| Age | Commit message (Collapse) | Author |
|---|
|
Use the new buffer-based WMI API to also support ACPI firmware
implementations that return a ACPI buffer instead of a ACPI integer.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://patch.msgid.link/20260116204116.4030-6-W_Armin@gmx.de
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
|
|
The string conversion frunctions provided by the WMI driver core
have no dependencies on the remaining WMI API, making them suitable
for unit tests.
Implement such a unit test using kunit. Those unit tests verify that
converting between WMI strings and UTF8 strings works as expected.
They also verify that edge cases are handled correctly.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://patch.msgid.link/20260116204116.4030-5-W_Armin@gmx.de
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
|
|
WMI strings are encoded using UTF16-LE characters, forcing WMI drivers
to manually convert them to/from standard UTF8 strings. Add a two
helper functions for those tasks.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://patch.msgid.link/20260116204116.4030-4-W_Armin@gmx.de
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
|
|
The marshalling code used by the WMI driver core is implemented as
a separate component, suitable for unit tests.
Implmented such a unit test using KUnit. Those unit tests verify that
ACPI objects are correctly converted into WMI buffers and that WMI
strings are correctly converted into ACPI strings. They also verify
that invalid ACPI data (like nested packages) is rejected.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://patch.msgid.link/20260116204116.4030-3-W_Armin@gmx.de
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
|
|
The Windows WMI-ACPI driver likely uses wmilib [1] to interact with
the WMI service in userspace. Said library uses plain byte buffers
for exchanging data, so the WMI-ACPI driver has to convert between
those byte buffers and ACPI objects returned by the ACPI firmware.
The format of the byte buffer is publicly documented [2], and after
some reverse eingineering of the WMI-ACPI driver using a set of custom
ACPI tables, the following conversion rules have been discovered:
- ACPI integers are always converted into a uint32
- ACPI strings are converted into special WMI strings
- ACPI buffers are copied as-is
- ACPI packages are unpacked
Extend the ACPI-WMI driver to also perform this kind of marshalling
for WMI data blocks, methods and events. Doing so gives us a number
of benefits:
- WMI drivers are not restricted to a fixed set of supported ACPI data
types anymore, see dell-wmi-aio (integer vs buffer) and
hp-wmi-sensors (string vs buffer)
- correct marshalling of WMI strings when data blocks are marked
as requiring ACPI strings instead of ACPI buffers
- development of WMI drivers without having to understand ACPI
This eventually should result in better compatibility with some
ACPI firmware implementations and in simpler WMI drivers. There are
however some differences between the original Windows driver and
the ACPI-WMI driver when it comes to ACPI object conversions:
- the Windows driver copies internal _ACPI_METHOD_ARGUMENT_V1 data
structures into the output buffer when encountering nested ACPI
packages. This is very likely an error inside the driver itself, so
we do not support nested ACPI packages.
- when converting WMI strings (UTF-16LE) into ACPI strings (ASCII),
the Windows driver replaces non-ascii characters (ä -> a, & -> ?)
instead of returning an error. This behavior is not documented
anywhere and might lead to severe errors in some cases (like
setting BIOS passwords over WMI), so we simply return an error.
As the current bus-based WMI API is based on ACPI buffers, a new
API is necessary. The legacy GUID-based WMI API is not extended to
support marshalling, as WMI drivers using said API are expected to
move to the bus-based WMI API in the future.
[1] https://learn.microsoft.com/de-de/windows-hardware/drivers/ddi/wmilib/
[2] https://learn.microsoft.com/en-us/windows-hardware/drivers/kernel/
driver-defined-wmi-data-items
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://patch.msgid.link/20260116204116.4030-2-W_Armin@gmx.de
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
|
|
The upper limit of the firmware queue fill state for each APQN
is reported by the hwinfo.qd field. This field shows the
numbers 0-7 for 1-8 queue spaces available. But the exploiting
code assumed the real boundary is stored there and thus stoppes
queuing in messages one tick too early.
Correct the limit calculation and thus offer a boost
of 12.5% performance for high traffic on one APQN.
