This commit creates a multi-queue mapping at device bring-up.
The driver first attempts to use the existing MSI-X interrupt
affinities (previously disabled), and if not present, will distribute
the request queues evenly over the CPUs.
If the latter fails as well, all CPUs are mapped to request queue zero.
When a request is handed from FUSE to the virtio-fs device driver, the
driver will use the current CPU to index into the multi-queue mapping
and determine the optimal request queue to use.
We measured the performance of this patch with the fio benchmarking
tool, increasing the number of queues results in a significant speedup
for both read and write operations, demonstrating the effectiveness
of multi-queue support.
Host:
- Dell PowerEdge R760
- CPU: Intel(R) Xeon(R) Gold 6438M, 128 cores
- VM: KVM with 32 cores
Virtio-fs device:
- BlueField-3 DPU
- CPU: ARM Cortex-A78AE, 16 cores
- One thread per queue, each busy polling on one request queue
- Each queue is 1024 descriptors deep
Workload:
- fio, sequential read or write, ioengine=libaio, numjobs=32,
4GiB file per job, iodepth=8, bs=256KiB, runtime=30s
Performance Results:
+===========================+==========+===========+
| Number of queues | Fio read | Fio write |
+===========================+==========+===========+
| 1 request queue (GiB/s) | 6.1 | 4.6 |
+---------------------------+----------+-----------+
| 8 request queues (GiB/s) | 25.8 | 10.3 |
+---------------------------+----------+-----------+
| 16 request queues (GiB/s) | 30.9 | 19.5 |
+---------------------------+----------+-----------+
| 32 request queue (GiB/s) | 33.2 | 22.6 |
+---------------------------+----------+-----------+
| Speedup | 5.5x | 5x |
+---------------=-----------+----------+-----------+
Signed-off-by: Peter-Jan Gootzen <pgootzen@nvidia.com>
Signed-off-by: Yoray Zack <yorayz@nvidia.com>
Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Virtio-fs devices might allocate significant resources to virtio queues
such as CPU cores that busy poll on the queue. The device indicates how
many request queues it can support and the driver should initialize the
number of queues that they want to utilize.
In this patch we limit the number of initialized request queues to the
number of CPUs, to limit the resource consumption on the device-side
and to prepare for the upcoming multi-queue patch.
Signed-off-by: Peter-Jan Gootzen <pgootzen@nvidia.com>
Signed-off-by: Yoray Zack <yorayz@nvidia.com>
Suggested-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Reviewed-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Merge series from Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>:
This patchset should be the last batch for this kernel cycle!
Brent Lu continued his cleanups to refactor and use fewer machine
drivers on Chrmebooks.
Bard Liao updated the sof-sdw machine driver to deal with UCM support
of the RT712 configuration. Note that this sof-sdw driver will be
refactored in the next kernel cycle to allow AMD and others to reuse
common SoundWire parts that are not Intel-specific. Initial changes
are described here: https://github.com/thesofproject/linux/pull/4967
Cortex-X4 and Neoverse-V3 suffer from errata whereby an MSR to the SSBS
special-purpose register does not affect subsequent speculative
instructions, permitting speculative store bypassing for a window of
time. This is described in their Software Developer Errata Notice (SDEN)
documents:
* Cortex-X4 SDEN v8.0, erratum 3194386:
https://developer.arm.com/documentation/SDEN-2432808/0800/
* Neoverse-V3 SDEN v6.0, erratum 3312417:
https://developer.arm.com/documentation/SDEN-2891958/0600/
To workaround these errata, it is necessary to place a speculation
barrier (SB) after MSR to the SSBS special-purpose register. This patch
adds the requisite SB after writes to SSBS within the kernel, and hides
the presence of SSBS from EL0 such that userspace software which cares
about SSBS will manipulate this via prctl(PR_GET_SPECULATION_CTRL, ...).
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240508081400.235362-5-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Add cputype definitions for Neoverse-V3. These will be used for errata
detection in subsequent patches.
