In amdxdna_remove(), all amdxdna_client structures are freed after
calling drm_dev_unplug(). However, drm_dev_unplug() does not force
existing file descriptors to be closed, so amdxdna_drm_close() may be
called after amdxdna_remove() has completed.
As a result, accessing client->pid for debug output in
amdxdna_drm_close() can lead to a use-after-free, since the access is
not protected by drm_dev_enter().
Fix this by decoupling hardware teardown from client cleanup.
amdxdna_remove() only performs hardware-related cleanup, while
per-client resources are released from amdxdna_drm_close() when the
corresponding file is closed.
Fixes: be462c97b7 ("accel/amdxdna: Add hardware context")
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://patch.msgid.link/20260611055150.3070216-1-lizhi.hou@amd.com
Convert the mtk-btcvsd-snd.txt DT binding to DT Schema format.
Signed-off-by: Luca Leonardo Scorcia <l.scorcia@gmail.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>
Link: https://patch.msgid.link/20260615185810.11804-1-l.scorcia@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
It seems some ARM header includes this and the build passes there, but
nowhere else. Note that the driver has COMPILE_TEST in depends.
Fixes: 37ae8501cd ("clk: stm32: introduce clocks for STM32MP21 platfor")
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The type will be used for Intel USB Audio Offload Link (UAOL) DAI.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://patch.msgid.link/20260615134439.1044872-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
aic26_probe() creates the keyclick sysfs file on the component device.
The file callback uses the ASoC component pointer stored in the driver
private data.
There is no matching remove callback, so the sysfs file can remain after
the component is removed while its backing component state is gone.
Add a component remove callback that removes the keyclick file.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Link: https://patch.msgid.link/20260615064549.34110-1-pengpeng@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
sma1303_i2c_probe() creates a sysfs group that exposes the fault-check
controls. The check_fault_status store callback can queue
check_fault_work.
sma1303_i2c_remove() only cancels the delayed work. It does not remove
the sysfs group, so the controls can remain published after remove while
their callbacks still use the driver data and can queue the work again.
Remove the sysfs group before cancelling the delayed work.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Link: https://patch.msgid.link/20260615064456.28615-1-pengpeng@iscas.ac.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
If request_irq() fails, report "can not request IRQ" rather than "can
not get IRQ" which may be misread as platform_get_irq() failure.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Link: https://patch.msgid.link/20260615044039.9750-3-jszhang@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Even with proper dma engine support, the first spi transfer always
fallback to PIO, the reason is the dws->n_bytes is 0 after
initialization, so the dw_spi_can_dma() calling from __spi_map_msg()
return false, thus both tx_sg_mapped and rx_sg_mapped are false, so
for the first spi transfer, the spi_xfer_is_dma_mapped() reports false
thus fallback to PIO.
Although this brings no harm, we can simply fix this issue by
calcuating the "n_bytes" from xfer->bits_per_word.
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Link: https://patch.msgid.link/20260615044039.9750-2-jszhang@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
ma35d1_clk_pll_determine_rate() called ma35d1_pll_find_closest()
unconditionally before the switch statement, and then every case
branch overwrote pll_freq by reading the current hardware registers.
For CAPLL and DDRPLL this means find_closest() ran unnecessarily
(and incorrectly, since those PLLs are read-only) and its result
was silently discarded.
Fix by moving the find_closest() call inside the APLL/EPLL/VPLL
branch where it belongs. Group CAPLL and DDRPLL together as
read-only PLLs that simply report their current rate; handle them
with an explicit if/else to keep the CAPLL (SMIC design) and DDRPLL
(standard design) paths distinct.
Fixes: 691521a367 ("clk: nuvoton: Add clock driver for ma35d1 clock controller")
Signed-off-by: Joey Lu <a0987203069@gmail.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
PLL_CTL1_FRAC was defined as GENMASK(31, 24), covering only 8 bits.
The hardware fractional field occupies bits [31:8] (24 bits), so the
mask must be GENMASK(31, 8).
The previous fractional-mode calculation used FIELD_MAX(PLL_CTL1_FRAC)
as the denominator to obtain 2 decimal places. With the corrected 24-bit
mask the old divisor is wrong; replace the arithmetic with a proper
24-bit fixed-point rounding to 3 decimal places using the kernel's
DIV_ROUND_CLOSEST_ULL helper:
n_frac = n * 1000 + DIV_ROUND_CLOSEST_ULL(x * 1000, 1 << 24)
Fixes: 691521a367 ("clk: nuvoton: Add clock driver for ma35d1 clock controller")
Signed-off-by: Joey Lu <a0987203069@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
div_u64() does not modify its argument in place; the return value must
be assigned. Both ma35d1_calc_smic_pll_freq() and ma35d1_calc_pll_freq()
called div_u64() and discarded the result, leaving pll_freq holding the
undivided product and thus returning a frequency orders of magnitude too
high.
