Jiri Pirko [Fri, 29 May 2026 13:43:09 +0000 (15:43 +0200)]
RDMA/mlx5: Use UMEM attributes for QP buffers in create_qp
Use the per-attribute UMEM helpers to pin QP buffer umems on
demand. The QP-type predicate selects between the BUF and RQ_BUF
attrs; raw-packet SQ uses its own dedicated SQ_BUF attr.
Jiri Pirko [Fri, 29 May 2026 13:43:08 +0000 (15:43 +0200)]
RDMA/uverbs: Use UMEM attributes for QP creation
Apply the per-attribute UMEM model to the QP create method. Add
three optional UMEM attributes that drivers pick from based on
how their user ABI lays out the QP rings:
- CREATE_QP_BUF_UMEM is a single user buffer that backs both
the SQ and RQ of one QP. This is the common case where
userspace pins one contiguous WQE region for the QP.
- CREATE_QP_SQ_BUF_UMEM and CREATE_QP_RQ_BUF_UMEM are a pair
of user buffers backing the SQ and RQ independently, used
when the two rings live in physically distinct user
allocations and must be pinned and addressed separately.
Existing drivers would map their current umems as follows:
- mlx5: BUF for normal QPs (one ucmd->buf_addr covers SQ+RQ);
for IB_QPT_RAW_PACKET and IB_QP_CREATE_SOURCE_QPN, the RQ
side comes from ucmd->buf_addr (RQ-sized) via RQ_BUF and
the SQ from ucmd->sq_buf_addr via SQ_BUF.
- mlx4: BUF, single ucmd.buf_addr covering SQ+RQ.
- hns: BUF, single ucmd.buf_addr covering SQ + ext-SGE + RQ.
- erdma: BUF, single ureq.qbuf_va sliced by the kernel into
SQ at offset 0 and RQ at rq_offset.
- bnxt_re: SQ_BUF (ureq->qpsva) + RQ_BUF (ureq->qprva, the
RQ side is skipped when the QP uses an SRQ).
- vmw_pvrdma: SQ_BUF (sbuf_addr) + RQ_BUF (rbuf_addr, the RQ
side is skipped when the QP uses an SRQ).
- qedr: SQ_BUF (sq_addr) + RQ_BUF (rq_addr) for whichever
side the QP type actually has (no SQ for XRC_TGT/GSI; no
RQ for XRC_INI/XRC_TGT/SRQ).
- ionic: SQ_BUF (req.sq.addr) + RQ_BUF (req.rq.addr); both
are skipped when the rings are placed in CMB instead of
host memory.
- mana: raw-packet QP uses SQ_BUF (sq_buf_addr) only; the RC
path uses multiple per-queue user buffers (ucmd.queue_buf[])
that do not fit the SQ/RQ pair semantics of these attrs and
stays on the legacy UHW path.
- efa, irdma, hfi1, ocrdma, mthca, cxgb4 and usnic do not pin
a QP WQE buffer via umem; none of these attributes apply.
Jiri Pirko [Fri, 29 May 2026 13:43:07 +0000 (15:43 +0200)]
RDMA/uverbs: Remove legacy umem field from struct ib_cq
Now that all drivers use helper to get umem and manage the lifetime,
legacy umem field in struct ib_cq is no longer needed. Remove it
along with ib_umem_get_cq_tmp() helper that populated it and both
error and destroy paths.
Jiri Pirko [Fri, 29 May 2026 13:43:06 +0000 (15:43 +0200)]
RDMA/mlx4: Use ib_umem_get_cq_buf() for user CQ buffer
Pin the user CQ buffer with ib_umem_get_cq_buf() and take
ownership of the umem in the driver; fall back to
ib_umem_get_va() for the legacy UHW VA path. Apply the same
ownership pattern to the resize path.
Jiri Pirko [Fri, 29 May 2026 13:43:05 +0000 (15:43 +0200)]
RDMA/bnxt_re: Use ib_umem_get_cq_buf_or_va() for user CQ buffer
Pin the user CQ buffer with ib_umem_get_cq_buf_or_va() and take
ownership of the umem in the driver. Apply the same ownership
pattern to the resize path.
Jiri Pirko [Fri, 29 May 2026 13:43:04 +0000 (15:43 +0200)]
RDMA/mlx5: Use ib_umem_get_cq_buf_or_va() for user CQ buffer
Pin the user CQ buffer with ib_umem_get_cq_buf_or_va() and take
ownership of the umem in the driver. Apply the same ownership
pattern to the resize path.
Jiri Pirko [Fri, 29 May 2026 13:43:02 +0000 (15:43 +0200)]
RDMA/uverbs: Add CQ buffer UMEM attribute and driver helpers
Add UVERBS_ATTR_CREATE_CQ_BUF_UMEM and two driver-facing
wrappers, ib_umem_get_cq_buf() and ib_umem_get_cq_buf_or_va(),
that pin a CQ buffer umem from it. The wrappers reuse the
existing legacy CQ buffer-attr filler.
Jiri Pirko [Fri, 29 May 2026 13:43:01 +0000 (15:43 +0200)]
RDMA/uverbs: Push out CQ buffer umem processing into a helper
Extract the UVERBS_ATTR_CREATE_CQ_BUFFER_* parser from the CQ
create handler into uverbs_create_cq_get_buffer_desc(), and wrap
it in ib_umem_get_cq_tmp(), the umem-producing helper the cq_create
handler now calls.
ib_umem_get_cq_tmp() is temporary; subsequent patches replace it
with driver-owned ib_umem_get_cq_buf*() wrappers built on the
same parser, and remove it once all CQ drivers have switched.
Jiri Pirko [Fri, 29 May 2026 13:43:00 +0000 (15:43 +0200)]
RDMA/umem: Route ib_umem_get_va() through ib_umem_get_attr_or_va()
ib_umem_get_va() is now redundant: ib_umem_get_attr_or_va() with
attrs=NULL and attr_id=0 covers the exact same path. Make it a static
inline wrapper instead of a separately exported symbol.
Jiri Pirko [Fri, 29 May 2026 13:42:59 +0000 (15:42 +0200)]
RDMA/core: Introduce generic buffer descriptor infrastructure for umem
Introduce a per-attribute UVERBS_ATTR_UMEM model so each uverbs
command's umem set is explicit in its UAPI definition. Add
driver-facing wrapper helpers that pin a umem on demand from an
attribute or a VA addr; the driver owns the returned umem and
releases it from its destroy/error paths.
Jiri Pirko [Fri, 29 May 2026 13:42:58 +0000 (15:42 +0200)]
RDMA/umem: Split ib_umem_get_va() into a thin wrapper around __ib_umem_get_va()
The follow-up patch is going to introduce ib_umem_get_desc(),
the canonical desc-to-umem helper, which needs to pin a userspace VA
without going through the exported ib_umem_get_va() helper so later on
ib_umem_get_va() would use the ib_umem_get_desc() flow too.
Move the existing ib_umem_get_va() to a static __ib_umem_get_va()
and have ib_umem_get_va() as a thin wrapper that calls it.
Jiri Pirko [Fri, 29 May 2026 13:42:57 +0000 (15:42 +0200)]
RDMA/umem: Rename ib_umem_get() to ib_umem_get_va()
The new umem getter family being introduced in follow-up patches
need a fitting name for the central all-source helper that resolves
attributes, legacy fillers and a UHW VA fallback.
Rename the existing VA-pinning helper ib_umem_get() to ib_umem_get_va()
so the name is freed up. The new name is consistent with names of rest
of the helpers that are about to be introduced.
Li RongQing [Thu, 28 May 2026 03:16:24 +0000 (23:16 -0400)]
KVM: x86: ioapic: Use old_dest_mode consistently in ioapic_write_indirect()
When reconstructing a temporary IRQ state from the historical/old IOAPIC
redirection table configuration in ioapic_write_indirect(), the code
previously assigned 'irq.dest_id' from 'old_dest_id', but incorrectly
queried the live/new 'e->fields.dest_mode' to populate 'irq.dest_mode'.
Mixing the old destination ID with the new destination mode creates an
inconsistent, hybrid IRQ state. This discrepancy leads to an incorrect
vCPU bitmap calculation via kvm_bitmap_or_dest_vcpus(), causing subsequent
interrupt routing updates (such as RTC interrupt handling) to target the
wrong set of virtual processors if both fields were modified simultaneously.
Fix this by using 'old_dest_mode' consistently alongside 'old_dest_id' to
ensure the historical IRQ structure is reconstructed accurately.