Fixes: d4c53ae8e4948 ("s390/ap: store TAPQ hwinfo in struct ap_card")
Cc: stable@vger.kernel.org
Reported-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
|
|
min_t(u8, a, b) casts both its arguments to u8 potentially discarding
signifinact bits.
Use min(a, b) instead as it cannot discard significant bits.
Detected by an extra check added to min_t().
Signed-off-by: David Laight <david.laight.linux@gmail.com>
Signed-off-by: Andi Shyti <andi.shyti@kernel.org>
Link: https://lore.kernel.org/r/20251119224140.8616-21-david.laight.linux@gmail.com
|
|
During normal playback, stream-on for input is followed by output, and
only after input stream-on does actual streaming begin. However, when
gst-play performs a seek, both input and output streams are stopped,
and on restart, output stream-on occurs first. At this point, firmware
has not yet started streaming. Queuing output buffers before the firmware
begins streaming causes it to process buffers in an invalid state, leading
to an error response. These buffers are returned to the driver as errors,
forcing the driver into an error state and stopping playback.
Fix this by deferring output buffer queuing until stream-on completes.
Input buffers can still be queued before stream-on as required.
Fixes: 92e007ca5ab6 ("media: iris: Add V4L2 streaming support for encoder video device")
Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
In iris_kill_session, inst->state is set to IRIS_INST_ERROR and
session_close is executed, which will kfree(inst_hfi_gen2->packet).
If stop_streaming is called afterward, it will cause a crash.
Add a NULL check for inst_hfi_gen2->packet before sendling STOP packet
to firmware to fix that.
Fixes: 11712ce70f8e ("media: iris: implement vb2 streaming ops")
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
After the firmware releases internal buffers, the driver was not
destroying them. This left stale allocations that were no longer used,
especially across resolution changes where new buffers are allocated per
the updated requirements. As a result, memory was wasted until session
close.
Destroy internal buffers once the release response is received from the
firmware.
Fixes: 73702f45db81 ("media: iris: allocate, initialize and queue internal buffers")
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
The resolution property is not supposed to be set during reconfig.
Existing iris_drc_pending(inst) check is insufficient, as it doesn't
cover the first port setting change.
Extend the conditional check to also skip resolution setting when
the instance is in IRIS_INST_SUB_FIRST_IPSC.
Fixes: caf205548769 ("media: iris: Avoid updating frame size to firmware during reconfig")
Reviewed-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
Move `list_add_tail()` to after `dma_alloc_attrs()` succeeds when creating
internal buffers. Previously, the buffer was enqueued in `buffers->list`
before the DMA allocation. If the allocation failed, the function returned
`-ENOMEM` while leaving a partially initialized buffer in the list, which
could lead to inconsistent state and potential leaks.
By adding the buffer to the list only after `dma_alloc_attrs()` succeeds,
we ensure the list contains only valid, fully initialized buffers.
Fixes: 73702f45db81 ("media: iris: allocate, initialize and queue internal buffers")
Reviewed-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
Two platform-data fields for SM8750 were missed:
- get_vpu_buffer_size = iris_vpu33_buf_size
Without this, the driver fails to allocate the required internal
buffers, leading to basic decode/encode failures during session
bring-up.
- max_core_mbps = ((7680 * 4320) / 256) * 60
Without this capability exposed, capability checks are incomplete and
v4l2-compliance for encoder fails.
Fixes: a5925a2ce077 ("media: iris: add VPU33 specific encoding buffer calculation")
Fixes: a6882431a138 ("media: iris: Add support for ENUM_FRAMESIZES/FRAMEINTERVALS for encoder")
Cc: stable@vger.kernel.org
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
handling
Currently v4l2_m2m_ioctl_{de,enc}coder_cmd is being invoked during STOP
command handling. However, this is not required as the iris driver has
its own drain and stop handling mechanism in place.
Using the m2m command API in this context leads to incorrect behavior,
where the LAST flag is prematurely attached to a capture buffer,
when there are no buffers in m2m source queue. But, in this scenario
even though the source buffers are returned to client, hardware might
still need to process the pending capture buffers.
Attaching LAST flag prematurely can result in the capture buffer being
removed from the destination queue before the hardware has finished
processing it, causing issues when the buffer is eventually returned by
the hardware.
