Josh Law [Wed, 18 Mar 2026 23:43:05 +0000 (08:43 +0900)]
lib/bootconfig: check xbc_init_node() return in override path
The ':=' override path in xbc_parse_kv() calls xbc_init_node() to
re-initialize an existing value node but does not check the return
value. If xbc_init_node() fails (data offset out of range), parsing
silently continues with stale node data.
Add the missing error check to match the xbc_add_node() call path
which already checks for failure.
In practice, a bootconfig using ':=' to override a value near the
32KB data limit could silently retain the old value, meaning a
security-relevant boot parameter override (e.g., a trace filter or
debug setting) would not take effect as intended.
Jakub Kicinski [Fri, 13 Mar 2026 23:20:47 +0000 (16:20 -0700)]
tools: ynl: rework policy access to support recursion
Donald points out that the current naive implementation using dicts
breaks if policy is recursive (child nest uses policy idx already
used by its parent).
Lean more into the NlPolicy class. This lets us "render" the policy
on demand, when user accesses it. If someone wants to do an infinite
walk that's on them :) Show policy info as attributes of the class
and use dict format to descend into sub-policies for extra neatness.
Linus Torvalds [Wed, 18 Mar 2026 22:50:29 +0000 (15:50 -0700)]
Merge tag 'libcrypto-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux
Pull crypto library fixes from Eric Biggers:
- Disable the "padlock" SHA-1 and SHA-256 driver on Zhaoxin
processors, since it does not compute hash values correctly
- Make a generated file be removed by 'make clean'
- Fix excessive stack usage in some of the arm64 AES code
* tag 'libcrypto-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux:
lib/crypto: powerpc: Add powerpc/aesp8-ppc.S to clean-files
crypto: padlock-sha - Disable for Zhaoxin processor
crypto: arm64/aes-neonbs - Move key expansion off the stack
William Roche [Tue, 17 Mar 2026 10:38:10 +0000 (10:38 +0000)]
x86/mce/amd: Check SMCA feature bit before accessing SMCA MSRs
People do effort to inject MCEs into guests in order to simulate/test
handling of hardware errors. The real use case behind it is testing the
handling of SIGBUS which the memory failure code sends to the process.
If that process is QEMU, instead of killing the whole guest, the MCE can
be injected into the guest kernel so that latter can attempt proper
handling and kill the user *process* in the guest, instead, which
caused the MCE. The assumption being here that the whole injection flow
can supply enough information that the guest kernel can pinpoint the
right process. But that's a different topic...
Regardless of virtualization or not, access to SMCA-specific registers
like MCA_DESTAT should only be done after having checked the smca
feature bit. And there are AMD machines like Bulldozer (the one before
Zen1) which do support deferred errors but are not SMCA machines.
Therefore, properly check the feature bit before accessing related MSRs.
Petr Pavlu [Fri, 13 Mar 2026 13:48:04 +0000 (14:48 +0100)]
module: Remove extern keyword from param prototypes
The external function declarations do not need the "extern" keyword. Remove
it to align with the Linux kernel coding style and to silence the
associated checkpatch warnings.
Signed-off-by: Petr Pavlu <petr.pavlu@suse.com> Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Petr Pavlu [Fri, 13 Mar 2026 13:48:03 +0000 (14:48 +0100)]
module: Clean up parse_args() arguments
* Use the preferred `unsigned int` over plain `unsigned` for the `num`
parameter.
* Synchronize the parameter names in moduleparam.h with the ones used by
the implementation in params.c.
Signed-off-by: Petr Pavlu <petr.pavlu@suse.com> Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Petr Pavlu [Fri, 13 Mar 2026 13:48:02 +0000 (14:48 +0100)]
module: Fix freeing of charp module parameters when CONFIG_SYSFS=n
When setting a charp module parameter, the param_set_charp() function
allocates memory to store a copy of the input value. Later, when the module
is potentially unloaded, the destroy_params() function is called to free
this allocated memory.
However, destroy_params() is available only when CONFIG_SYSFS=y, otherwise
only a dummy variant is present. In the unlikely case that the kernel is
configured with CONFIG_MODULES=y and CONFIG_SYSFS=n, this results in
a memory leak of charp values when a module is unloaded.
Fix this issue by making destroy_params() always available when
CONFIG_MODULES=y. Rename the function to module_destroy_params() to clarify
that it is intended for use by the module loader.
Fixes: e180a6b7759a ("param: fix charp parameters set via sysfs") Signed-off-by: Petr Pavlu <petr.pavlu@suse.com> Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Linus Torvalds [Wed, 18 Mar 2026 21:27:11 +0000 (14:27 -0700)]
Merge tag 'nfsd-7.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Pull nfsd fixes from Chuck Lever:
- Fix cache_request leak in cache_release()
- Fix heap overflow in the NFSv4.0 LOCK replay cache
- Hold net reference for the lifetime of /proc/fs/nfs/exports fd
- Defer sub-object cleanup in export "put" callbacks
* tag 'nfsd-7.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
nfsd: fix heap overflow in NFSv4.0 LOCK replay cache
sunrpc: fix cache_request leak in cache_release
NFSD: Hold net reference for the lifetime of /proc/fs/nfs/exports fd
NFSD: Defer sub-object cleanup in export put callbacks
The unlocked read of rtc->irq_data in rtc_dev_poll() can race with
the write in rtc_handle_legacy_irq() and also, theoretically, with
the write in rtc_dev_read().
These races should be safe (see inline comment), thus annotate the
read with data_race() for KCSAN.
Sanman Pradhan [Wed, 18 Mar 2026 19:40:19 +0000 (19:40 +0000)]
hwmon: (pmbus/isl68137) Fix unchecked return value and use sysfs_emit()
isl68137_avs_enable_show_page() uses the return value of
pmbus_read_byte_data() without checking for errors. If the I2C transaction
fails, a negative error code is passed through bitwise operations,
producing incorrect output.
Add an error check to propagate the return value if it is negative.
Additionally, modernize the callback by replacing sprintf()
with sysfs_emit().
Chen Pei [Tue, 17 Mar 2026 03:48:47 +0000 (11:48 +0800)]
perf symbol: Add RISCV case in get_plt_sizes
According to RISC-V psABI specification, the PLT (Program Linkage Table)
has the following layout:
- The first PLT entry occupies two 16-byte entries (32 bytes total)
- Subsequent PLT entries take up 16 bytes each
This aligns with the binutils-gdb implementation which defines the same
PLT sizes for RISC-V architecture.
