Save the shadow stack pointer in the sigcontext structure when
delivering a signal. Restore the shadow stack pointer from sigcontext
on sigreturn.
As part of the save operation, the kernel uses the 'ssamoswap'
instruction to save a snapshot of the current shadow stack on the
shadow stack itself (this can be called a "save token"). During
restore on sigreturn, the kernel retrieves the save token from the top
of the shadow stack and validates it. This ensures that user mode
can't arbitrarily pivot to any shadow stack address without having a
token and thus provides a strong security assurance during the window
between signal delivery and sigreturn.
Use an ABI-compatible way of saving/restoring the shadow stack pointer
into the signal stack. This follows the vector extension, where extra
registers are placed in a form of extension header + extension body in
the stack. The extension header indicates the size of the extra
architectural states plus the size of header itself, and a magic
identifier for the extension. Then, the extension body contains the
new architectural states in the form defined by uapi.
Signed-off-by: Andy Chiu <andy.chiu@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de>
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-17-b55691eacf4f@rivosinc.com
[pjw@kernel.org: cleaned patch description, code comments; resolved checkpatch warning]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
The Zicfiss and Zicfilp extensions introduce a new exception, the
'software check exception', in the privileged ISA, with cause code =
18. This patch implements support for software check exceptions.
Additionally, the patch implements a CFI violation handler which
checks the code in the xtval register. If xtval=2, the software check
exception happened because of an indirect branch that didn't land on a
4 byte aligned PC or on a 'lpad' instruction, or the label value
embedded in 'lpad' didn't match the label value set in the x7
register. If xtval=3, the software check exception happened due to a
mismatch between the link register (x1 or x5) and the top of shadow
stack (on execution of `sspopchk`).
In case of a CFI violation, SIGSEGV is raised with code=SEGV_CPERR.
SEGV_CPERR was introduced by the x86 shadow stack patches.
To keep uprobes working, handle the uprobe event first before
reporting the CFI violation in the software check exception
handler. This is because, when the landing pad is activated, if the
uprobe point is set at the lpad instruction at the beginning of a
function, the system triggers a software check exception instead of an
ebreak exception due to the exception priority. This would prevent
uprobe from working.
Reviewed-by: Zong Li <zong.li@sifive.com>
Co-developed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de> # QEMU, custom CVA6
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-15-b55691eacf4f@rivosinc.com
[pjw@kernel.org: cleaned up the patch description]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
This patch adds a RISC-V implementation of the following prctls:
PR_SET_INDIR_BR_LP_STATUS, PR_GET_INDIR_BR_LP_STATUS and
PR_LOCK_INDIR_BR_LP_STATUS.
Reviewed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de>
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-14-b55691eacf4f@rivosinc.com
[pjw@kernel.org: clean up patch description]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
Three architectures (x86, aarch64, riscv) have support for indirect
branch tracking feature in a very similar fashion. On a very high
level, indirect branch tracking is a CPU feature where CPU tracks
branches which use a memory operand to transfer control. As part of
this tracking, during an indirect branch, the CPU expects a landing
pad instruction on the target PC, and if not found, the CPU raises
some fault (architecture-dependent).
x86 landing pad instr - 'ENDBRANCH'
arch64 landing pad instr - 'BTI'
riscv landing instr - 'lpad'
Given that three major architectures have support for indirect branch
tracking, this patch creates architecture-agnostic 'prctls' to allow
userspace to control this feature. They are:
- PR_GET_INDIR_BR_LP_STATUS: Get the current configured status for indirect
branch tracking.
- PR_SET_INDIR_BR_LP_STATUS: Set the configuration for indirect branch
tracking.
The following status options are allowed:
- PR_INDIR_BR_LP_ENABLE: Enables indirect branch tracking on user
thread.
- PR_INDIR_BR_LP_DISABLE: Disables indirect branch tracking on user
thread.
- PR_LOCK_INDIR_BR_LP_STATUS: Locks configured status for indirect branch
tracking for user thread.
Reviewed-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de> # QEMU, custom CVA6
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-13-b55691eacf4f@rivosinc.com
[pjw@kernel.org: cleaned up patch description, code comments]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
Implement an architecture-agnostic prctl() interface for setting and
getting shadow stack status. The prctls implemented are
PR_GET_SHADOW_STACK_STATUS, PR_SET_SHADOW_STACK_STATUS and
PR_LOCK_SHADOW_STACK_STATUS.