Fixes: 691521a367 ("clk: nuvoton: Add clock driver for ma35d1 clock controller")
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Joey Lu <a0987203069@gmail.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
This patch allows policies to be compressed in userspace and be sent to
the kernel through the existing ".load" and ".replace" kernel interfaces.
The benefits of this approach are:
- Save kernel time when loading policies
- Allow userspace to provide a higher level of compression than the one
provided by the kernel (ZSTD_CLEVEL_DEFAULT), thus saving space.
- Allow small embedded systems to only store the compressed version of
policies in userspace, saving memory.
Userspace-compressed policies improve system time by up to ~30% for big
profiles.
Signed-off-by: Maxime Bélair <maxime.belair@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Besides of resolving clangd IDE warnings, self-contained headers will be
less likely to break if the surrounding includes in .c files using them
change.
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
While the #ifdef guards prevent the circular include from blowing up,
policy.h does not actually need anything from net.h. Remove, that include
and instead include net.h in the other files that need it.
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The MediaTek MT8196 vlpckgen clock driver uses
__devm_regmap_init_mmio_clk() by devm_regmap_init_mmio(),
which is defined in drivers/base/regmap/regmap-mmio.c.
However, the driver's Kconfig entry does not select REGMAP_MMIO.
This causes a linker error when REGMAP_MMIO is not enabled.
Fix this by selecting REGMAP_MMIO in the Kconfig entry.
Fixes: 2f8b3ae6f0 ("clk: mediatek: Add MT8196 vlpckgen clock support")
Cc: stable@vger.kernel.org
Signed-off-by: Akari Tsuyukusa <akkun11.open@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Not only these error checks are redundand,
those of_clk_get() return values weren't cleaned up via clk_put().
Fixes: c7bb4fc16e ("clk: add MOXA ART SoCs clock driver")
Signed-off-by: Alexander A. Klimov <grandmaster@al2klimov.de>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Named initializers are better readable and more robust to changes of the
struct definition. This robustness is relevant for a planned change to
struct platform_device_id replacing .driver_data by an anonymous union.
While touching these arrays unify spacing and usage of commas.
Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@baylibre.com>
Reviewed-by: Matti Vaittinen <mazziesaccount@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Stop defaulting to 0 on BAD_CLK_INDEX (U8_MAX). Passing the out-of-bounds
index directly allows the clk core to fail gracefully during parent
lookup.
Link: https://lore.kernel.org/lkml/177431305509.5403.15386021337517970667@lazor/
Signed-off-by: Yu-Chun Lin <eleanor.lin@realtek.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
device_reset() may return -EPROBE_DEFER. Switch to dev_err_probe() so
probe failures are reported consistently and deferred probing is handled
properly.
This matches the existing pattern used in aiu_probe().
No functional change intended.
Signed-off-by: bui duc phuc <phucduc.bui@gmail.com>
Link: https://patch.msgid.link/20260612020113.9557-1-phucduc.bui@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
zynq_clk_register_periph_clk() ignores the return value of the two
kasprintf() calls used to build the mux and divider clock names, and
passes the resulting (possibly NULL) pointers straight into
clk_register_mux(), clk_register_divider() and clk_register_gate() as
the clock '"'name'"' argument. On allocation failure that name later
gets dereferenced by the clock framework (e.g. in debugfs name
formatting), causing a NULL-pointer dereference.
Check both kasprintf() returns. On failure unwind any allocated name
buffer and the spinlock, then fall through to the existing err label
which sets clks[] to ERR_PTR(-ENOMEM). Freeing the spinlock on the
error path is correct here because no clk_register_*() call has had
a chance to take ownership of it; the success path intentionally
hands it off to the registered clocks.
The neighbouring zynq_clk_register_fclk() in the same file already
uses this per-allocation goto-label cleanup pattern; this change
brings periph_clk into line with it.
Signed-off-by: William Theesfeld <william@theesfeld.net>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
On multi-socket platforms each socket has its own BPMP which exposes the
same clock names. Fix this by using the NUMA ID as a prefix for the
clock names on multi-socket platforms.