Fixes: c96001c5702e ("KVM: X86: Use APIC_DEST_* macros properly in kvm_lapic_irq.dest_mode") Signed-off-by: Li RongQing <lirongqing@baidu.com> Link: https://patch.msgid.link/20260528031624.1929-1-lirongqing@baidu.com Signed-off-by: Sean Christopherson <seanjc@google.com>
Linus Torvalds [Fri, 29 May 2026 22:46:40 +0000 (15:46 -0700)]
Merge tag 'net-7.1-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull more networking fixes from Jakub Kicinski:
"Quick follow up, nothing super urgent here. Main reason I'm sending
this out is because the IPsec and Bluetooth PRs did not make it
yesterday. I don't want to have to send you all of this + whatever
comes next week, for rc7. The fixes under "Previous releases -
regressions" are for real user-reported regressions from v7.0.
Previous releases - regressions:
- Revert "ipv6: preserve insertion order for same-scope addresses"
- xfrm: move policy_bydst RCU sync, a fix which added a sync RCU on
netns exit got backported to stable and was causing serious
accumulation of dying netns's for real workloads
- pcs-mtk-lynxi: fix bpi-r3 serdes configuration
Previous releases - always broken:
- usual grab bag of race, locking and leak fixes for Bluetooth
- handful of page handling fixes for IPsec"
* tag 'net-7.1-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (36 commits)
wireguard: send: append trailer after expanding head
Revert "ipv6: preserve insertion order for same-scope addresses"
net: skbuff: fix pskb_carve leaking zcopy pages
ipv6: fix possible infinite loop in fib6_select_path()
ipv6: fix possible infinite loop in rt6_fill_node()
bpf: sockmap: fix tail fragment offset in bpf_msg_push_data
vsock/virtio: bind uarg before filling zerocopy skb
Revert "esp: fix page frag reference leak on skb_to_sgvec failure"
net: pcs: pcs-mtk-lynxi: fix bpi-r3 serdes configuration
sctp: fix race between sctp_wait_for_connect and peeloff
net: mana: Skip redundant detach on already-detached port
net: mana: Add NULL guards in teardown path to prevent panic on attach failure
Bluetooth: hci_sync: Reset device counters in hci_dev_close_sync()
Bluetooth: hci_sync: Set HCI_CMD_DRAIN_WORKQUEUE during device close
Bluetooth: hci_core: Rework hci_dev_do_reset() to use hci_sync functions
Bluetooth: ISO: serialize iso_sock_clear_timer with socket lock
Bluetooth: ISO: fix UAF in iso_recv_frame
Bluetooth: L2CAP: Fix possible crash on l2cap_ecred_conn_rsp
Bluetooth: l2cap: clear chan->ident on ECRED reconfiguration success
Bluetooth: hci_qca: Use 100 ms SSR delay for rampatch and NVM loading
...
Arnd Bergmann [Fri, 29 May 2026 22:17:56 +0000 (00:17 +0200)]
Merge tag 'ffa-updates-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux into soc/drivers
Arm FF-A + SMCCC updates for v7.2
1. The FF-A core is moved onto the driver model by reverting the earlier
rootfs initcall change and registering the core as a platform driver with
a synthetic arm-ffa platform device. Enumerated FF-A devices are now
parented below the FF-A core device, and probing is deferred until pKVM
has completed its FF-A proxy initialisation.
2. The platform-driver conversion is also adjusted so systems without FF-A
support treat early unsupported transport/version discovery as a quiet
probe miss rather than a failed matched probe.
3. The register-based partition discovery path now honors the descriptor
size reported by FFA_PARTITION_INFO_GET_REGS. This keeps parsing aligned
for newer FF-A descriptor layouts while still copying only the fields the
driver understands.
4. Also included is an Arm SMCCC fix for the optional SOC_ID name call: the
name query now uses the SMC64 function ID required for returning eight
characters per register.
* tag 'ffa-updates-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux:
firmware: arm_ffa: Treat missing FF-A feature on a platform as a probe miss
firmware: smccc: Fix Arm SMCCC SOC_ID name call
firmware: arm_ffa: Honor partition info descriptor size
firmware: arm_ffa: Defer probe until pKVM is initialized
firmware: arm_ffa: Set the core device as FF-A device parent
firmware: arm_ffa: Register core as a platform driver
Revert "firmware: arm_ffa: Change initcall level of ffa_init() to rootfs_initcall"
firmware: arm_ffa: Fix sched-recv callback partition lookup
firmware: arm_ffa: Snapshot notifier callbacks under lock
firmware: arm_ffa: Align RxTx buffer size before mapping
firmware: arm_ffa: Validate framework notification message layout
firmware: arm_ffa: Keep framework RX release under lock
firmware: arm_ffa: Bound PARTITION_INFO_GET_REGS copies
firmware: arm_ffa: Unregister bus notifier on teardown for FF-A v1.0
firmware: arm_ffa: Fix per-vcpu self notifications handling in workqueue
firmware: arm_ffa: Avoid collapsing NPI work from different CPUs
firmware: arm_ffa: Skip free_pages on RX buffer alloc failure
firmware: arm_ffa: Check for NULL FF-A ID table while driver registration
Linus Torvalds [Fri, 29 May 2026 22:17:53 +0000 (15:17 -0700)]
Merge tag 'clang-fixes-7.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/nathan/linux
Pull clang build fix from Nathan Chancellor:
"A small fix to disable -Wattribute-alias for clang in the few places
it is already disabled for GCC, now that tip of tree clang has
implemented -Wattribute-alias as GCC has"
* tag 'clang-fixes-7.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/nathan/linux:
Disable -Wattribute-alias for clang-23 and newer
Andrei Vagin [Tue, 26 May 2026 20:50:43 +0000 (20:50 +0000)]
Revert "x86/fpu: Refine and simplify the magic number check during signal return"
This reverts
dc8aa31a7ac2 ("x86/fpu: Refine and simplify the magic number check during signal return").
The aforementioned commit broke applications that construct signal frames in
userspace (such as CRIU and gVisor) if the frame's xstate size is smaller than
the kernel's fpstate->user_size.
Furthermore, this introduces a critical issue for checkpoint/restore tools
like CRIU. If a process is checkpointed while inside a signal handler, its
stack contains a signal frame formatted according to the source host's xstate
capabilities.
If that process is later restored on a destination host with larger xstate
capabilities (e.g., a newer CPU with more features enabled, resulting in
a larger fpstate->user_size), the kernel will look for FP_XSTATE_MAGIC2 at the
destination host's larger user_size offset instead of the offset encoded in
the frame's fx_sw->xstate_size.
This causes the magic2 check to fail, forcing sigreturn to silently fall back
to "FX-only" mode. Upon return from the signal handler, the process's extended
state is reset to initial values instead of being restored, leading to silent
data corruption.
The aforementioned commit cited
d877550eaf2d ("x86/fpu: Stop relying on userspace for info to fault in xsave buffer")
as justification to stop relying on userspace for the magic number check.
However, these two changes are fundamentally different. The last one only
changed how much memory the kernel ensures is paged-in before running XRSTOR
to prevent an infinite loop. It did not change the signal frame format or how
the layout is validated.
Reverting this change restores the use of fx_sw->xstate_size for
locating magic2 and restores the necessary sanity checks, ensuring that
the signal frame remains self-describing and portable.
[ bp: Massage commit message. ]
Fixes: dc8aa31a7ac2 ("x86/fpu: Refine and simplify the magic number check during signal return") Signed-off-by: Andrei Vagin <avagin@google.com> Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Acked-by: Chang S. Bae <chang.seok.bae@intel.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/all/20260429000623.3356606-1-avagin@google.com
Arnd Bergmann [Fri, 29 May 2026 22:04:45 +0000 (00:04 +0200)]
Merge tag 'scmi-updates-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux into soc/drivers
Arm SCMI updates for v7.2
1. Improve SCMI clock handling with a protocol-level determine_rate operation,
simplified per-clock rate properties, dynamic rate allocation, bounded
iterator support, lazy full-rate discovery, and hardened parent/rate
enumeration.
2. Fix several SCMI bounds and payload validation issues, including clock rate
discovery OOB handling, power domain name lookup, Powercap domain state
access, BASE_ERROR_EVENT and SENSOR_UPDATE payload sizing, and sensor config
width handling.
3. Rework SCMI transport probing for virtio and OP-TEE using per-instance
transport handles and a generic transport supplier, removing the need to
register SCMI core drivers from transport probe paths.
4. Add i.MX SCMI MISC reset reason support and print i.MX95 boot/shutdown
reasons via the System Manager interface.
5. Clean up SCMI core internals, including base-info naming, quirk parsing and
table iteration, and list iteration.
6. Fix SCPI clock provider removal so child clock providers are unregistered
using the same DT nodes used at registration time.
