Otherwise they are very hard to read in the kernel log.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250818045456.1482889-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Corrected a few spelling mistakes
functionalty ==> functionality
in Documentation/driver-api/cxl/devices/device-types.rst
adjascent ==> adjacent
in Documentation/driver-api/cxl/platform/example-configurations/one-dev-per-hb.rst
succeessful ==> successful
in Documentation/driver-api/thermal/exynos_thermal_emulation.rst
Signed-off-by: Ranganath V N <vnranganath.20@gmail.com>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Link: https://lore.kernel.org/r/20250814184304.20448-1-vnranganath.20@gmail.com
queue_limits_stack_integrity needs to handle the new pi_tuple_size field,
otherwise stacking PI-capable devices will always fail.
Fixes: 76e45252a4 ("block: introduce pi_tuple_size field in blk_integrity")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250818045456.1482889-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
-----BEGIN PGP SIGNATURE-----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=y2Q6
-----END PGP SIGNATURE-----
Merge tag 'for-6.17-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"Several zoned mode fixes, mount option printing fixups, folio state
handling fixes and one log replay fix.
- zoned mode:
- zone activation and finish fixes
- block group reservation fixes
- mount option fixes:
- bring back printing of mount options with key=value that got
accidentally dropped during mount option parsing in 6.8
- fix inverse logic or typos when printing nodatasum/nodatacow
- folio status fixes:
- writeback fixes in zoned mode
- properly reset dirty/writeback if submission fails
- properly handle TOWRITE xarray mark/tag
- do not set mtime/ctime to current time when unlinking for log
replay"
* tag 'for-6.17-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: fix printing of mount info messages for NODATACOW/NODATASUM
btrfs: restore mount option info messages during mount
btrfs: fix incorrect log message for nobarrier mount option
btrfs: fix buffer index in wait_eb_writebacks()
btrfs: subpage: keep TOWRITE tag until folio is cleaned
btrfs: clear TAG_TOWRITE from buffer tree when submitting a tree block
btrfs: do not set mtime/ctime to current time when unlinking for log replay
btrfs: clear block dirty if btrfs_writepage_cow_fixup() failed
btrfs: clear block dirty if submit_one_sector() failed
btrfs: zoned: limit active zones to max_open_zones
btrfs: zoned: fix write time activation failure for metadata block group
btrfs: zoned: fix data relocation block group reservation
btrfs: zoned: skip ZONE FINISH of conventional zones
The __einj_error_inject() function allocates memory via kmalloc()
without checking for allocation failure, which could lead to a
NULL pointer dereference.
Return -ENOMEM in case allocation fails.
Fixes: b47610296d ("ACPI: APEI: EINJ: Enable EINJv2 error injections")
Signed-off-by: Charles Han <hanchunchao@inspur.com>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Hanjun Guo <guohanjun@huawei.com>
Link: https://patch.msgid.link/20250815024207.3038-1-hanchunchao@inspur.com
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
On an EINJV2 capable system, users may still use the old injection
interface but einj_get_parameter_address() takes the EINJV2 path to map
the parameter structure. This results in the address the user supplied
being stored to the wrong location and the BIOS injecting based on an
uninitialized field (0x0 in the reported case).
Check the version of the request when mapping the EINJ parameter
structure in BIOS reserved memory.
Fixes: 691a0f0a55 ("ACPI: APEI: EINJ: Discover EINJv2 parameters")
Reported-by: Lai, Yi1 <yi1.lai@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Zaid Alali <zaidal@os.amperecomputing.com>
Reviewed-by: Hanjun Guo <gouhanjun@huawei.com>
Link: https://patch.msgid.link/20250814161706.4489-1-tony.luck@intel.com
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Change the resource cleanup order to align with reverse allocation order.
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Christoph Manszewski <christoph.manszewski@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250813101231.196632-3-christoph.manszewski@intel.com
* Fix fast commit checks for file systems with ea_inode enabled
* Don't drop the i_version mount option on a remount
* Fix FIEMAP reporting when there are holes in a bigalloc file system
* Don't fail when mounting read-only when there are inodes in the
orphan file
* Fix hole length overflow for indirect mapped files on file systems
with an 8k or 16k block file system.
