realtek: simplify RTL8214FC patches for ethtool copper/fiber switching
There is a patch/bug cascade in the realtek target phy code that must be resolved.
1. The phy_driver structure is patched to add features ONLY needed for RTL8214FC
2. The kernel is patched to allow switching fiber/copper port of phys through ethtool
by calling these new features.
3. With those patches applied the bootup always switches RTL8214FC ports to copper.
Even if a SFP module was found before and the phy driver switched to fibre before.
3. So another patch is needed that reprobes the SFP module to activate fiber again.
4. Because of the reprobing we need a fourth patch that avoid duplicate devices.
Simplify this by removing all patches and reusing the existing ethtool phy tunable
interface. The command line usage might be counterintuitive but it avoids tons of
problems in the code. In addition, this scenario is not used frequently.
Mathew McBride [Thu, 16 Jan 2025 01:27:58 +0000 (12:27 +1100)]
armsr: remove 'console=tty1' from kernel command line
We have modified the kernel to setup all "default" consoles,
including serial ports and framebuffers/screens, providing
no console= argument is supplied on the kernel command line.
Adding 'console=tty1' caused the 'default' serial port on
device tree systems to break, as the kernel would not carry
over the settings (like baud rate) from the bootloader.
The system administrator can still force the use of a
specific console by adding their own console= arguments.
Signed-off-by: Mathew McBride <matt@traverse.com.au> Fixes: c099523d66a3 ("armsr: use console=tty1 to make
console more readily available") Link: https://github.com/openwrt/openwrt/pull/17012 Signed-off-by: Robert Marko <robimarko@gmail.com>
Mathew McBride [Thu, 16 Jan 2025 01:20:56 +0000 (12:20 +1100)]
armsr: add patch (hack) to enable all "default" consoles
A previous change added 'console=tty1' to the default kernel command
line on armsr, in order to ensure the framebuffer console is enabled
on systems capable of graphics output.
Unfortunately, this change broke boards that used device tree
(DT) firmware with serial consoles, as the serial console
specified by the system firmware (stdout-path) was no longer
setup by the kernel.
A bit of probing determined that the SPCR (serial port console
direction table) on ACPI systems was preventing Linux from setting
up a default framebuffer console on these systems (which is why
console=tty1 was added).
(The affected ACPI systems are usually VMs using QEMU's
'virt' machine and EDK2 firmware. The firmware on these systems
does not remove the SPCR when a screen is present)
So to ensure all possible systems are setup correctly, we modify
the kernel so all "default" console types (serial and screen)
are setup when no console= arguments are specified on the kernel
command line.
Signed-off-by: Mathew McBride <matt@traverse.com.au> Fixes: c099523d66a3 ("use console=tty1 to make console more
readily available") Link: https://github.com/openwrt/openwrt/pull/17012 Signed-off-by: Robert Marko <robimarko@gmail.com>
Antti Seppälä [Fri, 16 May 2025 17:18:54 +0000 (20:18 +0300)]
adb: Switch to mbedtls
Currently adb uses libopenssl for certain authentication tasks between
the host and the target device such as certificate generation, hashing,
base64 encoding and pki signatures.
Add a patch to use functionalities available in mbedtls instead.
Also switch package makefile and dependency to libmbedtls and drop
patches and references to lib{crypto,openssl} as they are no longer
required.
This conserves considerable amount of space on the device as openwrt
ships with libmbedtls by default.
The MIPS CPC (Cluster Power Controller) is setup during boot and can take
its configuration from the devicetree. This is currently not possible
because the cpc probing happens before dt initialization. Call order
during startup is:
setup_arch()
prom_init() <- our function
mips_cpc_probe()
smp_stuff()
arch_mem_init()
device_tree_init() <- our function
unflatten_and_copy_device_tree()
To avoid ugly hacking and support a clean devicetree relocate the cps/smp
stuff to device_tree_init(). This is basically the same location as in
generic mips initialization.
Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de> Signed-off-by: Sebastian Gottschall <s.gottschall@dd-wrt.com> Link: https://github.com/openwrt/openwrt/pull/18888 Signed-off-by: Robert Marko <robimarko@gmail.com>
realtek: take over boot command line patch into driver
Make it easier to upgrade the kernel in the future. For this remove
the 320-harden-fw_init_cmdline.patch and add the logic into the
startup sequence at the right place.
Shiji Yang [Wed, 30 Apr 2025 16:33:41 +0000 (00:33 +0800)]
ath79: refresh kernel symbol configs
Manually move the common NVMEM Kconfig symbols to the shared
config file and disable them for the mikrotik subtarget. Other
changes are automatically completed by `make kernel_oldconfig`.
This is a preparation for introducing the 6.12 kernel support.
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Robert Marko [Sat, 24 May 2025 10:20:54 +0000 (12:20 +0200)]
image: only filter out images when ImageBuilder is used
Currently, we are filtering out images if DEFAULT:=n or BROKEN:=y are set,
so if you are building from scratch and want to build custom images that
are stripped down to fit, you must edit the image recipe or its just
filtered out.
So, to allow this behaviour when building from scratch as we can assume
that person doing that knows what they are attempting to do lets just limit
the filtering to ImageBuilder.
Fixes: f060615a78e5 ("image: respect DEFAULT and BROKEN when Default profile is selected") Signed-off-by: Robert Marko <robimarko@gmail.com>
RTL931x devices are dual core with two threads each. That is a total
of 4 VPEs (CPUs) in the kernel. Adapt the kernel config for that.
Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de> Signed-off-by: Sebastian Gottschall <s.gottschall@dd-wrt.com> Link: https://github.com/openwrt/openwrt/pull/18889 Signed-off-by: Robert Marko <robimarko@gmail.com>
Debian Changelogs from 20250512:
* New upstream microcode datafile 20250512
- Mitigations for INTEL-SA-01153 (ITS: Indirect Target Selection):
CVE-2024-28956: Processor may incompletely mitigate Branch Target
Injection due to indirect branch predictions that are not fully
constrained by eIBRS nor by the IBPB barrier. Part of the "Training
Solo" set of vulnerabilities.
- Mitigations for INTEL-SA-01244:
CVE-2025-20103: Insufficient resource pool in the core management
mechanism for some Intel Processors may allow an authenticated user
to potentially enable denial of service via local access.
CVE-2025-20054: Uncaught exception in the core management mechanism
for some Intel Processors may allow an authenticated user to
potentially enable denial of service via local access.
- Mitigations for INTEL-SA-01247:
CVE-2024-43420, CVE-2025-20623: Exposure of sensitive information
caused by shared microarchitectural predictor state that influences
transient execution for some Intel Atom and some Intel Core
processors (10th Generation) may allow an authenticated user to
potentially enable information disclosure via local access.
CVE-2024-45332 (Branch Privilege Injection): Exposure of sensitive
information caused by shared microarchitectural predictor state that
influences transient execution in the indirect branch predictors for
some Intel Processors may allow an authenticated user to potentially
enable information disclosure via local access.
- Mitigations for INTEL-SA-01322:
CVE-2025-24495 (Training Solo): Incorrect initialization of resource
in the branch prediction unit for some Intel Core Ultra Processors
may allow an authenticated user to potentially enable information
disclosure via local access (IBPB bypass)
CVE-2025-20012 (Training Solo): Incorrect behavior order for some
Intel Core Ultra Processors may allow an unauthenticated user to
potentially enable information disclosure via physical access.
- Improved fix for the Vmin Shift Instability for the Intel Core 13th
and 14th gen processors under low-activity scenarios (sig 0xb0671).
This microcode update is supposed to be delivered as a system
firmware update, but according to Intel it should be effective when
loaded by the operating system if the system firmware has revision
0x12e.