To prevent this, remove the m2m API usage in stop handling.
Fixes: d09100763bed ("media: iris: add support for drain sequence")
Fixes: 75db90ae067d ("media: iris: Add support for drain sequence in encoder video device")
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
According to 4.5.1.5 of the M2M stateful decoder UAPI documentation,
providing the width and the height to S_FMT is "required only if it
cannot be parsed from the stream", otherwise they can be left as 0
and the S_FMT implementation is expected to return a valid placeholder
resolution that would let REQBUFS succeed.
iris was missing the fallback, so clients like rpi-ffmpeg wouldn't work.
Fix by adding an explicit fallback to defaults.
Fixes: b530b95de22c ("media: iris: implement s_fmt, g_fmt and try_fmt ioctls")
Link: https://github.com/jc-kynesim/rpi-ffmpeg/issues/103
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Signed-off-by: Val Packett <val@packett.cool>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
iris_venc_s_param() uses do_div to divide two 64 bits operators, this is
wrong. Luckily for us, both of the operators fit in 32 bits, so we can use
a normal division.
Now that we are at it, mark the fps smaller than 1 as invalid, the code
does not seem to handle them properly.
The following cocci warning is fixed with this patch:
./platform/qcom/iris/iris_venc.c:378:1-7: WARNING: do_div() does a 64-by-32 division, please consider using div64_u64 instead
Fixes: 4ff586ff28e3 ("media: iris: Add support for G/S_PARM for encoder video device")
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
When the ffmpeg decoder is running, the driver receives the
V4L2_BUF_FLAG_KEYFRAME flag in the input buffer. The driver then forwards
this flag information to the firmware. The firmware, in turn, copies the
input buffer flags directly into the output buffer flags. Upon receiving
the output buffer from the firmware, the driver observes that the buffer
contains the HFI_BUFFERFLAG_DATACORRUPT flag. The root cause is that both
V4L2_BUF_FLAG_KEYFRAME and HFI_BUFFERFLAG_DATACORRUPT are the same value.
As a result, the driver incorrectly interprets the output frame as
corrupted, even though the frame is actually valid. This misinterpretation
causes the driver to report an error and skip good frames, leading to
missing frames in the final video output and triggering ffmpeg's "corrupt
decoded frame" error.
To resolve this issue, the input buffer flags should not be sent to the
firmware during decoding, since the firmware does not require this
information.
Fixes: 17f2a485ca67 ("media: iris: implement vb2 ops for buf_queue and firmware response")
Cc: stable@vger.kernel.org
Signed-off-by: Vishnu Reddy <quic_bvisredd@quicinc.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
This reverts commit ad699fa78b59241c9d71a8cafb51525f3dab04d4.
Revert the check that skipped stop_streaming when the instance was in
IRIS_INST_ERROR, as it caused multiple regressions:
1. Buffers were not returned to vb2 when the instance was already in
error state, triggering warnings in the vb2 core because buffer
completion was skipped.
2. If a session failed early (e.g. unsupported configuration), the
instance transitioned to IRIS_INST_ERROR. When userspace attempted
to stop streaming for cleanup, stop_streaming was skipped due to the
added check, preventing proper teardown and leaving the firmware
in an inconsistent state.
Fixes: ad699fa78b59 ("media: iris: Add sanity check for stop streaming")
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
The Venus encoder and decoder video devices currently report the same
bus_info string ("platform:qcom-venus").
Assign unique bus_info identifiers by appending ":dec" and ":enc" to the
parent device name. With this change v4l2-ctl will display two separate
logical devices
Signed-off-by: Jorge Ramirez-Ortiz <jorge.ramirez@oss.qualcomm.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
On SM8250 (IRIS2) with firmware older than 1.0.087, the firmware could
not handle a dummy device address for EOS buffers, so a NULL device
address is sent instead. The existing check used IS_V6() alongside a
firmware version gate:
if (IS_V6(core) && is_fw_rev_or_older(core, 1, 0, 87))
fdata.device_addr = 0;
else
fdata.device_addr = 0xdeadb000;
However, SC7280 which is also V6, uses a firmware string of the form
"1.0.<commit-hash>", which the version parser translates to 1.0.0. This
unintentionally satisfies the `is_fw_rev_or_older(..., 1, 0, 87)`
condition on SC7280. Combined with IS_V6() matching there as well, the
quirk is incorrectly applied to SC7280, causing VP9 decode failures.