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEK2m5VNv+CHkogTfJ8vlZVpUNgaMFAmijEhcACgkQ8vlZVpUN
gaNAhAf/YfJZ/S+cm1hNBEBA2l874MH6NGVjc19JMp9nXhBkFKXw7xsvMEM4IN5v
EogWLCkGxszbPjtDMJvvLTf0YU8sC4YsOaxZsx2PhYYNt7xCipEAvszQ0HsC1ItU
eUkk0uhQM6VzeHgO30i+vHdvpt9YvfvrJBtXsa8fFLi6+ORMArc4dttV8RWfkkTR
xFt5Y3MdBK9bmBG4AQGN4REL3YkTKbVoFq7zS0e8/5Gk4HO9ttduDPPWHGIrDRUh
T3pv/ADenHytkiLOVOnTnxioSD2QaYy0eI0TOqNyprLOfE1e8w3VtPAviJMXxKe3
cwaXyvtcjCJQiboRPt4S9w1hPWvKvA==
=PwYy
-----END PGP SIGNATURE-----
Merge tag 'ext4_for_linus-6.17-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull ext4 fixes from Ted Ts'o:
- Fix fast commit checks for file systems with ea_inode enabled
- Don't drop the i_version mount option on a remount
- Fix FIEMAP reporting when there are holes in a bigalloc file system
- Don't fail when mounting read-only when there are inodes in the
orphan file
- Fix hole length overflow for indirect mapped files on file systems
with an 8k or 16k block file system
* tag 'ext4_for_linus-6.17-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
jbd2: prevent softlockup in jbd2_log_do_checkpoint()
ext4: fix incorrect function name in comment
ext4: use kmalloc_array() for array space allocation
ext4: fix hole length calculation overflow in non-extent inodes
ext4: don't try to clear the orphan_present feature block device is r/o
ext4: fix reserved gdt blocks handling in fsmap
ext4: fix fsmap end of range reporting with bigalloc
ext4: remove redundant __GFP_NOWARN
ext4: fix unused variable warning in ext4_init_new_dir
ext4: remove useless if check
ext4: check fast symlink for ea_inode correctly
ext4: preserve SB_I_VERSION on remount
ext4: show the default enabled i_version option
This enables the pin control support of the onboard FPGA on AAEON UP
boards.
This FPGA acts as a level shifter between the Intel SoC pins and the pin
header, and also as a mux or switch.
+---------+ +--------------+ +---+
| | | | |
| PWM0 | \ | | H |
|----------|------ \-----|-------------| E |
| I2C0_SDA | | | A |
Intel SoC |----------|------\ | | D |
| GPIO0 | \------|-------------| E |
|----------|------ | | R |
| | FPGA | | |
----------+ +--------------+ +---+
For most of the pins, the FPGA opens/closes a switch to enable/disable
the access to the SoC pin from a pin header.
Each switch, has a direction flag that is set depending the status of the
SoC pin.
For some other pins, the FPGA acts as a mux, and routes one pin (or the
other one) to the header.
The driver also provides a GPIO chip. It requests SoC pins in GPIO mode,
and drives them in tandem with FPGA pins (switch/mux direction).
This commit adds support only for UP Squared board.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Thomas Richard <thomas.richard@bootlin.com>
Link: https://lore.kernel.org/20250811-aaeon-up-board-pinctrl-support-v9-10-29f0cbbdfb30@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Walking hardware contexts created by a process is duplicated in multiple
spots. Add a function, amdxdna_hwctx_walk(), and replace all spots.
hwctx_srcu and dev_lock are good enough to protect hardware context list.
Remove hwctx_lock.
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://lore.kernel.org/r/20250815171634.3417487-1-lizhi.hou@amd.com
When a BPF program which is being loaded reaches the map limit
(MAX_USED_MAPS) or the BTF limit (MAX_USED_BTFS) the -E2BIG is
returned. However, in the former case there is an accompanying
verifier verbose message, and in the latter case there is not.
Add a verbose message to make the behaviour symmetrical.