* tag 'scmi-updates-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/sudeep.holla/linux: (31 commits)
firmware: arm_scmi: optee: Rework transport probe sequence
firmware: arm_scmi: virtio: Rework transport probe sequence
firmware: arm_scmi: Add a generic transport supplier
firmware: arm_scmi: Add transport instance handles
firmware: arm_scmi: Fix OOB in scmi_power_name_get()
firmware: arm_scmi: Validate Powercap domains before state access
firmware: arm_scmi: Validate SENSOR_UPDATE payload size
firmware: arm_scmi: Validate BASE_ERROR_EVENT payload size
firmware: arm_scmi: Read sensor config as 32-bit value
clk: scpi: Unregister child clock providers on remove
firmware: arm_scmi: Introduce all_rates_get clock operation
firmware: arm_scmi: Fix OOB in scmi_clock_describe_rates_get_lazy()
firmware: arm_scmi: Use bound iterators to minimize discovered rates
firmware: arm_scmi: Use proper iter_response_bound_cleanup() name
firmware: arm_scmi: Fix bound iterators returning too many items
firmware: arm_scmi: Add bound iterators support
firmware: arm_scmi: Refactor iterators internal allocation
firmware: arm_scmi: Harden clock parents discovery
firmware: arm_scmi: Make clock rates allocation dynamic
firmware: arm_scmi: Drop unused clock rate interfaces
...
Arnd Bergmann [Fri, 29 May 2026 22:02:36 +0000 (00:02 +0200)]
Merge tag 'qcom-drivers-for-7.2' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux into soc/drivers
Qualcomm driver updates for v7.2
Enable QSEECOM and with that access to UEFI variables on the Surface Pro
12in laptop.
Refactor the Geni Serial-Engine helper code to allow the serial engine
drivers (such as I2C) to operate on targets where power and performance
is controlled through an SCMI server instead of individual resources in
Linux.
Extend the LLCC driver to support reading its data from a System Cache
Table (SCT) in memory instead of being hard coded per platform in the
driver. Also add support for the Eliza platform.
Add support for the Hawi platform to pd-mapper.
Switch the SMEM driver to track partitions using xarray to handle the
ever growing number of hosts better.
Extend the socinfo driver with knowledge about the Nord, SM7750,
IPQ9650, and Shikra SoCs, as well as PMAU0102, PMC1020H, PMIV0102, and
PMIV0104 PMICs.
Define UBWC 3.1 and add a couple of convenient helpers in the UBWC
library for MDSS and Adreno.
Fix a memory leak in the WCNSS firmware download mechanism.
* tag 'qcom-drivers-for-7.2' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux: (40 commits)
soc: qcom: geni-se: Introduce helper APIs for performance control
soc: qcom: geni-se: Introduce helper API for attaching power domains
soc: qcom: geni-se: Add resources activation/deactivation helpers
soc: qcom: geni-se: Handle core clk in geni_se_clks_off() and geni_se_clks_on()
soc: qcom: geni-se: Introduce helper API for resource initialization
soc: qcom: geni-se: Add geni_icc_set_bw_ab() function
soc: qcom: geni-se: Refactor geni_icc_get() and make qup-memory ICC path optional
soc: qcom: llcc-qcom: Fix NULL vs IS_ERR() bug in qcom_llcc_get_fw_config()
soc: qcom: llcc-qcom: Add support for Eliza
dt-bindings: cache: qcom,llcc: Document Eliza LLCC block
soc: qcom: ubwc: add helper controlling AMSBC enablement
soc: qcom: ubwc: define helper for MDSS and Adreno drivers
soc: qcom: ubwc: define UBWC 3.1
soc: qcom: socinfo: Add SoC ID for Nord SA8797P
dt-bindings: arm: qcom,ids: Add SoC ID for Nord SA8797P
soc: qcom: socinfo: Add SoC ID for SM7750
dt-bindings: arm: qcom,ids: Add SoC ID for SM7750
soc: qcom: socinfo: Add PMIC PMAU0102
soc: qcom: socinfo: Add PMIV0102 & PMIV0104 PMICs
firmware: qcom: scm: Allow QSEECOM on Surface Pro 12in
...
The probe ordering fix in the following patch ensures amd_gpio_probe()
completes before i2c-designware probes AMDI0010:02, allowing the
existing amd_gpio_irq_enable() flow to work correctly. The manual IRQ
restoration added by this patch is therefore no longer needed.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=221494 Signed-off-by: Hardik Prakash <hardikprakash.official@gmail.com> Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org> Fixes: 3812a9e84265a ("pinctrl-amd: enable IRQ for WACF2200 touchscreen on Lenovo Yoga 7 14AGP11") Signed-off-by: Linus Walleij <linusw@kernel.org>
Arnd Bergmann [Fri, 29 May 2026 21:39:45 +0000 (23:39 +0200)]
Merge tag 'memory-controller-drv-fixes-7.1' of https://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux-mem-ctrl into arm/fixes
Memory controller drivers - fixes for v7.1
Fix probing of Atmel EBI memory controller driver e.g. on at91
sam9x60-curiosity board due to usage of platform_driver_probe() which is
not handling deferred probe. Lack of EBI driver caused dependant NAND
controller to fail to probe, basically failing entire board boot.
* tag 'memory-controller-drv-fixes-7.1' of https://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux-mem-ctrl:
memory: atmel-ebi: Allow deferred probing
Arnd Bergmann [Fri, 29 May 2026 21:37:25 +0000 (23:37 +0200)]
Merge tag 'samsung-drivers-fixes-7.1-2' of https://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux into arm/fixes
Samsung SoC driver fixes for v7.1
Fix several concurrency issues present in Samsung ACPM firmware drivers,
used currently only on Google GS101. Tudor with help of Sashiko
identified several missing barriers and incomplete synchronization,
leading to possible transfer data corruption or use after free. Few
other issues related to probe, including missing mailbox cleanup, were
also fixed.
* tag 'samsung-drivers-fixes-7.1-2' of https://git.kernel.org/pub/scm/linux/kernel/git/krzk/linux:
firmware: samsung: acpm: Fix infinite loop on sequence number exhaustion
firmware: samsung: acpm: Fix missing LKMM barriers in sequence allocator
firmware: samsung: acpm: Fix false timeouts and Use-After-Free in polling
firmware: samsung: acpm: Fix mailbox channel leak on probe error
firmware: samsung: acpm: Fix cross-thread RX length corruption
Zhenghang Xiao [Tue, 26 May 2026 08:53:13 +0000 (16:53 +0800)]
drm/gem: fix race between change_handle and handle_delete
drm_gem_change_handle_ioctl leaves the old handle live in the IDR
during the window between spin_unlock(table_lock) and the final
spin_lock(table_lock). A concurrent drm_gem_handle_delete on the old
handle succeeds in this window, decrements handle_count to 0, and frees
the GEM object while the new handle's IDR entry still references it.
NULL the old handle's IDR entry before dropping table_lock so that any
concurrent GEM_CLOSE on the old handle sees NULL and returns -EINVAL.
Restore the old entry on the prime-bookkeeping error path.
Fixes: 5e28b7b94408 ("drm: Set old handle to NULL before prime swap in change_handle") Signed-off-by: Zhenghang Xiao <kipreyyy@gmail.com> Cc: stable@vger.kernel.org Signed-off-by: Dave Airlie <airlied@redhat.com> Link: https://patch.msgid.link/20260526085313.26791-1-kipreyyy@gmail.com
Gary Guo [Fri, 8 May 2026 15:29:49 +0000 (16:29 +0100)]
rust: pin_init: internal: use `loop {}` to produce never value
In the `init!`/`pin_init!` macros, we rely on a trick that assigns never
(`!`) values to all mentioned fields in never-executed code to let the
compiler check that all fields have been initialized.
Currently we use `::core::panic!()` to produce this value, but before Rust
1.91.0, it creates outlined `panic_cold_explicit` functions which do not
get removed by the optimizer, thus leaving dead code behind in the binary.
This has been fixed by [1], which lands in Rust 1.91.0+, higher than the
kernel minimum version 1.85.0.
This causes ~200 dead `panic_cold_explicit` instances being included in the
binary, with ~90 of them from nova-core's usage of pin-init.
Work around the issue by using `loop {}` which creates the never value
without macro expansion or function call at all. All instances of
`panic_cold_explicit` outside libcore are removed by this change in my
kernel build.
Linus Torvalds [Fri, 29 May 2026 20:47:55 +0000 (13:47 -0700)]
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"arm64:
- Restore CONFIG_PKVM_DISABLE_STAGE2_ON_PANIC to its former glory by
making sure the config symbol is correctly spelled out in the code
- Don't reset the AArch32 view of the PMU counters to zero when the
guest is writing to them
- Fix an assorted collection of memory leaks in the newly added
tracing code
- Fix the capping of ZCR_EL2 which could be used in an unsanitised
way by an L2 guest
x86:
- Include the kernel's linux/mman.h in KVM selftests to ensure
MADV_COLLAPSE is defined, as older libc versions may not provide
it.
