When the ene device is detaching, function ene_remove() will
be called. But there is no function to cancel tx_sim_timer
in ene_remove(), the timer handler ene_tx_irqsim() could race
with ene_remove(). As a result, the UAF bugs could happen,
the process is shown below.
Fix by adding del_timer_sync(&dev->tx_sim_timer) in ene_remove(),
The tx_sim_timer could stop before ene device is deallocated.
What's more, The rc_unregister_device() and del_timer_sync()
should be called first in ene_remove() and the deallocated
functions such as free_irq(), release_region() and so on
should be called behind them. Because the rc_unregister_device()
is well synchronized. Otherwise, race conditions may happen. The
situations that may lead to race conditions are shown below.
Firstly, the rx receiver is disabled with ene_rx_disable()
before rc_unregister_device() in ene_remove(), which means it
can be enabled again if a process opens /dev/lirc0 between
ene_rx_disable() and rc_unregister_device().
Secondly, the irqaction descriptor is freed by free_irq()
before the rc device is unregistered, which means irqaction
descriptor may be accessed again after it is deallocated.
Thirdly, the timer can call ene_tx_sample() that can write
to the io ports, which means the io ports could be accessed
again after they are deallocated by release_region().
Therefore, the rc_unregister_device() and del_timer_sync()
should be called first in ene_remove().
in the E.2.1 of Rec. ITU-T H.264 (06/2019),
0 of colour primaries is reserved, and 2 is unspecified.
driver can map V4L2_COLORSPACE_LAST to 0,
and map V4L2_COLORSPACE_DEFAULT to 2.
v4l2_xfer_func and v4l2_ycbcr_encoding are similar case.
using the api of clk_bulk can simplify the code.
and the clock of the jpeg codec may be changed,
the clk_bulk api can be compatible with the future change.
Fixes: 605d96ce1793 ("media: imx-jpeg: Add pm-runtime support for imx-jpeg") Signed-off-by: Ming Qian <ming.qian@nxp.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since f497f696d2c9a, enumerating framesize on format set with "MODE_NONE"
(any raw formats) is reporting an invalid frmsize.
Size: Stepwise 0x0 - 0x0 with step 0/0
Before this change, the driver would return EINVAL, which is also invalid
but worked in GStreamer. The original intent was not to implement it, hence
the -ENOTTY return in this change. While drivers should implement
ENUM_FRMSIZE for all formats and queues, this change is limited in scope to
fix the regression.
This fixes taking picture in Gnome Cheese software, or any software using
GSteamer to encode JPEG with hardware acceleration.
Fixes: f497f696d2c9 ("media: hantro: Use post processor scaling capacities") Reported-by: Robert Mader <robert.mader@collabora.com> Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com> Reviewed-by: Benjamin Gaignard <benjamin.gaignard@collabora.com> Reviewed-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar> Tested-by: Robert Mader <robert.mader@collabora.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The legal identifier of APP14 is "Adobe\0",
but sometimes it may be
"This is an unknown APP marker . Compliant decoders must ignore it."
In this case, just ignore it.
It won't affect the decode result.
Fixes: 403c6775b6f6 ("media: Add parsing for APP14 data segment in jpeg helpers") Signed-off-by: Ming Qian <ming.qian@nxp.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The new mdp3 driver uses 'select' to force-enable a couple of drivers
it depends on. This is error-prone and likely to cause dependency
loops as well as warnings like:
This specific warning was already addressed in a previous patch,
but there are similar unnecessary 'select' statements, so turn those
into 'depends on'. This also means the dependency on ARCH_MEDIATEK
is redundant and can be dropped.
Since REMOTEPROC completely replaces the VIDEO_MEDIATEK_VPU in MDP3,
unused config should be removed to avoid compilation warnings
reported on i386 or x86_64.
Marking a case of the switch statement as unreachable means the
compiler treats it as undefined behavior, which is then caught by
an objtool warning:
drivers/media/platform/qcom/camss/camss-csiphy-3ph-1-0.o: warning: objtool: csiphy_lanes_enable() falls through to next function csiphy_lanes_disable()
Instead of simply continuing execution at a random place of the
driver, print a warning and return from to the caller, which
makes it possible to understand what happens and avoids the
warning.
Fixes: 9017fa569b16 ("media: camss: csiphy-3ph: add support for SM8250 CSI DPHY") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Robert Foss <robert.foss@linaro.org> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In case of error, the function mtk_mutex_get()
returns ERR_PTR() and never returns NULL. The NULL test in the
return value check should be replaced with IS_ERR().
And also fix the err_free_mutex case.
Fixes: 0eba8ed8044f ("media: platform: mtk-mdp3: add MediaTek MDP3 driver") Signed-off-by: Qiheng Lin <linqiheng@huawei.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
2x2 binning works fine for RAW10 capture, but for RAW8 1232p mode it
leads to corrupted frames [1][2].
Using the special 2x2 analog binning mode fixes the issue, but causes
artefacts for RAW10 1232p capture. So here we choose the binning mode
depending upon the frame format selected.
As both binning modes work fine for 480p RAW8 and RAW10 capture, it can
share the same code path as 1232p for selecting binning mode.
There are four modes, and each mode has a table of registers.
Some of the registers are common to all modes, so create new
tables for these common registers to reduce duplicate code.
Signed-off-by: Adam Ford <aford173@gmail.com> Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Stable-dep-of: 33b2fbe4cbe1 ("media: i2c: imx219: Fix binning for RAW8 capture") Signed-off-by: Sasha Levin <sashal@kernel.org>
The reason is that if priv->hdl.error is set, ov772x_probe() jumps to the
error_mutex_destroy without doing v4l2_ctrl_handler_free(), and all
resources allocated in v4l2_ctrl_handler_init() and v4l2_ctrl_new_std()
are leaked.
Various functions access the media_device from a pad by going through
the entity the pad belongs to. Remove the level of indirection and get
the media_device from the pad directly.
Fixes: b34dbe1ba23b ("media: mc: convert pipeline funcs to take media_pad") Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some module manufacturers [1][2] don't expose the RESETB and PWDN pins
of the sensor directly through the 15-pin FFC connector. Instead wiring
~PWDN gpio to the sensor pins with appropriate delays.