Reported-by: Kevin Sheldrake <kevin.sheldrake@isovalent.com>
Signed-off-by: Anton Protopopov <a.s.protopopov@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250816151554.902995-1-a.s.protopopov@gmail.com
In the pin_config_set function, when handling PIN_CONFIG_BIAS_PULL_DOWN or
PIN_CONFIG_BIAS_PULL_UP, the function calls pcs_pinconf_clear_bias()
which writes the register. However, the subsequent operations continue
using the stale 'data' value from before the register write, effectively
causing the bias clear operation to be overwritten and not take effect.
Fix this by reading the 'data' value from the register after calling
pcs_pinconf_clear_bias().
This bug seems to have existed when this code was first merged in commit
9dddb4df90 ("pinctrl: single: support generic pinconf").
Signed-off-by: Chi Zhang <chizhang@asrmicro.com>
Link: https://lore.kernel.org/20250807062038.13610-1-chizhang@asrmicro.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The debug output for spacemit_pinconf_dbg_show() prints an extra newline
at the end. This is redundant as pinconf_pins_show() in pinconf.c already
adds a newline in its for loop.
Remove the newline to avoid the extra line in the output.
Example current output:
$ cat /sys/kernel/debug/pinctrl/d401e000.pinctrl/pinconf-pins
Pin config settings per pin
Format: pin (name): configs
pin 0 (GPIO_00): , bias pull disabled, io type (Fixed/1V8), drive strength (32 mA), register (0x1041)
pin 1 (GPIO_01): slew rate (0x0), bias pull disabled, io type (Fixed/1V8), drive strength (32 mA), register (0x1041)
pin 2 (GPIO_02): slew rate (0x0), bias pull disabled, io type (Fixed/1V8), drive strength (32 mA), register (0x1041)
...
Signed-off-by: Hendrik Hamerlinck <hendrik.hamerlinck@hammernet.be>
Reviewed-by: Yixun Lan <dlan@gentoo.org>
Link: https://lore.kernel.org/20250805150701.129113-1-hendrik.hamerlinck@hammernet.be
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Add support for the 'gpio-line-names' property in the MT8183 pinctrl
binding.
This allows naming the GPIOs, which is already done in several
device trees (e.g. mt8183-kukui-jacuzzi.dtsi, mt8183-kukui-kakadu.dtsi,
mt8183-kukui-krane.dtsi), but was previously generating DT schema
warnings.
Signed-off-by: Julien Massot <julien.massot@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://lore.kernel.org/20250801-mtk-dtb-warnings-v1-7-6ba4e432427b@collabora.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Support decoding AMD's XOP prefix encoded instructions.
These instructions are introduced for Bulldozer micro architecture, and not
supported on Intel's processors. But when compiling kernel with
CONFIG_X86_NATIVE_CPU on some AMD processor (e.g. -march=bdver2), these
instructions can be used.
Closes: https://lore.kernel.org/all/871pq06728.fsf@wylie.me.uk/
Reported-by: Alan J. Wylie <alan@wylie.me.uk>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Tested-by: Alan J. Wylie <alan@wylie.me.uk>
Link: https://lore.kernel.org/175386161199.564247.597496379413236944.stgit@devnote2
When a VM is marked as an usuable we disallow new submissions from it,
however submissions that where already scheduled on the ring would still
be re-sent.
Since this can lead to further hangs, avoid emitting the actual IBs.
Fixes: 6a4d287a1a ("drm/msm: Mark VM as unusable on GPU hangs")
Signed-off-by: Antonino Maniscalco <antomani103@gmail.com>
Reviewed-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/668314/
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
In the past %pK was preferable to %p as it would not leak raw pointer
values into the kernel log.
Since commit ad67b74d24 ("printk: hash addresses printed with %p")
the regular %p has been improved to avoid this issue.
Furthermore, restricted pointers ("%pK") were never meant to be used
through printk(). They can still unintentionally leak raw pointers or
acquire sleeping locks in atomic contexts.
Switch to the regular pointer formatting which is safer and
easier to reason about.
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Patchwork: https://patchwork.freedesktop.org/patch/667895/
Signed-off-by: Rob Clark <robin.clark@oss.qualcomm.com>
The reset reason value may be "all bits set", e.g. 0xFFFFFFFF. This is a
commonly used error response from hardware. This may occur due to a real
hardware issue or when running in a VM.
The user will see all reset reasons reported in this case.
