In the SoundWire probe, we store a pointer from the driver ops into
the 'slave' structure. This can lead to kernel oopses when unbinding
codec drivers, e.g. with the following sequence to remove machine
driver and codec driver.
The full details can be found in the BugLink below, for reference the
two following examples show different cases of driver ops/callbacks
being invoked after the driver .remove().
This was not detected earlier in Intel tests since the tests first
remove the parent PCI device and shut down the bus. The sequence
above is a corner case which keeps the bus operational but without a
driver bound.
While trying to solve this kernel oopses, it became clear that the
existing SoundWire bus does not deal well with the unbind case.
Commit a3e1041ee4bdb ("soundwire: sdw_slave: add probe_complete structure and new fields")
added a 'probed' status variable and a 'probe_complete'
struct completion. This status is however not reset on remove and
likewise the 'probe complete' is not re-initialized, so the
bind/unbind/bind test cases would fail. The timeout used before the
'update_status' callback was also a bad idea in hindsight, there
should really be no timing assumption as to if and when a driver is
bound to a device.
An initial draft was based on device_lock() and device_unlock() was
tested. This proved too complicated, with deadlocks created during the
suspend-resume sequences, which also use the same device_lock/unlock()
as the bind/unbind sequences. On a CometLake device, a bad DSDT/BIOS
caused spurious resumes and the use of device_lock() caused hangs
during suspend. After multiple weeks or testing and painful
reverse-engineering of deadlocks on different devices, we looked for
alternatives that did not interfere with the device core.
A bus notifier was used successfully to keep track of DRIVER_BOUND and
DRIVER_UNBIND events. This solved the bind-unbind-bind case in tests,
but it can still be defeated with a theoretical corner case where the
memory is freed by a .remove while the callback is in use. The
notifier only helps make sure the driver callbacks are valid, but not
that the memory allocated in probe remains valid while the callbacks
are invoked.
This patch suggests the introduction of a new 'sdw_dev_lock' mutex
protecting probe/remove and all driver callbacks. Since this mutex is
'local' to SoundWire only, it does not interfere with existing locks
and does not create deadlocks. In addition, this patch removes the
'probe_complete' completion, instead we directly invoke the
'update_status' from the probe routine. That removes any sort of
timing dependency and a much better support for the device/driver
model, the driver could be bound before the bus started, or eons after
the bus started and the hardware would be properly initialized in all
cases.
BugLink: https://github.com/thesofproject/linux/issues/3531 Fixes: 592fd752c764 ("soundwire: Add MIPI DisCo property helpers") Fixes: a3e1041ee4bdb ("soundwire: sdw_slave: add probe_complete structure and new fields") Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com> Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com> Link: https://lore.kernel.org/r/20220621225641.221170-2-pierre-louis.bossart@linux.intel.com Signed-off-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The bus sdw_drv_remove() and sdw_drv_shutdown() helpers are used
conditionally, if the driver provides these routines.
These helpers already test if the driver provides a .remove or
.shutdown callback, so there's no harm in invoking the
sdw_drv_remove() and sdw_drv_shutdown() unconditionally.
In addition, the current code is imbalanced with
dev_pm_domain_attach() called from sdw_drv_probe(), but
dev_pm_domain_detach() called from sdw_drv_remove() only if the driver
provides a .remove callback.
If the "snps,enable-cdm-check" property exists, we should enable the CDM
check. But previously dw_pcie_setup() could exit before doing so if the
"num-lanes" property was absent or invalid.
Move the CDM enable earlier so we do it regardless of whether "num-lanes"
is present.
If dw_pcie_ep_init() fails to perform any action after the EPC memory is
initialized and the MSI memory region is allocated, the latter parts won't
be undone thus causing a memory leak. Add a cleanup-on-error path to fix
these leaks.
We program the 64-bit ATU limit address (in PCIE_ATU_LIMIT/
PCIE_ATU_UPPER_LIMIT or PCIE_ATU_UNR_LOWER_LIMIT/PCIE_ATU_UNR_UPPER_LIMIT),
but in addition, the PCIE_ATU_INCREASE_REGION_SIZE bit must be set if the
upper 32 bits of the limit address differ from the upper 32 bits of the
base address (see [1,2]).
1190a4b54f55 ("PCI: dwc: Add upper limit address for outbound iATU") set
PCIE_ATU_INCREASE_REGION_SIZE, but only when the *size* was greater than
4GB. It did not set it when a smaller region crossed a 4GB boundary, e.g.,
[mem 0x0_f0000000-0x1_0fffffff].
Set PCIE_ATU_INCREASE_REGION_SIZE whenever PCIE_ATU_UPPER_LIMIT is
greater than PCIE_ATU_UPPER_BASE.
Some DWC-based controllers (e.g., pcie-al.c and pci-keystone.c, identified
by the fact that they override the default dw_child_pcie_ops) use their own
address translation approach instead of the DWC internal ATU (iATU). For
those controllers, skip disabling the iATU outbound windows.
dw_pcie_disable_atu() was introduced by 1ee611b62e8d ("PCI: dwc:
designware: Add EP mode support") and supported only the viewport version
of the iATU CSRs.
DW PCIe IP cores v4.80a and newer also support unrolled iATU/eDMA space.
Callers of dw_pcie_disable_atu(), including pci_epc_ops.clear_bar(),
pci_epc_ops.unmap_addr(), and dw_pcie_setup_rc(), don't work correctly when
it is enabled.