In such cases, reset_gpio will not be available to the driver, but it
will still be toggled when the sensor is powered on, and thus we should
still honor the wait time of >= 5ms + 1ms + 20ms (see figure 2-3 in [3])
before attempting any i/o operations over SCCB.
Also, rename the function to ov5640_powerup_sequence to better match the
datasheet (section 2.7).
Move the register-based reset out of the init_setting[] and into the
powerup_sequence function. The sensor is power cycled and reset using
the gpio pins so the soft reset is not always necessary.
This also ensures that soft reset honors the timing sequence
from the datasheet [1].
ov5675_init_controls() won't clean all the allocated resources in fail
path, which may causes the memleaks. Add v4l2_ctrl_handler_free() to
prevent memleak.
Fixes: cb63633b9e8c ("media: ov5675: Add support for OV5675 sensor") Signed-off-by: Shang XiaoJing <shangxiaojing@huawei.com> Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
ov2740_init_controls() won't clean all the allocated resources in fail
path, which may causes the memleaks. Add v4l2_ctrl_handler_free() to
prevent memleak.
max9286_v4l2_register() calls v4l2_ctrl_new_std(), but won't free the
created v412_ctrl when fwnode_graph_get_endpoint_by_id() failed, which
causes the memleak. Call v4l2_ctrl_handler_free() to free the v412_ctrl.
For each binary Debian package, a directory with the package name is
created in the debian directory. Correct the generated file matches in the
package's clean target, which were renamed without adjusting the target.
Fixes: 6ad3601e8bf9 ("builddeb: match temporary directory name to the package name") Signed-off-by: Bastian Germann <bage@linutronix.de> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When clang's -Qunused-arguments is dropped from KBUILD_CPPFLAGS, it
points out that there is a linking phase flag added to CFLAGS, which
will only be used for compiling
clang-16: error: argument unused during compilation: '-shared' [-Werror,-Wunused-command-line-argument]
'-shared' is already present in ldflags-y so it can just be dropped.
Fixes: cd55c5f2b221 ("s390/vdso: Use $(LD) instead of $(CC) to link vDSO") Signed-off-by: Nathan Chancellor <nathan@kernel.org> Acked-by: Heiko Carstens <hca@linux.ibm.com> Reviewed-by: Sven Schnelle <svens@linux.ibm.com> Tested-by: Linux Kernel Functional Testing <lkft@linaro.org> Tested-by: Anders Roxell <anders.roxell@linaro.org> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
This was likely supposed to be '-Wa,-a$(BITS)'. However, this change is
unnecessary, as all supported versions of clang and gcc will pass '-a64'
or '-a32' to GNU as based on the value of '-m'; the behavior of the
latest stable release of the oldest supported major version of each
compiler is shown below and each compiler's latest release exhibits the
same behavior (GCC 12.2.0 and Clang 15.0.6).
$ powerpc64-linux-gcc --version | head -1
powerpc64-linux-gcc (GCC) 5.5.0
The memory of ctx is allocated in cal_ctx_create(), but it will
not be freed when cal_ctx_v4l2_init() fails, so add kfree() when
cal_ctx_v4l2_init() fails to fix it.
Fixes: 463ac99c12f0 ("media: ti-vpe: cal: Split video device initialization and registration") Signed-off-by: Gaosheng Cui <cuigaosheng1@huawei.com> Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Any access to the dynamically allocated metadata region by the application
processor after assigning it to the remote Q6 will result in a XPU
violation. Fix this by replacing the dynamically allocated memory region
with a no-map carveout and unmap the modem metadata memory region before
passing control to the remote Q6.
Reported-and-tested-by: Amit Pundir <amit.pundir@linaro.org> Fixes: a2ed97bf0e19 ("remoteproc: qcom: Make secure world call for mem ownership switch") Signed-off-by: Sibi Sankar <quic_sibis@quicinc.com> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Signed-off-by: Bjorn Andersson <andersson@kernel.org> Link: https://lore.kernel.org/r/20230117085840.32356-7-quic_sibis@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit manages to do three API violations at once:
- dereference the return value of dma_alloc_attrs with the
DMA_ATTR_NO_KERNEL_MAPPING mapping, which is clearly forbidden and
will do the wrong thing on various dma mapping implementations. The
fact that dma-direct uses a struct page as a cookie is an undocumented
implementation detail
- include dma-map-ops.h and use pgprot_dmacoherent despite a clear
comment documenting that this is not acceptable
- use of the VM_DMA_COHERENT for something that is not the dma-mapping
code
- use of VM_FLUSH_RESET_PERMS for vmap, while it is only supported for
vmalloc
Acked-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sibi Sankar <quic_sibis@quicinc.com> Signed-off-by: Bjorn Andersson <andersson@kernel.org> Link: https://lore.kernel.org/r/20230117085840.32356-6-quic_sibis@quicinc.com
Stable-dep-of: 4e3cc27d2efd ("remoteproc: qcom_q6v5_mss: Use a carveout to authenticate modem headers") Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix three sources of error involving struct sdma_txreq.num_descs.
When _extend_sdma_tx_descs() extends the descriptor array, it uses the
value of tx->num_descs to determine how many existing entries from the
tx's original, internal descriptor array to copy to the newly allocated
one. As this value was incremented before the call, the copy loop will
access one entry past the internal descriptor array, copying its contents
into the corresponding slot in the new array.
If the call to _extend_sdma_tx_descs() fails, _pad_smda_tx_descs() then
invokes __sdma_tx_clean() which uses the value of tx->num_desc to drive a
loop that unmaps all descriptor entries in use. As this value was
incremented before the call, the unmap loop will invoke sdma_unmap_desc()
on a descriptor entry whose contents consist of whatever random data was
copied into it during (1), leading to cascading further calls into the
kernel and driver using arbitrary data.
_sdma_close_tx() was using tx->num_descs instead of tx->num_descs - 1.
Fix all of the above by:
- Only increment .num_descs after .descp is extended.
- Use .num_descs - 1 instead of .num_descs for last .descp entry.