Check for an error response value and return early to avoid decoding
invalid data.
Also, adjust the data variable type to match the hardware register size.
Fixes: ab81310287 ("x86/CPU/AMD: Print the reason for the last reset")
Reported-by: Libing He <libhe@redhat.com>
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/20250721181155.3536023-1-yazen.ghannam@amd.com
Since commit 2ca34b5087 ("staging: axis-fifo: Correct handling of
tx_fifo_depth for size validation"), write() operations with packets
larger than 'tx_fifo_depth - 4' words are no longer rejected with -EINVAL.
Fortunately, the packets are not actually getting transmitted to hardware,
otherwise they would be raising a 'Transmit Packet Overrun Error'
interrupt, which requires a reset of the TX circuit to recover from.
Instead, the request times out inside wait_event_interruptible_timeout()
and always returns -EAGAIN, since the wake up condition can never be true
for these packets. But still, they unnecessarily block other tasks from
writing to the FIFO and the EAGAIN return code signals userspace to retry
the write() call, even though it will always fail and time out.
According to the AXI4-Stream FIFO reference manual (PG080), the maximum
valid packet length is 'tx_fifo_depth - 4' words, so attempting to send
larger packets is invalid and should not be happening in the first place:
> The maximum packet that can be transmitted is limited by the size of
> the FIFO, which is (C_TX_FIFO_DEPTH–4)*(data interface width/8) bytes.
Therefore, bring back the old behavior and outright reject packets larger
than 'tx_fifo_depth - 4' with -EINVAL. Add a comment to explain why the
check is necessary. The dev_err() message was removed to avoid cluttering
the dmesg log if an invalid packet is received from userspace.
Fixes: 2ca34b5087 ("staging: axis-fifo: Correct handling of tx_fifo_depth for size validation")
Cc: stable@vger.kernel.org
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://lore.kernel.org/r/20250817171350.872105-1-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch corrects two minor spelling issues found in
Documentation/userspace-api/iommufd.rst:
- "primarly" -> "primarily"
- "sharable" -> "shareable"
Found using codespell(1).
Link: https://patch.msgid.link/r/20250815153840.188213-2-alessandro@0x65c.net
Signed-off-by: Alessandro Ratti <alessandro@0x65c.net>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Use kvfree() instead of kfree() to free pages allocated by kvcalloc()
in iommufs_hw_queue_alloc_phys() to fix potential memory corruption.
Ensure the memory is properly freed, as kvcalloc may internally use
vmalloc or kmalloc depending on available memory in the system.
Fixes: 2238ddc2b0 ("iommufd/viommu: Add IOMMUFD_CMD_HW_QUEUE_ALLOC ioctl")
Link: https://patch.msgid.link/r/aJifyVV2PL6WGEs6@bhairav-test.ee.iitb.ac.in
Signed-off-by: Akhilesh Patil <akhilesh@ee.iitb.ac.in>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Reviewed-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Pranjal Shrivastava <praan@google.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
LLVM/clang can not build the kernel for all architectures supported by
nolibc. The current setup uses the same compiler to build the kernel as
is used for nolibc-test. This prevents using the full qemu-system tests
for LLVM builds.
Instead always build the kernel with GCC. For the nolibc testsuite the
kernel does not need to be built with LLVM.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250719-nolibc-llvm-system-v1-3-1730216ce171@weissschuh.net
Various targets of the testsuite call back into the toplevel kernel
Makefile. These calls use various parameters and are quite long.
Introduce a common variable to make future changes smaller and the lines
shorter.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/r/20250719-nolibc-llvm-system-v1-1-1730216ce171@weissschuh.net
Only one of these variables is used.
$CC is preferred over $CROSS_COMPILE.
Make this clear in the help message.
Suggested-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/lkml/20250817093905.GA14213@1wt.eu/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
B2120 boards are internal boards which never were commercialised.
Drop them.
Signed-off-by: Raphael Gallais-Pou <rgallaispou@gmail.com>
Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
ST's STiH415 and STiH416 platforms have already been removed since
a while. Remove some remaining bits within the mach-sti.
Signed-off-by: Alain Volmat <avolmat@me.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Signed-off-by: Raphael Gallais-Pou <rgallaispou@gmail.com>
Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
B2120 boards are internal boards which never were commercialised.