Use 'sizeof(info->name)' in the strscpy() and snprintf() functions to
future proof against anyone changing the size of the 'name' array.
Co-developed-by: Timo Alho <talho@nvidia.com>
Signed-off-by: Timo Alho <talho@nvidia.com>
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
clk_divider_bestdiv() clamps maxdiv using:
maxdiv = min(ULONG_MAX / rate, maxdiv);
to avoid overflow in rate * i. However, requests like
clk_round_rate(clk, ULONG_MAX), which are used to determine the maximum
supported rate of a clock, result in maxdiv being clamped to 1. If no
valid divider of 1 exists in the table the loop is never entered and
bestdiv falls back to the maximum divider with the minimum parent rate,
causing clk_round_rate(clk, ULONG_MAX) to incorrectly return the minimum
supported rate instead of the maximum.
Fix this by removing the pre-loop maxdiv clamping and replacing the
unprotected rate * i multiplication with check_mul_overflow(). Guard
the exact-match short-circuit with !overflow to prevent a clamped
target_parent_rate of ULONG_MAX from falsely matching parent_rate_saved
and causing premature loop exit. Break out of the loop after evaluating
the first overflowing divider since clk_hw_round_rate(parent, ULONG_MAX)
returns a constant for all subsequent iterations, meaning no better
candidate can be found, and continuing would cause exponential recursive
calls in chained divider clocks.
Update the KUnit test expected values to reflect the corrected behaviour:
- clk_divider_bestdiv_ulong_max_returns_max_rate: PARENT_RATE_1GHZ / 8
(minimum rate, pre-fix) -> PARENT_RATE_1GHZ / 2 (maximum rate)
- clk_divider_bestdiv_mux_ulong_max_returns_max_rate: 0 (invalid,
pre-fix) -> PARENT_RATE_4GHZ / 2 (maximum rate with mux selecting
the 4 GHz parent and applying the smallest table divider of 2)
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add KUnit tests to verify the behaviour of clk_divider_bestdiv() when
clk_round_rate() is called with ULONG_MAX, which is the canonical way
to probe the maximum rate a clock can produce.
Two test cases are introduced:
- clk_divider_bestdiv_ulong_max_returns_max_rate: registers a 1 GHz
fixed-rate parent driving a table-based divider whose smallest entry
is div=2 (entries: 2, 4, 8). Calls clk_hw_round_rate(div_hw, ULONG_MAX)
and checks the result.
- clk_divider_bestdiv_mux_ulong_max_returns_max_rate: places a two-input
mux (4 GHz and 2 GHz fixed-rate parents, CLK_SET_RATE_PARENT) ahead of
the same table-based divider to verify correct parent selection under
ULONG_MAX.
Both tests use an explicit clk_div_table with a minimum divider of 2 so
that the pre-loop maxdiv clamping in clk_divider_bestdiv():
maxdiv = min(ULONG_MAX / rate, maxdiv);
clamps maxdiv to 1, causing _next_div() to return 2 on the first
iteration and skip the loop body entirely. This makes bestdiv fall back
to the maximum divider, returning the minimum rate rather than the
maximum.
The expected values intentionally reflect the buggy output:
- test 1: PARENT_RATE_1GHZ / 8 (minimum rate, not maximum)
- test 2: 0 (invalid, loop never populated bestdiv)
These will be corrected to PARENT_RATE_1GHZ / 2 and PARENT_RATE_4GHZ / 2
respectively once the fix to clk_divider_bestdiv() is applied.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add auxiliary driver to support ESWIN EIC7700 high-speed peripherals
system. The reset controller is created using the auxiliary device
framework and set up in the clock driver.
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add driver for the ESWIN EIC7700 high-speed peripherals system
clock controller and register an auxiliary device for system
reset controller which is named as "hsp-reset".
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add bindings for the high-speed peripherals clock and reset generator
on the ESWIN EIC7700 HSP.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The gate_noc_nsp_clk provides the essential clock source for NPU,
DSP, and PCIe subsystems. During kernel init, the clock framework
attempts to disable unused clocks when clk_ignore_unused kernel
parameter is not set.
However, gate_noc_nsp_clk is required to remain enabled for these
critical subsystems to function properly, causing PCIe boot failures
when auto-disabled.
Add CLK_IGNORE_UNUSED flag to gate_noc_nsp_clk to ensure it stays
enabled even when clk_ignore_unused is not specified in kernel
command line.