Fixes: f0b965fe4eaa ("staging/rdma/hfi1: Add coalescing support for SDMA TX descriptors") Link: https://lore.kernel.org/r/167656658879.2223096.10026561343022570690.stgit@awfm-02.cornelisnetworks.com Signed-off-by: Brendan Cunningham <bcunningham@cornelisnetworks.com> Signed-off-by: Patrick Kelsey <pat.kelsey@cornelisnetworks.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix arithmetic and logic errors in hfi1_can_pin_pages() that would allow
hfi1 to attempt pinning pages in cases where it should not because of
resource limits or lack of required capability.
Fixes: e25d07a42e77 ("IB/hfi1: Add pin query function") Link: https://lore.kernel.org/r/167656658362.2223096.10954762619837718026.stgit@awfm-02.cornelisnetworks.com Signed-off-by: Brendan Cunningham <bcunningham@cornelisnetworks.com> Signed-off-by: Patrick Kelsey <pat.kelsey@cornelisnetworks.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
An earlier patch which introduced smp_load_acquire/smp_store_release
into rxe_queue.h incorrectly assumed that surrounding spin-locks in
rxe_verbs.c around queue updates for kernel ulps was sufficient to
protect the passing of data through the queues between the ulp and
the rxe tasklets. But this was incorrect. The typical sequence was
queue_head() calls queue_empty() which calls smp_load_acquire()
For user space apps queue_advance_producer() calls smp_store_release()
so that there is a memory barrier between the producer and the
consumer but for kernel ulps queue_advance_produce() just incremented
the producer index because the lock function is a release function.
But to work the barrier has to come between filling in the wqe and
updating the producer index. This patch adds the missing barriers.
It also changes the enum names for the ulp queue types to
QUEUE_TYPE_FROM/TO_ULP instead of QUEUE_TYPE_TO/FROM_DRIVER
which is very ambiguous. This bug is suspected as the cause of very
rare lockups in a very high scale storage application. It is a bug
in any case and should be corrected.
Fixes: f90a51b1dfaf ("RDMA/rxe: Protect user space index loads/stores") Link: https://lore.kernel.org/r/20230214071053.5395-1-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit b799339cf7c0 ("iommu/vt-d: Do not use flush-queue when caching-mode
is on") forced default domains to be strict mode as long as IOMMU
caching-mode is flagged. The reason for doing this is that when vIOMMU
uses VT-d caching mode to synchronize shadowing page tables, the strict
mode shows better performance.
However, this optimization is orthogonal to the first-level page table
because the Intel VT-d architecture does not define the caching mode of
the first-level page table. Refer to VT-d spec, section 6.1, "When the
CM field is reported as Set, any software updates to remapping
structures other than first-stage mapping (including updates to not-
present entries or present entries whose programming resulted in
translation faults) requires explicit invalidation of the caches."
Exclude the first-level page table from this optimization.
Generally using first-stage translation in vIOMMU implies nested
translation enabled in the physical IOMMU. In this case the first-stage
page table is wholly captured by the guest. The vIOMMU only needs to
transfer the cache invalidations on vIOMMU to the physical IOMMU.
Forcing the default domain to strict mode will cause more frequent
cache invalidations, resulting in performance degradation. In a real
performance benchmark test measured by iperf receive, the performance
result on Sapphire Rapids 100Gb NIC shows:
w/ this fix ~51 Gbits/s, w/o this fix ~39.3 Gbits/s.
Theoretically a first-stage IOMMU page table can still be shadowed
in absence of the caching mode, e.g. with host write-protecting guest
IOMMU page table to synchronize changed PTEs with the physical
IOMMU page table. In this case the shadowing overhead is decoupled
from emulating IOTLB invalidation then the overhead of the latter part
is solely decided by the frequency of IOTLB invalidations. Hence
allowing guest default dma domain to be lazy can also benefit the
overall performance by reducing the total VM-exit numbers.
Fixes: b799339cf7c0 ("iommu/vt-d: Do not use flush-queue when caching-mode is on") Reported-by: Sanjay Kumar <sanjay.k.kumar@intel.com> Suggested-by: Sanjay Kumar <sanjay.k.kumar@intel.com> Signed-off-by: Tina Zhang <tina.zhang@intel.com> Reviewed-by: Kevin Tian <kevin.tian@intel.com> Link: https://lore.kernel.org/r/20230214025618.2292889-1-tina.zhang@intel.com Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com> Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Resolves a panic that can occur on AMD systems, typically during host
shutdown, after the PTDMA driver had been exercised. The issue was
the pt_issue_pending() function is mistakenly assuming that there will
be at least one descriptor in the Submitted queue when the function
is called. However, it is possible that both the Submitted and Issued
queues could be empty, which could result in pt_cmd_callback() being
mistakenly called with a NULL pointer.
Ref: Bugzilla Bug 216856.
Commit 9e0719624a40 ("dmaengine: sf-pdma: Add multithread support for a
DMA channel") changed sf_pdma_prep_dma_memcpy() to unconditionally
allocate a new sf_pdma_desc each time it is called.
The driver previously recycled descs, by checking the in_use flag, only
allocating additional descs if the existing one was in use. This logic
was removed in commit 9e0719624a40 ("dmaengine: sf-pdma: Add multithread
support for a DMA channel"), but sf_pdma_free_desc() was not changed to
handle the new behaviour.
As a result, each time sf_pdma_prep_dma_memcpy() is called, the previous
descriptor is leaked, over time leading to memory starvation:
Flow:
- Booted system with SNP enabled, memory encryption off and
IOMMU DMA translation mode
- AMD driver detects v2 capable device and amd_iommu_def_domain_type()
returns identity mode
- amd_iommu_domain_alloc() returns NULL an SNP is enabled
- System will fail to register device
On SNP enabled system, passthrough mode is not supported. IOMMU default
domain is set to translation mode. We need to return zero from
amd_iommu_def_domain_type() so that it allocates translation domain.