Update get_plt_sizes() to set plt_header_size=32 and plt_entry_size=16
for EM_RISCV, matching the architecture's standard ABI.
Since AARCH64, LOONGARCH, and RISCV have the same PLT size definition,
they are merged together.
Mike Leach [Wed, 18 Mar 2026 10:36:40 +0000 (10:36 +0000)]
perf: tools: cs-etm: Enhance raw Coresight trace debug display
When compiling perf with CORESIGHT=1, an additional build option may
be used: CSTRACE_RAW=1, which will cause the CoreSight formatted trace
frames to be printed out during a perf --dump command. This is useful
when investigating issues with trace generation, decode or possible
data corruption.
e.g. for ETMv4 trace source into a formatted ETR sink a dump -
Tested-by: Leo Yan <leo.yan@arm.com> Signed-off-by: Mike Leach <mike.leach@arm.com> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Mike Leach [Wed, 18 Mar 2026 10:36:39 +0000 (10:36 +0000)]
perf: tools: cs-etm: Fix print issue for Coresight debug in ETE/TRBE trace
Building perf with CORESIGHT=1 and the optional CSTRACE_RAW=1 enables
additional debug printing of raw trace data when using command:-
perf report --dump.
This raw trace prints the CoreSight formatted trace frames, which may be
used to investigate suspected issues with trace quality / corruption /
decode.
These frames are not present in ETE + TRBE trace.
This fix removes the unnecessary call to print these frames.
This fix also rationalises implementation - original code had helper
function that unnecessarily repeated initialisation calls that had
already been made.
Due to an addtional fault with the OpenCSD library, this call when ETE/TRBE
are being decoded will cause a segfault in perf. This fix also prevents
that problem for perf using older (<= 1.8.0 version) OpenCSD libraries.
Fixes: 68ffe3902898 ("perf tools: Add decoder mechanic to support dumping trace data") Reported-by: Leo Yan <leo.yan@arm.com> Signed-off-by: Mike Leach <mike.leach@arm.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Maíra Canal [Tue, 17 Mar 2026 22:41:49 +0000 (19:41 -0300)]
pmdomain: bcm: bcm2835-power: Increase ASB control timeout
The bcm2835_asb_control() function uses a tight polling loop to wait
for the ASB bridge to acknowledge a request. During intensive workloads,
this handshake intermittently fails for V3D's master ASB on BCM2711,
resulting in "Failed to disable ASB master for v3d" errors during
runtime PM suspend. As a consequence, the failed power-off leaves V3D in
a broken state, leading to bus faults or system hangs on later accesses.
As the timeout is insufficient in some scenarios, increase the polling
timeout from 1us to 5us, which is still negligible in the context of a
power domain transition. Also, replace the open-coded ktime_get_ns()/
cpu_relax() polling loop with readl_poll_timeout_atomic().
Cc: stable@vger.kernel.org Fixes: 670c672608a1 ("soc: bcm: bcm2835-pm: Add support for power domains under a new binding.") Signed-off-by: Maíra Canal <mcanal@igalia.com> Reviewed-by: Stefan Wahren <wahrenst@gmx.net> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Yuanshen Cao [Thu, 5 Mar 2026 03:34:07 +0000 (03:34 +0000)]
dt-bindings: power: Add Support for Allwinner A733 PCK600 Power Domain Controller
The A733 PCK600, similar to A523 PCK600, is likely a customized version
of ARM PCK-600 power controller. They share the same BSP drivers in the
package provided by Radxa, with the only difference being the lack of
resets.
Therefore, document A733 compatible and make resets required only for
the other models, as well as prepare the PD definitions for future
device trees.
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com> Signed-off-by: Yuanshen Cao <alex.caoys@gmail.com> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Document the dmas property to state it must be specified as TX/RX DMA
specifier pairs.
This clarifies the expected ordering and improves binding readability
without changing behavior.
Neil Armstrong [Wed, 18 Mar 2026 17:10:10 +0000 (12:10 -0500)]
PCI/pwrctrl: generic: Rename pci-pwrctrl-slot as generic
The driver is pretty generic and would fit for either PCI Slots or
endpoints connected to PCI ports, so rename the driver and module as
pci-pwrctrl-generic.
Olivier Sobrie [Tue, 17 Mar 2026 17:18:07 +0000 (18:18 +0100)]
mtd: rawnand: pl353: make sure optimal timings are applied
Timings of the nand are adjusted by pl35x_nfc_setup_interface() but
actually applied by the pl35x_nand_select_target() function.
If there is only one nand chip, the pl35x_nand_select_target() will only
apply the timings once since the test at its beginning will always be true
after the first call to this function. As a result, the hardware will
keep using the default timings set at boot to detect the nand chip, not
the optimal ones.
With this patch, we program directly the new timings when
pl35x_nfc_setup_interface() is called.
Fixes: 08d8c62164a3 ("mtd: rawnand: pl353: Add support for the ARM PL353 SMC NAND controller") Signed-off-by: Olivier Sobrie <olivier@sobrie.be> Cc: stable@vger.kernel.org Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Miquel Raynal [Tue, 17 Mar 2026 17:17:22 +0000 (18:17 +0100)]
mtd: spi-nor: Rename spi_nor_spimem_check_op()
This helper really is just a little helper for internal purposes, and is
I/O operation oriented, despite its name. It has already been misused
in commit 5008c3ec3f89 ("mtd: spi-nor: core: Check read CR support"), so
rename it to clarify its purpose: it is only useful for reads and page
programs.
Commit 5008c3ec3f89 ("mtd: spi-nor: core: Check read CR support") adds a
controller check to make sure the core will not use CR reads on
controllers not supporting them. The approach is valid but the fix is
incorrect. Unfortunately, the author could not catch it, because the
expected behavior was met. The patch indeed drops the RDCR capability,
but it does it for all controllers!
The issue comes from the use of spi_nor_spimem_check_op() which is an
internal helper dedicated to check read/write operations only, despite
its generic name.