Fixes: cd44f127c1 ("clk: eswin: Add eic7700 clock driver")
Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
When using driver on Zynq-7000 (32-bit) determine_rate calculation
overflows. For instance requesting 32MHz with 100MHz parent clock
results in 100000000*(4*1000+0) 32-bit multiplication.
Replace the expression with mult_frac which is already used in
clk_wzrd_recalc_ratef.
Cc: stable@vger.kernel.org
Fixes: 7681f64e64 ("clk: clocking-wizard: calculate dividers fractional parts")
Signed-off-by: Pale Löbl <pavel@loebl.cz>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The buffer length is known to be KUNIT_PARAM_DESC_SIZE
Signed-off-by: David Laight <david.laight.linux@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Simplify allocation by using a flexible array member and kzalloc_flex to
combine allocations.
Add __counted_by for extra runtime analysis. Move counting variable
assignment to right after allocation. kzalloc_flex does this
automatically with GCC >= 15.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
- Introduce 'clk-scmi-oem.c' to support vendor-specific OEM extensions
for the SCMI clock driver, allows clean integration of vendor-specific
features without impacting the core SCMI clock driver logic.
- Extend 'clk-scmi.h' with 'scmi_clk_oem' structure and related
declarations.
- Initialize OEM extensions via 'scmi_clk_oem_init()'.
- Support querying OEM-specific features and setting spread spectrum.
- Pass 'scmi_device' to 'scmi_clk_ops_select()' for OEM data access.
Reviewed-by: Sebin Francis <sebin.francis@ti.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Added a new header file 'clk-scmi.h' to define common structures and
interfaces for the SCMI clock driver. This header will also be used by
OEM-specific extensions to ensure consistency and reusability.
Moved relevant structure definitions from the driver implementation to
'clk-scmi.h' to facilitate shared usage.
Reviewed-by: Sebin Francis <sebin.francis@ti.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add KUnit test coverage for the assigned-clock-sscs DT property that
configures spread spectrum on clocks before they are used.
Extend the existing test infrastructure to support spread spectrum:
- Add struct clk_spread_spectrum field to clk_dummy_context and a
clk_dummy_set_spread_spectrum callback
- Wire set_spread_spectrum into all dummy clock ops
- Extend clk_assigned_rates_register_clk and test parameter struct
to propagate initial SSCS values
Add a new separate test suite clk_assigned_sscs with three categories:
1. clk_assigned_sscs_assigns_one — verifies that a single
assigned-clock-sscs entry correctly configures spread spectrum
on one clock, testing both provider and consumer paths
2. clk_assigned_sscs_assigns_multiple — verifies that multiple
assigned-clock-sscs entries configure spread spectrum on two
clocks, testing both provider and consumer paths
3. clk_assigned_sscs_skips — verifies that malformed DT properties
are correctly skipped without error: missing assigned-clocks,
zero-valued SSCS, and null phandles, tested for both provider
and consumer scenarios
New DT overlays are added for all test scenarios:
- kunit_clk_assigned_sscs_one{,consumer} — single valid entry
- kunit_clk_assigned_sscs_multiple{,consumer} — two valid entries
- kunit_clk_assigned_sscs_without{,consumer} — missing assigned-clocks
- kunit_clk_assigned_sscs_zero{,consumer} — all-zero SSCS values
- kunit_clk_assigned_sscs_null{,consumer} — null phandle
Co-developed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Parse the Spread Spectrum Configuration(SSC) from device tree and configure
them before using the clock.
Each SSC is three u32 elements which means '<modfreq spreaddepth
modmethod>', so assigned-clock-sscs is an array of multiple three u32
elements.
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Sebin Francis <sebin.francis@ti.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Add clk_hw_set_spread_spectrum to configure a clock to enable spread
spectrum feature. set_spread_spectrum ops is added for clk drivers to
have their own hardware specific implementation.
Reviewed-by: Brian Masney <bmasney@redhat.com>
Reviewed-by: Sebin Francis <sebin.francis@ti.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
Per dt-schema, the modulation methods are: down-spread(3), up-spread(2),
center-spread(1), no-spread(0). So define them in dt-bindings to avoid
write the magic number in device tree.
Reviewed-by: Brian Masney <bmasney@redhat.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Reviewed-by: Sebin Francis <sebin.francis@ti.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The "reg" property is an address-sized DT cell property. The AT91
compat clock parser only uses a small bus id from it, but reading it
with the u8 helper does not match the property encoding.