As part of PR_SET_SHADOW_STACK_STATUS/PR_GET_SHADOW_STACK_STATUS, only
PR_SHADOW_STACK_ENABLE is implemented because RISCV allows each mode to
write to their own shadow stack using 'sspush' or 'ssamoswap'.
PR_LOCK_SHADOW_STACK_STATUS locks the current shadow stack enablement
configuration.
Reviewed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de> # QEMU, custom CVA6
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-12-b55691eacf4f@rivosinc.com
[pjw@kernel.org: cleaned up patch description]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
Userspace specifies CLONE_VM to share address space and spawn new
thread. 'clone' allows userspace to specify a new stack for a new
thread. However there is no way to specify a new shadow stack base
address without changing the API. This patch allocates a new shadow
stack whenever CLONE_VM is given.
In case of CLONE_VFORK, the parent is suspended until the child
finishes; thus the child can use the parent's shadow stack. In case of
!CLONE_VM, COW kicks in because entire address space is copied from
parent to child.
'clone3' is extensible and can provide mechanisms for specifying the
shadow stack as an input parameter. This is not settled yet and is
being extensively discussed on the mailing list. Once that's settled,
this code should be adapted.
Reviewed-by: Zong Li <zong.li@sifive.com>
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Tested-by: Andreas Korb <andreas.korb@aisec.fraunhofer.de> # QEMU, custom CVA6
Tested-by: Valentin Haudiquet <valentin.haudiquet@canonical.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-11-b55691eacf4f@rivosinc.com
[pjw@kernel.org: cleaned up patch description]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
There are two unused percpu critical sections in the buffer management
code. These are remanents from when a more complex hold algorithm was
used. Remove them, as they serve no purpose.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The buffer hold is a measure of contention, but it is tracked per cpu
where the lock is a globabl resource. On some systems (eg. real time)
there is no guarantee that the code will be on the same cpu pre, and
post spinlock acquisition, nor that the buffer will be put back to
the same percpu cache when we are done with it.
Because of this the hold value can move asynchronous to the buffers on
the cache, meaning it is possible to underflow, and potentially in really
pathelogical cases overflow.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
When aa_get_buffer() pulls from the per-cpu list it unconditionally
decrements cache->hold. If hold reaches 0 while count is still non-zero,
the unsigned decrement wraps to UINT_MAX. This keeps hold non-zero for a
very long time, so aa_put_buffer() never returns buffers to the global
list, which can starve other CPUs and force repeated kmalloc(aa_g_path_max)
allocations.
Guard the decrement so hold never underflows.
Fixes: ea9bae12d0 ("apparmor: cache buffers on percpu list if there is lock contention")
Signed-off-by: Zhengmian Hu <huzhengmian@gmail.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
This patch doesn't change current functionality, it switches the two
uses of the in_ns fns and macros into the two semantically different
cases they are used for.
xxx_in_scope for checking mediation interaction between profiles
xxx_in_view to determine which profiles are visible.The scope will
always be a subset of the view as profiles that can not see each
other can not interact.
The split can not be completely done for label_match because it has to
distinct uses matching permission against label in scope, and checking
if a transition to a profile is allowed. The transition to a profile
can include profiles that are in view but not in scope, so retain this
distinction as a parameter.
While at the moment the two uses are very similar, in the future there
will be additional differences. So make sure the semantics differences
are present in the code.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
compound match is inconsistent in returning a state or an integer error
this is problemati if the error is ever used as a state in the state
machine
Fixes: f1bd904175 ("apparmor: add the base fns() for domain labels")
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The modes shouldn't be applied at the point of label match, it just
results in them being applied multiple times. Instead they should be
applied after which is already being done by all callers so it can
just be dropped from label_match.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The fast path cache check is incorrect forcing more slow path
revalidations than necessary, because the unix logic check is inverted.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Posix cpu timers requires an additional step beyond setting the rlimit.
Refactor the code so its clear when what code is setting the
limit and conditionally update the posix cpu timers when appropriate.
Fixes: baa73d9e47 ("posix-timers: Make them configurable")
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
aa_cred_raw_label() and cred_label() now do the same things so
consolidate to cred_label()
Document the crit section use and constraints better and refactor
__begin_current_label_crit_section() into a base fn
__begin_cred_crit_section() and a wrapper that calls the base with
current cred.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The label and profile debug macros were not correctly pasting their
var args.