These values can be found in Table B-249 ("MIDR_EL1 bit descriptions")
in issue 0001-04 of the Neoverse-V3 TRM, which can be found at:
https://developer.arm.com/documentation/107734/0001/?lang=en
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240508081400.235362-4-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Add cputype definitions for Cortex-X4. These will be used for errata
detection in subsequent patches.
These values can be found in Table B-249 ("MIDR_EL1 bit descriptions")
in issue 0002-05 of the Cortex-X4 TRM, which can be found at:
https://developer.arm.com/documentation/102484/0002/?lang=en
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240508081400.235362-3-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Upcoming errata workarounds will need to use SB from C code. Restore the
spec_bar() macro so that we can use SB.
This is effectively a revert of commit:
4f30ba1cce ("arm64: barrier: Remove spec_bar() macro")
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20240508081400.235362-2-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
add specific code to deal with the probe routine being called twice
(Geert Uytterhoeven)
- Remove unused field in the struct dmtimer in the TI driver
(Christophe JAILLET)
- Constify the hisi_161010101_oem_info variable in the ARM arch timer
(Stephen Boyd)
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEGn3N4YVz0WNVyHskqDIjiipP6E8FAmY94xsACgkQqDIjiipP
6E+L7gf/acsn2QDYpfJJOKNFo8cLeIHTQi3ZwuxM0LASocoEdwe8cDzj7LPOklsf
NMNg4UqQgo/iWHBhb+3PFy6R3WQiXnQXStbpTzh8rm2wNCD8zAP6RdizX6aYZGB9
+usmgC9y7JiC/lrChLNzcwAe4nheF51AQlHGbTE5OPpWGm4NGhn/IIvFyGMqKe5Z
G7LlBnVbrIKGEvFyQZhxUO/YQP/hwZU+HyPidkxBsL3SO1WJkww2px49q+vEIDI1
wyxw8qlGgSVCTqwasjmYWjHOhTo949PFmSM2O+YUegc+d8vTJjGINM0NBVBU52Jf
6Mkayymis4mdY9BPCZAjIf7QCGTA/g==
=gaGA
-----END PGP SIGNATURE-----
Merge tag 'timers-v6.10-rc1' of https://git.linaro.org/people/daniel.lezcano/linux into timers/core
Pull clockevent/source updates from Daniel Lezcano:
- Add the R9A09G057 compatible bindings in the DT documentation and
add specific code to deal with the probe routine being called twice
(Geert Uytterhoeven)
- Remove unused field in the struct dmtimer in the TI driver
(Christophe JAILLET)
- Constify the hisi_161010101_oem_info variable in the ARM arch timer
(Stephen Boyd)
Link: https://lore.kernel.org/lkml/7ca1c46a-93e6-4f67-bee3-623cb56764fa@linaro.org
If a serdev_device_driver is already loaded for a serdev_tty_port when it
gets registered by tty_port_register_device_attr_serdev() then that
driver's probe() method will be called immediately.
The serdev_device_driver's probe() method should then be able to call
serdev_device_open() successfully, but because UPF_DEAD is still dead
serdev_device_open() will fail with -ENXIO in this scenario:
serdev_device_open()
ctrl->ops->open() /* this callback being ttyport_open() */
tty->ops->open() /* this callback being uart_open() */
tty_port_open()
port->ops->activate() /* this callback being uart_port_activate() */
Find bit UPF_DEAD is set in uport->flags and fail with errno -ENXIO.
Fix this be clearing UPF_DEAD before tty_port_register_device_attr_serdev()
note this only moves up the UPD_DEAD clearing a small bit, before:
tty_port_register_device_attr_serdev();
mutex_unlock(&tty_port.mutex);
uart_port.flags &= ~UPF_DEAD;
mutex_unlock(&port_mutex);
after:
uart_port.flags &= ~UPF_DEAD;
tty_port_register_device_attr_serdev();
mutex_unlock(&tty_port.mutex);
mutex_unlock(&port_mutex);
Reported-by: Weifeng Liu <weifeng.liu.z@gmail.com>
Closes: https://lore.kernel.org/platform-driver-x86/20240505130800.2546640-1-weifeng.liu.z@gmail.com/
Tested-by: Weifeng Liu <weifeng.liu.z@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20240509141549.63704-1-hdegoede@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
At the default TX trigger level of 2 in non-DMA mode (meaning that an
interrupt is generated when less than 2 characters are left in the
FIFO), we have observed frequent buffer underruns at 115200 Baud on an
i.MX8M Nano. This can cause communication issues if the receiving side
expects a continuous transfer.