Constrain the check to IRIS2 (SM8250) only, which is the only platform
that needed this quirk, by replacing IS_V6() with IS_IRIS2(). This
restores correct behavior on SC7280 (no forced NULL EOS buffer address).
Fixes: 47f867cb1b63 ("media: venus: fix EOS handling in decoder stop command")
Cc: stable@vger.kernel.org
Reported-by: Mecid <mecid@mecomediagroup.de>
Closes: https://github.com/qualcomm-linux/kernel-topics/issues/222
Co-developed-by: Renjiang Han <renjiang.han@oss.qualcomm.com>
Signed-off-by: Renjiang Han <renjiang.han@oss.qualcomm.com>
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Tested-by: Renjiang Han <renjiang.han@oss.qualcomm.com>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
When hfi_session_flush is issued, all queued buffers are returned to
the V4L2 driver. Some of these buffers are not processed and have
bytesused = 0. Currently, the driver marks such buffers as error even
during drain operations, which can incorrectly flag EOS buffers.
Only capture buffers with zero payload (and not EOS) should be marked
with VB2_BUF_STATE_ERROR. The check is performed inside the non-EOS
branch to ensure correct handling.
Fixes: 51df3c81ba10b ("media: venus: vdec: Mark flushed buffers with error state")
Signed-off-by: Renjiang Han <renjiang.han@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
|
|
This applies the uapi change in commit ea78ec982653 ("drm/panthor:
Expose the selected coherency protocol to the UMD") to the Tyr driver as
well.
Once this is merged with drm-rust-next, this kind of change to the uapi
struct is automatically reflected in the Tyr driver's GpuInfo. [1]
Link: https://lore.kernel.org/r/aW8-oH7dtp-OTAZC@google.com [1]
Acked-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patch.msgid.link/20260120-tyr-pad0-coherency-v1-1-91f40e56c67a@google.com
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
|
|
Enable support for AF_XDP applications to operate on a netkit device.
The goal is that AF_XDP applications can natively consume AF_XDP
from network namespaces. The use-case from Cilium side is to support
Kubernetes KubeVirt VMs through QEMU's AF_XDP backend. KubeVirt is a
virtual machine management add-on for Kubernetes which aims to provide
a common ground for virtualization. KubeVirt spawns the VMs inside
Kubernetes Pods which reside in their own network namespace just like
regular Pods.
Raw QEMU AF_XDP backend example with eth0 being a physical device with
16 queues where netkit is bound to the last queue (for multi-queue RSS
context can be used if supported by the driver):
# ethtool -X eth0 start 0 equal 15
# ethtool -X eth0 start 15 equal 1 context new
# ethtool --config-ntuple eth0 flow-type ether \
src 00:00:00:00:00:00 \
src-mask ff:ff:ff:ff:ff:ff \
dst $mac dst-mask 00:00:00:00:00:00 \
proto 0 proto-mask 0xffff action 15
[ ... setup BPF/XDP prog on eth0 to steer into shared xsk map ... ]
# ip netns add foo
# ip link add numrxqueues 2 nk type netkit single
# ./pyynl/cli.py --spec ~/netlink/specs/netdev.yaml \
--do queue-create \
--json "{"ifindex": $(ifindex nk), "type": "rx", \
"lease": { "ifindex": $(ifindex eth0), \
"queue": { "type": "rx", "id": 15 } } }"
{'id': 1}
# ip link set nk netns foo
# ip netns exec foo ip link set lo up
# ip netns exec foo ip link set nk up
# ip netns exec foo qemu-system-x86_64 \
-kernel $kernel \
-drive file=${image_name},index=0,media=disk,format=raw \
-append "root=/dev/sda rw console=ttyS0" \
-cpu host \
-m $memory \
-enable-kvm \
-device virtio-net-pci,netdev=net0,mac=$mac \
-netdev af-xdp,ifname=nk,id=net0,mode=native,queues=1,start-queue=1,inhibit=on,map-path=$dir/xsks_map \
-nographic
We have tested the above against a dual-port Nvidia ConnectX-6 (mlx5)
100G NIC with successful network connectivity out of QEMU. An earlier
iteration of this work was presented at LSF/MM/BPF [0] and more
recently at LPC [1].