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
files with a dentry pointing aa_null.dentry where already rejected as
part of file_inheritance. Unfortunately the check in
common_file_perm() is insufficient to cover all cases causing
unnecessary audit messages without the original files context.
Eg.
[ 442.886474] audit: type=1400 audit(1704822661.616:329): apparmor="DENIED" operation="file_inherit" class="file" namespace="root//lxd-juju-98527a-0_<var-snap-lxd-common-lxd>" profile="snap.lxd.activate" name="/apparmor/.null" pid=9525 comm="snap-exec"
Further examples of this are in the logs of
https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/2120439https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1952084https://bugs.launchpad.net/snapd/+bug/2049099
These messages have no value and should not be sent to the logs.
AppArmor was already filtering the out in some cases but the original
patch did not catch all cases. Fix this by push the existing check
down into two functions that should cover all cases.
Link: https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/2122743
Fixes: 192ca6b55a ("apparmor: revalidate files during exec")
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
aa_free_data() and free_attachment() don't guard against having
a NULL parameter passed to them. Fix this.
Reviewed-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
In policy_unpack.c:unpack_perms_table, the perms struct is allocated via
kcalloc, with the position being reset if the allocation fails. However,
the error path results in -EPROTO being retured instead of -ENOMEM. Fix
this to return the correct error code.
Reported-by: Zygmunt Krynicki <zygmunt.krynicki@canonical.com>
Fixes: fd1b2b95a2 ("apparmor: add the ability for policy to specify a permission table")
Reviewed-by: Tyler Hicks <code@tyhicks.com>
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
If we are not in an atomic context in common_file_perm, then we don't have
to use the atomic versions, resulting in improved performance outside of
atomic contexts.
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
with the previous changes to mmap the in_atomic flag is now always
false, so drop it.
Suggested-by: Tyler Hicks <code@tyhicks.com>
Reviewed-by: Georgia Garcia <georgia.garcia@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
The previous value of GFP_ATOMIC is an int and not a bool, potentially
resulting in UB when being assigned to a bool. In addition, the mmap hook
is called outside of locks (i.e. in a non-atomic context), so we can pass
a fixed constant value of false instead to common_mmap.
Signed-off-by: Ryan Lee <ryan.lee@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
This new field allows reliable identification of the binary that
triggered a denial since the existing field (comm) only gives the name of
the binary, not its path. Thus comm doesn't work for binaries outside of
$PATH or works unreliably when two binaries have the same name.
Additionally comm can be modified by a program, for example, comm="(tor)"
or comm=4143504920506F6C6C6572 (= ACPI Poller).
Signed-off-by: Maxime Bélair <maxime.belair@canonical.com>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Add support for the per permission tag index for a given permission
set. This will be used by both meta-data tagging, to allow annotating
accept states with context and debug information. As well as by rule
tainting and triggers to specify the taint or trigger to be applied.
Since these are low frequency ancillary data items they are stored
in a tighter packed format to that allows for sharing and reuse of the
strings between permissions and accept states. Reducing the amount of
kernel memory use at the cost of having to go through a couple if
index based indirections.
The tags are just strings that has no meaning with out context. When
used as meta-data for auditing and debugging its entirely information
for userspace, but triggers, and tainting can be used to affect the
domain. However they all exist in the same packed data set and can
be shared between different uses.
Signed-off-by: John Johansen <john.johansen@canonical.com>
* Add a compatible to the reset-gpio driver, suppress the sysfs bind
attributes, and propagate GPIO API errors.
* Add support for the i.MX8ULP SIM LPAV reset controller.
* Add RZ/G3S USBPHY suspend/resume support.
* Enable reset-k230 by default on ARCH_CANAAN
* Add support for the SpacemiT K3 SoC reset controller.
* Merge the 'spacemit-clkrst-v6.20-3' tag, shared with the clk tree,
as a dependency for the SpacemiT changes.