Increasing the level to 8 makes the UART trigger an interrupt earlier,
giving the kernel enough time to refill the FIFO, at the cost of
triggering one interrupt per ~24 instead of ~30 bytes of transmitted
data (as the i.MX UART has a 32 byte FIFO).
Signed-off-by: Michael Krummsdorf <michael.krummsdorf@tq-group.com>
Signed-off-by: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Link: https://lore.kernel.org/r/20240508133744.35858-1-matthias.schiffer@ew.tq-group.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
8250_pnp sets drvdata to line + 1 if the probe is successful. The users
of drvdata are in remove, suspend and resume callbacks, none of which
will be called if probe failed. The line acquired from drvdata can
never be zero in those functions and the checks for that can be
removed.
Eliminate also +/-1 step because all users of line subtract 1 from the
value.
These might have been leftover from legacy PM callbacks that could
be called without probe being successful.
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20240506121202.11253-1-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The hrtimer for RXDMA timeout was unconditionally restarted in the RXDMA
complete handler ignoring the fact that setting up DMA may fail and PIO
is used instead. Explicitly stop the timer when DMA is completed and
only restart it when setting up DMA was successful. This makes the
intention of the timer much clearer, the driver easier to understand and
simplifies assumptions about the timer. The latter avoids race
conditions if these assumptions were not met or confused.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://lore.kernel.org/r/20240506114016.30498-9-wsa+renesas@sang-engineering.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The first thing i2c_get_match_data() does is calling
device_get_match_data(), which already checks if there is a fwnode.
Remove explicit usage of device_get_match_data() as it is already
included in i2c_get_match_data().
Signed-off-by: Javier Carrasco <javier.carrasco@wolfvision.net>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20240429-tps6598x_fix_event_handling-v3-3-4e8e58dce489@wolfvision.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current interrupt service routine of the tps6598x only reads the
first 64 bits of the INT_EVENT1 and INT_EVENT2 registers, which means
that any event above that range will be ignored, leaving interrupts
unattended. Moreover, those events will not be cleared, and the device
will keep the interrupt enabled.
This issue has been observed while attempting to load patches, and the
'ReadyForPatch' field (bit 81) of INT_EVENT1 was set.
Given that older versions of the tps6598x (1, 2 and 6) provide 8-byte
registers, a mechanism based on the upper byte of the version register
(0x0F) has been included. The manufacturer has confirmed [1] that this
byte is always 0 for older versions, and either 0xF7 (DH parts) or 0xF9
(DK parts) is returned in newer versions (7 and 8).
Read the complete INT_EVENT registers to handle all interrupts generated
by the device and account for the hardware version to select the
register size.
Link: https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1346521/tps65987d-register-command-to-distinguish-between-tps6591-2-6-and-tps65987-8 [1]
Fixes: 0a4c005bd1 ("usb: typec: driver for TI TPS6598x USB Power Delivery controllers")
Cc: stable@vger.kernel.org
Signed-off-by: Javier Carrasco <javier.carrasco@wolfvision.net>
Link: https://lore.kernel.org/r/20240429-tps6598x_fix_event_handling-v3-2-4e8e58dce489@wolfvision.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In its current form, the interrupt service routine of the tps25750
checks the event flags in the lowest 64 bits of the interrupt event
register (event[0]), but also in the upper part (event[1]).
Given that all flags are defined as BIT() or BIT_ULL(), they are
restricted to the first 64 bits of the INT_EVENT1 register. Including
the upper part of the register can lead to false positives e.g. if the
event 64 bits above the one being checked is set, but the one being
checked is not.
Restrict the flag checking to the first 64 bits of the INT_EVENT1
register.