Add dw_pcie_disable_atu() support for controllers with unrolled iATU CSRs
enabled.
It's logically correct to undo everything that was done when an error is
discovered or in the corresponding cleanup counterpart. Otherwise the host
controller will be left in an undetermined state. Since the link is set up
in the host_init method, deactivate it there in the cleanup-on-error block
and stop the link in the antagonistic routine - dw_pcie_host_deinit(). Link
deactivation is platform-specific and should be implemented in
dw_pcie_ops.stop_link().
Fixes: b337f432d9f8 ("PCI: dwc: Move link handling into common code") Link: https://lore.kernel.org/r/20220624143428.8334-2-Sergey.Semin@baikalelectronics.ru Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
On SM8250 two found VFE GDSC power domains shall not be operated, if
titan top is turned off, thus the former power domains will be set as
subdomains by a GDSC registration routine.
Fixes: bd1bc3e01295 ("clk: qcom: Add camera clock controller driver for SM8250") Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org> Reviewed-by: Robert Foss <robert.foss@linaro.org> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Link: https://lore.kernel.org/r/20220519214133.1728979-3-vladimir.zapolskiy@linaro.org Signed-off-by: Sasha Levin <sashal@kernel.org>
On SDM845 two found VFE GDSC power domains shall not be operated, if
titan top is turned off, thus the former power domains will be set as
subdomains by a GDSC registration routine.
Fixes: cb41ba691802 ("clk: qcom: Add camera clock controller driver for SDM845") Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org> Reviewed-by: Robert Foss <robert.foss@linaro.org> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Link: https://lore.kernel.org/r/20220519214133.1728979-2-vladimir.zapolskiy@linaro.org Signed-off-by: Sasha Levin <sashal@kernel.org>
NSS port 5 and 6 frequency tables are currently broken and are causing a
wide ranges of issue like 1G not working at all on port 6 or port 5 being
clocked with 312 instead of 125 MHz as UNIPHY1 gets selected.
So, update the frequency tables with the ones from the downstream QCA 5.4
based kernel which has already fixed this.
Fixes: 34e5b8de3c8f ("clk: qcom: ipq8074: add NSS ethernet port clocks") Signed-off-by: Robert Marko <robimarko@gmail.com> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Link: https://lore.kernel.org/r/20220515210048.483898-3-robimarko@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
UBI32 Huayra PLL fails to lock in 5 us in some SoC silicon and thus it
will cause the wait_for_pll() to timeout and thus return the error
indicating that the PLL failed to lock.
This is bug in Huayra PLL HW for which SW workaround
is to set bit 26 of TEST_CTL register.
This is ported from the QCA 5.4 based downstream kernel.
Like in IPQ6018 the NSS related Alpha PLL-s require initial configuration
to work.
So, obtain the regmap that is required for the Alpha PLL configuration
and thus utilize the qcom_cc_really_probe() as we already have the regmap.
Then utilize the Alpha PLL configs from the downstream QCA 5.4 based
kernel to configure them.
This fixes the UBI32 and NSS crypto PLL-s failing to get enabled by the
kernel.
When a local operation (invalidate mr, reg mr, bind mw) is finished there
will be no ack packet coming from a responder to cause the wqe to be
completed. This may happen anyway if a subsequent wqe performs
IO. Currently if the wqe is signalled the completer tasklet is scheduled
immediately but not otherwise.
This leads to a deadlock if the next wqe has the fence bit set in send
flags and the operation is not signalled. This patch removes the condition
that the wqe must be signalled in order to schedule the completer tasklet
which is the simplest fix for this deadlock and is fairly low cost. This
is the analog for local operations of always setting the ackreq bit in all
last or only request packets even if the operation is not signalled.
Link: https://lore.kernel.org/r/20220523223251.15350-1-rpearsonhpe@gmail.com Reported-by: Jenny Hack <jhack@hpe.com> Fixes: 3e1cf5f469ef ("RDMA/rxe: Move local ops to subroutine") 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 c8af135cd196 ("xhci: fix unsafe memory usage in xhci tracing")
apparently missed one sprintf() call in xhci_decode_trb() -- replace
it with the snprintf() call as well...
Found by Linux Verification Center (linuxtesting.org) with the SVACE static
analysis tool.
The msm8939 has an additional higher operating point for the multi-media
peripherals. The higher throughput MM componets operate off of the
system-mm noc not the system noc.
system_mm_noc_bfdcd_clk_src is the source clock for the higher frequency
capable system noc mm.
In __driver_attach function, There are also AA deadlock problem,
like the commit 0eb2b5aea27d ("driver core: fix deadlock in
__device_attach").
stack like commit 0eb2b5aea27d ("driver core: fix deadlock in
__device_attach").
list below:
In __driver_attach function, The lock holding logic is as follows:
...