-----BEGIN PGP SIGNATURE-----
iI0EABYKADUWIQRRO6F6WdpH1R0vGibVhaclGDdiwAUCaXiRShcccC56YWJlbEBw
ZW5ndXRyb25peC5kZQAKCRDVhaclGDdiwDWkAP0foLlNQzqjNj7uT6ysYoHwbixE
+F9iaLCqJ9SAS5ajcwEAvJdJOJJCHGLR/4ImjH4AdCC1RqENUPYTqivvjZ3ozAU=
=Z0D4
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=P4vq
-----END PGP SIGNATURE-----
Merge tag 'reset-for-v6.20' of https://git.pengutronix.de/git/pza/linux into soc/drivers
Reset controller updates for v6.20
* Add a compatible to the reset-gpio driver, suppress the sysfs bind
attributes, and propagate GPIO API errors.
* Add support for the i.MX8ULP SIM LPAV reset controller.
* Add RZ/G3S USBPHY suspend/resume support.
* Enable reset-k230 by default on ARCH_CANAAN
* Add support for the SpacemiT K3 SoC reset controller.
* Merge the 'spacemit-clkrst-v6.20-3' tag, shared with the clk tree,
as a dependency for the SpacemiT changes.
* tag 'reset-for-v6.20' of https://git.pengutronix.de/git/pza/linux:
reset: spacemit: Add SpacemiT K3 reset driver
reset: spacemit: Extract common K1 reset code
reset: Create subdirectory for SpacemiT drivers
dt-bindings: soc: spacemit: Add K3 reset support and IDs
reset: canaan: k230: drop OF dependency and enable by default
reset: rzg2l-usbphy-ctrl: Add suspend/resume support
reset: rzg2l-usbphy-ctrl: Propagate the return value of regmap_field_update_bits()
reset: gpio: check the return value of gpiod_set_value_cansleep()
reset: imx8mp-audiomix: Support i.MX8ULP SIM LPAV
reset: imx8mp-audiomix: Extend the driver usage
reset: imx8mp-audiomix: Switch to using regmap API
reset: imx8mp-audiomix: Drop unneeded macros
reset: gpio: suppress bind attributes in sysfs
clk: spacemit: k3: extract common header
reset: spacemit: fix auxiliary device id
clk: spacemit: prepare common ccu header
reset: gpio: add the "compatible" property
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Currently, dma_alloc_from_pool() unconditionally warns and dumps a stack
trace when an allocation fails, with the message "Failed to get suitable
pool".
This conflates two distinct failure modes:
1. Configuration error: No atomic pool is available for the requested
DMA mask (a fundamental system setup issue)
2. Resource Exhaustion: A suitable pool exists but is currently full (a
recoverable runtime state)
This lack of distinction prevents drivers from using __GFP_NOWARN to
suppress error messages during temporary pressure spikes, such as when
awaiting synchronous reclaim of descriptors.
Refactor the error handling to distinguish these cases:
- If no suitable pool is found, keep the unconditional WARN regarding
the missing pool.
- If a pool was found but is exhausted, respect __GFP_NOWARN and update
the warning message to explicitly state "DMA pool exhausted".
Fixes: 9420139f51 ("dma-pool: fix coherent pool allocations for IOMMU mappings")
Signed-off-by: Sai Sree Kartheek Adivi <s-adivi@ti.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Link: https://lore.kernel.org/r/20260128133554.3056582-1-s-adivi@ti.com
As discussed extensively in the changelog for the addition of this
syscall on x86 ("x86/shstk: Introduce map_shadow_stack syscall") the
existing mmap() and madvise() syscalls do not map entirely well onto the
security requirements for shadow stack memory since they lead to windows
where memory is allocated but not yet protected or stacks which are not
properly and safely initialised. Instead a new syscall map_shadow_stack()
has been defined which allocates and initialises a shadow stack page.
This patch implements this syscall for riscv. riscv doesn't require
tokens to be setup by kernel because user mode can do that by
itself. However to provide compatibility and portability with other
architectues, user mode can specify token set flag.
Signed-off-by: Deepak Gupta <debug@rivosinc.com>
Link: https://patch.msgid.link/20251112-v5_user_cfi_series-v23-10-b55691eacf4f@rivosinc.com
Link: https://lore.kernel.org/linux-riscv/aXfRPJvoSsOW8AwM@debug.ba.rivosinc.com/
[pjw@kernel.org: added allocate_shadow_stack() fix per Deepak; fixed bug found by sparse]
Signed-off-by: Paul Walmsley <pjw@kernel.org>
Kernel test robot has reported a regression in the patch "slab: refill
sheaves from all nodes". When taken in isolation like this, there is
indeed a tradeoff - we prefer to use remote objects prior to allocating
new local slabs. It is replicating a behavior that existed before
sheaves for replenishing cpu (partial) slabs - now called
get_from_any_partial() to allocate a single object.