Fixes: 7e7a3c815d ("USB: typec: tps6598x: Add TPS25750 support")
Cc: stable@vger.kernel.org
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Javier Carrasco <javier.carrasco@wolfvision.net>
Link: https://lore.kernel.org/r/20240429-tps6598x_fix_event_handling-v3-1-4e8e58dce489@wolfvision.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
*-objs suffix is reserved rather for (user-space) host programs while
usually *-y suffix is used for kernel drivers (although *-objs works
for that purpose for now).
Let's correct the old usages of *-objs in Makefiles.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20240508150406.1378672-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
of_gpio.h is deprecated and subject to remove.
The driver doesn't use it directly, replace it
with what is really being used.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20240508113809.926155-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The function ucsi_displayport_work() does not access the
connector, so it also must not acquire the connector lock.
This fixes a potential deadlock scenario:
ucsi_displayport_work() -> lock(&con->lock)
typec_altmode_vdm()
dp_altmode_vdm()
dp_altmode_work()
typec_altmode_enter()
ucsi_displayport_enter() -> lock(&con->lock)
Reported-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Fixes: af8622f6a5 ("usb: typec: ucsi: Support for DisplayPort alt mode")
Cc: stable@vger.kernel.org
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20240507134316.161999-1-heikki.krogerus@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If a probe function returns -EPROBE_DEFER after creating another device
there is a change of ending up in a probe deferral loop, (see commit
fbc35b45f9 ("Add documentation on meaning of -EPROBE_DEFER"). In case
of the qcom-pmic-typec driver the tcpm_register_port() function looks up
external resources (USB role switch and inherently via called
typec_register_port() USB-C muxes, switches and retimers).
In order to prevent such probe-defer loops caused by qcom-pmic-typec
driver, use the API added by Johan Hovold and move HPD bridge
registration to the end of the probe function.
The devm_drm_dp_hpd_bridge_add() is called at the end of the probe
function after all TCPM start functions. This is done as a way to
overcome a different problem, the DRM subsystem can not properly cope
with the DRM bridges being destroyed once the bridge is attached. Having
this function call at the end of the probe function prevents possible
DRM bridge device creation followed by destruction in case one of the
TCPM start functions returns an error.
Reported-by: Caleb Connolly <caleb.connolly@linaro.org>
Acked-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20240424-qc-pmic-typec-hpd-split-v4-1-f7e10d147443@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove the unused list head 'buffers' and the
'struct free_record' which is also unused below it.
To me it looks like this has always been unused, but I've
not dug into why.
Build test only.
Signed-off-by: "Dr. David Alan Gilbert" <linux@treblig.org>
Link: https://lore.kernel.org/r/20240504150315.77598-1-linux@treblig.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently all controller IP/revisions except DWC3_usb3 >= 310a
wait 1ms unconditionally for ENDXFER completion when IOC is not
set. This is because DWC_usb3 controller revisions >= 3.10a
supports GUCTL2[14: Rst_actbitlater] bit which allows polling
CMDACT bit to know whether ENDXFER command is completed.
Consider a case where an IN request was queued, and parallelly
soft_disconnect was called (due to ffs_epfile_release). This
eventually calls stop_active_transfer with IOC cleared, hence
send_gadget_ep_cmd() skips waiting for CMDACT cleared during
EndXfer. For DWC3 controllers with revisions >= 310a, we don't
forcefully wait for 1ms either, and we proceed by unmapping the
requests. If ENDXFER didn't complete by this time, it leads to
SMMU faults since the controller would still be accessing those
requests.
Fix this by ensuring ENDXFER completion by adding 1ms delay in
__dwc3_stop_active_transfer() unconditionally.
Cc: stable@vger.kernel.org
Fixes: b353eb6dc2 ("usb: dwc3: gadget: Skip waiting for CMDACT cleared during endxfer")
Signed-off-by: Prashanth K <quic_prashk@quicinc.com>
Acked-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/20240502044103.1066350-1-quic_prashk@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This includes following USB4/Thunderbolt changes for the v6.10 merge
window:
- Enable NVM firmare upgrade on Intel Maple Ridge Thunderbolt 4
controller
- Improve USB3 tunnel bandwidth calculation
- Improve sideband access
- Minor cleanups and fixes.
