As per the kernel documentation[1], hardlockup detector should
be disabled in KVM guests as it may give false positives. On
PPC, hardlockup detector is enabled inside KVM guests because
disable_hardlockup_detector() is marked as early_initcall and it
relies on kvm_guest static key (is_kvm_guest()) which is initialized
later during boot by check_kvm_guest(), which is a core_initcall.
check_kvm_guest() is also called in pSeries_smp_probe(), which is called
before initcalls, but it is skipped if KVM guest does not have doorbell
support or if the guest is launched with SMT=1.
Call check_kvm_guest() in disable_hardlockup_detector() so that
is_kvm_guest() check goes through fine and hardlockup detector can be
disabled inside the KVM guest.
Apply make_spte()'s optimization to skip trying to unsync shadow pages if
and only if the old SPTE was a leaf SPTE, as non-leaf SPTEs in direct MMUs
are always writable, i.e. could trigger a false positive and incorrectly
lead to KVM creating a SPTE without write-protecting or marking shadow
pages unsync.
This bug only affects the TDP MMU, as the shadow MMU only overwrites a
shadow-present SPTE when synchronizing SPTEs (and only 4KiB SPTEs can be
unsync). Specifically, mmu_set_spte() drops any non-leaf SPTEs *before*
calling make_spte(), whereas the TDP MMU can do a direct replacement of a
page table with the leaf SPTE.
Opportunistically update the comment to explain why skipping the unsync
stuff is safe, as opposed to simply saying "it's someone else's problem".
Cc: stable@vger.kernel.org Tested-by: Alex Bennée <alex.bennee@linaro.org> Signed-off-by: Sean Christopherson <seanjc@google.com> Tested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <20241010182427.1434605-5-seanjc@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kvm_vm_create_worker_thread() is meant to be used for kthreads that
can consume significant amounts of CPU time on behalf of a VM or in
response to how the VM behaves (for example how it accesses its memory).
Therefore it wants to charge the CPU time consumed by that work to
the VM's container.
However, because of these threads, cgroups which have kvm instances
inside never complete freezing. This can be trivially reproduced:
The cgroup freezing happens in the signal delivery path but
kvm_nx_huge_page_recovery_worker, while joining non-root cgroups, never
calls into the signal delivery path and thus never gets frozen. Because
the cgroup freezer determines whether a given cgroup is frozen by
comparing the number of frozen threads to the total number of threads
in the cgroup, the cgroup never becomes frozen and users waiting for
the state transition may hang indefinitely.
Since the worker kthread is tied to a user process, it's better if
it behaves similarly to user tasks as much as possible, including
being able to send SIGSTOP and SIGCONT. In fact, vhost_task is all
that kvm_vm_create_worker_thread() wanted to be and more: not only it
inherits the userspace process's cgroups, it has other niceties like
being parented properly in the process tree. Use it instead of the
homegrown alternative.
Incidentally, the new code is also better behaved when you flip recovery
back and forth to disabled and back to enabled. If your recovery period
is 1 minute, it will run the next recovery after 1 minute independent
of how many times you flipped the parameter.
Fix the AEGIS assembly code to access 'unsigned int' arguments as 32-bit
values instead of 64-bit, since the upper bits of the corresponding
64-bit registers are not guaranteed to be zero.
Note: there haven't been any reports of this bug actually causing
incorrect behavior. Neither gcc nor clang guarantee zero-extension to
64 bits, but zero-extension is likely to happen in practice because most
instructions that operate on 32-bit registers zero-extend to 64 bits.
Fixes: 1d373d4e8e15 ("crypto: x86 - Add optimized AEGIS implementations") Cc: stable@vger.kernel.org Reviewed-by: Ondrej Mosnacek <omosnace@redhat.com> Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the trace data buffer becomes full, a truncated flag [T] is reported
in PERF_RECORD_AUX. In some cases, the size reported is 0, even though
data must have been added to make the buffer full.
That happens when the buffer fills up from empty to full before the
Intel PT driver has updated the buffer position. Then the driver
calculates the new buffer position before calculating the data size.
If the old and new positions are the same, the data size is reported
as 0, even though it is really the whole buffer size.
Fix by detecting when the buffer position is wrapped, and adjust the
data size calculation accordingly.
Example
Use a very small buffer size (8K) and observe the size of truncated [T]
data. Before the fix, it is possible to see records of 0 size.
Before:
$ perf record -m,8K -e intel_pt// uname
Linux
[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.105 MB perf.data ]
$ perf script -D --no-itrace | grep AUX | grep -F '[T]'
Warning:
AUX data lost 2 times out of 3!
$ perf record -m,8K -e intel_pt// uname
Linux
[ perf record: Woken up 3 times to write data ]
[ perf record: Captured and wrote 0.040 MB perf.data ]
$ perf script -D --no-itrace | grep AUX | grep -F '[T]'
Warning:
AUX data lost 2 times out of 3!
On the Renesas RZ/G3S SMARC Carrier II board having a DA7212 codec (using
da7213 driver) connected to one SSIF-2 available on the Renesas RZ/G3S SoC
it has been discovered that using the runtime PM API for suspend/resume
(as will be proposed in the following commits) leads to the codec not
being propertly initialized after resume. This is because w/o
max_register populated to regmap_config the regcache_rbtree_sync()
breaks on base_reg > max condition and the regcache_sync_block() call is
skipped.
An atomicity violation occurs when the validity of the variables
da7219->clk_src and da7219->mclk_rate is being assessed. Since the entire
assessment is not protected by a lock, the da7219 variable might still be
in flux during the assessment, rendering this check invalid.
To fix this issue, we recommend adding a lock before the block
if ((da7219->clk_src == clk_id) && (da7219->mclk_rate == freq)) so that
the legitimacy check for da7219->clk_src and da7219->mclk_rate is
protected by the lock, ensuring the validity of the check.
This possible bug is found by an experimental static analysis tool
developed by our team. This tool analyzes the locking APIs
to extract function pairs that can be concurrently executed, and then
analyzes the instructions in the paired functions to identify possible
concurrency bugs including data races and atomicity violations.
Commit 7c55b78818cf ("jfs: xattr: fix buffer overflow for invalid xattr")
also addresses this issue but it only fixes it for positive values, while
ea_size is an integer type and can take negative values, e.g. in case of
a corrupted filesystem. This still breaks validation and would overflow
because of implicit conversion from int to size_t in print_hex_dump().
Fix this issue by clamping the ea_size value instead.
Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
The original implementation ext4's FS_IOC_GETFSMAP handling only
worked when the range of queried blocks included at least one free
(unallocated) block range. This is because how the metadata blocks
were emitted was as a side effect of ext4_mballoc_query_range()
calling ext4_getfsmap_datadev_helper(), and that function was only
called when a free block range was identified. As a result, this
caused generic/365 to fail.
Fix this by creating a new function ext4_getfsmap_meta_helper() which
gets called so that blocks before the first free block range in a
block group can get properly reported.
find_group_other() and find_group_orlov() read *_lo, *_hi with
ext4_free_inodes_count without additional locking. This can cause
data-race warning, but since the lock is held for most writes and free
inodes value is generally not a problem even if it is incorrect, it is
more appropriate to use READ_ONCE()/WRITE_ONCE() than to add locking.
==================================================================
BUG: KCSAN: data-race in ext4_free_inodes_count / ext4_free_inodes_set
write to 0xffff88810404300e of 2 bytes by task 6254 on cpu 1:
ext4_free_inodes_set+0x1f/0x80 fs/ext4/super.c:405
__ext4_new_inode+0x15ca/0x2200 fs/ext4/ialloc.c:1216
ext4_symlink+0x242/0x5a0 fs/ext4/namei.c:3391
vfs_symlink+0xca/0x1d0 fs/namei.c:4615
do_symlinkat+0xe3/0x340 fs/namei.c:4641
__do_sys_symlinkat fs/namei.c:4657 [inline]
__se_sys_symlinkat fs/namei.c:4654 [inline]
__x64_sys_symlinkat+0x5e/0x70 fs/namei.c:4654
x64_sys_call+0x1dda/0x2d60 arch/x86/include/generated/asm/syscalls_64.h:267
do_syscall_x64 arch/x86/entry/common.c:52 [inline]
do_syscall_64+0x54/0x120 arch/x86/entry/common.c:83
entry_SYSCALL_64_after_hwframe+0x76/0x7e
read to 0xffff88810404300e of 2 bytes by task 6257 on cpu 0:
ext4_free_inodes_count+0x1c/0x80 fs/ext4/super.c:349
find_group_other fs/ext4/ialloc.c:594 [inline]
__ext4_new_inode+0x6ec/0x2200 fs/ext4/ialloc.c:1017
ext4_symlink+0x242/0x5a0 fs/ext4/namei.c:3391
vfs_symlink+0xca/0x1d0 fs/namei.c:4615
do_symlinkat+0xe3/0x340 fs/namei.c:4641
__do_sys_symlinkat fs/namei.c:4657 [inline]
__se_sys_symlinkat fs/namei.c:4654 [inline]
__x64_sys_symlinkat+0x5e/0x70 fs/namei.c:4654
x64_sys_call+0x1dda/0x2d60 arch/x86/include/generated/asm/syscalls_64.h:267
do_syscall_x64 arch/x86/entry/common.c:52 [inline]
do_syscall_64+0x54/0x120 arch/x86/entry/common.c:83
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Cc: stable@vger.kernel.org Signed-off-by: Jeongjun Park <aha310510@gmail.com> Reviewed-by: Andreas Dilger <adilger@dilger.ca> Link: https://patch.msgid.link/20241003125337.47283-1-aha310510@gmail.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The unification of irq_domain_create_legacy() missed the fact that
interrupts must be associated even when the Linux interrupt number provided
in the first_irq argument is 0.
This breaks all call sites of irq_domain_create_legacy() which supply 0 as
the first_irq argument.
Enforce the association for legacy domains in __irq_domain_instantiate() to
cure this.
[ tglx: Massaged it slightly. ]
Fixes: 70114e7f7585 ("irqdomain: Simplify simple and legacy domain creation") Reported-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by Matti Vaittinen <mazziesaccount@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Jiaxun Yang <jiaxun.yang@flygoat.com> Link: https://lore.kernel.org/all/c3379142-10bc-4f14-b8ac-a46927aeac38@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In success case, the revision holds a non-null pointer. The current
logic incorrectly returns an error for a non-null pointer, whereas
it should return an error for a null pointer.
The socinfo driver for IPQ9574 and IPQ5332 is currently broken,
resulting in the following error message
qcom-socinfo qcom-socinfo: probe with driver qcom-socinfo failed with
error -12
Add a null check for the revision to ensure it returns an error only in
failure case (null pointer).
The commit 4c39529663b9 adds a warning about duplicate cache names if
CONFIG_DEBUG_VM is selected. These warnings are triggered by the dm-bufio
code. The dm-bufio code allocates a slab cache with each client. It is
not possible to preallocate the caches in the module init function
because the size of auxiliary per-buffer data is not known at this point.
So, this commit changes dm-bufio so that it appends a unique atomic value
to the cache name, to avoid the warnings.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Fixes: 4c39529663b9 ("slab: Warn on duplicate cache names when DEBUG_VM=y") Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
UCSI connector's indices start from 1 up to 3, PMIC_GLINK_MAX_PORTS.
Correct the condition in the pmic_glink_ucsi_connector_status()
callback, fixing Type-C orientation reporting for the third USB-C
connector.
Fixes: 76716fd5bf09 ("usb: typec: ucsi: glink: move GPIO reading into connector_status callback") Cc: stable@vger.kernel.org Reported-by: Abel Vesa <abel.vesa@linaro.org> Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org> Reviewed-by: Johan Hovold <johan+linaro@kernel.org> Tested-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Link: https://lore.kernel.org/r/20241109-ucsi-glue-fixes-v2-1-8b21ff4f9fbe@linaro.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current USB-audio driver code doesn't check bLength of each
descriptor at traversing for clock descriptors. That is, when a
device provides a bogus descriptor with a shorter bLength, the driver
might hit out-of-bounds reads.
For addressing it, this patch adds sanity checks to the validator
functions for the clock descriptor traversal. When the descriptor
length is shorter than expected, it's skipped in the loop.
For the clock source and clock multiplier descriptors, we can just
check bLength against the sizeof() of each descriptor type.
OTOH, the clock selector descriptor of UAC2 and UAC3 has an array
of bNrInPins elements and two more fields at its tail, hence those
have to be checked in addition to the sizeof() check.
This patch fixes an issue in the function xenbus_dev_probe(). In the
xenbus_dev_probe() function, within the if (err) branch at line 313, the
program incorrectly returns err directly without releasing the resources
allocated by err = drv->probe(dev, id). As the return value is non-zero,
the upper layers assume the processing logic has failed. However, the probe
operation was performed earlier without a corresponding remove operation.