- Include execinfo.h if and only if KVM selftests are building
against glibc, and provide a test_dump_stack() for non-glibc
builds.
- Silence an annoying RCU splat on (even non-KVM-related) panics.
The splat is technically legit, but in practice not an issue. To
have a race, you would need to unload the KVM modules at exactly
the time a panic happens; and speaking of incredibly rare races,
taking the locks risks introducing a deadlock if the module unload
code took the lock on a CPU that has been halted. Which seems
possibly more likely than the RCU grace period issue, so just shut
it up. This code used to be in KVM but is now outside it; but the
x86 maintainers haven't picked it up, so here we are.
- Rate-limit global clock updates once again (but without delayed
work), as KVM was subtly relying on the old rate-limiting for NPT
correction to guard against "update storms" when running without a
master clock on systems with overcommitted CPUs.
- Fix a brown paper bag goof where KVM checked if ERAPS is "dirty"
instead of marking it dirty when emulating INVPCID.
- Flush the TLB when transitioning from xAVIC => x2AVIC to ensure the
CPU TLB doesn't contain AVIC-tagged entries for the APIC base GPA.
- The top 10 commits fix buffer overflow (and potential TOC/TOU)
flaws in the page state change protocol for encrypted VMs. AI
models find it quite easily given it was reported three times, but
aren't as good at writing a comprehensive fix. There's more to
clean up in the area, which will come in 7.2"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (22 commits)
KVM: SEV: Use READ_ONCE() when reading entries/indices from PSC buffer
KVM: SEV: Check PSC request indices against the actual size of the buffer
KVM: SEV: Don't explicitly pass PSC buffer to snp_begin_psc()
KVM: SEV: WARN if KVM attempts to setup scratch area with min_len==0
KVM: SEV: Compute the correct max length of the in-GHCB scratch area
KVM: SEV: Use the size of the PSC header as the minimum size for PSC requests
KVM: SEV: Ignore Port I/O requests of length '0'
KVM: SEV: Reject MMIO requests larger than 8 bytes with GHCB v2+
KVM: SEV: Ignore MMIO requests of length '0'
KVM: SEV: Require in-GHCB scratch area if GHCB v2+ is in use
KVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor
KVM: arm64: Fix memory leak in hyp_trace_unload()
KVM: arm64: Fix rollback in hyp_trace_buffer_share_hyp()
KVM: arm64: Fix meta-page unsharing in pKVM hyp tracing
KVM: arm64: PMU: Preserve AArch32 counter low bits
KVM: SVM: Flush the current TLB when transitioning from xAVIC => x2AVIC
KVM: x86: Fix ERAPS RAP clear on INVPCID single-context invalidation
KVM: arm64: Fix CONFIG_PKVM_DISABLE_STAGE2_ON_PANIC
KVM: selftests: Guard execinfo.h inclusion for non-glibc builds
KVM: x86: Rate-limit global clock updates on vCPU load
...
pinctrl: max77620: Unify usage of space and comma in platform_device_id array
The most accepted style for the array terminator is to use a single
space between the curly braces and no trailing comma. Also don't use a comma
directly before a closing brace in the other entries.
pinctrl: Use named initializers for platform_device_id arrays
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 unit.
While touching these arrays unify spacing and usage of commas.
Rong Xu [Fri, 29 May 2026 18:53:46 +0000 (11:53 -0700)]
kbuild: distributed build support for Clang ThinLTO
Add distributed ThinLTO build support for the Linux kernel.
This new mode offers several advantages: (1) Increased
flexibility in handling user-specified build options.
(2) Improved user-friendliness for developers. (3) Greater
convenience for integrating with objtool and livepatch.
Note that "distributed" in this context refers to a term
that differentiates in-process ThinLTO builds by invoking
backend compilation through the linker, not necessarily
building in distributed environments.
Distributed ThinLTO is enabled via the
`CONFIG_LTO_CLANG_THIN_DIST` Kconfig option. For example:
> make LLVM=1 defconfig
> scripts/config -e LTO_CLANG_THIN_DIST
> make LLVM=1 oldconfig
> make LLVM=1 vmlinux -j <..>
The build flow proceeds in four stages:
1. Perform FE compilation, mirroring the in-process ThinLTO mode.
2. Thin-link the generated IR files and object files.
3. Find all IR files and perform BE compilation, using the flags
stored in the .*.o.cmd files.
4. Link the BE results to generate the final vmlinux.o.
NOTE: This patch currently implements the build for the main kernel
image (vmlinux) only. Kernel module support is planned for a
subsequent patch.
Tested on the following arch: x86, arm64, loongarch, and
riscv.
The earlier implementation details can be found here:
https://discourse.llvm.org/t/rfc-distributed-thinlto-build-for-kernel/85934
Masahiro Yamada [Fri, 29 May 2026 18:53:44 +0000 (11:53 -0700)]
kbuild: move vmlinux.a build rule to scripts/Makefile.vmlinux_a
Move the build rule for vmlinux.a to a separate file in preparation
for supporting distributed builds with Clang ThinLTO.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Tested-by: Rong Xu <xur@google.com> Tested-by: Piotr Gorski <piotrgorski@cachyos.org> Tested-by: Nathan Chancellor <nathan@kernel.org> Signed-off-by: Rong Xu <xur@google.com> Link: https://patch.msgid.link/20260529185347.2418373-2-xur@google.com
[nathan: Squash in forward fix from Rong around '--thin' to $(AR)
https://patch.msgid.link/20260529185347.2418373-3-xur@google.com] Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Maulik Shah [Fri, 29 May 2026 08:09:11 +0000 (13:39 +0530)]
pinctrl: qcom: Replace open coded eoi call with irq_chip_eoi_parent()
Before commit 14dbe186b9d4 ("pinctrl: msmgpio: Make the irqchip immutable")
msm gpio irqchip conditionally initialized pctrl->irq_chip.irq_eoi to
irq_chip_eoi_parent() only for the GPIO irqs having a wakeup capable irq.
In order to make gpio irqchip immutable pctrl->irq_chip.irq_eoi is
initialized to msm_gpio_irq_eoi() which now gets invoked for both wake up
and non-wakeup capable GPIO IRQs.
Replace open coded eoi call to parent irqchip with irq_chip_eoi_parent().
Since the irq_chip_*_parent() APIs internally do not check the valid parent
data is present to ensure irq_chip_eoi_parent() is only invoked for wakeup
capable GPIOs validate d->parent_data within msm_gpio_irq_eoi().
For non wakeup capable GPIOs d->parent_data will be NULL since parent
irqchip diconnects hierarchy using irq_domain_disconnect_hierarchy() and
later irq framework trims hierarchy using irq_domain_trim_hierarchy() which
makes d->parent_data as NULL.
Gary Guo [Wed, 27 May 2026 17:19:57 +0000 (18:19 +0100)]
rust: pin-init: remove `E` from `InitClosure`
Move `E` from type to trait impl block. This greatly shortens the
monomorphized type names. The `__pinned_init` function name is only
slightly shortened as it still encodes the `E` as part of `PinInit<T, E>`
in the symbol.
`T` cannot be moved to trait impl block otherwise it will start to conflict
with the `impl Init<T> for T` as Rust cannot deduce that there're no types
that fulfill `T: FnOnce(*mut T)`.
Gary Guo [Wed, 27 May 2026 17:19:56 +0000 (18:19 +0100)]
rust: pin-init: move `InitClosure` out from `__internal`
The `__internal` module is for exposing internal items publicly to
procedural macros (pin-init-internal). Types that are crate-local only can
just have proper visibility and does not need to be in `__internal`.
The type name of `InitClosure` can often shows up in symbol names, this
reduces the length slightly.
Mirko Adzic [Wed, 27 May 2026 17:19:54 +0000 (18:19 +0100)]
rust: pin-init: internal: suppress `non_snake_case` lint in `[pin_]init!`
Allows `non_snake_case` lint on local variables generated in `[pin_]init!`.
Conceptually the identifiers in `[pin_]init!` just references the field
names, and are not defining them, so the warning should not be generated,
similar to how constructing a struct with non-snake-case field names do no
generate these warnings.