Use of_property_read_reg() so the code goes through the helper for
"reg" properties, then keep the existing range check before passing
the bus id to the clock registration code.
Assisted-by: Codex:gpt-5-5
Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The rate set_rate helpers perform a read-modify-write on the divider
and multiplier registers but only ever OR the new value in, without
first masking off the existing field. The first write after reset lands
on a zeroed field and looks correct, but any later reprogramming leaves
the old bits set: the field becomes the bitwise OR of the previous and
new encodings, corrupting the divider or multiplier.
Mask off each field before writing the new value so reprogramming a
clock to a different rate produces the intended register contents.
Fixes: a7b7c7c6c0 ("clk: canaan: Add clock driver for Canaan K230")
Signed-off-by: David Carlier <devnexen@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Acked-by: Xukai Wang <kingxukai@zohomail.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
The driver was introduced in 2021 and since then only supports a single
chip variant. Simplify the driver by hard-coding the device properties
instead of using the id_table's abstraction for a single chip type.
While touching the id table, use a single space in the table terminator
to match the most used style.
Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@baylibre.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
vc7_probe() registers the APLL with clk_register_fixed_rate(), which is
not devm-managed and must be explicitly unregistered on probe failure.
Most later errors already unwind through err_clk, but a failure from
vc7_get_bank_clk() in the output registration loop returned directly.
That skipped clk_unregister_fixed_rate() and leaked the APLL clock.
Route that error through the existing err_clk label so the fixed-rate
clock is released consistently with the other probe failure paths.
This issue was identified during our ongoing static-analysis research while
reviewing kernel code.
Fixes: 48c5e98fed ("clk: Renesas versaclock7 ccf device driver")
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
Reviewed-by: Brian Masney <bmasney@redhat.com>
Signed-off-by: Brian Masney <bmasney@redhat.com>
of_find_compatible_node() returns a device node with its reference count
incremented. mxc_init_irq() looks up the i.MX25 CCM node for of_iomap()
and the AVIC node for irq_domain_create_legacy(), but does not release
either temporary reference.
of_iomap() does not consume the node reference, and
irq_domain_create_legacy() takes its own fwnode reference for the domain.
Drop the temporary OF node references after each use.
Fixes: 9b454d16e5 ("ARM: imx: avic: set low-power interrupt mask for imx25")
Fixes: 544496ab5c ("ARM: imx: move irq_domain_add_legacy call into avic driver")
Signed-off-by: Yuho Choi <dbgh9129@gmail.com>
Reviewed-by: Martin Kaiser <martin@kaiser.cx>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
ti_sci_probe() publishes the controller on the global ti_sci_list
before creating child devices. If of_platform_populate() fails
after creating some children, the probe error path leaves the
children around and leaves the failed controller visible through
ti_sci_get_handle().
Depopulate any children created by the failed populate call and remove
the controller from the global list before returning the probe error.
Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn>
Link: https://patch.msgid.link/20260616004741.1726-1-pengpeng@iscas.ac.cn
Signed-off-by: Nishanth Menon <nm@ti.com>
The manual page-program path used the UMA command byte for the
first byte of each continuation chunk. That forced command-lane
semantics onto chunked writes and blocked dual/quad payload
writes.
Switch continuation chunks to data-only UMA transactions so WDBPCK
controls the payload bus width.
Signed-off-by: Tomer Maimon <tmaimon77@gmail.com>
Link: https://patch.msgid.link/20260609071458.2692482-1-tmaimon77@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The stmmac hardware on Agilex5 supports 8 TX/RX queue pairs and is
dma-coherent. Update the schema to handle the hardware differences
between SoC variants.
Also make 'interrupts' and 'interrupt-names' as required properties.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
v2: Set the top level interrupts to include the widest contraints(1-17)
Just have 1 list of all interrupt-names
Update commit message to include 'interrupts' and 'interrupt-names'
as required properties
Change minItems 1 to maxItems 1 for non agilex5 variants
Add same constraint for interrupt-names
Flavien Solt reported lifetime issues with the CXL MCE notifier, which
can lead to NULL dereferences and use-after-free in the MCE handler.
The notifier was registered per memory device and stored in 'struct
cxl_memdev_state', even though it only needs the region state (the
region's SPA range and its extended linear cache size).