Fixes: ad377ef0c13a ("iommu/amd: Introduce function to check and enable SNP") CC: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com> Signed-off-by: Vasant Hegde <vasant.hegde@amd.com> Link: https://lore.kernel.org/r/20230207091752.7656-1-vasant.hegde@amd.com Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
This condition needs to match the previous "if (epcp->state == LISTEN) {"
exactly to avoid a NULL dereference of either "listen_ep" or "ep". The
problem is that "epcp" has been re-assigned so just testing
"if (epcp->state == LISTEN) {" a second time is not sufficient.
Fixes: f5ab704905b0 ("iw_cxgb4: provide detailed provider-specific CM_ID information") Signed-off-by: Dan Carpenter <error27@gmail.com> Link: https://lore.kernel.org/r/Y+usKuWIKr4dimZh@kili Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The existing logic in tcphy_get_mode() can cause the phy to be
incorrectly configured to USB UFP or DisplayPort mode when
extcon_get_state returns an error code.
extcon_get_state() can return 0, 1, or a negative error code.
It is possible to get into the failing state with an extcon driver
which does not support the extcon connector id specified as the
second argument to extcon_get_state().
When coalescing two resources for a contiguous aperture, the second
resource is enlarged to cover the full contiguous range, while the first
resource is marked invalid. This invalidation is done by clearing the
flags, start, and end members.
When adding the initial resources to the bus later, invalid resources are
skipped. Unfortunately, the check for an invalid resource considers only
the end member, causing false positives.
E.g. on r8a7791/koelsch, root bus resource 0 ("bus 00") is skipped, and no
longer registered with pci_bus_insert_busn_res() (causing the memory leak),
nor printed:
pci-rcar-gen2 ee090000.pci: host bridge /soc/pci@ee090000 ranges:
pci-rcar-gen2 ee090000.pci: MEM 0x00ee080000..0x00ee08ffff -> 0x00ee080000
pci-rcar-gen2 ee090000.pci: PCI: revision 11
pci-rcar-gen2 ee090000.pci: PCI host bridge to bus 0000:00
-pci_bus 0000:00: root bus resource [bus 00]
pci_bus 0000:00: root bus resource [mem 0xee080000-0xee08ffff]
Fix this by only skipping resources where all of the flags, start, and end
members are zero.
These "nvmem-cells" platform devices never get probed because there's no
platform driver for it and it's never used anywhere else. So it's a
waste of memory. These devices also cause fw_devlink to block nvmem
consumers of "nvmem-cells" partition from probing because the supplier
device never probes.
So stop creating platform devices for nvmem-cells partitions to avoid
wasting memory and to avoid blocking probing of consumers.
fw_devlink could only detect a single and simple cycle because it relied
mainly on device link cycle detection code that only checked for cycles
between devices. The expectation was that the firmware wouldn't have
complicated cycles and multiple cycles between devices. That expectation
has been proven to be wrong.
And this is without including parent child dependencies or nodes in the
cycle that are just firmware nodes that'll never have a struct device
created for them.
The proper way to treat these devices it to not force any probe ordering
between them, while still enforce dependencies between node in the
cycles (A, B and C) and their consumers.
So this patch goes all out and just deals with all types of cycles. It
does this by:
1. Following dependencies across device links, parent-child and fwnode
links.
2. When it find cycles, it mark the device links and fwnode links as
such instead of just deleting them or making the indistinguishable
from proxy SYNC_STATE_ONLY device links.
This way, when new nodes get added, we can immediately find and mark any
new cycles whether the new node is a device or firmware node.
fw_devlink shouldn't defer the probe of a device to wait on a supplier
that'll never have a struct device or will never be probed by a driver.
We currently check if a supplier falls into this category, but don't
check its ancestors. We need to check the ancestors too because if the
ancestor will never probe, then the supplier will never probe either.
Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Colin Foster <colin.foster@in-advantage.com> Tested-by: Sudeep Holla <sudeep.holla@arm.com> Tested-by: Douglas Anderson <dianders@chromium.org> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Tested-by: Luca Weiss <luca.weiss@fairphone.com> # qcom/sm7225-fairphone-fp4 Link: https://lore.kernel.org/r/20230207014207.1678715-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stable-dep-of: 9a2a98c0e4dc ("driver core: fw_devlink: Make cycle detection more robust") Signed-off-by: Sasha Levin <sashal@kernel.org>
To improve detection and handling of dependency cycles, we need to be
able to mark fwnode links as being part of cycles. fwnode links marked
as being part of a cycle should not block their consumers from probing.
When a device X is bound successfully to a driver, if it has a child
firmware node Y that doesn't have a struct device created by then, we
delete fwnode links where the child firmware node Y is the supplier. We
did this to avoid blocking the consumers of the child firmware node Y
from deferring probe indefinitely.
While that a step in the right direction, it's better to make the
consumers of the child firmware node Y to be consumers of the device X
because device X is probably implementing whatever functionality is
represented by child firmware node Y. By doing this, we capture the
device dependencies more accurately and ensure better
probe/suspend/resume ordering.
Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Colin Foster <colin.foster@in-advantage.com> Tested-by: Sudeep Holla <sudeep.holla@arm.com> Tested-by: Douglas Anderson <dianders@chromium.org> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Tested-by: Luca Weiss <luca.weiss@fairphone.com> # qcom/sm7225-fairphone-fp4 Link: https://lore.kernel.org/r/20230207014207.1678715-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stable-dep-of: 943adb0b55d2 ("driver core: fw_devlink: Allow marking a fwnode link as being part of a cycle") Signed-off-by: Sasha Levin <sashal@kernel.org>
fw_devlink uses DL_FLAG_SYNC_STATE_ONLY device link flag for two
purposes:
1. To allow a parent device to proxy its child device's dependency on a
supplier so that the supplier doesn't get its sync_state() callback
before the child device/consumer can be added and probed. In this
usage scenario, we need to ignore cycles for ensure correctness of
sync_state() callbacks.
2. When there are dependency cycles in firmware, we don't know which of
those dependencies are valid. So, we have to ignore them all wrt
probe ordering while still making sure the sync_state() callbacks
come correctly.
However, when detecting dependency cycles, there can be multiple
dependency cycles between two devices that we need to detect. For
example:
A -> B -> A and A -> C -> B -> A.