Reported-by: Gary Guo <gary@garyguo.net> Closes: https://github.com/Rust-for-Linux/pin-init/issues/125 Closes: https://lore.kernel.org/rust-for-linux/DGTBJBIVFZ2K.2F1ZEFGY0G7NK@garyguo.net/ Fixes: 42415d163e5d ("rust: pin-init: add references to previously initialized fields") Signed-off-by: Mirko Adzic <adzicmirko97@gmail.com>
[ Reworded commit message - Gary ] Link: https://patch.msgid.link/20260527-pin-init-sync-v1-3-e20335ed2501@garyguo.net Signed-off-by: Gary Guo <gary@garyguo.net>
Mirko Adzic [Wed, 27 May 2026 17:19:53 +0000 (18:19 +0100)]
rust: pin-init: internal: suppress `non_snake_case` lint in `#[pin_data]`
Allows `non_snake_case` lint on struct fields generated by `#[pin_data]`.
Since the same warning will be reported by the compiler on the struct
definition, having extra warnings for the generated code is unnecessary
and confusing.
rust: pin-init: internal: pin_data: filter non-`#[cfg]` attr in generated code
When using a macro with custom attributes in a `#[pin_data]` struct it
can mess up the generated code. The generated code needs nothing more than
the `#[cfg]` attribute, thus strip away all other attributes.
[ Rebased and updated to only include `#[cfg]` instead of both `#[cfg]` and
`#[doc]`; doc is not needed for the generated hidden items. - Gary ]
Xianwei Zhao [Thu, 7 May 2026 08:21:06 +0000 (08:21 +0000)]
dt-bindings: pinctl: amlogic,pinctrl-a4: Add compatible string for A9
Update dt-binding document for pinctrl of Amlogic A9.
In Amlogic A9 SoC, a bank mux register reuse other banks.
The multiplexed part requires special processing and is
therefore incompatible with the previous SoCs.
EDAC: Consistently define pci_device_ids using named initializers
... and PCI device helpers.
The various struct pci_device_id arrays were initialized mostly by one of the
PCI_DEVICE macros and then list expressions. The latter aren't easily readable
if you're not into PCI. Using named initializers is more explicit and thus
easier to parse.
Also use PCI_DEVICE* helper macros to assign .vendor, .device, .subvendor and
.subdevice where appropriate and skip explicit assignments of 0 (which the
compiler takes care of).
The secret plan is to make struct pci_device_id::driver_data an anonymous
union (similar to
Peter Griffin [Wed, 27 May 2026 21:08:45 +0000 (22:08 +0100)]
MAINTAINERS: add myself as co-maintainer for Samsung pinctrl drivers
As Google Tensor gs101 is based off a Samsung Exynos design I've been
working on the Samsung pinctrl drivers and have an interest in helping
maintain this code.
wireguard: send: append trailer after expanding head
With how this is currently written, we add the trailer, zero it out, and
then add the header space on. If that header space requires a
reallocation + copy, the zeros in the trailer aren't copied, because the
skb len hasn't actually been yet expanded to cover that. Instead add the
padding at the end of the process rather than at the beginning.
Revert "ipv6: preserve insertion order for same-scope addresses"
Chris Adams reported that preserving insertion order for same-scope
addresses is causing SSH connections to be dropped after stopping a VM
while running NetworkManager.
NetworkManager caches the IPv6 address configuration, when a RA arrives,
it determines the list of addresses to configure and checks if the
addresses are already in the right order in the kernel. If they aren't,
NetworkManager removes and re-adds them to achieve the desired order.
As the order changes, NetworkManager is confused and reconfigures the
addresses on every update. In addition, this would also affect to cloud
tooling that relies on IPv6 addresses order to identify primary and
secondaries addresses.
Add a device tree overlay enabling the Toradex Capacitive Touch Display
10.1" LVDS on the Verdin Development Board with Verdin AM62 Mezzanine
expansion board. The panel connects via the AM62 OLDI0 on the Mezzanine
LVDS interface (J10). The panel is a LogicTechno LT170410-2WHC 10.1" WXGA
IPS LCD and the touch input is provided by an Atmel MaxTouch capacitive
touch controller.
Vitor Soares [Fri, 22 May 2026 16:11:14 +0000 (17:11 +0100)]
arm64: dts: ti: k3-am62-verdin: Add Toradex Verdin Mezzanine CAN
Add a device tree overlay enabling AM62 MCU_MCAN1 on the Toradex Verdin
Development Board with Verdin AM62 Mezzanine expansion board. MCU_MCAN1
is exposed on the Mezzanine CAN Header (J13), Pin 3 (CAN1_CONN_N) and
Pin 4 (CAN1_CONN_P).
Add device tree overlays for the Toradex OV5640 CSI Cameras on Verdin
CSI_1. Two variants are supported: the current CSI Camera Set 5MP OV5640
with a 27 MHz oscillator and the legacy CSI Camera Module 5MP OV5640
with a 24 MHz oscillator.
Add a device tree overlay enabling Bridge Tied Load (BTL) mode on the
Nuvoton NAU8822 audio codec present on the Verdin Development Board.
In BTL mode, the two loudspeaker outputs are bridged to deliver higher
output power on the X28 speaker connector.
Add a device tree overlay for the Toradex Capacitive Touch Display 7"
DSI on the Verdin DSI_1 interface. The display features an internal
Texas Instruments SN65DSI83 DSI-to-LVDS bridge driving a Riverdi
RVT70HSLNWCA0 7" WSVGA IPS TFT LCD panel. The touch input is provided
by an Ilitek ILI2132 capacitive touch controller.
Add a device tree overlay for the Toradex Capacitive Touch Display 10.1"
on the Verdin DSI_1 interface. The display features an internal
Texas Instruments SN65DSI83 DSI-to-LVDS bridge driving a Riverdi
RVT101HVLNWC00 10.1" WXGA (1280x800) IPS TFT LCD panel. The touch input
is provided by an Ilitek ILI2132 capacitive touch controller.
The overlay is also combined with the Verdin AM62 Dahlia carrier board
device trees to provide ready-to-use DTBs in both WiFi and non-Wifi SoM
variants.
Add a device tree overlay for the Toradex Capacitive Touch Display 10.1"
LVDS connected via Verdin AM62 OLDI on carrier boards exposing LVDS
interface (e.g., Mallow). The panel is a LogicTechno LT170410-2WHC 10.1"
WXGA IPS LCD and the touch input is provided by an Atmel MaxTouch
capacitive touch controller.
Vitor Soares [Fri, 22 May 2026 16:11:05 +0000 (17:11 +0100)]
arm64: dts: ti: k3-am62-verdin: Add Toradex DSI to LVDS adapter with 10.1" display
Add a device tree overlay for the Toradex DSI to LVDS Adapter with the
Toradex Capacitive Touch Display 10.1" LVDS. The adapter connects to the
Verdin DSI_1 interface. It is based on the Texas Instruments SN65DSI84
DSI-to-LVDS bridge and drives a LogicTechno LT170410-2WHC 10.1" WXGA LVDS
panel. Touch input is provided by an Atmel MaxTouch capacitive touch
controller.
Arnd Bergmann [Fri, 29 May 2026 09:41:20 +0000 (11:41 +0200)]
pinctrl: tegra238: remove unused entries
The -Wunused-const-variable check points out a number of added
entries that are currently not referenced:
drivers/pinctrl/tegra/pinctrl-tegra238.c:1169:27: error: 'soc_gpio86_phh3_pins' defined but not used [-Werror=unused-const-variable=]
1169 | static const unsigned int soc_gpio86_phh3_pins[] = {
| ^~~~~~~~~~~~~~~~~~~~
drivers/pinctrl/tegra/pinctrl-tegra238.c:1165:27: error: 'uart5_cts_phh2_pins' defined but not used [-Werror=unused-const-variable=]
1165 | static const unsigned int uart5_cts_phh2_pins[] = {
| ^~~~~~~~~~~~~~~~~~~
drivers/pinctrl/tegra/pinctrl-tegra238.c:1161:27: error: 'uart5_rts_phh1_pins' defined but not used [-Werror=unused-const-variable=]
1161 | static const unsigned int uart5_rts_phh1_pins[] = {
| ^~~~~~~~~~~~~~~~~~~
Remove them for now, they can just be added back if they get
used in the future.
1) xfrm: route MIGRATE notifications to caller's netns
Thread the caller's netns through km_migrate() so that
MIGRATE notifications go to the issuing netns, fixing both the
init_net listener leak and MOBIKE notifications inside
non-init netns. From Maoyi Xie.
2) xfrm: ipcomp: Free destination pages on acomp errors
Move the out_free_req label up so that allocated destination
pages are released on decompression errors, not only on success.
From Herbert Xu.
3) xfrm: Check for underflow in xfrm_state_mtu
Reject configurations that cause xfrm_state_mtu() to underflow,
preventing a negative TFCPAD value from becoming a memset size
that triggers an out-of-bounds write of several terabytes.
From David Ahern.