__driver_attach
if (driver_allows_async_probing(drv))
device_lock(dev) // get lock dev
async_schedule_dev(__driver_attach_async_helper, dev); // func
async_schedule_node
async_schedule_node_domain(func)
entry = kzalloc(sizeof(struct async_entry), GFP_ATOMIC);
/* when fail or work limit, sync to execute func, but
__driver_attach_async_helper will get lock dev as
will, which will lead to A-A deadlock. */
if (!entry || atomic_read(&entry_count) > MAX_WORK) {
func;
else
queue_work_node(node, system_unbound_wq, &entry->work)
device_unlock(dev)
As above show, when it is allowed to do async probes, because of
out of memory or work limit, async work is not be allowed, to do
sync execute instead. it will lead to A-A deadlock because of
__driver_attach_async_helper getting lock dev.
Reproduce:
and it can be reproduce by make the condition
(if (!entry || atomic_read(&entry_count) > MAX_WORK)) untenable, like
below:
To fix the deadlock, move the async_schedule_dev outside device_lock,
as we can see, in async_schedule_node_domain, the parameter of
queue_work_node is system_unbound_wq, so it can accept concurrent
operations. which will also not change the code logic, and will
not lead to deadlock.
If an error occurs after a successful idr_alloc() call, the corresponding
resource must be released with idr_remove() as already done in the .remove
function.
Update the error handling path to add the missing idr_remove() call.
Access to I/O of SM8250 camera clock controller IP depends on enabled
GCC_CAMERA_AHB_CLK clock supplied by global clock controller, the latter
one is inited on subsys level, so, to satisfy the dependency, it would
make sense to deprive the init level of camcc-sm8250 driver.
If both drivers are compiled as built-in, there is a change that a board
won't boot up due to a race, which happens on the same init level.
Currently autoloading for SPI devices does not use the DT ID table, it uses
SPI modalises. Supporting OF modalises is going to be difficult if not
impractical, an attempt was made but has been reverted, so ensure that
module autoloading works for this driver by adding an id_table listing the
SPI IDs for everything.
In accordance with [1, 2] the DW eDMA controller has been created to be
part of the DW PCIe Root Port and DW PCIe End-point controllers and to
offload the transferring of large blocks of data between application and
remote PCIe domains leaving the system CPU free for other tasks. In the
first case (eDMA being part of DW PCIe Root Port) the eDMA controller is
always accessible via the CPU DBI interface and never over the PCIe wire.
The latter case is more complex. Depending on the DW PCIe End-Point IP-core
synthesize parameters it's possible to have the eDMA registers accessible
not only from the application CPU side, but also via mapping the eDMA CSRs
over a dedicated endpoint BAR. So based on the specifics denoted above the
eDMA driver is supposed to support two types of the DMA controller setups:
1) eDMA embedded into the DW PCIe Root Port/End-point and accessible over
the local CPU from the application side.
2) eDMA embedded into the DW PCIe End-point and accessible via the PCIe
wire with MWr/MRd TLPs generated by the CPU PCIe host controller.
Since the CPU memory resides different sides in these cases the semantics
of the MEM_TO_DEV and DEV_TO_MEM operations is flipped with respect to the
Tx and Rx DMA channels. So MEM_TO_DEV/DEV_TO_MEM corresponds to the Tx/Rx
channels in setup 1) and to the Rx/Tx channels in case of setup 2).
The DW eDMA driver has supported the case 2) since 417abf1be9a5
("dmaengine: Add Synopsys eDMA IP core driver") in the framework of the
drivers/dma/dw-edma/dw-edma-pcie.c driver.
The case 1) support was added later by e5933d438c32 ("dmaengine: dw-edma:
support local dma device transfer semantics"). Afterwards the driver was
supposed to cover the both possible eDMA setups, but the latter commit
turned out to be not fully correct.
The problem was that the commit together with the new functionality support
also changed the channel direction semantics so the eDMA Read-channel
(corresponding to the DMA_DEV_TO_MEM direction for case 1) now uses the
sgl/cyclic base addresses as the Source addresses of the DMA transfers and
dma_slave_config.dst_addr as the Destination address of the DMA transfers.
Similarly the eDMA Write-channel (corresponding to the DMA_MEM_TO_DEV
direction for case 1) now uses dma_slave_config.src_addr as a source
address of the DMA transfers and sgl/cyclic base address as the Destination
address of the DMA transfers. This contradicts the logic of the
DMA-interface, which implies that DEV side is supposed to belong to the
PCIe device memory and MEM - to the CPU/Application memory. Indeed it seems
irrational to have the SG-list defined in the PCIe bus space, while
expecting a contiguous buffer allocated in the CPU memory. Moreover the
passed SG-list and cyclic DMA buffers are supposed to be mapped in a way so
to be seen by the DW eDMA Application (CPU) interface.
So in order to have the correct DW eDMA interface we need to invert the
eDMA Rd/Wr-channels and DMA-slave directions semantics by selecting the
src/dst addresses based on the DMA transfer direction instead of using the
channel direction capability.
When the system is shutting down, iscsid is not running so we will not get
a response to the ISCSI_ERR_INVALID_HOST error event. The system shutdown
will then hang waiting on userspace to remove the session.
This has libiscsi force the destruction of the session from the kernel when
iscsi_host_remove() is called from a driver's shutdown callout.
This fixes a regression added in qedi boot with commit c55cd560a09b ("scsi:
qedi: Fix host removal with running sessions") which made qedi use the
common session removal function that waits on userspace instead of rolling
its own kernel based removal.
Link: https://lore.kernel.org/r/20220616222738.5722-7-michael.christie@oracle.com Fixes: c55cd560a09b ("scsi: qedi: Fix host removal with running sessions") Tested-by: Nilesh Javali <njavali@marvell.com> Reviewed-by: Lee Duncan <lduncan@suse.com> Reviewed-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
During qedi shutdown we need to stop the iSCSI layer from sending new nops
as pings and from responding to target ones and make sure there is no
running connection cleanups. Commit c55cd560a09b ("scsi: qedi: Fix host
removal with running sessions") converted the driver to use the libicsi
helper to drive session removal, so the above issues could be handled. The
problem is that during system shutdown iscsid will not be running so when
we try to remove the root session we will hang waiting for userspace to
reply.