To detect multiple cycles correct, we need to be able to differentiate
DL_FLAG_SYNC_STATE_ONLY device links used for (1) vs (2) above.
To allow this differentiation, add a DL_FLAG_CYCLE that can be use to
mark use case (2). We can then use the DL_FLAG_CYCLE to decide which
DL_FLAG_SYNC_STATE_ONLY device links to follow when looking for
dependency cycles.
There maybe pending USR interrupt before requesting irq, however
uart_add_one_port has not executed, so there will be kernel panic:
[ 0.795668] Unable to handle kernel NULL pointer dereference at virtual addre
ss 0000000000000080
[ 0.802701] Mem abort info:
[ 0.805367] ESR = 0x0000000096000004
[ 0.808950] EC = 0x25: DABT (current EL), IL = 32 bits
[ 0.814033] SET = 0, FnV = 0
[ 0.816950] EA = 0, S1PTW = 0
[ 0.819950] FSC = 0x04: level 0 translation fault
[ 0.824617] Data abort info:
[ 0.827367] ISV = 0, ISS = 0x00000004
[ 0.831033] CM = 0, WnR = 0
[ 0.833866] [0000000000000080] user address but active_mm is swapper
[ 0.839951] Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP
[ 0.845953] Modules linked in:
[ 0.848869] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 6.1.1+g56321e101aca #1
[ 0.855617] Hardware name: Freescale i.MX8MP EVK (DT)
[ 0.860452] pstate: 000000c5 (nzcv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 0.867117] pc : __imx_uart_rxint.constprop.0+0x11c/0x2c0
[ 0.872283] lr : imx_uart_int+0xf8/0x1ec
The issue only happends in the inmate linux when Jailhouse hypervisor
enabled. The test procedure is:
while true; do
jailhouse enable imx8mp.cell
jailhouse cell linux xxxx
sleep 10
jailhouse cell destroy 1
jailhouse disable
sleep 5
done
And during the upper test, press keys to the 2nd linux console.
When `jailhouse cell destroy 1`, the 2nd linux has no chance to put
the uart to a quiese state, so USR1/2 may has pending interrupts. Then
when `jailhosue cell linux xx` to start 2nd linux again, the issue
trigger.
In order to disable irqs before requesting them, both UCR1 and UCR2 irqs
should be disabled, so here fix that, disable the Ageing Timer interrupt
in UCR2 as UCR1 does.
Fixes: 53501201fa62 ("serial: imx: Disable irqs before requesting them") Suggested-by: Sherry Sun <sherry.sun@nxp.com> Reviewed-by: Sherry Sun <sherry.sun@nxp.com> Signed-off-by: Peng Fan <peng.fan@nxp.com> Acked-by: Jason Liu <jason.hui.liu@nxp.com> Link: https://lore.kernel.org/r/20230206013016.29352-1-sherry.sun@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The default polarity of RS485 DE signal is active high. This driver does
not handle such case properly. Currently, when a pin is multiplexed as a
UART CTS_B on boot, this pin is pulled HIGH by the i.MX UART CTS circuit,
which activates DE signal on the RS485 transceiver and thus behave as if
the RS485 was transmitting data, so the system blocks the RS485 bus when
it starts and until user application takes over. This behavior is not OK.
The problem consists of two separate parts.
First, the i.MX UART IP requires UCR1 UARTEN and UCR2 RXEN to be set for
UCR2 CTSC and CTS bits to have any effect. The UCR2 CTSC bit permits the
driver to set CTS (RTS_B or RS485 DE signal) to either level sychronous
to the internal UART IP clock. Compared to other options, like GPIO CTS
control, this has the benefit of being synchronous to the UART IP clock
and thus without glitches or bus delays. The reason for the CTS design
is likely because when the Receiver is disabled, the UART IP can never
indicate that it is ready to receive data by assering CTS signal, so
the CTS is always pulled HIGH by default.
When the port is closed by user space, imx_uart_stop_rx() clears UCR2
RXEN bit, and imx_uart_shutdown() clears UCR1 UARTEN bit. This disables
UART Receiver and UART itself, and forces CTS signal HIGH, which leads
to the RS485 bus being blocked because RS485 DE is incorrectly active.
The proposed solution for this problem is to keep the Receiver running
even after the port is closed, but in loopback mode. This disconnects
the RX FIFO input from the RXD external signal, and since UCR2 TXEN is
cleared, the UART Transmitter is disabled, so nothing can feed data in
the RX FIFO. Because the Receiver is still enabled, the UCR2 CTSC and
CTS bits still have effect and the CTS (RS485 DE) control is retained.
Note that in case of RS485 DE signal active low, there is no problem and
no special handling is necessary. The CTS signal defaults to HIGH, thus
the RS485 is by default set to Receive and the bus is not blocked.
Note that while there is the possibility to control CTS using GPIO with
either CTS polarity, this has the downside of not being synchronous to
the UART IP clock and thus glitchy and susceptible to slow DE switching.
Second, on boot, before the UART driver probe callback is called, the
driver core triggers pinctrl_init_done() and configures the IOMUXC to
default state. At this point, UCR1 UARTEN and UCR2 RXEN are both still
cleared, but UART CTS_B (RS485 DE) is configured as CTS function, thus
the RTS signal is pulled HIGH by the UART IP CTS circuit.
One part of the solution here is to enable UCR1 UARTEN and UCR2 RXEN and
UTS loopback in this driver probe callback, thus unblocking the CTSC and
CTS control early on. But this is still too late, since the pin control
is already configured and CTS has been pulled HIGH for a short period
of time.
When Linux kernel boots and this driver is bound, the pin control is set
to special "init" state if the state is available, and driver can switch
the "default" state afterward when ready. This state can be used to set
the CTS line as a GPIO in DT temporarily, and a GPIO hog can force such
GPIO to LOW, thus keeping the RS485 DE line LOW early on boot. Once the
driver takes over and UCR1 UARTEN and UCR2 RXEN and UTS loopback are all
enabled, the driver can switch to "default" pin control state and control
the CTS line as function instead. DT binding example is below:
The irdma driver can use a maximum number of msix vectors equal
to num_online_cpus() + 1 and the kernel warning stack below is shown
if that number is exceeded.