4) xfrm: ah: use skb_to_full_sk in async output callbacks
Convert the possibly-incomplete skb->sk to a full socket pointer
in async AH callbacks so that a request_sock or timewait_sock
never reaches xfrm_output_resume() downstream consumers.
From Michael Bommarito.
5) Add and revert: esp: fix page frag reference leak on skb_to_sgvec failure
The patch does not fix te issue completely.
6) xfrm: esp: restore combined single-frag length gate
Check the aligned post-trailer combined length against a page limit
in the fast path, preventing skb_page_frag_refill() from falling
back to a page too small for the destination scatterlist.
From Jingguo Tan.
7) xfrm: iptfs: reset runtime state when cloning SAs
Reinitialise the clone's mode_data runtime objects before
publishing it, preventing queued skbs from being freed with
list state copied from the original SA when migration fails.
From Shaomin Chen.
8) xfrm: move policy_bydst RCU sync from per-netns .exit to .pre_exit
Flush policy tables and drain the workqueue in a .pre_exit handler
so that cleanup_net() pays one RCU grace period per batch instead
of one per namespace, fixing stalls at high CLONE_NEWNET rates.
From Usama Arif.
9) xfrm: input: hold netns during deferred transport reinjection
Take a netns reference when queueing deferred transport reinjection
work and drop it after the callback completes, keeping the skb->cb
net pointer valid until the deferred work runs.
From Zhengchuan Liang.
* tag 'ipsec-2026-05-29' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec:
Revert "esp: fix page frag reference leak on skb_to_sgvec failure"
xfrm: input: hold netns during deferred transport reinjection
xfrm: move policy_bydst RCU sync from per-netns .exit to .pre_exit
xfrm: iptfs: reset runtime state when cloning SAs
xfrm: esp: restore combined single-frag length gate
esp: fix page frag reference leak on skb_to_sgvec failure
xfrm: ah: use skb_to_full_sk in async output callbacks
xfrm: Check for underflow in xfrm_state_mtu
xfrm: ipcomp: Free destination pages on acomp errors
xfrm: route MIGRATE notifications to caller's netns
====================
Pavel Begunkov [Thu, 28 May 2026 18:43:53 +0000 (19:43 +0100)]
net: skbuff: fix pskb_carve leaking zcopy pages
When SKBFL_MANAGED_FRAG_REFS is set, frag pages are not refcounted but
their lifetime is controlled by the attached ubuf_info. To make a copy
of the skb_shared_info, we either should clear the flag and reference
the frags, or keep the flag and have frags unreferenced.
pskb_carve_inside_header() and pskb_carve_inside_nonlinear() don't
follow the rule and thus can leak page references. Let's clear
SKBFL_MANAGED_FRAG_REFS from the original skb to fix it. It's the
simplest way to address it, but there are more performant ways to do
that if it ever becomes a problem.
Jiayuan Chen [Wed, 27 May 2026 05:31:31 +0000 (13:31 +0800)]
ipv6: fix possible infinite loop in fib6_select_path()
Found while auditing the same pattern Sashiko reported in
rt6_fill_node() [1]. Apply the same fix as
commit f8d8ce1b515a ("ipv6: fix possible infinite loop in fib6_info_uses_dev()").
Writers holding tb6_lock can list_del_rcu(&first->fib6_siblings)
without waiting for RCU readers; first->fib6_siblings.next then
still points into the old ring and this softirq-side walker never
reaches &first->fib6_siblings as its terminator. fib6_purge_rt()
always WRITE_ONCE()s first->fib6_nsiblings to 0 before
list_del_rcu(), so an inside-loop check is a reliable detach signal.
Fixes: d9ccb18f83ea ("ipv6: Fix soft lockups in fib6_select_path under high next hop churn") Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Link: https://patch.msgid.link/20260527053133.180695-2-jiayuan.chen@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jiayuan Chen [Wed, 27 May 2026 05:31:30 +0000 (13:31 +0800)]
ipv6: fix possible infinite loop in rt6_fill_node()
Sashiko reported this issue [1]. Apply the same fix as
commit f8d8ce1b515a ("ipv6: fix possible infinite loop in fib6_info_uses_dev()").
Writers holding tb6_lock can list_del_rcu(&rt->fib6_siblings)
without waiting for RCU readers; rt->fib6_siblings.next then still
points into the old ring and this softirq-side walker never reaches
&rt->fib6_siblings, causing a CPU stall. fib6_del_route() always
WRITE_ONCE()s rt->fib6_nsiblings to 0 before list_del_rcu(), so an
inside-loop check is a reliable detach signal.
Fixes: d9ccb18f83ea ("ipv6: Fix soft lockups in fib6_select_path under high next hop churn") Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Link: https://patch.msgid.link/20260527053133.180695-1-jiayuan.chen@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Yuqi Xu [Wed, 27 May 2026 03:48:15 +0000 (11:48 +0800)]
bpf: sockmap: fix tail fragment offset in bpf_msg_push_data
When bpf_msg_push_data() inserts data in the middle of a scatterlist
entry, it splits the original entry into a left fragment and a right
fragment.
The right fragment offset is page-local, but the code advances it with
`start`, which is the message-global insertion point. For inserts into a
non-first SG entry, this over-advances the offset and leaves the split
layout inconsistent.
Advance the right fragment offset by the fragment-local delta,
`start - offset`, which matches the length removed from the front of the
original entry.
Fixes: 6fff607e2f14 ("bpf: sk_msg program helper bpf_msg_push_data") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Zhengchuan Liang <zcliangcn@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Yuqi Xu <xuyq21@lenovo.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Link: https://patch.msgid.link/8b129d10566aa3eb43f61a8f9757bcf51707d324.1779636774.git.xuyq21@lenovo.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jingguo Tan [Wed, 27 May 2026 02:33:01 +0000 (10:33 +0800)]
vsock/virtio: bind uarg before filling zerocopy skb
virtio_transport_send_pkt_info() allocates or reuses the zerocopy uarg
before entering the send loop, but virtio_transport_alloc_skb() still
fills the skb before it inherits that uarg. When fixed-buffer vectored
zerocopy hits MAX_SKB_FRAGS, io_sg_from_iter() may partially attach
managed frags and return -EMSGSIZE. The rollback path call kfree_skb()
to free an skb that carries SKBFL_MANAGED_FRAG_REFS but no uarg, so
skb_release_data() falls through to ordinary frag unref.
Pass the uarg into virtio_transport_alloc_skb() and bind it immediately
before virtio_transport_fill_skb(). This keeps control or no-payload skbs
untouched while ensuring success and rollback share one lifetime rule.
Fixes: 581512a6dc93 ("vsock/virtio: MSG_ZEROCOPY flag support") Signed-off-by: Lin Ma <malin89@huawei.com> Signed-off-by: Rongzhen Cui <cuirongzhen@huawei.com> Signed-off-by: Jingguo Tan <tanjingguo@huawei.com> Acked-by: Arseniy Krasnov <avkrasnov@salutedevices.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Stefano Garzarella <sgarzare@redhat.com> Link: https://patch.msgid.link/20260527023301.1075581-1-malin89@huawei.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Sven Eckelmann [Tue, 12 May 2026 20:03:53 +0000 (22:03 +0200)]
batman-adv: use atomic_xchg() for gw.reselect check
batadv_gw_election() only needs to test whether gw.reselect was set and
clear it afterwards. Replace the batadv_atomic_dec_not_zero()
[atomic_add_unless(..., -1, 0)] call with atomic_xchg(..., 0) to simplify
the logic and make the intent more explicit.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Sun, 3 May 2026 20:46:14 +0000 (22:46 +0200)]
batman-adv: add missing includes
Some of the recent fixes required features from new header files. There is
currently no build problem because transitive includes take care of it. But
the batman-adv source code tries to avoid the dependency to
transitive/implicite includes because it has no control over them and they
might get removed at some point.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Wed, 6 May 2026 06:44:32 +0000 (08:44 +0200)]
MAINTAINERS: Don't send batman-adv patches to netdev
Do not send batman-adv patches directly to the networking maintainers or
the netdev mailing list for initial review. Keeping these patch iterations
off netdev reduces review load, especially for patchsets that require
multiple revisions before reaching consensus.
After the review was finished on the BATMAN ADVANCED mailing list, the
patches will be queued up and then submitted to netdev as PR, including the
full patchset in the same thread (for the last review).
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Tue, 5 May 2026 08:23:06 +0000 (10:23 +0200)]
MAINTAINERS: Rename batman-adv T(ree)
Replace the batman-adv tree name "linux-merge" with "batadv" to match the
patch prefix convention used in subject lines (e.g. "[PATCH batadv 1/10]").