Remove them from bindings.
Signed-off-by: Raphael Gallais-Pou <rgallaispou@gmail.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
Stick to the documentation and rename both SATA phy-names properties to
what is expected.
Signed-off-by: Raphael Gallais-Pou <rgallaispou@gmail.com>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>
Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
The get_next_cpu() function was only used in one place to find
the next possible CPU, which can be replaced by cpumask_next_wrap().
Signed-off-by: Fushuai Wang <wangfushuai@baidu.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250818032344.23229-1-wangfushuai@baidu.com
Clobbering a lot of registers and stack slots helps exposing tail call
counter overwrite bugs in JITs.
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250813121016.163375-5-iii@linux.ibm.com
The tailcall_bpf2bpf_hierarchy_fentry test hangs on s390. Its call
graph is as follows:
entry()
subprog_tail()
trampoline()
fentry()
the rest of subprog_tail() # via BPF_TRAMP_F_CALL_ORIG
return to entry()
The problem is that the rest of subprog_tail() increments the tail call
counter, but the trampoline discards the incremented value. This
results in an astronomically large number of tail calls.
Fix by making the trampoline write the incremented tail call counter
back.
Fixes: 528eb2cb87 ("s390/bpf: Implement arch_prepare_bpf_trampoline()")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250813121016.163375-4-iii@linux.ibm.com
The tailcall_bpf2bpf_hierarchy_1 test hangs on s390. Its call graph is
as follows:
entry()
subprog_tail()
bpf_tail_call_static(0) -> entry + tail_call_start
subprog_tail()
bpf_tail_call_static(0) -> entry + tail_call_start
entry() copies its tail call counter to the subprog_tail()'s frame,
which then increments it. However, the incremented result is discarded,
leading to an astronomically large number of tail calls.
Fix by writing the incremented counter back to the entry()'s frame.
Fixes: dd691e847d ("s390/bpf: Implement bpf_jit_supports_subprog_tailcalls()")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250813121016.163375-3-iii@linux.ibm.com
Only BPF functions make use of the tail call counter; helpers and
kfuncs ignore and most likely also clobber it. Writing it into these
functions' frames is pointless and misleading, so do not do it.
Fixes: dd691e847d ("s390/bpf: Implement bpf_jit_supports_subprog_tailcalls()")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20250813121016.163375-2-iii@linux.ibm.com
__ADDRESSABLE_ASM_STR() is where the necessary stringification happens.
As long as "sym" doesn't contain any odd characters, no quoting is
required for its use with .quad / .long. In fact the quotation gets in
the way with gas 2.25; it's only from 2.26 onwards that quoted symbols
are half-way properly supported.
However, assembly being different from C anyway, drop
__ADDRESSABLE_ASM_STR() and its helper macro altogether. A simple
.global directive will suffice to get the symbol "declared", i.e. into
the symbol table. While there also stop open-coding STATIC_CALL_TRAMP()
and STATIC_CALL_KEY().
Fixes: 0ef8047b73 ("x86/static-call: provide a way to do very early static-call updates")
Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Josh Poimboeuf <jpoimboe@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: Juergen Gross <jgross@suse.com>
Message-ID: <609d2c74-de13-4fae-ab1a-1ec44afb948d@suse.com>
Validate that all indirect calls adhere to kCFI rules. Notably doing
nocfi indirect call to a cfi function is broken.
Apparently some Rust 'core' code violates this and explodes when ran
with FineIBT.
All the ANNOTATE_NOCFI_SYM sites are prime targets for attackers.
- runtime EFI is especially henous because it also needs to disable
IBT. Basically calling unknown code without CFI protection at
runtime is a massice security issue.
- Kexec image handover; if you can exploit this, you get to keep it :-)
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Josh Poimboeuf <jpoimboe@kernel.org>
Acked-by: Sean Christopherson <seanjc@google.com>
Link: https://lkml.kernel.org/r/20250714103441.496787279@infradead.org
Now that FRED provides C-code entry points for handling IRQs, use the
FRED infrastructure for forwarding IRQs even if FRED is fully
disabled, e.g. isn't supported in hardware. Avoiding the non-FRED
assembly trampolines into the IDT handlers for IRQs eliminates the
associated non-CFI indirect call (KVM performs a CALL by doing a
lookup on the IDT using the IRQ vector).