- Fixes for unspecified functional issues on several processor models
* New microcodes or new extended signatures:
sig 0x000a06d1, pf_mask 0x95, 2025-02-07, rev 0x10003a2, size 1664000
sig 0x000a06d1, pf_mask 0x20, 2025-02-07, rev 0xa0000d1, size 1635328
sig 0x000b0650, pf_mask 0x80, 2025-03-18, rev 0x000a, size 136192
sig 0x000b06d1, pf_mask 0x80, 2025-03-18, rev 0x011f, size 79872
sig 0x000c0662, pf_mask 0x82, 2025-03-20, rev 0x0118, size 90112
sig 0x000c06a2, pf_mask 0x82, 2025-03-20, rev 0x0118
sig 0x000c0652, pf_mask 0x82, 2025-03-20, rev 0x0118
sig 0x000c0664, pf_mask 0x82, 2025-03-20, rev 0x0118
* Updated microcodes:
sig 0x00050657, pf_mask 0xbf, 2024-12-12, rev 0x5003901, size 39936
sig 0x0005065b, pf_mask 0xbf, 2024-12-12, rev 0x7002b01, size 30720
sig 0x000606a6, pf_mask 0x87, 2025-01-07, rev 0xd000404, size 309248
sig 0x000606c1, pf_mask 0x10, 2025-01-07, rev 0x10002d0, size 300032
sig 0x000706a8, pf_mask 0x01, 2024-12-05, rev 0x0026, size 76800
sig 0x000706e5, pf_mask 0x80, 2025-01-07, rev 0x00ca, size 115712
sig 0x000806c1, pf_mask 0x80, 2024-12-01, rev 0x00bc, size 112640
sig 0x000806c2, pf_mask 0xc2, 2024-12-01, rev 0x003c, size 99328
sig 0x000806d1, pf_mask 0xc2, 2024-12-11, rev 0x0056, size 105472
sig 0x000806ec, pf_mask 0x94, 2024-11-17, rev 0x0100, size 106496
sig 0x000806f8, pf_mask 0x87, 2025-01-28, rev 0x2b000639, size 591872
sig 0x000806f7, pf_mask 0x87, 2025-01-28, rev 0x2b000639
sig 0x000806f6, pf_mask 0x87, 2025-01-28, rev 0x2b000639
sig 0x000806f5, pf_mask 0x87, 2025-01-28, rev 0x2b000639
sig 0x000806f4, pf_mask 0x87, 2025-01-28, rev 0x2b000639
sig 0x000806f8, pf_mask 0x10, 2025-01-28, rev 0x2c0003f7, size 624640
sig 0x000806f6, pf_mask 0x10, 2025-01-28, rev 0x2c0003f7
sig 0x000806f5, pf_mask 0x10, 2025-01-28, rev 0x2c0003f7
sig 0x000806f4, pf_mask 0x10, 2025-01-28, rev 0x2c0003f7
sig 0x00090672, pf_mask 0x07, 2024-12-12, rev 0x003a, size 226304
sig 0x00090675, pf_mask 0x07, 2024-12-12, rev 0x003a
sig 0x000b06f2, pf_mask 0x07, 2024-12-12, rev 0x003a
sig 0x000b06f5, pf_mask 0x07, 2024-12-12, rev 0x003a
sig 0x000b06f6, pf_mask 0x07, 2024-12-12, rev 0x003a
sig 0x000b06f7, pf_mask 0x07, 2024-12-12, rev 0x003a
sig 0x000906a3, pf_mask 0x80, 2024-12-12, rev 0x0437, size 224256
sig 0x000906a4, pf_mask 0x80, 2024-12-12, rev 0x0437
sig 0x000906a4, pf_mask 0x40, 2024-12-06, rev 0x000a, size 119808
sig 0x000906ed, pf_mask 0x22, 2024-11-14, rev 0x0104, size 106496
sig 0x000a0652, pf_mask 0x20, 2024-11-14, rev 0x0100, size 97280
sig 0x000a0653, pf_mask 0x22, 2024-11-14, rev 0x0100, size 98304
sig 0x000a0655, pf_mask 0x22, 2024-11-14, rev 0x0100, size 97280
sig 0x000a0660, pf_mask 0x80, 2024-11-14, rev 0x0102, size 98304
sig 0x000a0661, pf_mask 0x80, 2024-11-14, rev 0x0100, size 97280
sig 0x000a0671, pf_mask 0x02, 2024-12-01, rev 0x0064, size 108544
sig 0x000a06a4, pf_mask 0xe6, 2025-02-13, rev 0x0024, size 140288