Since the probe actually allocates resources, failing to perform the remove
operation could lead to problems.
To fix this issue, we followed the resource release logic of the
xenbus_dev_remove() function by adding a new block fail_remove before the
fail_put block. After entering the branch if (err) at line 313, the
function will use a goto statement to jump to the fail_remove block,
ensuring that the previously acquired resources are correctly released,
thus preventing the reference count leak.
This bug was identified by an experimental static analysis tool developed
by our team. The tool specializes in analyzing reference count operations
and detecting potential issues where resources are not properly managed.
In this case, the tool flagged the missing release operation as a
potential problem, which led to the development of this patch.
Replacing the qdisc with pfifo makes retransmissions go away.
It appears that a flow may have a delayed packet with a very near
Tx time. Later, we may get busy processing Rx and the target Tx time
will pass, but we won't service Tx since the CPU is busy with Rx.
If Rx sees an ACK and we try to push more data for the delayed flow
we may fastpath the skb, not realizing that there are already "ready
to send" packets for this flow sitting in the qdisc.
Don't trust the fastpath if we are "behind" according to the projected
Tx time for next flow waiting in the Qdisc. Because we consider anything
within the offload window to be okay for fastpath we must consider
the entire offload window as "now".
where "xx_behind" counts how many times we hit the new "return false".
CC: stable@vger.kernel.org Fixes: 076433bd78d7 ("net_sched: sch_fq: add fast path for mostly idle qdisc") Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20241124022148.3126719-1-kuba@kernel.org Signed-off-by: Paolo Abeni <pabeni@redhat.com>
[stable: drop the offload horizon, it's not supported / 0] Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit 6ee600bfbe0f ("s390/pci: remove hotplug slot when releasing the
device") the zpci_exit_slot() was moved from zpci_device_reserved() to
zpci_release_device() with the intention of keeping the hotplug slot
around until the device is actually removed.
Now zpci_release_device() is only called once all references are
dropped. Since the zPCI subsystem only drops its reference once the
device is in the reserved state it follows that zpci_release_device()
must only deal with devices in the reserved state. Despite that it
contains code to tear down from both configured and standby state. For
the standby case this already includes the removal of the hotplug slot
so would cause a double removal if a device was ever removed in
either configured or standby state.
Instead of causing a potential double removal in a case that should
never happen explicitly WARN_ON() if a device in non-reserved state is
released and get rid of the dead code cases.
Following commit 13f58267cda3 ("ASoC: soc.h: don't create dummy
Component via COMP_DUMMY()"), COMP_DUMMY() became an array with zero
length, and only gets populated with the dummy struct after the card is
registered. Since the sound card driver's probe happens before the card
registration, accessing any of the members of a dummy component during
probe will result in undefined behavior.
This can be observed in the mt8188 and mt8195 machine sound drivers. By
omitting a dai link subnode in the sound card's node in the Devicetree,
the default uninitialized dummy codec is used, and when its dai_name
pointer gets passed to strcmp() it results in a null pointer dereference
and a kernel panic.
In addition to that, set_card_codec_info() in the generic helpers file,
mtk-soundcard-driver.c, will populate a dai link with a dummy codec when
a dai link node is present in DT but with no codec property.
The result is that at probe time, a dummy codec can either be
uninitialized with num_codecs = 0, or be an initialized dummy codec,
with num_codecs = 1 and dai_name = "snd-soc-dummy-dai". In order to
accommodate for both situations, check that num_codecs is not zero
before accessing the codecs' fields but still check for the codec's dai
name against "snd-soc-dummy-dai" as needed.
While at it, also drop the check that dai_name is not null in the mt8192
driver, introduced in commit 4d4e1b6319e5 ("ASoC: mediatek: mt8192:
Check existence of dai_name before dereferencing"), as it is actually
redundant given the preceding num_codecs != 0 check.
Fixes: 13f58267cda3 ("ASoC: soc.h: don't create dummy Component via COMP_DUMMY()") Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com> Reviewed-by: Fei Shao <fshao@chromium.org> Acked-by: Trevor Wu <trevor.wu@mediatek.com> Link: https://patch.msgid.link/20241126-asoc-mtk-dummy-panic-v1-1-42d53e168d2e@collabora.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add condition check to register ACP PDM sound card by reading
_WOV acpi entry.
Fixes: 5426f506b584 ("ASoC: amd: Add support for enabling DMIC on acp6x via _DSD") Signed-off-by: Venkata Prasad Potturu <venkataprasad.potturu@amd.com> Link: https://patch.msgid.link/20241127112227.227106-1-venkataprasad.potturu@amd.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
kunit_kzalloc() may return a NULL pointer, dereferencing it without
NULL check may lead to NULL dereference.
Add NULL checks for all the kunit_kzalloc() in sound_kunit.c
Multiple profiles shared 'ent->caps', so some logs missed.
Fixes: 0ed3b28ab8bf ("AppArmor: mediation of non file objects") Signed-off-by: chao liu <liuzgyid@outlook.com> Signed-off-by: John Johansen <john.johansen@canonical.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a previous ufence is not signalled, vm_bind will return -EBUSY.
Delaying the modification of ufence->signalled can cause issues if the
UMD reuses the same ufence so update ufence->signalled before waking up
waiters.
devm_kasprintf() can return a NULL pointer on failure,but this
returned value in imx_audmix_probe() is not checked.
Add NULL check in imx_audmix_probe(), to handle kernel NULL
pointer dereference error.
Fixes: 05d996e11348 ("ASoC: imx-audmix: Split capture device for audmix") Signed-off-by: Charles Han <hanchunchao@inspur.com> Link: https://patch.msgid.link/20241118084553.4195-1-hanchunchao@inspur.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit addresses a null pointer dereference issue in
hwss_setup_dpp(). The issue could occur when pipe_ctx->plane_state is
null. The fix adds a check to ensure `pipe_ctx->plane_state` is not null
before accessing. This prevents a null pointer dereference.
Fixes: 0baae6246307 ("drm/amd/display: Refactor fast update to use new HWSS build sequence") Reviewed-by: Tom Chung <chiahsuan.chung@amd.com> Signed-off-by: Zicheng Qu <quzicheng@huawei.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit addresses a null pointer dereference issue in
dcn20_program_pipe(). Previously, commit 8e4ed3cf1642 ("drm/amd/display:
Add null check for pipe_ctx->plane_state in dcn20_program_pipe")
partially fixed the null pointer dereference issue. However, in
dcn20_update_dchubp_dpp(), the variable pipe_ctx is passed in, and
plane_state is accessed again through pipe_ctx. Multiple if statements
directly call attributes of plane_state, leading to potential null
pointer dereference issues. This patch adds necessary null checks to
ensure stability.