Add a helper that will drive the destruction of sessions like these during
system shutdown.
Link: https://lore.kernel.org/r/20220616222738.5722-5-michael.christie@oracle.com Tested-by: Nilesh Javali <njavali@marvell.com> Reviewed-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
iscsi_if_stop_conn() is only called from the userspace interface but in a
subsequent commit we will want to call it from the kernel interface to
allow drivers like qedi to remove sessions from inside the kernel during
shutdown. This removes the iscsi_uevent code from iscsi_if_stop_conn() so we
can call it in a new helper.
Link: https://lore.kernel.org/r/20220616222738.5722-3-michael.christie@oracle.com Tested-by: Nilesh Javali <njavali@marvell.com> Reviewed-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
There are sleep in atomic context bugs when uploading device dump
data in mwifiex. The root cause is that dev_coredumpv could not
be used in atomic contexts, because it calls dev_set_name which
include operations that may sleep. The call tree shows execution
paths that could lead to bugs:
This patch uses delayed work to replace timer and moves the operations
that may sleep into a delayed work in order to mitigate bugs, it was
tested on Marvell 88W8801 chip whose port is usb and the firmware is
usb8801_uapsta.bin. The following is the result after using delayed
work to replace timer.
[ 134.936453] usb 1-1: == mwifiex dump information to /sys/class/devcoredump start
[ 135.043344] usb 1-1: == mwifiex dump information to /sys/class/devcoredump end
The firmware of the 88W8897 PCIe+USB card sends those events very
unreliably, sometimes bluetooth together with 2.4ghz-wifi is used and no
COEX event comes in, and sometimes bluetooth is disabled but the
coexistance mode doesn't get disabled.
This means we sometimes end up capping the rx/tx window size while
bluetooth is not enabled anymore, artifically limiting wifi speeds even
though bluetooth is not being used.
Since we can't fix the firmware, let's just ignore those events on the
88W8897 device. From some Wireshark capture sessions it seems that the
Windows driver also doesn't change the rx/tx window sizes when bluetooth
gets enabled or disabled, so this is fairly consistent with the Windows
driver.
Don't set Accessed/Dirty bits for a struct page with PG_reserved set,
i.e. don't set A/D bits for the ZERO_PAGE. The ZERO_PAGE (or pages
depending on the architecture) should obviously never be written, and
similarly there's no point in marking it accessed as the page will never
be swapped out or reclaimed. The comment in page-flags.h is quite clear
that PG_reserved pages should be managed only by their owner, and
strictly following that mandate also simplifies KVM's logic.
Fixes: 148bda399795 ("KVM: fix overflow of zero page refcount with ksm running") Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20220429010416.2788472-4-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
CPU1 CPU2
memunmap_pages
percpu_ref_exit
__percpu_ref_exit
free_percpu(percpu_count);
/* percpu_count is freed here! */
get_dev_pagemap
xa_load(&pgmap_array, PHYS_PFN(phys))
/* pgmap still in the pgmap_array */
percpu_ref_tryget_live(&pgmap->ref)
if __ref_is_percpu
/* __PERCPU_REF_ATOMIC_DEAD not set yet */
this_cpu_inc(*percpu_count)
/* access freed percpu_count here! */
ref->percpu_count_ptr = __PERCPU_REF_ATOMIC_DEAD;
/* too late... */
pageunmap_range
To fix the issue, do percpu_ref_exit() after pgmap_array is emptied. So
we won't do percpu_ref_tryget_live() against a being freed percpu_ref.
Link: https://lkml.kernel.org/r/20220609121305.2508-1-linmiaohe@huawei.com Fixes: ae01e73b9802 ("mm/memremap_pages: support multiple ranges per invocation") Signed-off-by: Miaohe Lin <linmiaohe@huawei.com> Cc: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
If make_device_exclusive_range() fails or returns pages marked for
exclusive access less than required, remaining fields of pages will left
uninitialized. So dmirror_atomic_map() will access those yet
uninitialized fields of pages. To fix it, do dmirror_atomic_map() iff all
pages are marked for exclusive access (we will break if mapped is less
than required anyway) so we won't access those uninitialized fields of
pages.
Link: https://lkml.kernel.org/r/20220609130835.35110-1-linmiaohe@huawei.com Fixes: 42ee4beb23cd ("mm: selftests for exclusive device memory") Signed-off-by: Miaohe Lin <linmiaohe@huawei.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Original assert/deassert bit is BIT(0), but it's more resonable to modify
them to BIT(id % 32) which is based on id.
This patch will not influence any previous driver because the reset is
only used for thermal. The id (MT8183_INFRACFG_AO_THERM_SW_RST) is 0.