sig 0x000a06f3, pf_mask 0x01, 2025-02-10, rev 0x3000341, size 1542144
sig 0x000b0671, pf_mask 0x32, 2025-03-17, rev 0x012f, size 219136
sig 0x000b0674, pf_mask 0x32, 2025-03-17, rev 0x012f
sig 0x000b06a2, pf_mask 0xe0, 2025-01-15, rev 0x4128, size 224256
sig 0x000b06a3, pf_mask 0xe0, 2025-01-15, rev 0x4128
sig 0x000b06a8, pf_mask 0xe0, 2025-01-15, rev 0x4128
sig 0x000b06e0, pf_mask 0x19, 2024-12-06, rev 0x001d, size 139264
sig 0x000c06f2, pf_mask 0x87, 2025-03-14, rev 0x210002a9, size 563200
sig 0x000c06f1, pf_mask 0x87, 2025-03-14, rev 0x210002a9
* Removed microcodes (ES/QS steppings):
sig 0x00050656, pf_mask 0xbf, 2023-07-28, rev 0x4003605, size 38912
sig 0x000c06f1, pf_mask 0x87, 2025-03-14, rev 0x210002a9 [EXCLUDED]
Rosen Penev [Tue, 20 May 2025 22:06:33 +0000 (15:06 -0700)]
tools/mold: remove static only dependencies
It seems that MOLD_MOSTLY_STATIC was required before to avoid packaging
every single dependency. But nowadays, mold uses whatever dependency it
can find and builds its own if it cannot.
realtek: fix MAC_FORCE_MODE_CTRL register for RTL931x
Ports will not work as expected on the RTL931x devices. This comes
from a typo in the address. Fixed according to the documentation:
https://svanheule.net/realtek/mango/register/mac_force_mode_ctrl
Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de> Signed-off-by: Sebastian Gottschall <s.gottschall@dd-wrt.com> Link: https://github.com/openwrt/openwrt/pull/18864 Signed-off-by: Robert Marko <robimarko@gmail.com>
This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
Shiji Yang [Sun, 18 May 2025 11:29:30 +0000 (19:29 +0800)]
uboot-bcm4908: fix build with GCC14
A lot of warnings were treated as errors after the default compiler
switched to GCC14. It's hard to fix them one by one, and this u-boot
is not maintained by upstream, so let's just silence these warnings.
Since the early beginning of the Realtek DSA driver there is an uncovered
locking issue between the standard (parent) mdio bus and the DSA (child)
mdio bus. This comes from the fact that the DSA bus simply links to the
parent read and write functions and calls them directly. This leads to
the following lock issue.
- Child bus calls phy_read/write functions and uses its internal lock
- Parent bus calls phy_read/write functions and uses its internal lock
It becomes clear that critical section can be accessed twice without
knowing that a operation from the other bus is currently active. This
can lead to critical malfunctions because the mdio driver needs a lot of
internal magic to get page selection done right. Effects are:
- The original page is lost after a phy_write/read_paged() call
- dmesg like "Realtek RTL8218B (external) rtl838x slave mii-0:00:
Expected external RTL8218B, found PHY-ID 6b23"
Other DSA drivers simply use the read/write functions from the parent bus
and thus avoid locking issues. Do it the same way.