Fixes: 8e4ed3cf1642 ("drm/amd/display: Add null check for pipe_ctx->plane_state in dcn20_program_pipe") Reviewed-by: Tom Chung <chiahsuan.chung@amd.com> Signed-off-by: Zicheng Qu <quzicheng@huawei.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On several HP models (tested on HP 3125 and HP Probook 455 G2),
spurious unplug events are emitted upon login on Chrome OS.
This is likely due to the way Chrome OS restarts graphics
upon login, so it's possible it's an issue on other
distributions but not as common, though I haven't
reproduced the issue elsewhere.
Use logic from an earlier version of the merged change
(see link below) which iterates over connectors and finds
matching encoders, rather than the other way around.
Also fixes an issue with screen mirroring on Chrome OS.
I've deployed this patch on Fedora and did not observe
any regression on these devices.
Add helper rdev_to_drm(rdev), similar to amdgpu, most function should
access the "drm_device" with "rdev_to_drm(rdev)" instead, where amdgpu has
"adev_to_drm(adev)". It also makes changing from "*drm_device" to "drm_device"
in "radeon_devicce" later on easier.
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de> Tested-by: Thomas Zimmermann <tzimmermann@suse.de> Signed-off-by: Wu Hoi Pok <wuhoipok@gmail.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: 7037bb04265e ("drm/radeon: Fix spurious unplug event on radeon HDMI") Signed-off-by: Sasha Levin <sashal@kernel.org>
The "command" variable can be controlled by the user via debugfs. The
worry is that if con_index is zero then "&uc->ucsi->connector[con_index
- 1]" would be an array underflow.
Fixes: 170a6726d0e2 ("usb: typec: ucsi: add support for separate DP altmode devices") Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com> Link: https://lore.kernel.org/r/c69ef0b3-61b0-4dde-98dd-97b97f81d912@stanley.mountain Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Each of the __field() macros were triggering sparse warnings similar to:
trace.h:87:1: sparse: sparse: cast to restricted __le64
trace.h:87:1: sparse: sparse: restricted __le64 degrades to integer
trace.h:87:1: sparse: sparse: restricted __le64 degrades to integer
Change each little endian type to its similarly sized native integer.
Convert inputs into native endian.
Return -EPROBE_DEFER from interconnect provider incase probe defer is
received from devm_clk_bulk_get_all(). This would help in reattempting
the inteconnect driver probe, once the required QoS clocks are
available.
Since the req_free list will updated if enqueuing one request was not
possible it will be added back to the free list. With every available
free request in the queue it is a valid case for the pump worker to use
it and continue the pending bufferdata into requests for the req_ready
list.
The devres ftrace event logs the name of the devres node, which is often a
function name (e.g., "devm_work_drop") stringified by macros like
devm_add_action. Currently, ftrace stores this name as a string literal
address, which can become invalid when the module containing the string is
unloaded. This results in page faults when ftrace tries to access the name.
This behavior is problematic because the devres ftrace event is designed to
trace resource management throughout a device driver's lifecycle, including
during module unload. The event should be available even after the module
is unloaded to properly diagnose resource issues.
Fix the issue by copying the devres node name into the ftrace ring buffer
using __assign_str(), instead of storing just the address. This ensures
that ftrace can always access the name, even if the module is unloaded.
This change increases the memory usage for each of the ftrace entry by
12-16 bytes assuming the average devres node name is 20 bytes long,
depending on the size of const char *.
Note that this change does not affect anything unless all of following
conditions are met.
- CONFIG_DEBUG_DEVRES is enabled
- ftrace tracing is enabled
- The devres event is enabled in ftrace tracing
[Syzbot reported two possible deadlocks]
The first possible deadlock is:
WARNING: possible recursive locking detected 6.12.0-rc1-syzkaller-00027-g4a9fe2a8ac53 #0 Not tainted
--------------------------------------------
syz-executor363/2651 is trying to acquire lock: ffffffff89b120e8 (chaoskey_list_lock){+.+.}-{3:3}, at: chaoskey_release+0x15d/0x2c0 drivers/usb/misc/chaoskey.c:322
but task is already holding lock: ffffffff89b120e8 (chaoskey_list_lock){+.+.}-{3:3}, at: chaoskey_release+0x7f/0x2c0 drivers/usb/misc/chaoskey.c:299
other info that might help us debug this:
Possible unsafe locking scenario:
The second possible deadlock is:
WARNING: possible circular locking dependency detected 6.12.0-rc1-syzkaller-00027-g4a9fe2a8ac53 #0 Not tainted
------------------------------------------------------
kworker/0:2/804 is trying to acquire lock: ffffffff899dadb0 (minor_rwsem){++++}-{3:3}, at: usb_deregister_dev+0x7c/0x1e0 drivers/usb/core/file.c:186
but task is already holding lock: ffffffff89b120e8 (chaoskey_list_lock){+.+.}-{3:3}, at: chaoskey_disconnect+0xa8/0x2a0 drivers/usb/misc/chaoskey.c:235
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
*** DEADLOCK ***
[Analysis]
The first is AA lock, it because wrong logic, it need a unlock.
The second is AB lock, it needs to rearrange the order of lock usage.
Fixes: 422dc0a4d12d ("USB: chaoskey: fail open after removal") Reported-by: syzbot+685e14d04fe35692d3bc@syzkaller.appspotmail.com Reported-by: syzbot+1f8ca5ee82576ec01f12@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=685e14d04fe35692d3bc Signed-off-by: Edward Adam Davis <eadavis@qq.com> Tested-by: syzbot+685e14d04fe35692d3bc@syzkaller.appspotmail.com Reported-by: syzbot+5f1ce62e956b7b19610e@syzkaller.appspotmail.com Tested-by: syzbot+5f1ce62e956b7b19610e@syzkaller.appspotmail.com Tested-by: syzbot+1f8ca5ee82576ec01f12@syzkaller.appspotmail.com Link: https://lore.kernel.org/r/tencent_84EB865C89862EC22EE94CB3A7C706C59206@qq.com Cc: Oliver Neukum <oneukum@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
chaoskey_open() takes the lock only to increase the
counter of openings. That means that the mutual exclusion
with chaoskey_disconnect() cannot prevent an increase
of the counter and chaoskey_open() returning a success.