This helper looks for the biggest number of address bytes that can be
used for a page operation and tries 4 then 3. It then calls the usual
spi-mem helpers to do the checks. These will always fail because there
is now an inconsistency: the address cycles are forced to 4 (then 3)
bytes, but the bus width during the address cycles rightfully remains
0. There is a non-zero address length but a zero address bus width,
which is an invalid combination.
The correct check in this case is to directly call spi_mem_supports_op()
which doesn't messes up with the operation content.
Qiuxu Zhuo [Wed, 18 Mar 2026 02:31:18 +0000 (10:31 +0800)]
EDAC/i10nm: Add driver decoder for Granite Rapids server
Current i10nm_edac only supports the firmware decoder (ACPI DSM methods)
for Granite Rapids servers. Add the driver decoder, which directly extracts
topology information from the IMC machine check bank IA32_MCi_MISC MSRs, to
improve decoding performance for Granite Rapids.
Jiucheng Xu [Wed, 11 Mar 2026 09:11:31 +0000 (17:11 +0800)]
erofs: add GFP_NOIO in the bio completion if needed
The bio completion path in the process context (e.g. dm-verity)
will directly call into decompression rather than trigger another
workqueue context for minimal scheduling latencies, which can
then call vm_map_ram() with GFP_KERNEL.
Due to insufficient memory, vm_map_ram() may generate memory
swapping I/O, which can cause submit_bio_wait to deadlock
in some scenarios.
Trimmed down the call stack, as follows:
f2fs_submit_read_io
submit_bio //bio_list is initialized.
mmc_blk_mq_recovery
z_erofs_endio
vm_map_ram
__pte_alloc_kernel
__alloc_pages_direct_reclaim
shrink_folio_list
__swap_writepage
submit_bio_wait //bio_list is non-NULL, hang!!!
Use memalloc_noio_{save,restore}() to wrap up this path.
First, pin_user_pages_fast() can return a short pin count (less than
requested but greater than zero) when it cannot pin all requested pages.
This is treated as success, leading to partially pinned regions being
used, which causes memory corruption.
Second, when an error occurs mid-loop, already pinned pages from the
current batch are not properly accounted for before calling
mshv_region_invalidate_pages(), causing a page reference leak.
Treat short pins as errors and fix partial batch accounting before
cleanup.
Signed-off-by: Stanislav Kinsburskii <skinsburskii@linux.microsoft.com> Reviewed-by: Michael Kelley <mhklinux@outlook.com> Signed-off-by: Wei Liu <wei.liu@kernel.org>
Cássio Gabriel [Wed, 18 Mar 2026 14:08:46 +0000 (11:08 -0300)]
ALSA: usb-audio: validate full match when resolving quirk aliases
get_alias_quirk() resolves a quirk for an aliased USB ID by scanning
usb_audio_ids[], but it currently checks only the vendor/product pair.
This is weak for quirk table entries that also depend on additional
USB_DEVICE_ID match fields, such as device or interface class,
subclass, protocol, interface number, or bcdDevice range.
Keep the aliased vid:pid as the lookup key, then validate only the
remaining match criteria of each candidate entry against the real
device/interface descriptors by clearing USB_DEVICE_ID_MATCH_DEVICE
from a temporary copy and passing it to usb_match_one_id().
Arnd Bergmann [Wed, 18 Mar 2026 12:40:07 +0000 (13:40 +0100)]
ALSA: asihpi: avoid write overflow check warning
clang-22 rightfully warns that the memcpy() in adapter_prepare() copies
between different structures, crossing the boundary of nested
structures inside it:
In file included from sound/pci/asihpi/hpimsgx.c:13:
In file included from include/linux/string.h:386:
include/linux/fortify-string.h:569:4: error: call to '__write_overflow_field' declared with 'warning' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Werror,-Wattribute-warning]
569 | __write_overflow_field(p_size_field, size);
The two structures seem to refer to the same layout, despite the
separate definitions, so the code is in fact correct.
Avoid the warning by copying the two inner structures separately.
I see the same pattern happens in other functions in the same file,
so there is a chance that this may come back in the future, but
this instance is the only one that I saw in practice, hitting it
multiple times per day in randconfig build.
staging: rtl8723bs: simplify return checks in validate_recv_data_frame()
Combine the return value checks for _FAIL and RTW_RX_HANDLED into a
single logical OR statement and remove unnecessary braces. This improves
code readability and resolves the following checkpatch.pl warning:
"WARNING: braces {} are not necessary for any arm of this statement"
Marcos Andrade [Mon, 16 Mar 2026 22:19:24 +0000 (19:19 -0300)]
staging: rtl8723bs: Replace msleep() with fsleep()
Replace msleep(10) with fsleep(10 * USEC_PER_MSEC)
in _rtw_init_xmit_priv().
fsleep() is the new standard API for delays, as it automatically chooses
the best sleep method based on the duration.
Marcos Andrade [Mon, 16 Mar 2026 22:19:23 +0000 (19:19 -0300)]
staging: rtl8723bs: Replace network magic numbers with EtherType macros
Replace hardcoded magic numbers for network protocols (e.g., 0x0806
for ARP, 0x888e for EAPOL) with their standard EtherType macro
equivalents (ETH_P_ARP, ETH_P_PAE) defined in <linux/if_ether.h>.
This change improves code readability and aligns the driver with
standard Linux networking definitions.
Marcos Andrade [Mon, 16 Mar 2026 22:04:35 +0000 (19:04 -0300)]
staging: rtl8723bs: remove unused global efuse variables
Remove several global efuse variables from rtw_efuse.c and their
corresponding extern declarations in rtw_efuse.h.
These variables (fakeEfuseBank, BTEfuseUsedBytes, etc.) are completely
unused legacy code. The driver currently maintains the efuse state
properly within the 'efuse_hal' structure, which is encapsulated
inside 'hal_com_data'.
The removal of this dead code cleans up the global namespace and
resolves multiple checkpatch.pl warnings regarding CamelCase naming
conventions. Verified by compilation that no functional code references
these variables.
Nikolay Kulikov [Mon, 16 Mar 2026 11:33:58 +0000 (14:33 +0300)]
staging: rtl8723bs: remove unusual 'NDIS_802_11_MAC_ADDRESS' type
Remove the 'NDIS_802_11_MAC_ADDRESS' type and replace all variables of
this type with an array of 'u8', which will make the code more
understandable and get rid of unnecessary 'typedef'.