So the possibility of allocating remote objects is intended even if
remote accesses are then slower. But the profiles in the report also
suggested a contention on the list_lock spinlock. And that's something
we can try to avoid without much tradeoff - if someone else has the
spin_lock, it's more likely they are allocating from the node than
freeing to it, so we can skip it even if it means allocating a new local
slab - contributing to that lock's contention isn't worth it. It should
not result in partial slabs accumulating on the remote node.
Thus add an allow_spin parameter to __refill_objects_node() and
get_partial_node_bulk() to make the attempts from __refill_objects_any()
use only a trylock.
Reported-by: kernel test robot <oliver.sang@intel.com>
Link: https://lore.kernel.org/oe-lkp/202601132136.77efd6d7-lkp@intel.com
Link: https://patch.msgid.link/20260129-b4-refill_any_trylock-v1-1-de7420b25840@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
This adds:
- A socinfo entry for the MT8371 Genio 520 SoC
- Support for the Dynamic Voltage and Frequency Scaling
Resource Controller (DVFSRC) version 4, found in the
new MediaTek Kompanio Ultra (MT8196) SoC
- Initial support for the CMDQ mailbox found in the MT8196.
- A memory leak fix in the MediaTek SVS driver's debug ops.
-----BEGIN PGP SIGNATURE-----
iLoEABYKAGIWIQQn3Xxr56ypAcSHzXSaNgTPrZeEeAUCaXNXhhsUgAAAAAAEAA5t
YW51MiwyLjUrMS4xMSwyLDIoHGFuZ2Vsb2dpb2FjY2hpbm8uZGVscmVnbm9AY29s
bGFib3JhLmNvbQAKCRCaNgTPrZeEeF2oAQDcnPsR1I6J92rmK75Dufy4ayA79Ko0
j2cYEOzAoO+mzgD/SDq0VGMpPhyatUv5Lj8md69gARhNlylJVDIpC0jBQAQ=
=v4Rp
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=i7PU
-----END PGP SIGNATURE-----
Merge tag 'mtk-soc-for-v6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/mediatek/linux into soc/drivers
MediaTek soc driver updates
This adds:
- A socinfo entry for the MT8371 Genio 520 SoC
- Support for the Dynamic Voltage and Frequency Scaling
Resource Controller (DVFSRC) version 4, found in the
new MediaTek Kompanio Ultra (MT8196) SoC
- Initial support for the CMDQ mailbox found in the MT8196.
- A memory leak fix in the MediaTek SVS driver's debug ops.
* tag 'mtk-soc-for-v6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/mediatek/linux:
soc: mediatek: mtk-cmdq: Add mminfra_offset adjustment for DRAM addresses
soc: mediatek: mtk-cmdq: Extend cmdq_pkt_write API for SoCs without subsys ID
soc: mediatek: mtk-cmdq: Add pa_base parsing for hardware without subsys ID support
soc: mediatek: mtk-cmdq: Add cmdq_get_mbox_priv() in cmdq_pkt_create()
mailbox: mtk-cmdq: Add driver data to support for MT8196
mailbox: mtk-cmdq: Add mminfra_offset configuration for DRAM transaction
mailbox: mtk-cmdq: Add GCE hardware virtualization configuration
mailbox: mtk-cmdq: Add cmdq private data to cmdq_pkt for generating instruction
soc: mediatek: mtk-dvfsrc: Rework bandwidth calculations
soc: mediatek: mtk-dvfsrc: Get and Enable DVFSRC clock
soc: mediatek: mtk-dvfsrc: Add support for DVFSRCv4 and MT8196
soc: mediatek: mtk-dvfsrc: Write bandwidth to EMI DDR if present
soc: mediatek: mtk-dvfsrc: Add a new callback for calc_dram_bw
soc: mediatek: mtk-dvfsrc: Add and propagate DVFSRC bandwidth type
soc: mediatek: mtk-dvfsrc: Change error check for DVFSRCv4 START cmd
dt-bindings: soc: mediatek: dvfsrc: Document clock
soc: mediatek: mtk-socinfo: Add entry for MT8371AV/AZA Genio 520
soc: mediatek: svs: Fix memory leak in svs_enable_debug_write()
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Amir Goldstein <amir73il@gmail.com> says:
Following the syzbot ovl bug report and a fix by Qing Wang,
I decided to follow up with a small vfs cleanup of some
open coded version of checking "." and ".." name in readdir.