Fixes: 2b88563ee5aafd9 ("realtek: update the tree to the latest refactored version") Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de> Link: https://github.com/openwrt/openwrt/pull/18824 Signed-off-by: Robert Marko <robimarko@gmail.com>
Shiji Yang [Sun, 18 May 2025 12:42:19 +0000 (20:42 +0800)]
ramips: fix wrg-header image recipe
Before generating the factory image, check if the input file
exists. Fix the build error when sysupgrade image is too big:
[mkwrgimg] *** error: stat failed on /builder/shared-workdir/build/build_dir/target-mipsel_24kc_musl/linux-ramips_rt288x/tmp/openwrt-ramips-rt288x-airlink101_ar670w-squashfs-factory.bin, No such file or directory
Zoltan HERPAI [Sat, 3 May 2025 22:29:54 +0000 (00:29 +0200)]
kernel/d1: Restore kernel files for v6.6
This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
There are currently two power functions for the RTL8214FC in the driver
- rtl8380_rtl8214fc_on_off(): powers on/off both ports
- rtl8214fc_power_set(): powers on/off fibre or copper individually
While it seems obvious to merge these, one of them uses extended register
29 and the other register 30 to achieve the update. From looking at the
GPL source drops both methods exist with exactly the same operations. So
where to go from here? The descision was made based on:
- register 30 is only used during patching, 29 during normal operation
- Avoid raw page if possible to use function for RTL839x
So the attached patch was created. The result in the power down bits
was compared for the two old and the new implementation. Result is
always the same.
RTL9311 switches (like the Linksys LGS352C) will boot but have not set
the right family id in the soc_info structure. Fix that so drivers do
their work right when checking for that.
Commit 0912d3f59547 ("kernel: bump 6.12 to 6.12.29") was merged right after
bcm27xx 6.12 support commits, causing conflicts.
Refresh patches and removed upstreamed patch in v6.12.29:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-6.12.y&id=a5f162727b91e480656da1876247a91f651f76de
realtek: refactor net rx interrupt handler rtl83xx_net_irq()
Cleanup the code of the RTL83xx packet receive interrupt handler. Not
only for better readability but to avoid inconsistencies and stalls on
the RTL839x targets.
The current implementation seems to come from the GPL source code.
Calling the existing cleanup() function inside the interrupt context
without any locks conflicts with SMP & NAPI polling and makes things
worse instead of giving any benefit. Simply ignore RX buffer overruns
and let the device handle packet dropping itself.
Konstantin Demin [Wed, 21 May 2025 05:34:16 +0000 (08:34 +0300)]
toolchain: glibc: Update glibc 2.38 to recent HEAD
3be3728d elf: Ignore LD_LIBRARY_PATH and debug env var for setuid for static 454f24e9 support: Use const char * argument in support_capture_subprogram_self_sgid 4335cd9b support: Add support_record_failure_barrier 31fa0f73 elf: Test case for bug 32976 (CVE-2025-4802) ca41fe44 x86_64: Fix typo in ifunc-impl-list.c.
These patches were generated from:
https://github.com/raspberrypi/linux/commits/rpi-6.12.y
With the following command:
git format-patch -N v6.12.27..HEAD
(HEAD -> 8d3206ee456a5ecdf9ddbfd8e5e231e4f0cd716e)
This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
Import pending PXA I2C recovery fixes so that if I2C recovery is enabled in
the DTS it does not completely break I2C as it currently does since kernel
6.6.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
INAGAKI Hiroshi [Sat, 17 May 2025 11:22:57 +0000 (20:22 +0900)]
ramips: mt7621: fix broken hexdump expression in iodata.sh
Use the "1/1" as a iteration count of hexdump expression instead of the
broken "1", to fix the issue that the invalid bootnum will be obtained.
Currently, the hexdump command always outputs "0" when the decimal format
and the iteration count "1" are specified[0]. This is unexpected
behaviour, but the cause is unknown and use this fix as a workaround.
Konstantin Demin [Tue, 20 May 2025 07:04:14 +0000 (10:04 +0300)]
toolchain: glibc: set min kernel version to 6.6
Minimum kernel version is 6.6 so it's safe to increase minimum kernel version to 6.6 for glibc too.
This change effectively removes last compatibility quirk in [1] which leads to better/faster detection of SVE support in string ifuncs for aarch64.
Furthermore, this change will allow to use "fchmodat2" syscall within fchmodat() wrapper in future glibc versions (starting from 2.39) as noted in [2] and [3].
projects / thirdparty / openwrt.git / log