Davidlohr Bueso [Sun, 1 Mar 2026 22:17:39 +0000 (14:17 -0800)]
cxl/mbox: Use proper endpoint validity check upon sanitize
Fuzzying CXL triggered:
BUG: KASAN: null-ptr-deref in cxl_num_decoders_committed+0x3e/0x80 drivers/cxl/core/port.c:49
Read of size 4 at addr 0000000000000642 by task syz.0.97/2282
This goes away using the correct check instead of abusing cxlmd->endpoint,
which is unusable (ENXIO) until the driver has probed. During that window
the memdev sysfs attributes are already visible, as soon as device_add()
completes.
Fixes: 29317f8dc6ed ("cxl/mem: Introduce cxl_memdev_attach for CXL-dependent operation") Signed-off-by: Davidlohr Bueso <dave@stgolabs.net> Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com> Reviewed-by: Gregory Price <gourry@gourry.net> Link: https://patch.msgid.link/20260301221739.1726722-1-dave@stgolabs.net Signed-off-by: Dave Jiang <dave.jiang@intel.com>
staging: greybus: loopback: remove unused argument from macro
The gb_dev_loopback_ro_attr macro accepted a 'conn' argument which
was never used in its expansion. Remove it from both the macro
definition and its invocation.
staging: greybus: loopback: use sysfs_emit in sysfs show functions
As per the kernel's documentation, sysfs_emit() is the preferred way
to format strings for sysfs attributes. It handles buffer overruns
safely. Replace sprintf calls with sysfs_emit across all loopback
sysfs show macros.
Dave Hansen [Thu, 5 Mar 2026 21:22:15 +0000 (13:22 -0800)]
MAINTAINERS: Add references to tip tree handbook
The tip tree handbook is a pretty handy reference for keeping the
various tip maintainers happy. But there are no references to it
in the MAINTAINERS file.
Add a references to the handbook for a number of subsystems. I've
added these for all of the obvious subsystems that list T:.../tip.git
and are likely to follow the tip rules.
This will be useful both for humans digging in the MAINTAINERS file
and our new AI bot overlords.
I'm cc'ing all the M:'s and L:'s listed for the subsystems but leaving
off the R:'s because the cc list is already out of control. I'll
commit this to *some* tip branch if there's general consensus that
this is a good idea.
[ dhansen: add objtool and futexes at peterz's urging ]
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com> Acked-by: Boqun Feng <boqun@kernel.org> Acked-by: Daniel Lezcano <daniel.lezcano@kernel.org> Acked-by: Borislav Petkov (AMD) <bp@alien8.de> Link: https://patch.msgid.link/20260305212215.3188399-1-dave.hansen@linux.intel.com Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Linus Torvalds [Wed, 18 Mar 2026 15:28:54 +0000 (08:28 -0700)]
Merge tag 'soc-fixes-7.0' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Pull SoC fixes from Arnd Bergmann:
"The firmware drivers for ARM SCMI, FF-A and the Tee subsystem, as
well as the reset controller and cache controller subsystem all see
small bugfixes for reference ounting errors, ABI correctness, and
NULL pointer dereferences.
Similarly, there are multiple reference counting fixes in drivers/soc/
for vendor specific drivers (rockchips, microchip), while the
freescale drivers get a fix for a race condition and error handling.
The devicetree fixes for Rockchips and NXP got held up, so for
the moment there is only Renesas fixing problesm with SD card
initialization, a boot hang on one board and incorrect descriptions
for interrupts and clock registers on some SoCs. The Microchip
polarfire gets a dts fix for a boot time warning.
A defconfig fix avoids a warning about a conflicting assignment"
* tag 'soc-fixes-7.0' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (21 commits)
ARM: multi_v7_defconfig: Drop duplicate CONFIG_TI_PRUSS=m
firmware: arm_scmi: Spelling s/mulit/multi/, s/currenly/currently/
firmware: arm_scmi: Fix NULL dereference on notify error path
firmware: arm_scpi: Fix device_node reference leak in probe path
firmware: arm_ffa: Remove vm_id argument in ffa_rxtx_unmap()
arm64: dts: renesas: r8a78000: Fix out-of-range SPI interrupt numbers
arm64: dts: renesas: rzg3s-smarc-som: Set bypass for Versa3 PLL2
arm64: dts: renesas: r9a09g087: Fix CPG register region sizes
arm64: dts: renesas: r9a09g077: Fix CPG register region sizes
arm64: dts: renesas: r9a09g057: Remove wdt{0,2,3} nodes
arm64: dts: renesas: rzv2-evk-cn15-sd: Add ramp delay for SD0 regulator
arm64: dts: renesas: rzt2h-n2h-evk: Add ramp delay for SD0 card regulator
tee: shm: Remove refcounting of kernel pages
reset: rzg2l-usbphy-ctrl: Check pwrrdy is valid before using it
soc: fsl: cpm1: qmc: Fix error check for devm_ioremap_resource() in qmc_qe_init_resources()
soc: fsl: qbman: fix race condition in qman_destroy_fq
soc: rockchip: grf: Add missing of_node_put() when returning
cache: ax45mp: Fix device node reference leak in ax45mp_cache_init()
cache: starfive: fix device node leak in starlink_cache_init()
riscv: dts: microchip: add can resets to mpfs
...
If a runtime suspend is executed followed by a system suspend, the driver
may invoke dwc3_imx_suspend() twice, which causes enable_irq() to be
called twice as well. This leads to an unbalanced IRQ state and may
trigger warnings or malfunction. Prevent this by checking the pm_suspended
flag before running the imx suspend/resume path.
Fixes: 76fc9452a6bf ("usb: dwc3: introduce flatten model driver of i.MX Soc") Signed-off-by: Xu Yang <xu.yang_2@nxp.com> Acked-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com> Link: https://patch.msgid.link/20260317030546.1665206-1-xu.yang_2@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Sebastian Urban [Sun, 15 Mar 2026 15:10:45 +0000 (16:10 +0100)]
usb: gadget: dummy_hcd: fix premature URB completion when ZLP follows partial transfer
When a gadget request is only partially transferred in transfer()
because the per-frame bandwidth budget is exhausted, the loop advances
to the next queued request. If that next request is a zero-length
packet (ZLP), len evaluates to zero and the code takes the
unlikely(len == 0) path, which sets is_short = 1. This bypasses the
bandwidth guard ("limit < ep->ep.maxpacket && limit < len") that
lives in the else branch and would otherwise break out of the loop for
non-zero requests. The is_short path then completes the URB before all
data from the first request has been transferred.