The kernel throws a warning as the driver tries to update the affinity
hint with a CPU mask greater than the max CPU IDs. Fix this by capping
the MSIX vectors to num_online_cpus() + 1.
To update the I/O pins, the registers are read/modified/written. The
read operation incorrectly always read the first register. Although
wrong, there wasn't any impact as all the output pins are always
written, and the inputs are read only anyway.
Fixes: 6cb52f04231c ("Add support for using a MAX3421E chip as a host driver.") Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz> Link: https://lore.kernel.org/r/20230207033337.18112-1-mark.tomlinson@alliedtelesis.co.nz Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
If get_ep_from_tid() fails to lookup non-NULL value for ep, ep is
dereferenced later regardless of whether it is empty.
This patch adds a simple sanity check to fix the issue.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 3de6fa6a58d8 ("RDMA/iw_cxgb4: atomically lookup ep and get a reference") Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru> Link: https://lore.kernel.org/r/20230202184850.29882-1-n.zhandarovich@fintech.ru Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The rn5t618 power driver fails to register
a cooling device because POWER_SUPPLY_PROP_CHARGE_CONTROL_LIMIT_MAX
is missing but availability is not checked before registering
cooling device. After improved error checking in the thermal
code, the registration of the power supply fails entirely.
Checking for availability of _MAX before registering cooling device
fixes the rn5t618 problem. But the whole logic feels questionable.
First, the logic is inverted here:
the code tells: max_current = max_cooling but
0 = max_cooling, so there needs to be some inversion
in the code which cannot be found. Comparing with other
cooling devices, it can be found that value for fan speed is not
inverted, value for cpufreq cooling is inverted (similar situation
as here lowest frequency = max cooling)
Second, analyzing usage of _MAX: it is seems that maximum capabilities
of charging controller are specified and not of the battery. Probably
there is not too much mismatch in the drivers actually implementing
that. So nothing has exploded yet. So there is no easy and safe way
to specifify a max cooling value now.
Conclusion for now (as a regression fix) just remove the cooling device
registration and do it properly later on.
Fixes: 4e2729ef5aa8 ("thermal/core: fix error code in __thermal_cooling_device_register()") Fixes: 3710a22cd124 ("power_supply: Register power supply for thermal cooling device") Signed-off-by: Andreas Kemnade <andreas@kemnade.info> Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
pci_device_group() can return an already existing IOMMU group if the PCI
device's pagetables have to be shared with another one due to bus
toplogy, isolation features and/or DMA alias quirks.
apple_dart_device_group() however assumes that the group has just been
created and overwrites its iommudata which will eventually lead to
apple_dart_release_group leaving stale entries in sid2group.
Fix that by merging the iommudata if the returned group already exists.
Fixes: af264fcce65f ("iommu/dart: Clear sid2group entry when a group is freed") Signed-off-by: Sven Peter <sven@svenpeter.dev> Reviewed-by: Eric Curtin <ecurtin@redhat.com> Link: https://lore.kernel.org/r/20230128113532.94651-1-sven@svenpeter.dev Signed-off-by: Joerg Roedel <jroedel@suse.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some devices like ZBT WE1326 and ZBT WF3526-P and some Netgear models need
to delay phy port initialization after calling the mt7621_pcie_init_port()
driver function to get into reliable boots for both warm and hard resets.
The delay required to detect the ports seems to be in the range [75-100]
milliseconds.
If the ports are not detected the controller is not functional.
There is no datasheet or something similar to really understand why this
extra delay is needed only for these devices and it is not for most of
the boards that are built on mt7621 SoC.
This issue has been reported by openWRT community and the complete
discussion is in [0]. The 100 milliseconds delay has been tested in all
devices to validate it.
Add the extra 100 milliseconds delay to fix the issue.
Remove the temporary @mask_, this may cause build warning when use clang
compiler for powerpc, but can't reproduce it when compile for arm64.
the build warning is caused by:
"warning: result of comparison of constant 18446744073709551615 with
expression of type (aka 'unsigned long') is always false
[-Wtautological-constant-out-of-range-compare]"
More information provided in below lore link.
After removing @mask_, there is a "CHECK:MACRO_ARG_REUSE" when run
checkpatch.pl, but due to @mask is constant, no reuse problem will happen.
During enumeration or composition switch,a userspace process
agnostic of the conventions of configs can try to create function
symlinks even after the UDC is bound to current config which is
not correct. Potentially it can create duplicates within the
current config.
Prevent this by adding a check if udc_name already exists, then bail
out of cfg_link.
both the lists corresponds to the same function instance ffs.a
but the usb_function* pointer is different because in step 3
ffs_alloc has created a new reference to usb_function* for
ffs.a and added it to cfg_list.
Step4:
Now a composition switch involving <ffs.b,ffs.a> is executed.
the composition switch will involve 3 things:
1. unlinking the previous functions existing
2. creating new symlinks
3. writing UDC
However, the composition switch is generally taken care by
userspace process which creates the symlinks in its own
nomenclature(X*) and removes only those.
So it won't be able to remove Y1 which user had created
by own.
Due to this the new symlinks cannot be created for ffs.a
since the entry already exists in CFG->FUNC_LIST.
The state of the CFG->FUNC_LIST is as follows:
CFG->FUNC_LIST: <ffs.a>
Fixes: 3349f588ef24 ("usb: gadget: the start of the configfs interface") Signed-off-by: Krishna Kurapati PSSNV <quic_kriskura@quicinc.com> Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> Link: https://lore.kernel.org/r/20230201132308.31523-1-quic_ugoswami@quicinc.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
This function only calls mtk_otg_switch_init() when the ->port_mode
is MUSB_OTG so the clean up code should only call mtk_otg_switch_exit()
for that mode.
Fixes: 2140fcf788bf ("usb: musb: Add support for MediaTek musb controller") Signed-off-by: Dan Carpenter <error27@gmail.com> Link: https://lore.kernel.org/r/Y8/3TqpqiSr0RxFH@kili Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
ip_dev_find() may return NULL and assign it to pdev which is
dereferenced later.