The previous name "linux-merge" was ambiguous and was not suitable as a
easy-to-recognize prefix. Routing of patches to net-next vs. net remains at
maintainer discretion.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Fri, 15 May 2026 06:41:19 +0000 (08:41 +0200)]
batman-adv: drop batman-adv specific version
Bumping the version number on the first pull request after each merge
window was deemed inappropriate for an in-tree component. The version
number carries little meaningful information in the context of the Linux
kernel release model, where stable and distribution might all carry
slightly different patches (without any change to the batman-adv version).
Instead, expose a UTS_RELEASE-based string to consumers of the netlink and
ethtool interfaces. To avoid recompilation for each (re)generate of
generated/utsrelease.h, init_utsname()->release is used in code which can
dynamically retrieve the version string. The MODULE_VERSION is moved to a
separate file because it doesn't support dynamic retrieval of the version
string (but constant "at compile time" string) and it is required for the
/sys/module/batman_adv/version. The latter is unfortunately still required
by userspace tools.
KVM: SEV: Use READ_ONCE() when reading entries/indices from PSC buffer
Use READ_ONCE() when reading entries/indices from the guest-accessible
Page State Change buffer to defend against TOCTOU bugs.
Don't bother with READ_ONCE()/WRITE_ONCE() for cases where KVM is writing
(and not consuming the result!), as the guest isn't supposed to touch the
buffer while it's being processed. I.e. using READ_ONCE() is all about
protecting against misbehaving guests.
Fixes: 9b54e248d264 ("KVM: SEV: Add support to handle Page State Change VMGEXIT") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-11-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: SEV: Check PSC request indices against the actual size of the buffer
When processing Page State Change (PSC) requests, validate the PSC buffer
against the effective size of the scratch area, which could be less than
the maximum size if the guest provided a pointer that isn't exactly at the
start of the GHCB shared buffer.
Fixes: 9b54e248d264 ("KVM: SEV: Add support to handle Page State Change VMGEXIT") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Reviewed-by: Michael Roth <michael.roth@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-10-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: SEV: Don't explicitly pass PSC buffer to snp_begin_psc()
Stop explicitly passing the PSC buffer to snp_begin_psc(): it *must*
be the scratch area. This will allow fixing a variety of bugs without
further complicating the code.
No functional change intended.
Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Reviewed-by: Michael Roth <michael.roth@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-9-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: SEV: WARN if KVM attempts to setup scratch area with min_len==0
Now that all paths in KVM properly validate the length needed for the
scratch area, and are guaranteed to pass in a non-zero length, WARN if KVM
attempts to configured the scratch area with min_len==0 to guard against
future bugs.
Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Reviewed-by: Michael Roth <michael.roth@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-8-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: SEV: Compute the correct max length of the in-GHCB scratch area
When setting the length of the GHCB scratch area, and the area is in the
GHCB shared buffer, set the effective length of the scratch area to the max
possible size given the start of the guest-provided pointer, and the end of
the shared buffer.
The code was "fine" when first introduced, as KVM doesn't consult the
length of the buffer when emulating MMIO, because the passed in @len always
specifies the *max* size required. But for PSC requests, the incoming @len
is just the minimum length (to process the header), and KVM needs to know
the full size of the scratch area to avoid buffer overflows (spoiler alert).
Opportunistically rename @len => @min_len to better reflect its role.
Fixes: 9b54e248d264 ("KVM: SEV: Add support to handle Page State Change VMGEXIT") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Reviewed-by: Michael Roth <michael.roth@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-7-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: SEV: Use the size of the PSC header as the minimum size for PSC requests
When handling a Page State Change (PSC) #VMGEXIT use the size of the PSC
header as the minimum size for the scratch area. Per the GHCB spec, PSC
requests do NOT provide the length, i.e. using control->exit_info_2 for the
length is completely made up behavior. The existing code "works", e.g.
even though Linux-as-a-guest always passes '0', because KVM doesn't do
anything with the length when the request is in the GHCB's shared buffer.
Use the header as the min length. Once the header is retrieved, KVM can
use the specified indices to compute the full size of the request.
Fixes: 9b54e248d264 ("KVM: SEV: Add support to handle Page State Change VMGEXIT") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Reviewed-by: Michael Roth <michael.roth@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-6-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Explicitly ignore Port I/O requests of length '0' (or count '0'), so that
setting up the software scratch area (and other code) doesn't have to
worry about underflowing the length, and to allow for WARNing on trying
to configure the scratch area with len==0.
Fixes: 291bd20d5d88 ("KVM: SVM: Add initial support for a VMGEXIT VMEXIT") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-5-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Explicitly ignore MMIO requests of length '0', so that setting up the
software scratch area (and other code) doesn't have to worry about
underflowing the length, and to allow for special casing '0' in the
future.
Fixes: 8f423a80d299 ("KVM: SVM: Support MMIO for an SEV-ES guest") Cc: stable@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260501202250.2115252-3-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Michael Roth [Fri, 1 May 2026 20:22:26 +0000 (13:22 -0700)]
KVM: SEV: Require in-GHCB scratch area if GHCB v2+ is in use
As per the GHCB spec, when using GHCB v2+ require the software scratch area
to reside in the GHCB's shared buffer. Note, things like Page State Change
(PSC) requests _rely_ on this behavior, as the guest can't provide a length
when making the request, i.e. the size of the guest payload is bounded by
the size of the shared buffer.
Failure to force usage of the GHCB, and a slew of other flaws, lets a
malicious SNP guest corrupt host kernel heap memory, and leak host heap
layout information.
setup_vmgexit_scratch() allocates a buffer via kvzalloc(exit_info_2),
where exit_info_2 is guest-controlled. With exit_info_2=24, this yields
a 24-byte allocation in kmalloc-cg-32 (32-byte slab objects). The buffer
holds an 8-byte psc_hdr followed by 8-byte psc_entry structs, so only
entries[0] and entries[1] are in-bounds.
snp_begin_psc() validates end_entry against VMGEXIT_PSC_MAX_COUNT (253)
but NOT against the actual buffer size:
idx_end = hdr->end_entry;
if (idx_end >= VMGEXIT_PSC_MAX_COUNT) { // checks 253, not buffer
snp_complete_psc(svm, ...);
return 1;
}
for (idx = idx_start; idx <= idx_end; idx++) {
entry_start = entries[idx]; // OOB when idx >= 2
The guest sets end_entry=10+, causing the host to iterate entries[2+]
which are OOB into adjacent slab objects. For each OOB entry:
- The host reads 8 bytes (OOB READ / info leak oracle)
- If the data passes PSC validation, __snp_complete_one_psc() writes
cur_page = 1 or 512 into the entry (OOB WRITE, sev.c:3806)
- If validation fails, the error response reveals whether adjacent
memory is zero vs non-zero (information disclosure to guest)
The guest controls allocation size (exit_info_2), entry range
(cur_entry/end_entry), and can fire unlimited VMGEXITs to repeatedly
hit different slab positions.
By exploiting the variety of bugs, a malicious SEV-SNP guest can:
- OOB read adjacent kmalloc-cg-32 objects (heap layout disclosure)
- OOB write cur_page bits into adjacent objects (heap corruption)
- Trigger use-after-free conditions across VMGEXITs
E.g. with KASAN enabled, a single insmod of the PoC guest module
produces 73 KASAN reports:
BUG: KASAN: slab-out-of-bounds in snp_begin_psc+0x126/0x890
Read of size 8 at addr ffff888219ffb5e0 by task qemu-system-x86/2199
BUG: KASAN: slab-out-of-bounds in snp_begin_psc+0x468/0x890
Write of size 8 at addr ffff888351566648 by task qemu-system-x86/2199
The buggy address belongs to the object at ffff888XXXXXXXXX
which belongs to the cache kmalloc-cg-32 of size 32
The buggy address is located N bytes to the right of
allocated 32-byte region [ffff888XXXXXXXXX, ffff888XXXXXXXXX)
Guopeng Zhang [Thu, 28 May 2026 09:37:42 +0000 (17:37 +0800)]
cgroup/cpuset: Free sched domains on rebuild guard failure
generate_sched_domains() returns sched-domain masks and optional
attributes that are normally handed to partition_sched_domains(), which
takes ownership of them.
rebuild_sched_domains_locked() has a WARN guard after
generate_sched_domains() and before partition_sched_domains() to avoid
passing offline CPUs into the scheduler domain rebuild path. If that
guard fires, the function currently returns directly without freeing
the generated doms and attr.
Free the generated sched-domain masks and attributes before returning
from the guard failure path.
Marco Crivellari [Fri, 29 May 2026 13:06:39 +0000 (15:06 +0200)]
workqueue: Add warnings and fallback if system_{unbound}_wq is used
Currently many users transitioned already to the new introduced workqueue
(system_percpu_wq, system_dfl_wq), but there are new users who still use the
older system_wq and system_unbound_wq.