Reproducer (bulk IN, high speed):
Device side (FunctionFS with Linux AIO):
1. Queue a 65024-byte write via io_submit (127 * 512, i.e. a
multiple of the HS bulk max packet size).
2. Immediately queue a zero-length write (ZLP) via io_submit.
Host side:
3. Submit a 65536-byte bulk IN URB.
Expected: URB completes with actual_length = 65024.
Actual: URB completes with actual_length = 53248, losing 11776
bytes that leak into subsequent URBs.
At high speed the per-frame budget is 53248 bytes (512 * 13 * 8).
The 65024-byte request exhausts this budget after 53248 bytes, leaving
the request incomplete (req->req.actual < req->req.length). Neither
the request nor the URB is finished, and rescan is 0, so the loop
advances to the ZLP. For the ZLP, dev_len = 0, so len = min(12288, 0)
= 0, taking the unlikely(len == 0) path and setting is_short = 1.
The is_short handler then sets *status = 0, completing the URB with
only 53248 of the expected 65024 bytes.
Fix this by breaking out of the loop when the current request has
remaining data (req->req.actual < req->req.length). The request
resumes on the next timer tick, preserving correct data ordering.
Signed-off-by: Sebastian Urban <surban@surban.net> Cc: stable <stable@kernel.org> Reviewed-by: Alan Stern <stern@rowland.harvard.edu> Link: https://patch.msgid.link/20260315151045.1155850-1-surban@surban.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Felix Gu [Mon, 16 Mar 2026 15:30:06 +0000 (23:30 +0800)]
dwc3: google: Fix PM domain leak in dwc3_google_probe()
When syscon_regmap_lookup_by_phandle_args() fails, the function was
returning directly without cleaning up the power domain initialized
earlier by dwc3_google_pm_domain_init().
Fix by jumping to err_deinit_pdom to properly clean up resources.
Oliver Neukum [Tue, 17 Mar 2026 08:41:10 +0000 (09:41 +0100)]
cdc-acm: new quirk for EPSON HMD
This device has a union descriptor that is just garbage
and needs a custom descriptor.
In principle this could be done with a (conditionally
activated) heuristic. That would match more devices
without a need for defining a new quirk. However,
this always carries the risk that the heuristics
does the wrong thing and leads to more breakage.
Defining the quirk and telling it exactly what to do
is the safe and conservative approach.
Xu Yang [Mon, 16 Mar 2026 09:48:11 +0000 (17:48 +0800)]
usb: hcd: queue wakeup_work to system_freezable_wq workqueue
After commit 4fb352df14de ("PM: sleep: Do not flag runtime PM workqueue
as freezable"), pm_wq workqueue will be unfreezable during system pm. This
brings issue as below:
The reason is if the host controller resume routine xhci_resume() call
usb_hcd_resume_root_hub(), wakeup_work will be queued and run immediately.
Then usb_autoresume_device() will be called and usb device will exit
runtime suspended state (if it was suspended before). For a hub device,
hub_resume()/hub_reset_resume() will be called accordingly.
After the host controller device system resume is finished, the root hub
usb device "usb1" will do system resume too. Then hub_resume() will be
called again.
Above sequence will cause hub->urb to be submitted twice. To avoid this
issue, restore the previous behavior by queuing wakeup_work to
system_freezable_wq workqueue.
Acked-by: Alan Stern <stern@rowland.harvard.edu> Fixes: 4fb352df14de ("PM: sleep: Do not flag runtime PM workqueue as freezable") Cc: stable <stable@kernel.org> Signed-off-by: Xu Yang <xu.yang_2@nxp.com> Link: https://patch.msgid.link/20260316094811.1559471-1-xu.yang_2@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Kuen-Han Tsai [Mon, 16 Mar 2026 07:49:09 +0000 (15:49 +0800)]
usb: gadget: u_ether: Fix NULL pointer deref in eth_get_drvinfo
Commit ec35c1969650 ("usb: gadget: f_ncm: Fix net_device lifecycle with
device_move") reparents the gadget device to /sys/devices/virtual during
unbind, clearing the gadget pointer. If the userspace tool queries on
the surviving interface during this detached window, this leads to a
NULL pointer dereference.
Add a NULL check for dev->gadget in eth_get_drvinfo(). When detached,
skip copying the fw_version and bus_info strings, which is natively
handled by ethtool_get_drvinfo for empty strings.
This fixes an error in synchronization in the dummy-hcd driver. The
error has a somewhat involved history. The synchronization mechanism
was introduced by commit 7dbd8f4cabd9 ("USB: dummy-hcd: Fix erroneous
synchronization change"), which added an emulated "interrupts enabled"
flag together with code emulating synchronize_irq() (it waits until
all current handler callbacks have returned).
But the emulated interrupt-disable occurred too late, after the driver
containing the handler callback routines had been told that it was
unbound and no more callbacks would occur. Commit 4a5d797a9f9c ("usb:
gadget: dummy_hcd: fix gpf in gadget_setup") tried to fix this by
moving the synchronize_irq() emulation code from dummy_stop() to
dummy_pullup(), which runs before the unbind callback.
There still were races, though, because the emulated interrupt-disable
still occurred too late. It couldn't be moved to dummy_pullup(),
because that routine can be called for reasons other than an impending
unbind. Therefore commits 7dc0c55e9f30 ("USB: UDC core: Add
udc_async_callbacks gadget op") and 04145a03db9d ("USB: UDC: Implement
udc_async_callbacks in dummy-hcd") added an API allowing the UDC core
to tell dummy-hcd exactly when emulated interrupts and their callbacks
should be disabled.
That brings us to the current state of things, which is still wrong
because the emulated synchronize_irq() occurs before the emulated
interrupt-disable! That's no good, beause it means that more emulated
interrupts can occur after the synchronize_irq() emulation has run,
leading to the possibility that a callback handler may be running when
the gadget driver is unbound.
To fix this, we have to move the synchronize_irq() emulation code yet
again, to the dummy_udc_async_callbacks() routine, which takes care of
enabling and disabling emulated interrupt requests. The
synchronization will now run immediately after emulated interrupts are
disabled, which is where it belongs.