Fix this by checking the return value of ip_dev_find() for NULL
similar to the way it is done with other instances of said function.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 3d3a6d2a20f8 ("RDMA/cxgb4: Fix LE hash collision bug for passive open connection") Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru> Link: https://lore.kernel.org/r/20230201172103.17261-1-n.zhandarovich@fintech.ru Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The RXWATER value must be greater than 0 according to the LPUART
reference manual. And when the number of datawords in the receive
FIFO is greater than RXWATER, an interrupt or a DMA request is
generated, so no need to set the different value for lpuart interrupt
case and dma case. Here delete the wrong RXWATER setting for dma case
directly.
Fixes: 3ccc30b6013b ("serial: fsl_lpuart: DMA support for 32-bit variant") Signed-off-by: Sherry Sun <sherry.sun@nxp.com> Link: https://lore.kernel.org/r/20230130064449.9564-4-sherry.sun@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Original busy loop with retries count in mpf_poll_status() is not too
reliable, as it takes different times on different systems. Replace it
with read_poll_timeout() macro.
While at it, fix polling stop condition to met function's original
intention declared in the comment. The issue with original polling stop
condition is that it stops if any of mask bits is set, while intention
was to stop if all mask bits is set. This was not noticible because only
MPF_STATUS_READY is passed as mask argument and it is BIT(1).
As spi-summary doc says:
> I/O buffers use the usual Linux rules, and must be DMA-safe.
> You'd normally allocate them from the heap or free page pool.
> Don't use the stack, or anything that's declared "static".
Replace spi_write() with spi_write_then_read(), which is dma-safe for
on-stack buffers. Use cacheline aligned buffers for transfers used in
spi_sync_transfer().
Although everything works OK with stack-located I/O buffers, better
follow the doc to be safe.
Interleaved DMA transfer support was added by 2f21da333266 ("dmaengine:
dw-edma: Add device_prep_interleave_dma() support"), but depending on the
selected channel, either source or destination address are left
uninitialized which was obviously wrong.
Initialize the destination address of the eDMA burst descriptors for
DEV_TO_MEM interleaved operations and the source address for MEM_TO_DEV
operations.
The normal call sequence of using transport class is:
Add path:
transport_setup_device()
transport_setup_classdev() // call sas_host_setup() here
transport_add_device() // if fails, need call transport_destroy_device()
transport_configure_device()
Remove path:
transport_remove_device()
transport_remove_classdev // call sas_host_remove() here
transport_destroy_device()
If transport_add_device() fails, need call transport_destroy_device()
to free memory, but in this case, ->remove() is not called, and the
resources allocated in ->setup() are leaked. So fix these leaks by
calling ->remove() in transport_add_class_device() if it returns error.
Current some drivers(like iscsi) call transport_register_device()
failed, they don't call transport_destroy_device() to release the
memory allocated in transport_setup_device(), because they don't
know what was done, it should be internal thing to release the
resource in register function. So fix this leak by calling destroy
function inside register function.
When calling kobject_add() failed in device_add(), it will call
cleanup_glue_dir() to free resource. But in kobject_add(),
dev->kobj.parent has been set to NULL. This will cause resource leak.
The process is as follows:
device_add()
get_device_parent()
class_dir_create_and_add()
kobject_add() //kobject_get()
...
dev->kobj.parent = kobj;
...
kobject_add() //failed, but set dev->kobj.parent = NULL
...
glue_dir = get_glue_dir(dev) //glue_dir = NULL, and goto
//"Error" label
...
cleanup_glue_dir() //becaues glue_dir is NULL, not call
//kobject_put()
The preceding problem may cause insmod mac80211_hwsim.ko to failed.
sysfs: cannot create duplicate filename '/devices/virtual/mac80211_hwsim'
Call Trace:
<TASK>
dump_stack_lvl+0x8e/0xd1
sysfs_warn_dup.cold+0x1c/0x29
sysfs_create_dir_ns+0x224/0x280
kobject_add_internal+0x2aa/0x880
kobject_add+0x135/0x1a0
get_device_parent+0x3d7/0x590
device_add+0x2aa/0x1cb0
device_create_groups_vargs+0x1eb/0x260
device_create+0xdc/0x110
mac80211_hwsim_new_radio+0x31e/0x4790 [mac80211_hwsim]
init_mac80211_hwsim+0x48d/0x1000 [mac80211_hwsim]
do_one_initcall+0x10f/0x630
do_init_module+0x19f/0x5e0
load_module+0x64b7/0x6eb0
__do_sys_finit_module+0x140/0x200
do_syscall_64+0x35/0x80
entry_SYSCALL_64_after_hwframe+0x46/0xb0
</TASK>
kobject_add_internal failed for mac80211_hwsim with -EEXIST, don't try to
register things with the same name in the same directory.
Fixes: b76a4f94368e ("driver core: fix race between creating/querying glue dir and its cleanup") Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com> Link: https://lore.kernel.org/r/20221123012042.335252-1-shaozhengchao@huawei.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
GUID_INIT() is for internal guid_t type and shouldn't be used
for the uuid_le. I.o.w. relying on the implementation details
is layering violation. Use correct macros to initialize uuid_le.
Fixes: 7ce0cd512fbd ("misc/mei/hdcp: Client driver for HDCP application") Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Acked-by: Tomas Winkler <tomas.winkler@intel.com> Link: https://lore.kernel.org/r/20221228160500.21220-1-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
GUID_INIT() is for internal guid_t type and shouldn't be used
for the uuid_le. I.o.w. relying on the implementation details
is layering violation. Use correct macros to initialize uuid_le.