Fixes: 992fce56fe57 ("clk: reset: Modify reset-controller driver") Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com> Reviewed-by: Chen-Yu Tsai <wenst@chromium.org> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Reviewed-by: Nícolas F. R. A. Prado <nfraprado@collabora.com> Tested-by: Nícolas F. R. A. Prado <nfraprado@collabora.com> Link: https://lore.kernel.org/r/20220523093346.28493-3-rex-bc.chen@mediatek.com Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The last resume result exposing logic in cros_ec_sleep_event()
incorrectly requires S0ix support, which doesn't work on ARM based
systems where S0ix doesn't exist. That's because cros_ec_sleep_event()
only reports the last resume result when the EC indicates the last sleep
event was an S0ix resume. On ARM systems, the last sleep event is always
S3 resume, but the EC can still detect sleep hang events in case some
other part of the AP is blocking sleep.
Always expose the last resume result if the EC supports it so that this
works on all devices regardless of S0ix support. This fixes sleep hang
detection on ARM based chromebooks like Trogdor.
Cc: Rajat Jain <rajatja@chromium.org> Cc: Matthias Kaehlcke <mka@chromium.org> Cc: Hsin-Yi Wang <hsinyi@chromium.org> Cc: Tzung-Bi Shih <tzungbi@kernel.org> Reviewed-by: Guenter Roeck <groeck@chromium.org> Reviewed-by: Evan Green <evgreen@chromium.org> Fixes: e8a936107ac9 ("platform/chrome: Add support for v1 of host sleep event") Signed-off-by: Stephen Boyd <swboyd@chromium.org> Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org> Link: https://lore.kernel.org/r/20220614075726.2729987-1-swboyd@chromium.org Signed-off-by: Sasha Levin <sashal@kernel.org>
The scenario is this: User loaded driver but has not started authentication
app. All sessions to secure device will exhaust all login attempts, fail,
and in stay in deleted state. Then some time later the app is started. The
driver will replenish the login retry count, trigger delete to prepare for
secure login. After deletion, relogin is triggered.
For the session that is already deleted, the delete trigger is a no-op. If
none of the sessions trigger a relogin, no progress is made.
If the session is down and the local port continues to receive AUTH ELS
messages, the driver needs to send back LOGO so that the remote device
knows to tear down its session. Terminate and clean up the AUTH ELS
exchange followed by a passthrough LOGO.
Link: https://lore.kernel.org/r/20220608115849.16693-3-njavali@marvell.com Fixes: ce4d4dd912f3 ("scsi: qla2xxx: edif: Reject AUTH ELS on session down") Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com> Signed-off-by: Quinn Tran <qutran@marvell.com> Signed-off-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit cdb51be7ebd5 ("License cleanup: add SPDX license identifier to
uapi header files with a license") added the correct SPDX identifier to
include/uapi/linux/netfilter/xt_IDLETIMER.h.
A subsequent commit removed it for no reason and reintroduced the UAPI
license incorrectness as the file is now missing the UAPI exception
again.
Add it back and remove the GPLv2 boilerplate while at it.
In the function tegra_xusb_powerdomain_init(),
dev_pm_domain_attach_by_name() may return NULL in some cases,
so IS_ERR() doesn't meet the requirements. Thus fix it.
of_parse_phandle() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
Fixes: 0775db4aa75f ("USB: ohci-nxp: Use isp1301 driver") Acked-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Link: https://lore.kernel.org/r/20220603141231.979-1-linmq006@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
of_find_compatible_node() returns a node pointer with refcount
incremented, we should use of_node_put() on it when done.
Add missing of_node_put() to avoid refcount leak.
Fixes: 2c9458645465 ("USB: powerpc: Workaround for the PPC440EPX USBH_23 errata [take 3]") Acked-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Link: https://lore.kernel.org/r/20220602110849.58549-1-linmq006@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
of_find_node_by_path() returns a node pointer with refcount incremented,
we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
of_get_child_by_name() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
Fixes: f0e4c7650b8f ("mtd: partitions: redboot: seek fis-index-block in the right node") Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com> Link: https://lore.kernel.org/linux-mtd/20220526110652.64849-1-linmq006@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
There is a deadlock between sm_release and sm_cache_flush_work
which is a work item. The cancel_work_sync in sm_release will
not return until sm_cache_flush_work is finished. If we hold
mutex_lock and use cancel_work_sync to wait the work item to
finish, the work item also requires mutex_lock. As a result,
the sm_release will be blocked forever. The race condition is
shown below:
Smatch warnings:
drivers/hid/hid-cp2112.c:793 cp2112_xfer() error: __memcpy()
'data->block[1]' too small (33 vs 255)
drivers/hid/hid-cp2112.c:793 cp2112_xfer() error: __memcpy() 'buf' too
small (64 vs 255)
The 'read_length' variable is provided by 'data->block[0]' which comes
from user and it(read_length) can take a value between 0-255. Add an
upper bound to 'read_length' variable to prevent a buffer overflow in
memcpy().
pm_runtime_enable() will increase power disable depth. If
dw_pcie_ep_init() fails, we should use pm_runtime_disable() to balance it
with pm_runtime_enable().
Add missing pm_runtime_disable() for tegra_pcie_config_ep().
Fixes: ae0487c2a0af ("PCI: tegra: Add support for PCIe endpoint mode in Tegra194") Link: https://lore.kernel.org/r/20220602031910.55859-1-linmq006@gmail.com Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Vidya Sagar <vidyas@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
of_get_next_child() returns a node pointer with refcount incremented, so we
should use of_node_put() on it when we don't need it anymore.
mc_pcie_init_irq_domains() only calls of_node_put() in the normal path,
missing it in some error paths. Add missing of_node_put() to avoid
refcount leak.