Alan Stern [Sun, 15 Mar 2026 18:30:43 +0000 (14:30 -0400)]
USB: dummy-hcd: Fix locking/synchronization error
Syzbot testing was able to provoke an addressing exception and crash
in the usb_gadget_udc_reset() routine in
drivers/usb/gadgets/udc/core.c, resulting from the fact that the
routine was called with a second ("driver") argument of NULL. The bad
caller was set_link_state() in dummy_hcd.c, and the problem arose
because of a race between a USB reset and driver unbind.
These sorts of races were not supposed to be possible; commit 7dbd8f4cabd9 ("USB: dummy-hcd: Fix erroneous synchronization change"),
along with a few followup commits, was written specifically to prevent
them. As it turns out, there are (at least) two errors remaining in
the code. Another patch will address the second error; this one is
concerned with the first.
The error responsible for the syzbot crash occurred because the
stop_activity() routine will sometimes drop and then re-acquire the
dum->lock spinlock. A call to stop_activity() occurs in
set_link_state() when handling an emulated USB reset, after the test
of dum->ints_enabled and before the increment of dum->callback_usage.
This allowed another thread (doing a driver unbind) to sneak in and
grab the spinlock, and then clear dum->ints_enabled and dum->driver.
Normally this other thread would have to wait for dum->callback_usage
to go down to 0 before it would clear dum->driver, but in this case it
didn't have to wait since dum->callback_usage had not yet been
incremented.
The fix is to increment dum->callback_usage _before_ calling
stop_activity() instead of after. Then the thread doing the unbind
will not clear dum->driver until after the call to
usb_gadget_udc_reset() safely returns and dum->callback_usage has been
decremented again.
Nathan Rebello [Fri, 13 Mar 2026 22:24:53 +0000 (18:24 -0400)]
usb: typec: ucsi: validate connector number in ucsi_notify_common()
The connector number extracted from CCI via UCSI_CCI_CONNECTOR() is a
7-bit field (0-127) that is used to index into the connector array in
ucsi_connector_change(). However, the array is only allocated for the
number of connectors reported by the device (typically 2-4 entries).
A malicious or malfunctioning device could report an out-of-range
connector number in the CCI, causing an out-of-bounds array access in
ucsi_connector_change().
Add a bounds check in ucsi_notify_common(), the central point where CCI
is parsed after arriving from hardware, so that bogus connector numbers
are rejected before they propagate further.
usb: usbtmc: Flush anchored URBs in usbtmc_release
When calling usbtmc_release, pending anchored URBs must be flushed or
killed to prevent use-after-free errors (e.g. in the HCD giveback
path). Call usbtmc_draw_down() to allow anchored URBs to be completed.
Fixes: 4f3c8d6eddc2 ("usb: usbtmc: Support Read Status Byte with SRQ per file") Reported-by: syzbot+9a3c54f52bd1edbd975f@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=9a3c54f52bd1edbd975f Cc: stable <stable@kernel.org> Signed-off-by: Heitor Alves de Siqueira <halves@igalia.com> Link: https://patch.msgid.link/20260312-usbtmc-flush-release-v1-1-5755e9f4336f@igalia.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Kuen-Han Tsai [Wed, 11 Mar 2026 09:12:15 +0000 (17:12 +0800)]
usb: gadget: u_ether: Fix race between gether_disconnect and eth_stop
A race condition between gether_disconnect() and eth_stop() leads to a
NULL pointer dereference. Specifically, if eth_stop() is triggered
concurrently while gether_disconnect() is tearing down the endpoints,
eth_stop() attempts to access the cleared endpoint descriptor, causing
the following NPE:
Because eth_stop() crashes while holding the dev->lock, the thread
running gether_disconnect() fails to acquire the same lock and spins
forever, resulting in a hardlockup:
Core - Debugging Information for Hardlockup core(7)
Call trace:
queued_spin_lock_slowpath+0x94/0x488
_raw_spin_lock+0x64/0x6c
gether_disconnect+0x19c/0x1e8
ncm_set_alt+0x68/0x1a0
composite_setup+0x6a0/0xc50
The root cause is that the clearing of dev->port_usb in
gether_disconnect() is delayed until the end of the function.
Move the clearing of dev->port_usb to the very beginning of
gether_disconnect() while holding dev->lock. This cuts off the link
immediately, ensuring eth_stop() will see dev->port_usb as NULL and
safely bail out.
tcpm: Implement sink support for PD SPR AVS negotiation
Add support to enable TCPM to negotiate with
USB PD Standard Power Range Adjustable Voltage Supply (SPR AVS) when
acting as a power sink.
* Added support to the tcpm power supply properties, allowing userspace
to enable and control the dynamic limits (voltage and current)
specific to the SPR AVS contract.
* Implemented tcpm_pd_select_spr_avs_apdo() to select the appropriate
APDO and validate the requested voltage/current against both the
Source and Sink capabilities.
* Implemented tcpm_pd_build_spr_avs_request() to construct the
Request Data Object (RDO) for SPR AVS.
* Added SNK_NEGOTIATE_SPR_AVS_CAPABILITIES state to the state machine to
handle negotiation for SPR AVS.
* Updated the SNK_TRANSITION_SINK state to implement the SPR
AVS-specific VBUS transition rules, including reducing current draw to
PD_I_SNK_STBY_MA for large voltage changes, as required by USB PD spec.
power: supply: Add PD SPR AVS support to USB type enum
Add two new members to the power_supply_usb_type to represent the
USB Power Delivery (PD) Standard Power Range (SPR) Adjustable Voltage
Supply (AVS) charging types:
POWER_SUPPLY_USB_TYPE_PD_SPR_AVS: For devices supporting only the
PD SPR AVS type.
POWER_SUPPLY_USB_TYPE_PD_PPS_SPR_AVS: For devices that support both
PD Programmable Power Supply (PPS) and PD SPR AVS.
USB Power Delivery 3.2 introduces a new power supply type SPR AVS.
Add macro definitions for the USB Power Delivery (PD)
Standard Power Range (SPR) Adjustable Voltage Supply (AVS) as a
Sink Augmented Power Data Object (APDO) in the device tree bindings.
Aldo Conte [Wed, 11 Mar 2026 16:33:20 +0000 (17:33 +0100)]
usb: typec: Document priority and mode_selection fields in struct typec_altmode
The fields 'priority' and 'mode_selection' in struct typec_altmode are
missing from the kernel-doc comment, which results in warnings when
building the documentation with 'make htmldocs'.