If that race is hit, chaoskey_disconnect() will happily
free all resources associated with the device after
it has dropped the lock, as it has read the counter
as zero.
To prevent this race chaoskey_open() has to check
the presence of the device under the lock.
However, the current per device lock cannot be used,
because it is a part of the data structure to be
freed. Hence an additional global mutex is needed.
The issue is as old as the driver.
The IO yurex_write() needs to wait for in order to have a device
ready for writing again can take a long time time.
Consequently the sleep is done in an interruptible state.
Therefore others waiting for yurex_write() itself to finish should
use mutex_lock_interruptible.
iowarrior_read() uses the iowarrior dev structure, but does not use any
lock on the structure. This can cause various bugs including data-races,
so it is more appropriate to use a mutex lock to safely protect the
iowarrior dev structure. When using a mutex lock, you should split the
branch to prevent blocking when the O_NONBLOCK flag is set.
In addition, it is unnecessary to check for NULL on the iowarrior dev
structure obtained by reading file->private_data. Therefore, it is
better to remove the check.
If i2c_smbus_write_byte_data() fails in al3010_init(),
al3010_set_pwr(false) is not called.
In order to avoid such a situation, move the devm_add_action_or_reset()
witch calls al3010_set_pwr(false) right after a successful
al3010_set_pwr(true).
Similar to the previous patch, plumb the RCU lock inside
the ipmr_get_table(), provided a lockless variant and apply
the latter in the few spots were the lock is already held.
Fixes: 709b46e8d90b ("net: Add compat ioctl support for the ipv4 multicast ioctl SIOCGETSGCNT") Fixes: f0ad0860d01e ("ipv4: ipmr: support multiple tables") Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Several places call ip6mr_get_table() with no RCU nor RTNL lock.
Add RCU protection inside such helper and provide a lockless variant
for the few callers that already acquired the relevant lock.
Note that some users additionally reference the table outside the RCU
lock. That is actually safe as the table deletion can happen only
after all table accesses are completed.
Fixes: e2d57766e674 ("net: Provide compat support for SIOCGETMIFCNT_IN6 and SIOCGETSGCNT_IN6.") Fixes: d7c31cbde4bc ("net: ip6mr: add RTM_GETROUTE netlink op") Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The cited commit replaced inet_csk_reqsk_queue_drop_and_put() with
__inet_csk_reqsk_queue_drop() and reqsk_put() in reqsk_timer_handler().
Then, oreq should be passed to reqsk_put() instead of req; otherwise
use-after-free of nreq could happen when reqsk is migrated but the
retry attempt failed (e.g. due to timeout).
Let's pass oreq to reqsk_put().
Fixes: e8c526f2bdf1 ("tcp/dccp: Don't use timer_pending() in reqsk_queue_unlink().") Reported-by: Liu Jian <liujian56@huawei.com> Closes: https://lore.kernel.org/netdev/1284490f-9525-42ee-b7b8-ccadf6606f6d@huawei.com/ Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev> Reviewed-by: Liu Jian <liujian56@huawei.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Martin KaFai Lau <martin.lau@kernel.org> Link: https://patch.msgid.link/20241123174236.62438-1-kuniyu@amazon.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
copy_from_sockptr() does not return negative value on error; instead, it
reports the number of bytes that failed to copy. Since it's deprecated,
switch to copy_safe_from_sockptr().
Note: Keeping the `optlen != sizeof(unsigned int)` check as
copy_safe_from_sockptr() by itself would also accept
optlen > sizeof(unsigned int). Which would allow a more lenient handling
of inputs.
Fixes: 17926a79320a ("[AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both") Signed-off-by: Michal Luczaj <mhal@rbox.co> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
copy_from_sockptr() is used incorrectly: return value is the number of
bytes that could not be copied. Since it's deprecated, switch to
copy_safe_from_sockptr().
Note: Keeping the `optlen != sizeof(int)` check as copy_safe_from_sockptr()
by itself would also accept optlen > sizeof(int). Which would allow a more
lenient handling of inputs.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Suggested-by: David Wei <dw@davidwei.uk> Signed-off-by: Michal Luczaj <mhal@rbox.co> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Tested-by: Kiran K <kiran.k@intel.com> Fixes: f53e1c9c726d ("Bluetooth: MGMT: Fix possible crash on mgmt_index_removed") Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
==================================================================
BUG: KASAN: slab-use-after-free in set_powered_sync+0x3a/0xc0 net/bluetooth/mgmt.c:1353
Read of size 8 at addr ffff888029b4dd18 by task kworker/u9:0/54
If we go through the PCI shutdown or suspend path, we shutdown the
NIC but PTP remains registered. If the kernel continues to run for
a little bit, the periodic PTP .do_aux_work() function may be called
and it will read the PHC from the BAR register. Since the device
has already been disabled, it will cause a PCIe completion timeout.
Fix it by calling bnxt_ptp_clear() in the PCI shutdown/suspend
handlers. bnxt_ptp_clear() will unregister from PTP and
.do_aux_work() will be canceled.
In bnxt_resume(), we need to re-initialize PTP.
Fixes: a521c8a01d26 ("bnxt_en: Move bnxt_ptp_init() from bnxt_open() back to bnxt_init_one()") Cc: Richard Cochran <richardcochran@gmail.com> Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com> Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Instead of passing the 2nd parameter phc_cfg to bnxt_ptp_init().
Store it in bp->ptp_cfg so that the caller doesn't need to know what
the value should be.
In the next patch, we'll need to call bnxt_ptp_init() in bnxt_resume()
and this will make it easier.
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com> Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stable-dep-of: 3661c05c54e8 ("bnxt_en: Unregister PTP during PCI shutdown and suspend") Signed-off-by: Sasha Levin <sashal@kernel.org>
The MTU setting at the time an XDP multi-buffer is attached
determines whether the aggregation ring will be used and the
rx_skb_func handler. This is done in bnxt_set_rx_skb_mode().
If the MTU is later changed, the aggregation ring setting may need
to be changed and it may become out-of-sync with the settings
initially done in bnxt_set_rx_skb_mode(). This may result in
random memory corruption and crashes as the HW may DMA data larger
than the allocated buffer size, such as:
To address the issue, we now call bnxt_set_rx_skb_mode() within
bnxt_change_mtu() to properly set the AGG rings configuration and
update rx_skb_func based on the new MTU value.
Additionally, BNXT_FLAG_NO_AGG_RINGS is cleared at the beginning of
bnxt_set_rx_skb_mode() to make sure it gets set or cleared based on
the current MTU.