The fix patch is applied at the start of this cleanup series to allow
for easy backporting, but it is not an urgent fix so I don't think
there is a need to fast track it.
* patches from https://patch.msgid.link/20260128132406.23768-1-amir73il@gmail.com:
ovl: use name_is_dot* helpers in readdir code
fs: add helpers name_is_dot{,dot,_dotdot}
ovl: Fix uninit-value in ovl_fill_real
Link: https://patch.msgid.link/20260128132406.23768-1-amir73il@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Use the helpers in place of all the different open coded variants.
This makes the code more readable and robust.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://patch.msgid.link/20260128132406.23768-4-amir73il@gmail.com
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
- Add a poweroff function to the RTKit library which will be required
for the first USB4/Thunderbolt series I hope to submit next cycle.
-----BEGIN PGP SIGNATURE-----
iHUEABYKAB0WIQS3vz815OHsEaWy0u9EEX0kKnUe6QUCaXaPiQAKCRBEEX0kKnUe
6WxsAP4yy1tXANQgHcOP34gsbvp4LIpQw+OORSMxJ1rH9SerdgD+PlvJKmBXK185
gzxJInYEF+E5vL7DJ0eZWDHDNHrP+wU=
=iswd
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=PFGj
-----END PGP SIGNATURE-----
Merge tag 'apple-soc-drivers-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/sven/linux into soc/drivers
Apple SoC driver updates for 6.20
- Add a poweroff function to the RTKit library which will be required
for the first USB4/Thunderbolt series I hope to submit next cycle.
* tag 'apple-soc-drivers-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/sven/linux:
soc: apple: rtkit: Add function to poweroff
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Rename the helper is_dot_dotdot() into the name_ namespace
and add complementary helpers to check for dot and dotdot
names individually.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://patch.msgid.link/20260128132406.23768-3-amir73il@gmail.com
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Syzbot reported a KMSAN uninit-value issue in ovl_fill_real.
This iusse's call chain is:
__do_sys_getdents64()
-> iterate_dir()
...
-> ext4_readdir()
-> fscrypt_fname_alloc_buffer() // alloc
-> fscrypt_fname_disk_to_usr // write without tail '\0'
-> dir_emit()
-> ovl_fill_real() // read by strcmp()
The string is used to store the decrypted directory entry name for an
encrypted inode. As shown in the call chain, fscrypt_fname_disk_to_usr()
write it without null-terminate. However, ovl_fill_real() uses strcmp() to
compare the name against "..", which assumes a null-terminated string and
may trigger a KMSAN uninit-value warning when the buffer tail contains
uninit data.
Reported-by: syzbot+d130f98b2c265fae5297@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=d130f98b2c265fae5297
Fixes: 4edb83bb10 ("ovl: constant d_ino for non-merge dirs")
Signed-off-by: Qing Wang <wangqing7171@gmail.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://patch.msgid.link/20260128132406.23768-2-amir73il@gmail.com
Acked-by: Miklos Szeredi <mszeredi@redhat.com>
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
This update includes:
- drop the use of of_platform_default_populate() from the
machine specific code, as it is already handled by the DT core
-----BEGIN PGP SIGNATURE-----
iHUEABYKAB0WIQTsZ8eserC1pmhwqDmejrg/N2X7/QUCaXSQZwAKCRCejrg/N2X7
/TjtAPsHUI7hPm8xBuoK0edLsbGCUIFIsIqQOJF+cmSMEVPR0QD9E5Y20x+C4yfL
vouBeu20N2MJj8iCU8RR/c9fqWyqZwQ=
=DFnK
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=LuV6
-----END PGP SIGNATURE-----
Merge tag 'at91-soc-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/at91/linux into soc/arm
Microchip AT91 SoC updates for v6.20
This update includes:
- drop the use of of_platform_default_populate() from the
machine specific code, as it is already handled by the DT core
* tag 'at91-soc-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/at91/linux:
ARM: at91: remove unnecessary of_platform_default_populate calls
ARM: at91: Move PM init functions to .init_late hook
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
The changes to the logic in the Qualcomm SMEM driver for separating
"failed to probe" from "not yet probed", did not change the
qcom_smem_is_available() function, with the result that clients sees
SMEM as always available. Clients might then proceed to interact with
SMEM in codepaths that aren't suited to cope with -EPROBE_DEFER.