WARNING: ./include/linux/usb/typec_altmode.h:44 struct member 'priority' not described in 'typec_altmode'
WARNING: ./include/linux/usb/typec_altmode.h:44 struct member 'mode_selection' not described in 'typec_altmode'
Document both fields to keep the kernel-doc comment aligned with the
structure definition.
Linus Torvalds [Wed, 18 Mar 2026 15:06:30 +0000 (08:06 -0700)]
Merge tag 'loongarch-fixes-7.0-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson
Pull LoongArch fixes from Huacai Chen:
- only use SC.Q when supported by the assembler to fix a build failure
- fix calling smp_processor_id() in preemptible code
- make a BPF helper arch_protect_bpf_trampoline() return 0 to fix a
kernel memory access failure
- fix a typo issue in kvm_vm_init_features()
* tag 'loongarch-fixes-7.0-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
LoongArch: KVM: Fix typo issue in kvm_vm_init_features()
LoongArch: BPF: Make arch_protect_bpf_trampoline() return 0
LoongArch: No need to flush icache if text copy failed
LoongArch: Check return values for set_memory_{rw,rox}
LoongArch: Give more information if kmem access failed
LoongArch: Fix calling smp_processor_id() in preemptible code
LoongArch: Only use SC.Q when supported by the assembler
Static structure "ti_tusb8041_data" is exactly the same as
"ti_tusb8020b_data" and "cypress_hx2vl_data" is the same as
"microchip_usb424_data". Drop the duplicated structures to reduce
driver size and memory usage without affecting functionality.
That's because the ps883x_retimer_remove() retrieves the driver data
from i2c_get_clientdata() which was never set at probe. So, add
i2c_set_clientdata() at the end of the probe.
Signed-off-by: Mostafa Saleh <smostafa@google.com> Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com> Fixes: 257a087c8b52 ("usb: typec: Add support for Parade PS8830 Type-C Retimer") Link: https://patch.msgid.link/20260313155534.1916773-1-smostafa@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Xu Yang [Mon, 16 Mar 2026 09:50:42 +0000 (17:50 +0800)]
usb: port: add delay after usb_hub_set_port_power()
When a port is disabled, an attached device will be disconnected. This
causes a port-status-change event, which will race with hub autosuspend
(if the disabled port was the only connected port on its hub), causing
an immediate resume and a second autosuspend. Both of these can be
avoided by adding a short delay after the call to
usb_hub_set_port_power().
Below log shows what is happening:
$ echo 1 > usb1-port1/disable
[ 37.958239] usb 1-1: USB disconnect, device number 2
[ 37.964101] usb 1-1: unregistering device
[ 37.970070] hub 1-0:1.0: hub_suspend
[ 37.971305] hub 1-0:1.0: state 7 ports 1 chg 0000 evt 0002
[ 37.974412] usb usb1: bus auto-suspend, wakeup 1
[ 37.988175] usb usb1: suspend raced with wakeup event <---
[ 37.993947] usb usb1: usb auto-resume
[ 37.998401] hub 1-0:1.0: hub_resume
[ 38.105688] usb usb1-port1: status 0000, change 0000, 12 Mb/s
[ 38.112399] hub 1-0:1.0: state 7 ports 1 chg 0000 evt 0000
[ 38.118645] hub 1-0:1.0: hub_suspend
[ 38.122963] usb usb1: bus auto-suspend, wakeup 1
[ 38.200368] usb usb1: usb wakeup-resume
[ 38.204982] usb usb1: usb auto-resume
[ 38.209376] hub 1-0:1.0: hub_resume
[ 38.213676] usb usb1-port1: status 0101 change 0001
[ 38.321552] hub 1-0:1.0: state 7 ports 1 chg 0002 evt 0000
[ 38.327978] usb usb1-port1: status 0101, change 0000, 12 Mb/s
[ 38.457429] usb 1-1: new high-speed USB device number 3 using ci_hdrc
Then, port change bit will be fixed to the final state and
usb_clear_port_feature() can correctly clear it after this period. This
will also avoid usb runtime suspend routine to run because
usb_autopm_put_interface() not run yet.
Fixes: f061f43d7418 ("usb: hub: port: add sysfs entry to switch port power") Cc: stable@kernel.org Signed-off-by: Xu Yang <xu.yang_2@nxp.com> Link: https://patch.msgid.link/20260316095042.1559882-1-xu.yang_2@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Sean Rhodes [Sun, 15 Mar 2026 22:34:33 +0000 (22:34 +0000)]
usb: core: allow ACPI-managed hard-wired ports to power off
USB core only relaxes the default PM_QOS_FLAG_NO_POWER_OFF policy when
an upstream hub reports switchable port power. That misses internal
ports whose power is managed by platform firmware instead of the USB
hub descriptor.
Allow the port-poweroff policy to be exposed for hard-wired ports with
an ACPI-managed power resource. The existing runtime PM path still
requires the child usage count to drop and remote wakeup to be clear
before it will power the port down.
This lets internal devices such as CNVi Bluetooth use the existing USB
ACPI runtime power path even when the root hub reports no USB-standard
port power switching.
Bence Csókás [Sun, 15 Mar 2026 11:24:44 +0000 (12:24 +0100)]
USB: core: Use krealloc() in usb_cache_string()
Instead of "shrinking" the allocation by kmalloc()ing a new, smaller
buffer, utilize krealloc() to shrink the existing allocation. This saves
a memcpy(), as well as eliminates the temporary `smallbuf` allocation,
which guards against allocation failure under extreme memory pressure.
Alexey Charkov [Wed, 11 Mar 2026 15:20:23 +0000 (19:20 +0400)]
usb: misc: onboard_usb_dev: Add Corechips SL6341 USB 2.0/3.0 hub
Add the ID entries and platform data for the Corechips SL6341 onboard
USB 2.0/3.0 hub controller, which requires a reset pin and a power supply
for proper operation.
Corechips SL6341 is a 4-port low-power USB 3.2 Gen 1x1 hub controller
supporting SS, HS, FS and LS connections and integrating a 5V to 3.3V
built-in LDO to enable its IO to be powered directly from the 5V USB VBUS.