Keep NMIs on the legacy IDT path, as the FRED NMI entry code relies on
FRED's architectural behavior with respect to NMI blocking, i.e. doesn't
jump through the myriad hoops needed to deal with IRET "unexpectedly"
unmasking NMIs. KVM's NMI path already makes a direct CALL to C-code,
i.e. isn't problematic for CFI. KVM does make a short detour through
assembly code to build the stack frame, but the "FRED entry from KVM"
path does the same.
Force FRED for 64-bit kernels if KVM_INTEL is enabled, as the benefits of
eliminating the IRQ trampoline usage far outwieghts the code overhead for
FRED.
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250714103441.381946911@infradead.org
Modify asm_fred_entry_from_kvm() to allow it to be invoked by KVM even
when FRED isn't fully enabled, e.g. when running with
CONFIG_X86_FRED=y on non-FRED hardware. This will allow forcing KVM
to always use the FRED entry points for 64-bit kernels, which in turn
will eliminate a rather gross non-CFI indirect call that KVM uses to
trampoline IRQs by doing IDT lookups.
The point of asm_fred_entry_from_kvm() is to bridge between C
(vmx:handle_external_interrupt_irqoff()) and more C
(__fred_entry_from_kvm()) while changing the calling context to appear
like an interrupt (pt_regs). Making the whole thing bound by C ABI.
All that remains for non-FRED hardware is to restore RSP (to undo the
redzone and alignment). However the trivial change would result in
code like:
push %rbp
mov %rsp, %rbp
sub $REDZONE, %rsp
and $MASK, %rsp
PUSH_AND_CLEAR_REGS
push %rbp
POP_REGS
pop %rbp <-- *objtool fail*
mov %rbp, %rsp
pop %rbp
ret
And this will confuse objtool something wicked -- it gets confused by
the extra pop %rbp, not realizing the push and pop preserve the value.
Rather than trying to each objtool about this, recognise that since
the code is bound by C ABI on both ends and interrupts are not allowed
to change pt_regs (only exceptions are) it is sufficient to PUSH_REGS
in order to create pt_regs, but there is no reason to POP_REGS --
provided the callee-saved registers are preserved.
So avoid clearing callee-saved regs and skip POP_REGS.
[Original patch by Sean; much of this version by Josh; Changelog,
comments and final form by Peterz]
Originally-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Sean Christopherson <seanjc@google.com>
Link: https://lkml.kernel.org/r/20250714103441.245417052@infradead.org
Install the system vector IRQ handlers for FRED even if FRED isn't fully
enabled in hardware. This will allow KVM to use the FRED IRQ path even
on non-FRED hardware, which in turn will eliminate a non-CFI indirect CALL
(KVM currently invokes the IRQ handler via an IDT lookup on the vector).
[sean: extract from diff, drop stub, write changelog]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250714103441.121251108@infradead.org
Instead of using an indirect call to the hypercall page, use a direct
call instead. This avoids all CFI problems, including the one where
the hypercall page doesn't have IBT on.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Michael Kelley <mhklinux@outlook.com>
Acked-by: Sean Christopherson <seanjc@google.com>
Link: https://lkml.kernel.org/r/20250714103441.011387946@infradead.org
What used to be a simple few instructions has turned into a giant mess
(for x86_64). Not only does it use static_branch wrong, it mixes it
with dynamic branches for no apparent reason.
Notably it uses static_branch through an out-of-line function call,
which completely defeats the purpose, since instead of a simple
JMP/NOP site, you get a CALL+RET+TEST+Jcc sequence in return, which is
absolutely idiotic.
Add to that a dynamic test of hyperv_paravisor_present, something
which is set once and never changed.
Replace all this idiocy with a single direct function call to the
right hypercall variant.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Michael Kelley <mhklinux@outlook.com>
Tested-by: Michael Kelley <mhklinux@outlook.com>
Acked-by: Sean Christopherson <seanjc@google.com>
Link: https://lkml.kernel.org/r/20250714103440.897136093@infradead.org