For getting to a first starting point to connect all things with
KubeVirt, bind mounting the xsk map from Cilium into the VM launcher
Pod which acts as a regular Kubernetes Pod while not perfect, is not
a big problem given its out of reach from the application sitting
inside the VM (and some of the control plane aspects are baked in
the launcher Pod already), so the isolation barrier is still the VM.
Eventually the goal is to have a XDP/XSK redirect extension where
there is no need to have the xsk map, and the BPF program can just
derive the target xsk through the queue where traffic was received
on.
The exposure through netkit is because Cilium should not act as a
proxy handing out xsk sockets. Existing applications expect a netdev
from kernel side and should not need to rewrite just to implement
against a CNI's protocol. Also, all the memory should not be accounted
against Cilium but rather the application Pod itself which is consuming
AF_XDP. Further, on up/downgrades we expect the data plane to being
completely decoupled from the control plane; if Cilium would own the
sockets that would be disruptive. Another use-case which opens up and
is regularly asked from users would be to have DPDK applications on
top of AF_XDP in regular Kubernetes Pods.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Co-developed-by: David Wei <dw@davidwei.uk>
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://bpfconf.ebpf.io/bpfconf2025/bpfconf2025_material/lsfmmbpf_2025_netkit_borkmann.pdf [0]
Link: https://lpc.events/event/19/contributions/2275/ [1]
Link: https://patch.msgid.link/20260115082603.219152-13-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add a netdevice notifier in netkit to watch for NETDEV_UNREGISTER events.
If the target device is indeed NETREG_UNREGISTERING and previously leased
a queue to a netkit device, then collect the related netkit devices and
batch-unregister_netdevice_many() them.
If this would not be done, then the netkit device would hold a reference
on the physical device preventing it from going away. However, in case of
both io_uring zero-copy as well as AF_XDP this situation is handled
gracefully and the allocated resources are torn down.
In the case where mentioned infra is used through netkit, the applications
have a reference on netkit, and netkit in turn holds a reference on the
physical device. In order to have netkit release the reference on the
physical device, we need such watcher to then unregister the netkit ones.
This is generally quite similar to the dependency handling in case of
tunnels (e.g. vxlan bound to a underlying netdev) where the tunnel device
gets removed along with the physical device.
# ip a
[...]
4: enp10s0f0np0: <BROADCAST,MULTICAST> mtu 1500 qdisc mq state DOWN group default qlen 1000
link/ether e8:eb:d3:a3:43:f6 brd ff:ff:ff:ff:ff:ff
inet 10.0.0.2/24 scope global enp10s0f0np0
valid_lft forever preferred_lft forever
[...]
8: nk@NONE: <BROADCAST,MULTICAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
[...]
# rmmod mlx5_ib
# rmmod mlx5_core
[ 309.261822] mlx5_core 0000:0a:00.0 mlx5_0: Port: 1 Link DOWN
[ 344.235236] mlx5_core 0000:0a:00.1: E-Switch: Unload vfs: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 344.246948] mlx5_core 0000:0a:00.1: E-Switch: Disable: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 344.463754] mlx5_core 0000:0a:00.1: E-Switch: Disable: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 344.770155] mlx5_core 0000:0a:00.1: E-Switch: cleanup
[ 345.345709] mlx5_core 0000:0a:00.0: E-Switch: Unload vfs: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 345.357524] mlx5_core 0000:0a:00.0: E-Switch: Disable: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 350.995989] mlx5_core 0000:0a:00.0: E-Switch: Disable: mode(LEGACY), nvfs(0), necvfs(0), active vports(0)
[ 351.574396] mlx5_core 0000:0a:00.0: E-Switch: cleanup
# ip a
[...]
[ both enp10s0f0np0 and nk gone ]
[...]