External 1v1 VDD supply is still required for its core power.
Zhaoyang Yu [Thu, 12 Mar 2026 11:03:29 +0000 (11:03 +0000)]
USB: pxa27x_udc: check return value of clk_enable
clk_enable() may fail according to the API contract.
Previously, udc_enable() ignored its return value and returned void.
Modify udc_enable() to return the error code. Additionally, update
all of its callers (pxa_udc_pullup, pxa_udc_vbus_session,
pxa27x_udc_start, pxa_udc_probe, and pxa_udc_resume) to check
this return value and handle the failure properly with necessary
cleanups or rollbacks.
Krishna Kurapati [Thu, 12 Mar 2026 10:14:31 +0000 (15:44 +0530)]
usb: typec: ucsi: Add UCSI_USB4_IMPLIES_USB quirk for X1E80100
On X1E80100, when we connect a USB4 capable dock, the PARTNER_FLAGS
indicate USB4_GEN3 being set whilst keeping the PARTNER_FLAGS_USB
cleared. Due to this, during ucsi_partner_change call, the usb role
is marked as ROLE_NONE and passed to DWC3 controller the same way.
Fix this by adding UCSI_USB4_IMPLIES_USB quirk and check for it to
decide and pass on proper ROLE information to DWC3 layer.
Oliver Neukum [Thu, 12 Mar 2026 09:45:28 +0000 (10:45 +0100)]
iowarrior: use interruptible lock in iowarrior_write()
The function itself, if it has to wait to perform
IO, use interruptible sleep. Hence the sleep needed
to avoid the write code path racing with itself should also
use interruptible sleep.
Oliver Neukum [Thu, 12 Mar 2026 09:45:27 +0000 (10:45 +0100)]
usb: iowarrior: remove inherent race with minor number
The driver saves the minor number it gets upon registration
in its descriptor for debugging purposes. However, there is
inevitably a window between registration and saving the correct
minor in a descriptor. During this window the debugging output
will be wrong.
As wrong debug output is worse than no debug output, just
remove it.
Oliver Neukum [Thu, 12 Mar 2026 12:34:27 +0000 (13:34 +0100)]
USB: uas: give the error handler the correct name
A UAS device can in principle contain multiple busses.
A reset on the USB level will reset them all. We
cannot reset a single bus.
In practical terms this does not matter, as only
one method of reset is implemented, but we should
not lie.
Johan Hovold [Thu, 5 Mar 2026 13:38:51 +0000 (14:38 +0100)]
USB: usbip: drop redundant device reference
Driver core holds a reference to the USB device while it is bound to a
driver and there is no need to take additional references unless the
structure is needed after disconnect.
Drop the redundant device reference to reduce cargo culting, make it
easier to spot drivers where an extra reference is needed, and reduce
the risk of memory leaks when drivers fail to release it.
Johan Hovold [Thu, 5 Mar 2026 13:38:50 +0000 (14:38 +0100)]
USB: apple-mfi-fastcharge: drop redundant device reference
Driver core holds a reference to the USB device while it is bound to a
driver and there is no need to take additional references unless the
structure is needed after disconnect.
Drop the redundant device reference to reduce cargo culting, make it
easier to spot drivers where an extra reference is needed, and reduce
the risk of memory leaks when drivers fail to release it.
Mikulas Patocka [Mon, 16 Mar 2026 14:04:15 +0000 (15:04 +0100)]
dm: don't report warning when doing deferred remove
If dm_hash_remove_all was called from dm_deferred_remove, it would write
a warning "remove_all left %d open device(s)" if there are some other
devices active.
The warning is bogus, so let's disable it in this case.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Reported-by: Zdenek Kabelac <zkabelac@redhat.com> Cc: stable@vger.kernel.org Fixes: 2c140a246dc0 ("dm: allow remove to be deferred")
Guillaume Gonnet [Tue, 17 Mar 2026 21:32:28 +0000 (22:32 +0100)]
dm init: ensure device probing has finished in dm-mod.waitfor=
The early_lookup_bdev() function returns successfully when the disk
device is present but not necessarily its partitions. In this situation,
dm_early_create() fails as the partition block device does not exist
yet.
In my case, this phenomenon occurs quite often because the device is
an SD card with slow reading times, on which kernel takes time to
enumerate available partitions.
Fortunately, the underlying device is back to "probing" state while
enumerating partitions. Waiting for all probing to end is enough to fix
this issue.
That's also the reason why this problem never occurs with rootwait=
parameter: the while loop inside wait_for_root() explicitly waits for
probing to be done and then the function calls async_synchronize_full().
These lines were omitted in 035641b, even though the commit says it's
based on the rootwait logic...
Anyway, calling wait_for_device_probe() after our while loop does the
job (it both waits for probing and calls async_synchronize_full).
Loic Poulain [Mon, 2 Mar 2026 14:26:03 +0000 (15:26 +0100)]
arm64: dts: qcom: monaco: Reserve full Gunyah metadata region
We observe spurious "Synchronous External Abort" exceptions
(ESR=0x96000010) and kernel crashes on Monaco-based platforms.
These faults are caused by the kernel inadvertently accessing
hypervisor-owned memory that is not properly marked as reserved.
>From boot log, The Qualcomm hypervisor reports the memory range
at 0x91a80000 of size 0x80000 (512 KiB) as hypervisor-owned:
qhee_hyp_assign_remove_memory: 0x91a80000/0x80000 -> ret 0
However, the EFI memory map provided by firmware only reserves the
subrange 0x91a40000–0x91a87fff (288 KiB). The remaining portion
(0x91a88000–0x91afffff) is incorrectly reported as conventional
memory (from efi debug):
efi: 0x000091a40000-0x000091a87fff [Reserved...]
efi: 0x000091a88000-0x0000938fffff [Conventional...]
As a result, the allocator may hand out PFNs inside the hypervisor
owned region, causing fatal aborts when the kernel accesses those
addresses.
Add a reserved-memory carveout for the Gunyah hypervisor metadata
at 0x91a80000 (512 KiB) and mark it as no-map so Linux does not
map or allocate from this area.
For the record:
Hyp version: gunyah-e78adb36e debug (2025-11-17 05:38:05 UTC)
UEFI Ver: 6.0.260122.BOOT.MXF.1.0.c1-00449-KODIAKLA-1
projects / thirdparty / linux.git / log