-----BEGIN PGP SIGNATURE-----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=cI29
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=Ynp6
-----END PGP SIGNATURE-----
Merge tag 'qcom-drivers-fixes-for-6.19' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux into arm/fixes
Qualcomm driver fix for v6.19
The changes to the logic in the Qualcomm SMEM driver for separating
"failed to probe" from "not yet probed", did not change the
qcom_smem_is_available() function, with the result that clients sees
SMEM as always available. Clients might then proceed to interact with
SMEM in codepaths that aren't suited to cope with -EPROBE_DEFER.
* tag 'qcom-drivers-fixes-for-6.19' of https://git.kernel.org/pub/scm/linux/kernel/git/qcom/linux:
soc: qcom: smem: fix qcom_smem_is_available and check if __smem is valid
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Add myself as maintainer of HiSilicon SoC HHA driver as I will be
responsible for this driver.
Signed-off-by: Yushan Wang <wangyushan12@huawei.com>
Acked-by: Jonathan Cameron <jonathan.cameron@huawei.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20260128-audition-professed-31d1b66d5b4e@spud
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
- Shawn Guo steps down and Frank Li steps up as i.MX platform
maintainer
-----BEGIN PGP SIGNATURE-----
iQEzBAABCgAdFiEEFmJXigPl4LoGSz08UFdYWoewfM4FAml2w5gACgkQUFdYWoew
fM71qwgApSyKWYpBI+v067vSpOb4uUEBHK2b/pmW5cKmvz0qdDWYDhVUMwCMQxSX
5ZOYsFgNHyjQuOzFXgIhmCITo+b6lKAvaQqQ2H1Rz8XvIoijD5ooQnRo0vERQbzt
O+rE/zYGaTOEnZNIAqLtwepmBHJ8JNkIO9Le/J2YSYS8+jHdzJ1Cm/Ou6FIFhPaX
k9Gr9d5trBo6bNUWqHnzRkC8w/9KdCqMihu0ueSgNPxeRYDZF2KUrXyByuVPTj5T
hkfnbwYmqK/bQdjfKZ7XXKTgosdc6UgWhEunPJEr7smWKbqAYgU09vns3RICENNd
bxVSnxzx8sYAAcoeiNnicldQNZdocw==
=sc5O
-----END PGP SIGNATURE-----
gpgsig -----BEGIN PGP SIGNATURE-----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=LOM4
-----END PGP SIGNATURE-----
Merge tag 'imx-maintainers-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/shawnguo/linux into arm/fixes
i.MX MAINTAINERS changes for 6.20:
- Shawn Guo steps down and Frank Li steps up as i.MX platform
maintainer
* tag 'imx-maintainers-6.20' of https://git.kernel.org/pub/scm/linux/kernel/git/shawnguo/linux:
MAINTAINERS: Replace Shawn with Frank as i.MX platform maintainer
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
The current implementation uses a linear list to find queued packets by
ID when processing verdicts from userspace. With large queue depths and
out-of-order verdicting, this O(n) lookup becomes a significant
bottleneck, causing userspace verdict processing to dominate CPU time.
Replace the linear search with a hash table for O(1) average-case
packet lookup by ID. A global rhashtable spanning all network
namespaces attributes hash bucket memory to kernel but is subject to
fixed upper bound.
Signed-off-by: Scott Mitchell <scott.k.mitch1@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Use the is_leap_year() helper from rtc.h instead of
writing it by hand
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Similar to IPIP, introduce specific selftest for IP6IP6 flowtable SW
acceleration in nft_flowtable.sh
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Introduce sw acceleration for tx path of IP6IP6 tunnels relying on the
netfilter flowtable infrastructure.