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Co-developed-by: David Wei <dw@davidwei.uk>
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/20260115082603.219152-12-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Implement rtnl_link_ops->alloc that allows the number of rx queues to be
set when netkit is created. By default, netkit has only a single rxq (and
single txq). The number of queues is deliberately not allowed to be changed
via ethtool -L and is fixed for the lifetime of a netkit instance.
For netkit device creation, numrxqueues with larger than one rxq can be
specified. These rxqs are leasable to real rxqs in physical netdevs:
ip link add type netkit peer numrxqueues 64 # for device pair
ip link add numrxqueues 64 type netkit single # for single device
The limit of numrxqueues for netkit is currently set to 1024, which allows
leasing multiple real rxqs from physical netdevs.
The implementation of ndo_queue_create() adds a new rxq during the queue
lease operation. We allow to create queues either in single device mode
or for the case of dual device mode for the netkit peer device which gets
placed into the target network namespace. For dual device mode the lease
against the primary device does not make sense for the targeted use cases,
and therefore gets rejected.
We also need to add a lockdep class for netkit, such that lockdep does
not trip over us, similarly done as in commit 0bef512012b1 ("net: add
netdev_lockdep_set_classes() to virtual drivers").
This is also the last missing bit to netkit for supporting io_uring with
zero-copy mode [0]. Up until this point it was not possible to consume the
latter out of containers or Kubernetes Pods where applications are in their
own network namespace.
io_uring example with eth0 being a physical device with 16 queues where
netkit is bound to the last queue, iou-zcrx.c is binary from selftests.
Flow steering to that queue is based on the service VIP:port of the
server utilizing io_uring:
# ethtool -X eth0 start 0 equal 15
# ethtool -X eth0 start 15 equal 1 context new
# ethtool --config-ntuple eth0 flow-type tcp4 dst-ip 1.2.3.4 dst-port 5000 action 15
# ip netns add foo
# ip link add type netkit peer numrxqueues 2
# ./pyynl/cli.py --spec ~/netlink/specs/netdev.yaml \
--do queue-create \
--json "{"ifindex": $(ifindex nk0), "type": "rx", \
"lease": { "ifindex": $(ifindex eth0), \
"queue": { "type": "rx", "id": 15 } } }"
{'id': 1}
# ip link set nk0 netns foo
# ip link set nk1 up
# ip netns exec foo ip link set lo up
# ip netns exec foo ip link set nk0 up
# ip netns exec foo ip addr add 1.2.3.4/32 dev nk0
[ ... setup routing etc to get external traffic into the netns ... ]
# ip netns exec foo ./iou-zcrx -s -p 5000 -i nk0 -q 1
Remote io_uring client:
# ./iou-zcrx -c -h 1.2.3.4 -p 5000 -l 12840 -z 65536
We have tested the above against a Broadcom BCM957504 (bnxt_en) 100G NIC,
supporting TCP header/data split.
Similarly, this also works for devmem which we tested using ncdevmem:
# ip netns exec foo ./ncdevmem -s 1.2.3.4 -l -p 5000 -f nk0 -t 1 -q 1
And on the remote client:
# ./ncdevmem -s 1.2.3.4 -p 5000 -f eth0
For Cilium, the plan is to open up support for the various memory providers
for regular Kubernetes Pods when Cilium is configured with netkit datapath
mode.
Signed-off-by: David Wei <dw@davidwei.uk>
Co-developed-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://kernel-recipes.org/en/2024/schedule/efficient-zero-copy-networking-using-io_uring [0]
Link: https://patch.msgid.link/20260115082603.219152-11-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add a single device mode for netkit instead of netkit pairs. The primary
target for the paired devices is to connect network namespaces, of course,
and support has been implemented in projects like Cilium [0]. For the rxq
leasing the plan is to support two main scenarios related to single device
mode:
* For the use-case of io_uring zero-copy, the control plane can either
set up a netkit pair where the peer device can perform rxq leasing which
is then tied to the lifetime of the peer device, or the control plane
can use a regular netkit pair to connect the hostns to a Pod/container
and dynamically add/remove rxq leasing through a single device without
having to interrupt the device pair. In the case of io_uring, the memory
pool is used as skb non-linear pages, and thus the skb will go its way
through the regular stack into netkit. Things like the netkit policy when
no BPF is attached or skb scrubbing etc apply as-is in case the paired
devices are used, or if the backend memory is tied to the single device
and traffic goes through a paired device.