This change try to push this transition forward, by warning whether the old
workqueues are used.
Linus Torvalds [Fri, 29 May 2026 17:36:57 +0000 (10:36 -0700)]
Merge tag 'io_uring-7.1-20260529' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux
Pull io_uring fix from Jens Axboe:
"Just a single fix for a regression introduced in this cycle, where
we should ensure the node is visible before the entry is added to
the tctx list"
* tag 'io_uring-7.1-20260529' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux:
io_uring/tctx: set ->io_uring before publishing the tctx node
Paolo Bonzini [Fri, 29 May 2026 17:28:16 +0000 (19:28 +0200)]
Merge tag 'kvm-x86-fixes-7.1-rc6' of https://github.com/kvm-x86/linux into HEAD
KVM x86 fixes for 7.1-rcN
- Include the kernel's linux/mman.h in KVM selftests to ensure MADV_COLLAPSE
is defined, as older libc versions may not provide it.
- Include execinfo.h if and only if KVM selftests are building against glibc,
and provide a test_dump_stack() for non-glibc builds.
- Fudge around an RCU splat in the emegerncy reboot code that is technically
a legitimate flaw, but in practice is a non-issue and fixing the flaw, e.g.
by adding locking, would incur meaningful risk, i.e. do more harm than good.
- Rate-limit global clock updates once again (but without delayed work), as
KVM was subtly relying on the old rate-limiting for NPT correction to guard
against "update storms" when running without a master clock on systems with
overcommitted CPUs.
- Fix a brown paper bag goof where KVM checked if ERAPS is "dirty" instead of
marking it dirty when emulating INVPCID.
- Flush the TLB when transitioning from xAVIC => x2AVIC to ensure the CPU TLB
doesn't contain AVIC-tagged entries for the APIC base GPA.
Linus Torvalds [Fri, 29 May 2026 17:04:09 +0000 (10:04 -0700)]
Merge tag 'cxl-fixes-7.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl
Pull Compute Express Link (CXL) fixes from Dave Jiang:
- cxl/test: update mock dev array before calling platform_device_add()
* tag 'cxl-fixes-7.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl:
cxl/test: Update mock dev array before calling platform_device_add()
Marek Vasut [Sat, 11 Apr 2026 13:02:35 +0000 (15:02 +0200)]
arm64: dts: st: Fix SAI addresses on stm32mp251
The second field of SAI register addresses should be within 0x3f0 bytes
from the start of the SAI register addresses, the second field describes
the ID registers which are at that addrses. Currently, the second field
does not match RM, fix it.
Fixes: bf26d75a95f1 ("arm64: dts: st: add sai support on stm32mp251") Signed-off-by: Marek Vasut <marex@nabladev.com> Reviewed-by: Olivier Moysan <olivier.moysan@foss.st.com> Link: https://lore.kernel.org/r/20260411130300.19603-1-marex@nabladev.com Signed-off-by: Alexandre Torgue <alexandre.torgue@foss.st.com>
* tag 'iommu-fixes-v7.1-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/iommu/linux:
MAINTAINERS: Add my employer to my entries
MAINTAINERS: Add Vasant Hegde to reviewers of AMD IOMMU
iommu, debugobjects: avoid gcc-16.1 section mismatch warnings
iommu/vt-d: Simplify calculate_psi_aligned_address()
Linus Torvalds [Fri, 29 May 2026 15:55:41 +0000 (08:55 -0700)]
Merge tag 'sound-7.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound fixes from Takashi Iwai:
"A collection of recent small fixes and quirks.
We still see a bit more changes than wished, but most of them are
device-specific ones that are pretty safe to apply, while a core fix
is a typical UAF fix for PCM core that was recently caught by fuzzer;
so overall nothing looks really worrisome.
Core:
- Fix a UAF in PCM OSS proc interface
HD-audio:
- Fix memory leaks in CS35L56 driver
- Various device-specific quirks for Realtek and CS420x codecs
USB-audio:
- Quirk for TAE1160 USB Audio
- Fix for Scarlett2 Gen4 direct monitor gain
ASoC:
- Fixes for QCom q6asm-dai, Intel bytcht_es8316, and simple-mux codec
FireWire:
- Fix for Motu DSP event queue protection"
* tag 'sound-7.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ASoC: codecs: simple-mux: Fix enum control bounds check
ALSA: usb-audio: Add iface reset and delay quirk for TAE1160 USB Audio
ALSA: hda/cs420x: Add CS4208 fixup for iMac16,1
ALSA: hda/realtek: add quirk for HP Dragonfly Folio G3 2-in-1
ALSA: hda/realtek: Fix speaker output on ASUS ROG Strix G615LP
ASoC: qcom: q6asm-dai: use pointer type with kzalloc_obj()
ASoC: qcom: q6asm-dai: remove unnecessary braces
ASoC: qcom: q6asm-dai: fix error handling in prepare and set_params
ASoC: qcom: q6asm-dai: close stream only when running
ASoC: qcom: q6asm-dai: do not set stream state in event and trigger callbacks
ASoC: Intel: bytcht_es8316: Fix MCLK leak on init errors
ALSA: hda/realtek: Limit mic boost on Positivo DN140
ALSA: scarlett2: Fix 2i2 Gen 4 direct monitor gain on firmware 2417
ALSA: pcm: oss: Fix setup list UAF on proc write error
ALSA: hda: cs35l56: Fix system name string leaks
ALSA: hda/realtek: Add HDA_CODEC_QUIRK for Lenovo Yoga Slim 7 14AGP11
ALSA: hda/realtek: Fix incorrect comment for ALC299_FIXUP_PREDATOR_SPK
ALSA: firewire-motu: Protect register DSP event queue positions
Qiuxu Zhuo [Thu, 21 May 2026 12:38:12 +0000 (20:38 +0800)]
EDAC/igen6: Add Intel Nova Lake-H SoC support
Nova Lake-H SoCs share similar memory controller registers and IBECC
(In-Band ECC) registers with Panther Lake-H SoCs but use a new memory
subsystem register for IBECC presence detection.
Add Nova Lake-H SoC compute die IDs and create a new configuration
structure for Nova Lake-H SoCs to enable EDAC support.
Qiuxu Zhuo [Thu, 21 May 2026 12:38:11 +0000 (20:38 +0800)]
EDAC/igen6: Make registers for detecting IBECC configurable
Some Intel CPUs with IBECC (In-Band ECC) capability use different registers
to indicate IBECC presence. Make IBECC detection registers CPU-model
specific and configure them properly for scalable IBECC detection.
Qiuxu Zhuo [Thu, 21 May 2026 07:31:12 +0000 (15:31 +0800)]
EDAC/imh: Add RRL support for Intel Diamond Rapids server
Compared to previous generations, Diamond Rapids RRL (Retry Read error Log)
operates at DDR sub-channel granularity and adds an extra register per set.
It also increases the CORRERRCNT register width from 4 to 8 bytes while
reducing the number of registers from 8 to 4.
Add the Diamond Rapids RRL register configuration table and enable support.
Qiuxu Zhuo [Thu, 21 May 2026 07:31:11 +0000 (15:31 +0800)]
EDAC/{skx_common,i10nm}: Prepare RRL for sub-channel granularity
To prepare for enabling Diamond Rapids server RRL (Retry Read error Log),
which operates at sub-channel granularity by converting struct
res_config::reg_rrl_ddr from a single pointer to an array (reg_rrl_ddr[2])
and updating all users in i10nm_edac and skx_common accordingly.
Initialize only reg_rrl_ddr[0] for existing platforms and prepare for
supporting two RRL set groups per DDR channel (one per sub-channel)
when present.
Qiuxu Zhuo [Thu, 21 May 2026 07:31:10 +0000 (15:31 +0800)]
EDAC/skx_common: Add SubChannel support to ADXL decode
Diamond Rapids server RRL (Retry Read error Log) operates at sub-channel
granularity. Add SubChannel support to ADXL decoding in preparation for
enabling this feature.
Also introduce adxl_component_required() to validate mandatory ADXL
components to improve code readability.
Qiuxu Zhuo [Thu, 21 May 2026 07:31:08 +0000 (15:31 +0800)]
EDAC/{skx_common,i10nm}: Introduce rrl_ctrl_mode
RRL (Retry Read error Log) ownership is currently inferred from
retry_rd_err_log magic values, making control semantics implicit
and harder to understand.
Introduce rrl_ctrl_mode to explicitly describe whether RRL is
controlled by none, BIOS, or Linux, and replace direct checks with
named control states to improve readability and maintainability.
projects / thirdparty / linux.git / log