IP6IP6 tx sw acceleration can be tested running the following scenario
where the traffic is forwarded between two NICs (eth0 and eth1) and an
IP6IP6 tunnel is used to access a remote site (using eth1 as the underlay
device):
ETH0 -- TUN0 <==> ETH1 -- [IP network] -- TUN1 (2001:db8:3::2)
$ip addr show
6: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:00:22:33:11:55 brd ff:ff:ff:ff:ff:ff
inet6 2001:db8:1::2/64 scope global nodad
valid_lft forever preferred_lft forever
7: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:11:22:33:11:55 brd ff:ff:ff:ff:ff:ff
inet6 2001:db8:2::1/64 scope global nodad
valid_lft forever preferred_lft forever
8: tun0@NONE: <POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1480 qdisc noqueue state UNKNOWN group default qlen 1000
link/tunnel6 2001:db8:2::1 peer 2001:db8:2::2 permaddr ce9c:2940:7dcc::
inet6 2002:db8:1::1/64 scope global nodad
valid_lft forever preferred_lft forever
$ip -6 route show
2001:db8:1::/64 dev eth0 proto kernel metric 256 pref medium
2001:db8:2::/64 dev eth1 proto kernel metric 256 pref medium
2002:db8:1::/64 dev tun0 proto kernel metric 256 pref medium
default via 2002:db8:1::2 dev tun0 metric 1024 pref medium
$nft list ruleset
table inet filter {
flowtable ft {
hook ingress priority filter
devices = { eth0, eth1 }
}
chain forward {
type filter hook forward priority filter; policy accept;
meta l4proto { tcp, udp } flow add @ft
}
}
Reproducing the scenario described above using veths I got the following
results:
- TCP stream received from the IPIP tunnel:
- net-next: (baseline) ~93Gbps
- net-next + IP6IP6 flowtbale support: ~98Gbps
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Introduce sw acceleration for rx path of IP6IP6 tunnels relying on the
netfilter flowtable infrastructure. Subsequent patches will add sw
acceleration for IP6IP6 tunnels tx path.
IP6IP6 rx sw acceleration can be tested running the following scenario
where the traffic is forwarded between two NICs (eth0 and eth1) and an
IP6IP6 tunnel is used to access a remote site (using eth1 as the underlay
device):
ETH0 -- TUN0 <==> ETH1 -- [IP network] -- TUN1 (2001:db8:3::2)
$ip addr show
6: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:00:22:33:11:55 brd ff:ff:ff:ff:ff:ff
inet6 2001:db8:1::2/64 scope global nodad
valid_lft forever preferred_lft forever
7: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
link/ether 00:11:22:33:11:55 brd ff:ff:ff:ff:ff:ff
inet6 2001:db8:2::1/64 scope global nodad
valid_lft forever preferred_lft forever
8: tun0@NONE: <POINTOPOINT,NOARP,UP,LOWER_UP> mtu 1480 qdisc noqueue state UNKNOWN group default qlen 1000
link/tunnel6 2001:db8:2::1 peer 2001:db8:2::2 permaddr ce9c:2940:7dcc::
inet6 2002:db8:1::1/64 scope global nodad
valid_lft forever preferred_lft forever
$ip -6 route show
2001:db8:1::/64 dev eth0 proto kernel metric 256 pref medium
2001:db8:2::/64 dev eth1 proto kernel metric 256 pref medium
2002:db8:1::/64 dev tun0 proto kernel metric 256 pref medium
default via 2002:db8:1::2 dev tun0 metric 1024 pref medium
$nft list ruleset
table inet filter {
flowtable ft {
hook ingress priority filter
devices = { eth0, eth1 }
}
chain forward {
type filter hook forward priority filter; policy accept;
meta l4proto { tcp, udp } flow add @ft
}
}
Reproducing the scenario described above using veths I got the following
results:
- TCP stream received from the IPIP tunnel:
- net-next: (baseline) ~ 81Gbps
- net-next + IP6IP6 flowtbale support: ~112Gbps
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Add tunnel hdr_size and tunnel proto fields in nf_flowtable_ctx struct
in order to store IP tunnel header size and protocol used during IPIP
and IP6IP6 tunnel sw offloading decapsulation and avoid recomputing them
during tunnel header pop since this is constant for IPv6.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Rely on nf_flowtable_ctx struct pointer in nf_flow_ip4_tunnel_proto and
nf_flow_skb_encap_protocol routine signature. This is a preliminary patch
to introduce IP6IP6 flowtable acceleration since nf_flowtable_ctx will
be used to store IP6IP6 tunnel info.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