* For the use-case of AF_XDP, the control plane needs to use netkit in the
single device mode. The single device mode currently enforces only a
pass policy when no BPF is attached, and does not yet support BPF link
attachments for AF_XDP. skbs sent to that device get dropped at the
moment. Given AF_XDP operates at a lower layer of the stack tying this
to the netkit pair did not make sense. In future, the plan is to allow
BPF at the XDP layer which can: i) process traffic coming from the AF_XDP
application (e.g. QEMU with AF_XDP backend) to filter egress traffic or
to push selected egress traffic up to the single netkit device to the
local stack (e.g. DHCP requests), and ii) vice-versa skbs sent to the
single netkit into the AF_XDP application (e.g. DHCP replies). Also,
the control-plane can dynamically manage rxq leasing for the single
netkit device without having to interrupt (e.g. down/up cycle) the main
netkit pair for the Pod which has traffic going in and out.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Co-developed-by: David Wei <dw@davidwei.uk>
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Jordan Rife <jordan@jrife.io>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://docs.cilium.io/en/stable/operations/performance/tuning/#netkit-device-mode [0]
Link: https://patch.msgid.link/20260115082603.219152-10-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
rtsx_pci_sdmmc does not have an sdmmc_card_busy function, so any voltage
switches cause a kernel warning, "mmc0: cannot verify signal voltage
switch."
Copy the sdmmc_card_busy function from rtsx_pci_usb to rtsx_pci_sdmmc to
fix this.
Fixes: ff984e57d36e ("mmc: Add realtek pcie sdmmc host driver")
Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
Tested-by: Ricky WU <ricky_wu@realtek.com>
Reviewed-by: Ricky WU <ricky_wu@realtek.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
|
|
If there is a plane state to destroy when doing a plane reset, destroy
it using the atmel_hlcdc_plane_destroy_state() function. So we call
__drm_atomic_helper_plane_destroy_state() and avoid code duplication.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-8-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Use the drmm_universal_plane_alloc() helper to simplify the code. Using
it, we no longer need to register the destroy callback for
drm_plane_funcs.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-7-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Use drmm_crtc_alloc_with_planes() to simplify the code. As we no longer
have to take care about cleanup, we can get rid of
atmel_hlcdc_crtc_destroy().
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-6-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Get rid of drm_of_find_panel_or_bridge() as it is deprecated and use
devm_drm_of_get_bridge() instead.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-5-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Prefer using the drm_crtc_mask() helper instead of a raw value. It
involves reordering components initialization as we need a valid crtc.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-4-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Simplify the code using drmm_simple_encoder_alloc to handle allocation
and initialization at once.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-3-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
According to Documentation/admin-guide/kernel-parameters.txt, this
parameter can be used to disable kernel modesetting.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Dharma Balasubiramani <dharma.b@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-2-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Take benefit of managed device resources to reduce the risk of memory
leak and to simplify error paths.
Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Reviewed-by: Manikandan Muralidharan <manikandan.m@microchip.com>
Link: https://patch.msgid.link/20251218-lcd_cleanup_mainline-v2-1-df837aba878f@microchip.com
Signed-off-by: Manikandan Muralidharan <manikandan.m@microchip.com>
|
|
Add support for both panels used in the Anbernic RG-DS. These panels
are physically identical and differ only with a single instruction
in the init sequence. The init sequence commands suggest it uses
an identical controller as the jd9365da.
Additionally, allow specifying per-panel dsi->mode_flags that can
override the default values.
Co-developed-by: Alexander Weinzerl <aweinzerl13@yahoo.com>
Signed-off-by: Alexander Weinzerl <aweinzerl13@yahoo.com>
Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20260113195721.151205-4-macroalpha82@gmail.com
|
|
Change instances of gpiod_set_value() to gpiod_set_value_cansleep().