If NRIPS is supported in hardware but disabled in KVM, set next_rip to
the next RIP when advancing RIP as part of emulating INT3 injection.
There is no flag to tell the CPU that KVM isn't using next_rip, and so
leaving next_rip is left as is will result in the CPU pushing garbage
onto the stack when vectoring the injected event.
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Fixes: 397076fd8c2e ("KVM: SVM: Emulate nRIP feature when reinjecting INT3") Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <cd328309a3b88604daa2359ad56f36cb565ce2d4.1651440202.git.maciej.szmigiero@oracle.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Unwind the RIP advancement done by svm_queue_exception() when injecting
an INT3 ultimately "fails" due to the CPU encountering a VM-Exit while
vectoring the injected event, even if the exception reported by the CPU
isn't the same event that was injected. If vectoring INT3 encounters an
exception, e.g. #NP, and vectoring the #NP encounters an intercepted
exception, e.g. #PF when KVM is using shadow paging, then the #NP will
be reported as the event that was in-progress.
Note, this is still imperfect, as it will get a false positive if the
INT3 is cleanly injected, no VM-Exit occurs before the IRET from the INT3
handler in the guest, the instruction following the INT3 generates an
exception (directly or indirectly), _and_ vectoring that exception
encounters an exception that is intercepted by KVM. The false positives
could theoretically be solved by further analyzing the vectoring event,
e.g. by comparing the error code against the expected error code were an
exception to occur when vectoring the original injected exception, but
SVM without NRIPS is a complete disaster, trying to make it 100% correct
is a waste of time.
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Fixes: 397076fd8c2e ("KVM: SVM: Emulate nRIP feature when reinjecting INT3") Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <450133cf0a026cb9825a2ff55d02cb136a1cb111.1651440202.git.maciej.szmigiero@oracle.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
After initiator has burned up all login retries, target authentication
application begins to run. This triggers a link bounce on target side.
Initiator will attempt another login. Due to N2N, the PRLI [nvme | fcp] can
fail because of the mode mismatch with target. This patch add a few more
login retries to revive the connection.
Link: https://lore.kernel.org/r/20220607044627.19563-11-njavali@marvell.com Fixes: 03d1f5917d23 ("scsi: qla2xxx: edif: Add N2N support for EDIF") Signed-off-by: Quinn Tran <qutran@marvell.com> Signed-off-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
User failed to see disk via n2n topology. Driver used up all login retries
before authentication application started. When authentication application
started, driver did not have enough login retries to connect securely. On
app_start, driver will reset the login retry attempt count.
Link: https://lore.kernel.org/r/20220607044627.19563-10-njavali@marvell.com Fixes: 03d1f5917d23 ("scsi: qla2xxx: edif: Add N2N support for EDIF") Signed-off-by: Quinn Tran <qutran@marvell.com> Signed-off-by: Nilesh Javali <njavali@marvell.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Relating to EDIF, when sending IKE message, updating key or deleting key,
driver can encounter IOCB queue full. Add additional retries to reduce
higher level recovery.
On ipsec start by remote port, target port may use RSCN to trigger
initiator to relogin. If driver is already in the process of a relogin,
then ignore the RSCN and allow the current relogin to continue. This
reduces thrashing of the connection.
When a thread is in the process of reestablish a session, a flag is set to
prevent multiple threads/triggers from doing the same task. This flag was
left on, and any attempt to relogin was locked out. Clear this flag if the
attempt has failed.
This patch uses GFFID switch command to scan whether remote device is
Target or Initiator mode. Based on that info, driver will not pass up
Initiator info to authentication application. This helps reduce unnecessary
stress for authentication application to deal with unused connections.
Presently ima_get_kexec_buffer() doesn't check if the previous kernel's
ima-kexec-buffer lies outside the addressable memory range. This can result
in a kernel panic if the new kernel is booted with 'mem=X' arg and the
ima-kexec-buffer was allocated beyond that range by the previous kernel.
The panic is usually of the form below:
of_find_matching_node() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
Fixes: e5bea735e41c ("mtd: physmap_of: add a hook for Versatile write protection") Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com> Link: https://lore.kernel.org/linux-mtd/20220523143255.4376-1-linmq006@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
of_find_matching_node_and_match() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
Fixes: e5bea735e41c ("mtd: physmap_of: add a hook for Versatile write protection") Signed-off-by: Miaoqian Lin <linmq006@gmail.com> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com> Link: https://lore.kernel.org/linux-mtd/20220523140205.48625-1-linmq006@gmail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
There are two UART clock groups, each having a mux to select its
upstream clock source. The register/bit definitions for accessing these
two muxes appear to have been reversed since introduction. Correct them
so as to match the hardware manual.
When sensor discovery fails, this means that the system doesn't have
any sensors connected and a user should only be notified at most one time.
A message is already displayed at WARN level of "failed to discover,
sensors not enabled". It's pointless to show that the client init failed
at ERR level for the same condition.
Check the return code and don't display this message in those conditions.
Fixes: 432a5bd3fc29 ("HID: amd_sfh: Add support for sensor discovery") Reported-by: David Chang <David.Chang@amd.com> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com> Acked-by: Basavaraj Natikar <Basavaraj.Natikar@amd.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: Sasha Levin <sashal@kernel.org>
In case push_rcu() and related functions are buggy, there's a
WARN_ON(len >= 128), which the selftest tries to hit by being tricky. In
case it is hit, we shouldn't corrupt the kernel's stack, though;
otherwise it may be hard to even receive the report that it's buggy. So
conditionalize the stack write based on that WARN_ON()'s return value.