All these have been in linux-next with no reported issues.
-----BEGIN PGP SIGNATURE-----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=fiCC
-----END PGP SIGNATURE-----
Merge tag 'thunderbolt-for-v6.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt into usb-next
Mika writes:
thunderbolt: Changes for v6.10 merge window
This includes following USB4/Thunderbolt changes for the v6.10 merge
window:
- Enable NVM firmare upgrade on Intel Maple Ridge Thunderbolt 4
controller
- Improve USB3 tunnel bandwidth calculation
- Improve sideband access
- Minor cleanups and fixes.
All these have been in linux-next with no reported issues.
* tag 'thunderbolt-for-v6.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/westeri/thunderbolt:
thunderbolt: Correct trace output of firmware connection manager packets
thunderbolt: Fix kernel-doc for tb_tunnel_alloc_dp()
thunderbolt: Fix uninitialized variable in tb_tunnel_alloc_usb3()
thunderbolt: There are only 5 basic router registers in pre-USB4 routers
thunderbolt: No need to loop over all retimers if access fails
thunderbolt: Increase sideband access polling delay
thunderbolt: Get rid of TB_CFG_PKG_PREPARE_TO_SLEEP
thunderbolt: Use correct error code with ERROR_NOT_SUPPORTED
thunderbolt: Allow USB3 bandwidth to be lower than maximum supported
thunderbolt: Fix calculation of consumed USB3 bandwidth on a path
thunderbolt: Enable NVM upgrade support on Intel Maple Ridge
This reverts commit 9801b5b28c.
This patch introduced a potential deadlock scenario:
[Wed May 8 10:02:06 2024] Possible unsafe locking scenario:
[Wed May 8 10:02:06 2024] CPU0 CPU1
[Wed May 8 10:02:06 2024] ---- ----
[Wed May 8 10:02:06 2024] lock(vivid_ctrls:1620:(hdl_vid_cap)->_lock);
[Wed May 8 10:02:06 2024] lock(vivid_ctrls:1608:(hdl_user_vid)->_lock);
[Wed May 8 10:02:06 2024] lock(vivid_ctrls:1620:(hdl_vid_cap)->_lock);
[Wed May 8 10:02:06 2024] lock(vivid_ctrls:1608:(hdl_user_vid)->_lock);
For now just revert.
Fixes: 9801b5b28c ("media: v4l2-ctrls: show all owned controls in log_status")
Cc: stable@vger.kernel.org
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Before this commit on probe() the driver would do:
reset=1 // from probe() calling gpiod_get(GPIOD_OUT_HIGH)
reset=0 // from resume()
msleep(20) // from resume()
So if reset was 0 before getting the GPIO the reset line would only be
driven high for a very short time and sometimes there would be errors
reading the id register afterwards.
Add a msleep(20) after getting the reset line to ensure the sensor is
properly reset:
reset=1 // from probe() calling gpiod_get(GPIOD_OUT_HIGH)
msleep(20) // from probe()
reset=0 // from resume()
msleep(20) // from resume()
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Use dev_dbg() for printing messages on user-triggerable conditions that
have no relation to driver or hardware issues.
Fixes: 3c1dfb5a69 ("media: intel/ipu6: input system video nodes and buffer queues")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Set up sysfs for the Atmel SHA204a. Provide the content of the otp zone as
an attribute field on the sysfs entry. Thereby make sure that if the chip
is locked, not connected or trouble with the i2c bus, the sysfs device is
not set up. This is mostly already handled in atmel-i2c.
Signed-off-by: Lothar Rubusch <l.rubusch@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Provide a read function reading the otp zone. The otp zone can be used for
storing serial numbers. The otp zone, as also data zone, are only
accessible if the chip was locked before. Locking the chip is a post
production customization and has to be done manually i.e. not by this
driver. Without this step the chip is pretty much not usable, where
putting or not putting data into the otp zone is optional.