Instead of keeping the memory device and endpoint alive, the correct fix
is to move the notifier into 'struct cxl_region' and register it from
cxl_region_probe() as it should be a per-region notifier. Setup the
registration to only happen for regions that have an extended linear
cache as that is the only current usage.
Remove cxl_port_get_spa_cache_alias() as it is now dead code.
[ dj: Update dev_warn() when notifier fails due to kconfig. (Ben) ]
Reported-by: Flavien Solt <flavien@nus.edu.sg>
Suggested-by: Dan Williams <djbw@kernel.org>
Fixes: 516e5bd0b6 ("cxl: Add mce notifier to emit aliased address for extended linear cache")
Cc: stable@vger.kernel.org
Assisted-by: Claude:claude-opus-4-8
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Ben Cheatham <benjamin.cheatham@amd.com>
Link: https://patch.msgid.link/20260616224912.2567474-1-dave.jiang@intel.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
The driver's design intent is that missing or malformed component
registers should not prevent mailbox initialization. cxl_pci_probe()
already reflects this: the CXL_REGLOC_RBI_COMPONENT setup path only
emits a dev_warn() and continues when component registers are absent,
rather than returning an error.
The check 'if (!cxlds->reg_map.device_map.mbox.valid)' violates this
intent and is also technically incorrect for two reasons:
1. Wrong struct: the MEMDEV register block is enumerated into a local
variable 'map', not into 'cxlds->reg_map'. The device_map.mbox.valid
field inside cxlds->reg_map is never written by the MEMDEV probe and
will always read as zero regardless of actual hardware capability.
2. Already validated: cxl_pci_setup_regs(CXL_REGLOC_RBI_MEMDEV) calls
cxl_probe_regs() which explicitly checks mbox.valid and returns
-ENXIO if the mailbox is absent. If that check passes, the mailbox is
guaranteed to be present by the time cxl_pci_type3_init_mailbox() is
called.
The value that the check actually reads is component_map.ras.valid,
which aliases device_map.mbox.valid in the union. This is populated by
the COMPONENT probe, not the MEMDEV probe. On devices where the
component register BAR does not implement a CXL Component Capability
Array (e.g. certain DCD devices), cxl_probe_component_regs() returns
early leaving ras.valid=false. Through the union, this makes mbox.valid
read as false, causing cxl_pci_type3_init_mailbox() to return -ENODEV
(-19) even though the mailbox hardware is fully functional.
Remove the check. Mailbox presence has already been validated by
cxl_pci_setup_regs(CXL_REGLOC_RBI_MEMDEV). The presence or absence of
component registers is irrelevant to mailbox initialization.
Fixes: 8d8081cecf ("cxl: Move mailbox related bits to the same context")
Reviewed-by: Richard Cheng <icheng@nvidia.com>
Signed-off-by: Wei Hou <wei.hou@scaleflux.com>
Reviewed-by: Li Ming <ming.li@zohomail.com>
Link: https://patch.msgid.link/20260628155857.239866-1-wei.hou@scaleflux.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Fix the PCIe outbound memory ranges for both pcie1 controllers on i.MX943.
The memory window size was incorrectly set to 256MB during initial
bring-up, but the hardware supports up to 4GB of outbound address space
per controller.
Additionally, the ECAM region cannot be mapped as I/O space. Use a
memory-mapped region for I/O space instead, and relocate the 1MB I/O
region to immediately follow the memory region at offset 0xf0000000
within each window.
Update the outbound address space layout per controller as follows:
- 3.5GB 64-bit prefetchable memory
- 256MB 32-bit non-prefetchable memory
- 1MB I/O
Fixes: fa6067fd8e ("arm64: dts: imx943: Add pcie1 and pcie1-ep supports")
Signed-off-by: Richard Zhu <hongxing.zhu@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Fix the PCIe outbound memory ranges for both pcie0 controllers on i.MX94.
The memory window size was incorrectly set to 256MB during initial
bring-up, but the hardware supports up to 4GB of outbound address space
per controller.
Additionally, the ECAM region cannot be mapped as I/O space. Use a
memory-mapped region for I/O space instead, and relocate the 1MB I/O
region to immediately follow the memory region at offset 0xf0000000
within each window.
Update the outbound address space layout per controller as follows:
- 3.5GB 64-bit prefetchable memory
- 256MB 32-bit non-prefetchable memory
- 1MB I/O
Fixes: 8cd439f177 ("arm64: dts: imx94: Add pcie0 and pcie0-ep supports")
Signed-off-by: Richard Zhu <hongxing.zhu@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