Note that this never *actually* happens anyway. The WARN_ON() in the
first place is bounded by IS_ENABLED(DEBUG), and isn't expected to ever
actually hit. This is just a debugging sanity check.
Additionally, hoist the constant 128 into a named enum,
MAX_ALLOWEDIPS_BITS, so that it's clear why this value is chosen.
Using msleep() is problematic because it's compared against
ratelimiter.c's ktime_get_coarse_boottime_ns(), which means on systems
with slow jiffies (such as UML's forced HZ=100), the result is
inaccurate. So switch to using schedule_hrtimeout().
However, hrtimer gives us access only to the traditional posix timers,
and none of the _COARSE variants. So now, rather than being too
imprecise like jiffies, it's too precise.
One solution would be to give it a large "range" value, but this will
still fire early on a loaded system. A better solution is to align the
timeout to the actual coarse timer, and then round up to the nearest
tick, plus change.
So add the timeout to the current coarse time, and then
schedule_hrtimer() until the absolute computed time.
This should hopefully reduce flakes in CI as well. Note that we keep the
retry loop in case the entire function is running behind, because the
test could still be scheduled out, by either the kernel or by the
hypervisor's kernel, in which case restarting the test and hoping to not
be scheduled out still helps.
Fixes: 4dff5496e367 ("net: WireGuard secure network tunnel") Suggested-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Avoid r8153_ecm is compiled as built-in, if r8152 driver is compiled
as modules. Otherwise, the r8153_ecm would be used, even though the
device is supported by r8152 driver.
this commit amounted to:
drivers/net/usb/Kconfig:
+config USB_RTL8153_ECM
+ tristate "RTL8153 ECM support"
+ depends on USB_NET_CDCETHER && (USB_RTL8152 || USB_RTL8152=n)
+ default y
+ help
+ This option supports ECM mode for RTL8153 ethernet adapter, when
+ CONFIG_USB_RTL8152 is not set, or the RTL8153 device is not
+ supported by r8152 driver.
And as can be seen it pulls a piece of the cdc_ether driver out into
a separate config option to be able to make this piece modular in case
cdc_ether is builtin, while r8152 is modular.
While in general, device drivers should indeed not be enabled by default:
this isn't a device driver per say, but rather this is support code for
the CDCETHER (ECM) driver, and should thus be enabled if it is enabled.
See also email thread at:
https://www.spinics.net/lists/netdev/msg767649.html
Jakub wrote:
And when we say "removed" we can just hide it from what's prompted
to the user (whatever such internal options are called)? I believe
this way we don't bring back Marek's complaint.
Side note: these incorrect defaults will result in Android 13
on 5.15 GKI kernels lacking USB_RTL8153_ECM support while having
USB_NET_CDCETHER (luckily we also have USB_RTL8150 and USB_RTL8152,
so it's probably only an issue for very new RTL815x hardware with
no native 5.15 driver).
Fixes: f6744f38a132f6d0 ("nt: usb: USB_RTL8153_ECM should not default to y") Cc: Geert Uytterhoeven <geert+renesas@glider.be> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Hayes Wang <hayeswang@realtek.com> Cc: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Maciej Żenczykowski <maze@google.com> Link: https://lore.kernel.org/r/20220730230113.4138858-1-zenczykowski@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In the case of sk->dccps_qpolicy == DCCPQ_POLICY_PRIO, dccp_qpolicy_full
will drop a skb when qpolicy is full. And the lock in dccp_sendmsg is
released before sock_alloc_send_skb and then relocked after
sock_alloc_send_skb. The following conditions may lead dccp_qpolicy_push
to add skb to an already full sk_write_queue:
thread1--->lock
thread1--->dccp_qpolicy_full: queue is full. drop a skb
thread1--->unlock
thread2--->lock
thread2--->dccp_qpolicy_full: queue is not full. no need to drop.
thread2--->unlock
thread1--->lock
thread1--->dccp_qpolicy_push: add a skb. queue is full.
thread1--->unlock
thread2--->lock
thread2--->dccp_qpolicy_push: add a skb!
thread2--->unlock
Fix this by moving dccp_qpolicy_full.
Fixes: bc2fb404a31d ("[DCCP]: Set TX Queue Length Bounds via Sysctl") Signed-off-by: Hangyu Hua <hbh25y@gmail.com> Link: https://lore.kernel.org/r/20220729110027.40569-1-hbh25y@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The prototype of input features of ionic_set_nic_features() is
netdev_features_t, but the vlan_flags is using the private
definition of ionic drivers. It should use the variable
ctx.cmd.lif_setattr.features, rather than features to check
the vlan flags. So fixes it.
Fixes: 1c7e71712563 ("ionic: Add the basic NDO callbacks for netdev support") Signed-off-by: Jian Shen <shenjian15@huawei.com> Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com> Acked-by: Shannon Nelson <snelson@pensando.io> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Bernard reported that trying to unload rose module would lead
to infamous messages:
unregistered_netdevice: waiting for rose0 to become free. Usage count = xx
This patch solves the issue, by making sure each socket referring to
a netdevice holds a reference count on it, and properly releases it
in rose_release().
rose_dev_first() is also fixed to take a device reference
before leaving the rcu_read_locked section.
