hack that
was required for the arm32 DTS dir change.
Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
Upgrade the OpenSBI firmware used by RISC-V CPUs to 1.4.
Runtime-tested:
- d1 (Lichee RV)
- sifiveu (SiFive Unleashed)
Updates since last release:
1.4:
Synopsys DesignWare APB GPIO driver
Zicntr and Zihpm support
Console print improvements
Smepmp support
Simple FDT based syscon regmap driver
Syscon based reboot and poweroff driver
Non-contiguous hpm counters
Smcntrpmf support
Full sparse hartid support
IPI improvements
RFENCE improvements
Zkr support
Andes custom PMU support
1.3.1:
ACLINT driver fix for disabled CPUs
SBI PMU fix for out-of-bound access
Designware GPIO driver
1.3:
Allow platform to influence cold boot HART selection
Starfive JH7110 platform support
Split RX and RW firmware regions
Advertise non-retentive suspend for allwinner D1 platform
Byteorder/endianness conversion macros
SBI debug console extension (Experimental)
Configure the PMA regions for RZ/Five platform
SBI system suspend extension (Experimental)
SBI PMU platform firmware events (Experimental)
SBI CPPC extension (Experimental)
Optimized remote TLB flushes
Simple heap for boot time memory allocations
Bring back no-map DT property for reserved memory nodes
Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>
AVM FRITZ!Repeater 1200 does not use QCA807x PHY at all and thus it
disables all of the individual PHY nodes, however this is not enough
anymore since the conversion to PHY package.
Now its now enough to disable the PHY-s in the package alone, but the PHY
package node itself must also be disabled.
Fixes: 1b931c33a2 ("ipq40xx: adapt to new Upstream QCA807x PHY driver")
Fixes: #15355
Link: https://github.com/openwrt/openwrt/pull/15365
Signed-off-by: Robert Marko <robimarko@gmail.com>
Increasing the size of the rootfs_data filesystem has become a ever
repeating discussion and seems to be the most important thing for
users of the MediaTek-based BananaPi boards.
Using the whole remaining size of a microSD or the eMMC for rootfs_data
doesn't make sense for many reasons, but neither does the current
default of 104 MiB for the 'rootfs' partition size.
Increase the 'rootfs' partition size to 448 MiB which will result in
the sdcard image being exactly 512 MiB. Finding a microSD card smaller
than 512 MiB and still working could anyway be difficult in 2024.
That will allow users to install even bloatware written in Go or other
space-hungry languages while still leaving most of the space unallocated
for additional partitions or volumes to be used for persistent user
data.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
It appears that the CFE boot loader found in the XG6846
cannot load kernels over a certain size, and the old
relocate hack is not working.
What to do? We can build a small U-Boot into the image,
make CFE boot that, place the kernel immediately after
U-Boot, and use U-Boot to boot the system instead.
The compiled u-boot.bin becomes around ~300KB and with
LZMA compression it will swiftly fit into 128KB, so
we use two 64KB erase blocks right after the CFE to
store an imagetag:ed U-Boot.
Reviewed-by: Paul Donald <newtwen+github@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
This is needed to boot the BCM6238-based Inteno XG6846.
Currently this is restricted to the XG6846 board.
Reviewed-by: Paul Donald <newtwen+github@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
This adds a device tree and build options for the XG6846
switch/router to the BMIPS target.
Hardware:
- SoC: Broadcom BCM6328
- CPU: BMIPS4350 V7.5
- RAM: 64 MB DDR
- NOR Flash: 16 MB parallel (CFE and OS)
- Ethernet LAN: 4x 1Gbit
- Ethernet WAN: 2x 1Gbit, fiber and TP
- Buttons: reset
- LEDs: 7 or 8, power and USB LEDs are GPIO-based, the
LAN LEDs are controlled by the Marvell DSA Switch.
- USB: on some versions
- UART: yes
The device ODM (original device manufacturer) is XAVi
http://www.xavi.com.tw/
It is possible to boot the initramfs version
openwrt-bmips-bcm6328-inteno_xg6846-initramfs.elf from
CFE by interrupting the boot on the UART console and downloading
it from a TFTP server e.g.:
CFE> r 192.168.1.2:openwrt-bmips-bcm6328-inteno_xg6846-initramfs.elf
Installation to target flash is not possible using CFE because
the image becomes too big for the CFE version found in these
devices. A separate U-Boot two-stage solution exists for
actually booting the device.
This device is called a "managed ethernet switch" by the vendor
and "media converter" or "fiber modem" by some of the ISPs
using it: the main purpose is to convert fiber connections to
ethernet, most devices just act as switches bridging the
fiber SFP to ethernet TP.
The device has a Marvell MV88E6352 DSA switch managed by
a BCM6328 BMIPS SoC.
This port makes it possible to use the XG6846 to grab an IP
number from the fiber connection and use all four LAN
connections out, turning it into a proper router.
This support is based mostly on the observations by the people on
the forum thread "Help with Inteno XG6846" where users NPeca75,
mrhaav, systemcrash and csom helped out to reverse engineer the
device. Then I made it work on the BMIPS target, figured out
the two-level switch hierarchy and settings.
Link: https://forum.openwrt.org/t/help-with-inteno-xg6846/68276/14
Signed-off-by: Paul Donald <newtwen+github@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Since we split the Inteno XG6846 "firmware" partition with the
uImage MTD splitter, we need to compile in support for this
splitting method into the BCM6328.