Fixes: eb40256c3032 ("mei: pxp: export pavp client to me client bus") Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Acked-by: Tomas Winkler <tomas.winkler@intel.com> Link: https://lore.kernel.org/r/20221228160558.21311-1-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The call to get_user_pages_fast() in vmci_host_setup_notify() can return
NULL context->notify_page causing a GPF. To avoid GPF check if
context->notify_page == NULL and return error if so.
general protection fault, probably for non-canonical address
0xe0009d1000000060: 0000 [#1] PREEMPT SMP KASAN NOPTI
KASAN: maybe wild-memory-access in range [0x0005088000000300-
0x0005088000000307]
CPU: 2 PID: 26180 Comm: repro_34802241 Not tainted 6.1.0-rc4 #1
Hardware name: Red Hat KVM, BIOS 1.15.0-2.module+el8.6.0 04/01/2014
RIP: 0010:vmci_ctx_check_signal_notify+0x91/0xe0
Call Trace:
<TASK>
vmci_host_unlocked_ioctl+0x362/0x1f40
__x64_sys_ioctl+0x1a1/0x230
do_syscall_64+0x3a/0x90
entry_SYSCALL_64_after_hwframe+0x63/0xcd
As comment of pci_get_class() says, it returns a pci_device with its
refcount increased and decreased the refcount for the input parameter
@from if it is not NULL.
If we break the loop in applicom_init() with 'dev' not NULL, we need to
call pci_dev_put() to decrease the refcount. Add the missing
pci_dev_put() to avoid refcount leak.
A problem about idt_89hpesx create debugfs failed is triggered with the
following log given:
[ 4973.269647] debugfs: Directory 'idt_csr' with parent '/' already present!
The reason is that idt_init() returns i2c_add_driver() directly without
checking its return value, if i2c_add_driver() failed, it returns without
destroy the newly created debugfs, resulting the debugfs of idt_csr can
never be created later.
The function monitor_card() is a timer handler that runs in an
atomic context, but it calls usleep_range() that can sleep.
As a result, the sleep-in-atomic-context bugs will happen.
The process is shown below:
Add the missing clk_disable_unprepare() before return from
tegra_uart_hw_init() in the error handling path.
When request_irq() fails in tegra_uart_startup(), 'tup->uart_clk'
has been enabled, fix it by adding clk_disable_unprepare().
Fixes: 46c51708df77 ("serial: tegra: Only print FIFO error message when an error occurs") Fixes: 8b32d9ebd3f7 ("serial: tegra: report clk rate errors") Signed-off-by: Yi Yang <yiyang13@huawei.com> Link: https://lore.kernel.org/r/20221126020852.113378-1-yiyang13@huawei.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
UARTBAUD_RDMAE and UARTBAUD_TDMAE are enabled in lpuart32_startup(), but
lpuart32_shutdown() not disable them, only free the dma ring buffer and
release the dma channels, so here disable the Rx/Tx DMA first in
lpuart32_shutdown().
Fixes: 3ccc30b6013b ("serial: fsl_lpuart: DMA support for 32-bit variant") Signed-off-by: Sherry Sun <sherry.sun@nxp.com> Link: https://lore.kernel.org/r/20221125101953.18753-3-sherry.sun@nxp.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The PTT device can only support the devices on the same PCIe core,
within BDF range [lower_bdf, upper_bdf]. It's not correct to assume
the devices on the root bus are from the same PCIe core, there are
cases that root ports from different PCIe core are sharing the same
bus. So check when initializing the filters list.
Fixes: a5342fd38922 ("hwtracing: hisi_ptt: Add trace function support for HiSilicon PCIe Tune and Trace device") Signed-off-by: Yicong Yang <yangyicong@hisilicon.com> Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Link: https://lore.kernel.org/r/20230112112201.16283-1-yangyicong@huawei.com Signed-off-by: Sasha Levin <sashal@kernel.org>
28255722357e ("PCI: endpoint: pci-epf-vntb: fix error handle in
epf_ntb_mw_bar_init()") added a "num_mws" parameter to
epf_ntb_mw_bar_clear() but failed to add kernel-doc for num_mws.
Add kernel-doc for num_mws on epf_ntb_mw_bar_clear().
switchtec_dev_read() didn't handle copy_to_user() errors correctly: it
assigned "rc = -EFAULT", but actually returned either "size", -ENXIO, or
-EBADMSG instead.
Update the failure cases to unlock mrpc_mutex and return -EFAULT directly.
The sysfs link name "virtfn%u" constructed by pci_iov_sysfs_link() requires
17 bytes to contain the longest possible string. Increase VIRTFN_ID_LEN to
accommodate that.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
[bhelgaas: commit log, comment at #define] Fixes: 27f88952acc5 ("PCI: add SR-IOV API for Physical Function driver") Link: https://lore.kernel.org/r/20221218033347.23743-1-gremlin@altlinux.org Signed-off-by: Alexey V. Vissarionov <gremlin@altlinux.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When acpi_dev_get_memory_resources() fails, the reference count is
left bumped. Drop it as it's done in the other error paths.
Fixes: dc42a35a5125 ("usb: typec: intel_pmc_mux: Check the port status before connect") Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com> Link: https://lore.kernel.org/r/20230102202933.15968-1-andriy.shevchenko@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In commit bd78bc0786e4 ("coresight: cti: Fix hang in cti_disable_hw()")
PM runtime calls are removed from cti_enable_hw/cti_disable_hw. When
enabling CTI by writing enable sysfs node, clock for accessing CTI
register won't be enabled. Device will crash due to register access
issue. Add PM runtime call in enable_store to fix this issue.
Fixes: bd78bc0786e4 ("coresight: cti: Fix hang in cti_disable_hw()") Signed-off-by: Mao Jinlong <quic_jinlmao@quicinc.com>
[Change to only call pm_runtime_put if a disable happened] Tested-by: Jinlong Mao <quic_jinlmao@quicinc.com> Signed-off-by: James Clark <james.clark@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Link: https://lore.kernel.org/r/20230110110736.2709917-3-james.clark@arm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Writing 0 to the enable control repeatedly results in a negative value
for enable_req_count. After this, writing 1 to the enable control
appears to not work until the count returns to positive.
Change it so that it's impossible for enable_req_count to be < 0.
Return an error to indicate that the disable request was invalid.
Fixes: 510d6ba8ef8f ("coresight: cti: Initial CoreSight CTI Driver") Tested-by: Jinlong Mao <quic_jinlmao@quicinc.com> Signed-off-by: James Clark <james.clark@arm.com> Reviewed-by: Mike Leach <mike.leach@linaro.org> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Link: https://lore.kernel.org/r/20230110110736.2709917-2-james.clark@arm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / kernel.git / log