Following patch will add ref_tracker annotations to ease
future bug hunting.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: Bernard Pidoux <f6bvp@free.fr> Signed-off-by: Eric Dumazet <edumazet@google.com> Tested-by: Bernard Pidoux <f6bvp@free.fr> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
We need to suppress warnings from sily map sizes. Also switch
from GFP_USER to GFP_KERNEL_ACCOUNT, I'm pretty sure I misunderstood
the flags when writing this code.
Fix max_rate option in TC, check for proper quanta boundaries.
Check for minimum value provided and if it fits expected 50Mbps
quanta.
Without this patch, iavf could send settings for max_rate limiting
that would be accepted from by PF even the max_rate option is less
than expected 50Mbps quanta. It results in no rate limiting
on traffic as rate limiting will be floored to 0.
The function alloc_workqueue() in rtw_core_init() can fail, but
there is no check of its return value. To fix this bug, its return value
should be checked with new error handling code.
Fixes: 15678325112d6 ("rtw88: replace tx tasklet with work queue") Reported-by: Hacash Robot <hacashRobot@santino.com> Signed-off-by: William Dean <williamsukatube@gmail.com> Reviewed-by: Ping-Ke Shih <pkshih@realtek.com> Signed-off-by: Kalle Valo <kvalo@kernel.org> Link: https://lore.kernel.org/r/20220723063756.2956189-1-williamsukatube@163.com Signed-off-by: Sasha Levin <sashal@kernel.org>
As part of FIB offload simulation, netdevsim stores IPv4 and IPv6 routes
and holds a reference on FIB info structures that in turn hold a
reference on the associated nexthop device(s).
In the unlikely case where we are unable to allocate memory to process a
route deletion request, netdevsim will not release the reference from
the associated FIB info structure, thereby preventing the associated
nexthop device(s) from ever being removed [1].
Fix this by scheduling a work item that will flush netdevsim's FIB table
upon route deletion failure. This will cause netdevsim to release its
reference from all the FIB info structures in its table.
Reported by Lucas Leong of Trend Micro Zero Day Initiative.
Fixes: 56f51c8d9d8b ("netdevsim: fib: Perform the route programming in a non-atomic context") Signed-off-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Amit Cohen <amcohen@nvidia.com> Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit bcca38376a9d ("net: allow binding socket in a VRF when
there's an unbound socket") changed the inet socket lookup to avoid
packets in a VRF from matching an unbound socket. This is to ensure the
necessary isolation between the default and other VRFs for routing and
forwarding. VRF-unaware processes running in the default VRF cannot
access another VRF and have to be run with 'ip vrf exec <vrf>'. This is
to be expected with tcp_l3mdev_accept disabled, but could be reallowed
when this sysctl option is enabled. So instead of directly checking dif
and sdif in inet[6]_match, here call inet_sk_bound_dev_eq(). This
allows a match on unbound socket for non-zero sdif i.e. for packets in
a VRF, if tcp_l3mdev_accept is enabled.
Fixes: bcca38376a9d ("net: allow binding socket in a VRF when there's an unbound socket") Signed-off-by: Mike Manning <mvrmanning@gmail.com> Link: https://lore.kernel.org/netdev/a54c149aed38fded2d3b5fdb1a6c89e36a083b74.camel@lasnet.de/ Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The authentication algorithm supports a maximum of 128-byte keys.
The allocated key memory is insufficient.
Fixes: f50836e1f4d1 ("crypto: hisilicon - Add aead support on SEC2") Signed-off-by: Kai Ye <yekai13@huawei.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Sasha Levin <sashal@kernel.org>
Without MODULE_DEVICE_TABLE, crypto_safexcel.ko module is not automatically
loaded on platforms where inside-secure crypto HW is specified in device
tree (e.g. Armada 3720). So add missing MODULE_DEVICE_TABLE for of.
The cited commit limited log_max_qp to be 17 due to FW capabilities.
Recently, it turned out that there are old FW versions that supported
more than 17, so the cited commit caused a degradation.
Thus, set the maximum log_max_qp back to 18 as it was before the
cited commit.
Fixes: 3c04dea24290 ("net/mlx5: Update log_max_qp value to be 17 at most") Signed-off-by: Maher Sanalla <msanalla@nvidia.com> Reviewed-by: Maor Gottlieb <maorg@nvidia.com> Signed-off-by: Saeed Mahameed <saeedm@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
MLX5E_MAX_RQ_NUM_MTTS should be the maximum value, so that
MLX5_MTT_OCTW(MLX5E_MAX_RQ_NUM_MTTS) fits into u16. The current value of
1 << 17 results in MLX5_MTT_OCTW(1 << 17) = 1 << 16, which doesn't fit
into u16. This commit replaces it with the maximum value that still
fits u16.
The driver reports whether TX/RX TLS device offloads are supported, but
not which ciphers/versions, these should be handled by returning
-EOPNOTSUPP when .tls_dev_add() is called.
Remove the WARN_ON kernel trace when the driver gets a request to
offload a cipher/version that is not supported as it is expected.
Not all DPB entries will be used most of the time. Unused entries will
thus have invalid timestamps. They will produce negative buffer index
which is not specifically handled. This works just by chance in current
code. It will even produce bogus pointer, but since it's not used, it
won't do any harm.
Let's fix that brittle design by skipping writing DPB entry altogether
if timestamp is invalid.
projects / kernel.git / log