Reviewed-by: Paul Donald <newtwen+github@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Commit [ca8c30208d5e][1] updates procd to handle muliple "console=" on the
kernel command line. This affects Raspberry Pi builds because cmdline.txt
specifies a UART console and a virtual console on HDMI, in that order.
When procd finds multiple consoles on the command line, it attempts to
open /dev/console. Linux uses the [last console][2] for /dev/console, so
procd opens the virtual console on Raspberry Pi. This completely disables
the UART console and causes [strange behavior][3] on the virtual console.
Prior to ca8c30208d5e, procd would always open the first console, which is
the UART console.
The simplest fix without reverting ca8c30208d5e is to swap the order of
console options in cmdline.txt. By putting the UART console last, procd
handles the serial console correctly as before.
[1]: https://git.openwrt.org/?p=project/procd.git;a=commit;h=ca8c30208d5e1aaa2c0e3f732c4c9944735e9850
[2]: https://www.kernel.org/doc/html/latest/admin-guide/serial-console.html
[3]: https://forum.openwrt.org/t/rasberry-pi-4-model-b-keyboards-gone-wild/195594
Signed-off-by: Elbert Mai <code@elbertmai.com>
Currently it's needed to have gcc-multilib on the host to correctly
compile xdp-tools. This is wrong and means that we are using host header
to compile a tool.
By some searching in how the makefile works it was discovered that
BPF_CFLAGS were not used and required to be appended to config.mk
Only one single header was added but we should include each BPF_CFLAGS
from bpf.mk. To make this some patching to bpf-header were required and
some patches to xdp-tools were required.
Also it's needed to pass the correct target to BPF_CFLAGS.
With the following changes xdp-tools can correctly compile with each
header from bpf-headers and should not use any host header.
Co-Developed-by: Andre Heider <a.heider@gmail.com>
Signed-off-by: Andre Heider <a.heider@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/11825
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
netlink.h header have NL_SET_ERR_MSG_MOD that is tied to kmods. We don't
need kmods on bpf tools and this cause compilation error if the header
is included. Fix it by dropping NL_SET_ERR_MSG_MOD.
Link: https://github.com/openwrt/openwrt/pull/11825
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Currently, the MT7530 DSA subdriver configures the MT7530 switch to provide
direct access to switch PHYs, meaning, the switch PHYs listen on the MDIO
bus the switch listens on. The PHY muxing feature makes use of this.
This is problematic as the PHY may be attached before the switch is
initialised, in which case, the PHY will fail to be attached.
Since commit 91374ba537bd ("net: dsa: mt7530: support OF-based registration
of switch MDIO bus") on mainline Linux, we can describe the switch PHYs on
the MDIO bus of the switch on the device tree.
When the PHY is described this way, the switch will be initialised first,
then the switch MDIO bus will be registered. Only after these steps, the
PHY will be attached.
Describe the switch PHYs on mt7621.dtsi and remove defining the switch PHY
on the SoC's mdio bus node. When the PHY muxing is in use, the interrupts
for the muxed PHY won't work, therefore delete the "interrupts" property on
the devices where the PHY muxing feature is in use.
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
The ethernet-phy@4 node doesn't exist for WAVLINK WL-WN573HX1. Remove it
and the duplicate gmac0 node.
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
Currently, the pinctrl-0 property on the ethernet node is modified to
exclude the rgmii1 and rgmii2 pin groups to be claimed with rgmii1 and
rgmii2 functions, respectively. Remove the modification of this property as
we need these pin groups to be claimed with the said functions for this
device.
Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>
1. Enable this feature only for 32-bit CPUs as MIPS64 can not
access the full range unmapped uncached memory.
2. Backport this fix to the 6.1 old LTS kernel.
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
There was no config in the uboot-envtools package, so there is no
generated /etc/fw_env.config for the fw_printenv and fw_setenv utils.
Since uboot-envtools 2024.01, there is a way to make these utils work
without /etc/fw_env.config if the DT has an env partition with the prop.:
compatible = "u-boot,env";
So, this commit adds the prop. above to the appsblenv:0 partition
in the yuncore ax880 DTS file.
Signed-off-by: Isaev Ruslan <legale.legale@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15305
Signed-off-by: Robert Marko <robimarko@gmail.com>
In the 0.191 update dirname was used instead of <libgen.h> to fix the
poisoned basename error:
/usr/include/libgen.h:35:9: error: attempt to use poisoned "basename"
35 | #define basename __xpg_basename
However, doing this has lead to libelf.a pulling in xmalloc, xstrdup and
friends and statically linking them thus leading to a symbol name conflict
with FRR host build and anything else that links against libelf and uses
xmalloc and friends.
Well, it turns out that upstream has added a helper[1] for basename so it
can compile with musl 1.2.5 which dropped the basename declaration, but it
also means that we must NOT include <libgen.h> and that poisoned error is
intentional and added to prevent duplicate basename definitions.
This also means that for macOS we dont need to do any additional header
inclusions as the new helper takes care of basename.
So, to fix the symbol conflict we can simply drop the <dirname.h> inclusion
and build from elfutils.
Tested on Fedora 40 as well as macOS 14.4.1.
[1] https://sourceware.org/git/?p=elfutils.git;a=commit;h=a2194f6b305bf0d0b9dd49dccd0a5c21994c8eeaFixes: #24030
Fixes: b6f025b424 ("tools/elfutils: update to 1.91")
Link: https://github.com/openwrt/openwrt/pull/15337
Signed-off-by: Robert Marko <robimarko@gmail.com>