Use the return value from bnxt_get_media() to determine the port and
link modes. bnxt_get_media() returns the proper BNXT_MEDIA_KR when
the PHY is backplane. This will correct the ethtool settings for
backplane devices.
Fixes: 5d4e1bf60664 ("bnxt_en: extend media types to supported and autoneg modes") Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com> Signed-off-by: Shravya KN <shravya.k-n@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
After successful PCIe AER recovery, FW will reset all resource
reservations. If it is IF_UP, the driver will call bnxt_open() and
all resources will be reserved again. It it is IF_DOWN, we should
call bnxt_reserve_rings() so that we can reserve resources including
RoCE resources to allow RoCE to resume after AER. Without this
patch, RoCE fails to resume in this IF_DOWN scenario.
Later, if it becomes IF_UP, bnxt_open() will see that resources have
been reserved and will not reserve again.
`atmel_qspi_reg_name()` is used for pretty-printing register offsets
for verbose logging of register accesses. However, due to a typo
(likely a copy-paste error), QSPI_RD's offset prints as "MR", the
name of the previous register. Fix this typo.
Fixes: c528ecfbef04 ("spi: atmel-quadspi: Add verbose debug facilities to monitor register accesses") Signed-off-by: Csókás, Bence <csokas.bence@prolan.hu> Reviewed-by: Alexander Dahl <ada@thorsis.com> Link: https://patch.msgid.link/20241122141302.2599636-1-csokas.bence@prolan.hu Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
During initialization, the AF driver resets all blocks. The RPM (MAC)
block and NIX block operate on a credit-based model. When the NIX block
resets during active traffic flow, it doesn't release credits to the RPM
block. This causes the RPM FIFO to overflow, leading to receive traffic
struck.
To address this issue, the patch introduces the following changes:
1. Stop receiving traffic at the MAC level during AF driver
initialization.
2. Perform an X2P reset (prevents RXFIFO of all LMACS from pushing data)
3. Reset the NIX block.
4. Clear the X2P reset and re-enable receiving traffic.
Fixes: 54d557815e15 ("octeontx2-af: Reset all RVU blocks") Signed-off-by: Hariprasad Kelam <hkelam@marvell.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The corrected words register(FCFECX_VL0_CCW_LO)/Uncorrected words
register (FCFECX_VL0_NCCW_LO) of FCFEC counter has different LMAC
offset which needs to be accessed differently.
The earlier patch sets the 'Stats control register' for RPM
receive/transmit statistics instead of RSFEC statistics,
causing the driver to return stale FEC counters.
The current design stores the FIFO length in a common structure for all
RPMs (mac_ops). As a result, the FIFO length of the last RPM is applied
to all RPMs, leading to reduced network performance.
This patch resolved the problem by storing the fifo length in per MAC
structure (cgx).
Fixes: b9d0fedc6234 ("octeontx2-af: cn10kb: Add RPM_USX MAC support") Signed-off-by: Hariprasad Kelam <hkelam@marvell.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On DWMAC3 and later, there's a RX Watchdog interrupt that's used for
interrupt coalescing. It's known to be buggy on some platforms, and
dwmac-socfpga appears to be one of them. Changing the interrupt
coalescing from ethtool doesn't appear to have any effect here.
Without disabling RIWT (Received Interrupt Watchdog Timer, I
believe...), we observe latencies while receiving traffic that amount to
around ~0.4ms. This was discovered with NTP but can be easily reproduced
with a simple ping. Without this patch :
64 bytes from 192.168.5.2: icmp_seq=1 ttl=64 time=0.657 ms
With this patch :
64 bytes from 192.168.5.2: icmp_seq=1 ttl=64 time=0.254 ms
If an optional resource is found but fails to remap, return on failure.
Avoids any potential problems when using the iomapped resource as the
assumption is that it's available.
Fixes: 23a890d493e3 ("net: mdio: Add the reset function for IPQ MDIO driver") Signed-off-by: Rosen Penev <rosenp@gmail.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://patch.msgid.link/20241121193152.8966-1-rosenp@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
RFC8981 section 3.4 says that existing temporary addresses must have their
lifetimes adjusted so that no temporary addresses should ever remain "valid"
or "preferred" longer than the incoming SLAAC Prefix Information. This would
strongly imply in Linux's case that if the "mngtmpaddr" address is deleted or
un-flagged as such, its corresponding temporary addresses must be cleared out
right away.
But now the temporary address is renewed even after ‘mngtmpaddr’ is removed
or becomes unmanaged as manage_tempaddrs() set temporary addresses
prefered/valid time to 0, and later in addrconf_verify_rtnl() all checkings
failed to remove the addresses. Fix this by deleting the temporary address
directly for these situations.
Fixes: 778964f2fdf0 ("ipv6/addrconf: fix timing bug in tempaddr regen") Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Passing MSG_PEEK flag to skb_recv_datagram() increments skb refcount
(skb->users) and iucv_sock_recvmsg() does not decrement skb refcount
at exit.
This results in skb memory leak in skb_queue_purge() and WARN_ON in
iucv_sock_destruct() during socket close. To fix this decrease
skb refcount by one if MSG_PEEK is set in order to prevent memory
leak and WARN_ON.
Unaligned direct writes are invalid and should return an error
without making any changes, rather than extending ->valid_size
and then returning an error. Therefore, alignment checking is
required before extending ->valid_size.
Fixes: 11a347fb6cef ("exfat: change to get file size from DataLength") Signed-off-by: Yuezhang Mo <Yuezhang.Mo@sony.com> Co-developed-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit under fixes extended extack reporting to dumps.
It works under normal conditions, because extack errors are
usually reported during ->start() or the first ->dump(),
it's quite rare that the dump starts okay but fails later.
If the dump does fail later, however, the input skb will
already have the initiating message pulled, so checking
if bad attr falls within skb->data will fail.
Switch the check to using nlh, which is always valid.
syzbot found a way to hit that scenario by filling up
the receive queue. In this case we initiate a dump
but don't call ->dump() until there is read space for
an skb.