Signed-off-by: Lothar Rubusch <l.rubusch@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Make the memory read function name more specific to the read memory zone.
The Atmel SHA204 chips provide config, otp and data zone. The implemented
read function in fact only reads some fields in zone config. The function
renaming allows for a uniform naming scheme when reading from other memory
zones.
Signed-off-by: Lothar Rubusch <l.rubusch@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Fixes the following two Coccinelle/coccicheck warnings reported by
memdup.cocci:
iaa_crypto_main.c:350:19-26: WARNING opportunity for kmemdup
iaa_crypto_main.c:358:18-25: WARNING opportunity for kmemdup
Signed-off-by: Thorsten Blum <thorsten.blum@toblux.com>
Reviewed-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
There is a confusing pattern in the kernel to use a variable named 'timeout' to
store the result of wait_for_completion_timeout() causing patterns like:
timeout = wait_for_completion_timeout(...)
if (!timeout) return -ETIMEDOUT;
with all kinds of permutations. Use 'time_left' as a variable to make the code
self explaining.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
There is a confusing pattern in the kernel to use a variable named 'timeout' to
store the result of wait_for_completion_killable_timeout() causing patterns like:
timeout = wait_for_completion_killable_timeout(...)
if (!timeout) return -ETIMEDOUT;
with all kinds of permutations. Use 'time_left' as a variable to make the code
self explaining.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
iMX8ULP have a secure-enclave hardware IP called EdgeLock Enclave(ELE),
that control access to caam controller's register page, i.e., page0.
At all, if the ELE release access to CAAM controller's register page,
it will release to secure-world only.
Clocks are turned on automatically for iMX8ULP. There exists the caam
clock gating bit, but it is not advised to gate the clock at linux, as
optee-os or any other entity might be using it.
Signed-off-by: Pankaj Gupta <pankaj.gupta@nxp.com>
Reviewed-by: Gaurav Jain <gaurav.jain@nxp.com>
Reviewed-by: Horia Geanta <horia.geanta@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
CAAM clock initializat is done based on the basis of soc specific
info stored in struct caam_imx_data:
- caam-page0-access flag
- num_clks
CAAM driver needs to be aware of access rights to CAAM control page
i.e., page0, to do things differently.
Signed-off-by: Pankaj Gupta <pankaj.gupta@nxp.com>
Reviewed-by: Gaurav Jain <gaurav.jain@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dma address mapping fails on unaligned scatterlist offset. Use sw
fallback for these cases.
Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
RSA text data uses variable length buffer allocated in software stack.
Calling kfree on it causes undefined behaviour in subsequent operations.
Cc: <stable@vger.kernel.org> #6.7+
Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Skip sw fallback allocation if RSA module failed to get device handle.
Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Skip dma setup and mapping for AES driver if plaintext is empty.
Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Sven Auhagen reports transaction failures with following error:
./main.nft:13:1-26: Error: Could not process rule: Cannot allocate memory
percpu: allocation failed, size=16 align=8 atomic=1, atomic alloc failed, no space left
This points to failing pcpu allocation with GFP_ATOMIC flag.
However, transactions happen from user context and are allowed to sleep.
One case where we can call into percpu allocator with GFP_ATOMIC is
nft_counter expression.
Normally this happens from control plane, so this could use GFP_KERNEL
instead. But one use case, element insertion from packet path,
needs to use GFP_ATOMIC allocations (nft_dynset expression).
At this time, .clone callbacks always use GFP_ATOMIC for this reason.
Add gfp_t argument to the .clone function and pass GFP_KERNEL or
GFP_ATOMIC flag depending on context, this allows all clone memory
allocations to sleep for the normal (transaction) case.
Cc: Sven Auhagen <sven.auhagen@voleatech.de>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Add a new test script that uses packetdrill tool to exercise conntrack
state machine.
Needs ip/ip6tables and conntrack tool (to check if we have an entry in
the expected state).