Reported-by: syzbot+d4373fa8042c06cefa84@syzkaller.appspotmail.com Fixes: 8af4f60472fc ("netlink: support all extack types in dumps") Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119224432.1713040-1-kuba@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
VCAP API unit tests fail randomly with errors such as
# vcap_api_iterator_init_test: EXPECTATION FAILED at drivers/net/ethernet/microchip/vcap/vcap_api_kunit.c:387
Expected 134 + 7 == iter.offset, but
134 + 7 == 141 (0x8d)
iter.offset == 17214 (0x433e)
# vcap_api_iterator_init_test: EXPECTATION FAILED at drivers/net/ethernet/microchip/vcap/vcap_api_kunit.c:388
Expected 5 == iter.reg_idx, but
iter.reg_idx == 702 (0x2be)
# vcap_api_iterator_init_test: EXPECTATION FAILED at drivers/net/ethernet/microchip/vcap/vcap_api_kunit.c:389
Expected 11 == iter.reg_bitpos, but
iter.reg_bitpos == 15 (0xf)
# vcap_api_iterator_init_test: pass:0 fail:1 skip:0 total:1
Comments in the code state that "A typegroup table ends with an all-zero
terminator". Add the missing terminators.
Some of the typegroups did have a terminator of ".offset = 0, .width = 0,
.value = 0,". Replace those terminators with "{ }" (no trailing ',') for
consistency and to excplicitly state "this is a terminator".
Fixes: 67d637516fa9 ("net: microchip: sparx5: Adding KUNIT test for the VCAP API") Cc: Steen Hegelund <steen.hegelund@microchip.com> Signed-off-by: Guenter Roeck <linux@roeck-us.net> Reviewed-by: Daniel Machon <daniel.machon@microchip.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20241119213202.2884639-1-linux@roeck-us.net Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Validate Wake-on-LAN (WoL) options in `lan78xx_set_wol` before calling
`usb_autopm_get_interface`. This prevents USB autopm refcounting issues
and ensures the adapter can properly enter autosuspend when invalid WoL
options are provided.
The hardware on Broadcom 1G chipsets have a known limitation
where they cannot handle DMA addresses that cross over 4GB.
When such an address is encountered, the hardware sets the
address overflow error bit in the DMA status register and
triggers a reset.
However, BCM57766 hardware is setting the overflow bit and
triggering a reset in some cases when there is no actual
underlying address overflow. The hardware team analyzed the
issue and concluded that it is happening when the status
block update has an address with higher (b16 to b31) bits
as 0xffff following a previous update that had lowest bits
as 0xffff.
To work around this bug in the BCM57766 hardware, set the
coherent dma mask from the current 64b to 31b. This will
ensure that upper bits of the status block DMA address are
always at most 0x7fff, thus avoiding the improper overflow
check described above. This work around is intended for only
status block and ring memories and has no effect on TX and
RX buffers as they do not require coherent memory.
Add calls to `phy_device_free` after `fixed_phy_unregister` to fix a
memory leak that occurs when the device is unplugged. This ensures
proper cleanup of pseudo fixed-link PHYs.
Fixes: 89b36fb5e532 ("lan78xx: Lan7801 Support for Fixed PHY") Cc: Raghuram Chary J <raghuramchary.jallipalli@microchip.com> Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de> Link: https://patch.msgid.link/20241116130558.1352230-2-o.rempel@pengutronix.de Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In lan78xx_probe(), the buffer `buf` was being freed twice: once
implicitly through `usb_free_urb(dev->urb_intr)` with the
`URB_FREE_BUFFER` flag and again explicitly by `kfree(buf)`. This caused
a double free issue.
To resolve this, reordered `kmalloc()` and `usb_alloc_urb()` calls to
simplify the initialization sequence and removed the redundant
`kfree(buf)`. Now, `buf` is allocated after `usb_alloc_urb()`, ensuring
it is correctly managed by `usb_fill_int_urb()` and freed by
`usb_free_urb()` as intended.
Correct bq27426 registers, according to technical reference manual
it does not have Design Capacity register so it is not register
compatible with bq27421.
Fixes: 5ef6a16033b47 ("power: supply: bq27xxx: Add support for BQ27426") Signed-off-by: Barnabás Czémán <barnabas.czeman@mainlining.org> Link: https://lore.kernel.org/r/20241016-fix_bq27426-v2-1-aa6c0f51a9f6@mainlining.org Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The put_device() call in power_supply_put() may call
power_supply_dev_release(). The latter function does not sleep so
power_supply_put() doesn't sleep either. Hence, remove the might_sleep()
call from power_supply_put(). This patch suppresses false positive
complaints about calling a sleeping function from atomic context if
power_supply_put() is called from atomic context.
Cc: Kyle Tso <kyletso@google.com> Cc: Krzysztof Kozlowski <krzk@kernel.org> Fixes: 1a352462b537 ("power_supply: Add power_supply_put for decrementing device reference counter") Signed-off-by: Bart Van Assche <bvanassche@acm.org> Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Link: https://lore.kernel.org/r/20240917193914.47566-1-bvanassche@acm.org Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
There are no failed test cases compiled with the lower version of GCC
such as 13.3.0, while the problems only appear with higher version of
GCC such as 14.2.0.
This is because the problems were hidden by the lower version of GCC due
to redundant sign extension instructions generated by compiler, but with
optimization of higher version of GCC, the sign extension instructions
have been removed.
(4) Root Cause Analysis:
The LoongArch architecture does not expose sub-registers, and hold all
32-bit values in a sign-extended format. While BPF, on the other hand,
exposes sub-registers, and use zero-extension (similar to arm64/x86).
This has led to some subtle bugs, where a BPF JITted program has not
sign-extended the a0 register (return value in LoongArch land), passed
the return value up the kernel, for example:
| int from_bpf(void);
|
| long foo(void)
| {
| return from_bpf();
| }
Here, a0 would be 0xffffffff instead of the expected 0xffffffffffffffff.
Internally, the LoongArch JIT uses a5 as a dedicated register for BPF
return values. That is to say, the LoongArch BPF uses a5 for BPF return
values, which are zero-extended, whereas the LoongArch ABI uses a0 which
is sign-extended.
(5) Final Solution:
Keep a5 zero-extended, but explicitly sign-extend a0 (which is used
outside BPF land). Because libbpf currently defines the return value
of an ebpf program as a 32-bit unsigned integer, just use addi.w to
extend bit 31 into bits 63 through 32 of a5 to a0. This is similar to
commit 2f1b0d3d7331 ("riscv, bpf: Sign-extend return values").
Whenever I try to build the kernel with upcoming GCC 15 which defaults
to -std=gnu23 I get a build failure:
CC arch/loongarch/vdso/vgetcpu.o
In file included from ./include/uapi/linux/posix_types.h:5,
from ./include/uapi/linux/types.h:14,
from ./include/linux/types.h:6,
from ./include/linux/kasan-checks.h:5,
from ./include/asm-generic/rwonce.h:26,
from ./arch/loongarch/include/generated/asm/rwonce.h:1,
from ./include/linux/compiler.h:317,
from ./include/asm-generic/bug.h:5,
from ./arch/loongarch/include/asm/bug.h:60,
from ./include/linux/bug.h:5,
from ./include/linux/mmdebug.h:5,
from ./include/linux/mm.h:6,
from ./arch/loongarch/include/asm/vdso.h:10,
from arch/loongarch/vdso/vgetcpu.c:6:
./include/linux/stddef.h:11:9: error: expected identifier before 'false'
11 | false = 0,
| ^~~~~
./include/linux/types.h:35:33: error: two or more data types in declaration specifiers
35 | typedef _Bool bool;
| ^~~~
./include/linux/types.h:35:1: warning: useless type name in empty declaration
35 | typedef _Bool bool;
| ^~~~~~~
The kernel builds explicitly with -std=gnu11 in top Makefile, but
arch/loongarch/vdso does not use KBUILD_CFLAGS from the rest of the
kernel, just add -std=gnu11 flag to arch/loongarch/vdso/Makefile.
By the way, commit e8c07082a810 ("Kbuild: move to -std=gnu11") did a
similar change for arch/arm64/kernel/vdso32/Makefile.
Fixes: c6b99bed6b8f ("LoongArch: Add VDSO and VSYSCALL support") Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn> Signed-off-by: Huacai Chen <chenhuacai@loongson.cn> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add the missing 'name' parameter to the mount_api documentation for
fs_validate_description().
Fixes: 96cafb9ccb15 ("fs_parser: remove fs_parameter_description name field") Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Link: https://lore.kernel.org/r/20241125215021.231758-1-rdunlap@infradead.org Cc: Eric Sandeen <sandeen@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christian Brauner <brauner@kernel.org> Cc: Jan Kara <jack@suse.cz> Cc: Jonathan Corbet <corbet@lwn.net> Cc: linux-doc@vger.kernel.org Signed-off-by: Christian Brauner <brauner@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
There are cases where a PCIe extended capability should be hidden from
the user. For example, an unknown capability (i.e., capability with ID
greater than PCI_EXT_CAP_ID_MAX) or a capability that is intentionally
chosen to be hidden from the user.
Hiding a capability is done by virtualizing and modifying the 'Next
Capability Offset' field of the previous capability so it points to the
capability after the one that should be hidden.
The special case where the first capability in the list should be hidden
is handled differently because there is no previous capability that can
be modified. In this case, the capability ID and version are zeroed
while leaving the next pointer intact. This hides the capability and
leaves an anchor for the rest of the capability list.
However, today, hiding the first capability in the list is not done
properly if the capability is unknown, as struct
vfio_pci_core_device->pci_config_map is set to the capability ID during
initialization but the capability ID is not properly checked later when
used in vfio_config_do_rw(). This leads to the following warning [1] and
to an out-of-bounds access to ecap_perms array.
Fix it by checking cap_id in vfio_config_do_rw(), and if it is greater
than PCI_EXT_CAP_ID_MAX, use an alternative struct perm_bits for direct
read only access instead of the ecap_perms array.
Note that this is safe since the above is the only case where cap_id can
exceed PCI_EXT_CAP_ID_MAX (except for the special capabilities, which
are already checked before).
Currently the mount_setattr_test fails on machines with a 64K PAGE_SIZE,
with errors such as:
# RUN mount_setattr_idmapped.invalid_fd_negative ...
mkfs.ext4: No space left on device while writing out and closing file system
# mount_setattr_test.c:1055:invalid_fd_negative:Expected system("mkfs.ext4 -q /mnt/C/ext4.img") (256) == 0 (0)
# invalid_fd_negative: Test terminated by assertion
# FAIL mount_setattr_idmapped.invalid_fd_negative
not ok 12 mount_setattr_idmapped.invalid_fd_negative
At first glance it seems like that should never work, after all 2MB is
larger than 100,000 bytes. However the filesystem image doesn't actually
occupy 2MB on "disk" (actually RAM, due to tmpfs). On 4K kernels the
ext4.img uses ~84KB of actual space (according to du), which just fits.
However on 64K PAGE_SIZE kernels the ext4.img takes at least 256KB,
which is too large to fit in the tmpfs, hence the errors.
It seems fraught to rely on the ext4.img taking less space on disk than
the allocated size, so instead create the tmpfs with a size of 2MB. With
that all 21 tests pass on 64K PAGE_SIZE kernels.
Fix unwind flows in mlx5vf_pci_save_device_data() and
mlx5vf_pci_resume_device_data() to avoid freeing the migf pointer at the
'end' label, as this will be handled by fput(migf->filp) through
mlx5vf_release_file().
To ensure mlx5vf_release_file() functions correctly, move the
initialization of migf fields (such as migf->lock) to occur before any
potential unwind flow, as these fields may be accessed within
mlx5vf_release_file().
The starting iova address to iterate iotlb map entry within a range
was set to an irrelevant value when passing to the itree_next()
iterator, although luckily it doesn't affect the outcome of finding
out the granule of the smallest iotlb map size. Fix the code to make
it consistent with the following for-loop.
Fixes: 94abbccdf291 ("vdpa/mlx5: Add shared memory registration code") Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com> Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Message-Id: <20241021134040.975221-3-dtatulea@nvidia.com> Signed-off-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Jason Wang <jasowang@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix the following register definitions for REG_CSR_2L_RX{0,1}_REV0
registers:
- CSR_2L_PXP_VOS_PNINV
- CSR_2L_PXP_FE_GAIN_NORMAL_MODE
- CSR_2L_PXP_FE_GAIN_TRAIN_MODE
Commit 120584c728a6 ("hwmon: (aquacomputer_d5next) Add support for Octo
flow sensor") added support for reading Octo flow sensor, but didn't
update the priv->speed_input array length. Since Octo has 8 fans, with
the addition of the flow sensor the proper length for speed_input is 9.
Reported by Arne Schwabe on Github [1], who received a UBSAN warning.
Fixes: 120584c728a6 ("hwmon: (aquacomputer_d5next) Add support for Octo flow sensor") Closes: https://github.com/aleksamagicka/aquacomputer_d5next-hwmon/issues/100 [1] Reported-by: Arne Schwabe <arne@rfc2549.org> Signed-off-by: Aleksa Savic <savicaleksa83@gmail.com>
Message-ID: <20241124152725.7205-1-savicaleksa83@gmail.com> Signed-off-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