Test cases added here cover following scenarios:
1. already-acked (retransmitted) packets are not tagged as INVALID
2. RST packet coming when conntrack is already closing (FIN/CLOSE_WAIT)
transitions conntrack to CLOSE even if the RST is not an exact match
3. RST packets with out-of-window sequence numbers are marked as INVALID
4. SYN+Challenge ACK: check that challenge ack is allowed to pass
5. Old SYN/ACK: check conntrack handles the case where SYN is answered
with SYN/ACK for an old, previous connection attempt
6. Check SYN reception while in ESTABLISHED state generates a challenge
ack, RST response clears 'outdated' state + next SYN retransmit gets
us into 'SYN_RECV' conntrack state.
Tests get run twice, once with ipv4 and once with ipv6.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The following warning was reported by lee bruce:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 8264 at fs/fuse/dev.c:300
fuse_request_end+0x685/0x7e0 fs/fuse/dev.c:300
Modules linked in:
CPU: 0 PID: 8264 Comm: ab2 Not tainted 6.9.0-rc7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996)
RIP: 0010:fuse_request_end+0x685/0x7e0 fs/fuse/dev.c:300
......
Call Trace:
<TASK>
fuse_dev_do_read.constprop.0+0xd36/0x1dd0 fs/fuse/dev.c:1334
fuse_dev_read+0x166/0x200 fs/fuse/dev.c:1367
call_read_iter include/linux/fs.h:2104 [inline]
new_sync_read fs/read_write.c:395 [inline]
vfs_read+0x85b/0xba0 fs/read_write.c:476
ksys_read+0x12f/0x260 fs/read_write.c:619
do_syscall_x64 arch/x86/entry/common.c:52 [inline]
do_syscall_64+0xce/0x260 arch/x86/entry/common.c:83
entry_SYSCALL_64_after_hwframe+0x77/0x7f
......
</TASK>
The warning is due to the FUSE_NOTIFY_RESEND notify sent by the write()
syscall in the reproducer program and it happens as follows:
(1) calls fuse_dev_read() to read the INIT request
The read succeeds. During the read, bit FR_SENT will be set on the
request.
(2) calls fuse_dev_write() to send an USE_NOTIFY_RESEND notify
The resend notify will resend all processing requests, so the INIT
request is moved from processing list to pending list again.
(3) calls fuse_dev_read() with an invalid output address
fuse_dev_read() will try to copy the same INIT request to the output
address, but it will fail due to the invalid address, so the INIT
request is ended and triggers the warning in fuse_request_end().
Fix it by clearing FR_SENT when re-adding requests into pending list.
Acked-by: Miklos Szeredi <mszeredi@redhat.com>
Reported-by: xingwei lee <xrivendell7@gmail.com>
Reported-by: yue sun <samsun1006219@gmail.com>
Closes: https://lore.kernel.org/linux-fsdevel/58f13e47-4765-fce4-daf4-dffcc5ae2330@huaweicloud.com/T/#m091614e5ea2af403b259e7cea6a49e51b9ee07a7
Fixes: 760eac73f9 ("fuse: Introduce a new notification type for resend pending requests")
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When fuse_resend() moves the requests from processing lists to pending
list, it uses __set_bit() to set FR_PENDING bit in req->flags.
Using __set_bit() is not safe, because other functions may update
req->flags concurrently (e.g., request_wait_answer() may call
set_bit(FR_INTERRUPTED, &flags)).
Fix it by using set_bit() instead.
Fixes: 760eac73f9 ("fuse: Introduce a new notification type for resend pending requests")
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
commit c6f597bc59 ("ASoC: audio-graph-card2: remove unneeded
of_node_get()") removed of_node_get(), but it need to keep original
reference, so of_node_get() itself is needed.
Because of_get_next_child() will call of_node_put() inside, if the
reference count dropped to zero, then of_node_get() after that will be
use afer free. Need to call of_node_get() *before* that.
Fixes: c6f597bc59 ("ASoC: audio-graph-card2: remove unneeded of_node_get()")
Link: https://lore.kernel.org/r/f930862e-9d30-4ea3-b3e7-b4b4f411f6d1@moroto.mountain
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/875xvmqzs8.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