Uses of gpiod_set_value() generates warnings when used in instances
where desc->gdev->can_sleep is true.
Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20260113195721.151205-2-macroalpha82@gmail.com
|
|
The different Video Ports support different maximum resolutions.
Reject resolutions that are not supported by a specific VP.
Only the output width is checked because the hardware itself does
not have a hard output height limit.
Filter the mode that can't output by the VP/crtc.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Tested-by: Sebastian Reichel <sebastian.reichel@collabora.com> # Sige5
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20260117020738.294825-1-andyshrk@163.com
|
|
The threaded interrupt handler on RK3588 checks HPD IRQ status before
deciding to continue with interrupt clearing and unmasking. However,
this is not really necessary, since the hard interrupt handler already
performs the very same verification before waking the handler thread.
Get rid of the redundant verification of the HPD interrupt status in the
threaded interrupt handler.
Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20260115-dw-hdmi-qp-hpd-v1-2-e59c166eaa65@collabora.com
|
|
The threaded interrupt handler on RK3576 checks HPD IRQ status before
deciding to continue with interrupt clearing and unmasking.
This is not only redundant, since a similar verification has been
already performed by the hard IRQ handler before masking the interrupt,
but is also error prone, because it might happen that hardware clears
the status register right after the masking operation completes, and
before the threaded handler reads its value.
The consequence is that HPD IRQ gets never unmasked, which breaks
hotplug detection until reloading the driver or rebooting the system.
Drop the unnecessary verification of the HPD interrupt status from the
threaded interrupt handler.
Fixes: 36439120efbd ("drm/rockchip: dw_hdmi_qp: Add basic RK3576 HDMI output support")
Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patch.msgid.link/20260115-dw-hdmi-qp-hpd-v1-1-e59c166eaa65@collabora.com
|
|
Fix missing MODULE_LICENSE() and MODULE_DESCRIPTION() in the revocable
Kunit test module.
Reported-by: Mark Brown <broonie@kernel.org>
Closes: https://lore.kernel.org/all/aW6GNvuQVNCUcoy-@sirena.org.uk/
Fixes: cd7693419bb5 ("revocable: Add Kunit test cases")
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://patch.msgid.link/20260119195141.12843-1-dakr@kernel.org
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
|
|
On error handling paths, gpiolib_cdev_register() doesn't free the
allocated resources which results leaks. Fix it.
Cc: stable@vger.kernel.org
Fixes: 7b9b77a8bba9 ("gpiolib: add a per-gpio_device line state notification workqueue")
Fixes: d83cee3d2bb1 ("gpio: protect the pointer to gpio_chip in gpio_device with SRCU")
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20260120092650.2305319-1-tzungbi@kernel.org
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
|
|
Add HINIC3_INIT_UP flags to trace netdev open status.
Add port module event handler.
Add link status event type(FAULT, PCIE link down, heart lost, mgmt
watchdog).
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/53a2b928136998f740d597bbd45ca1740b95538f.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add ops to support unicast and multicast to filter mac address in
packets sending and receiving.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/9ea618ad5d6c404e222e9114e08e913a73177705.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
DIM offers a way to adjust the coalescing settings based
on load. As hinic3 rx and tx share interrupts, we only need
to base dim on rx stats.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/af96c20a836800a5972a09cdaf520029d976ad48.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Implement following callback function:
.ndo_vlan_rx_add_vid
.ndo_vlan_rx_kill_vid
.ndo_validate_addr
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/70be187afcaa7b38981d114c088ffdc2cba0b4f1.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
As we cannot solve packets with multiple stacked vlan, so we use
.ndo_features_check to check for these packets and return a smaller
feature without offload features.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/3879b20b7ffa20106a3f8f56dbf2d5eb389f260a.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Implement following callback function:
.ndo_set_features
.ndo_fix_features
The .ndo_set_features function includes five features: rx_csum,
tso, lro, rx_cvlan and vlan_filter.
Add these new features in netdev_feature_init.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/682734a08fde421413048bf70057dafe3cbe8497.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Implement following callback function:
.ndo_tx_timeout
.ndo_get_stats64
Use a work queue to trace tx_timeout callback and dump necessary
debug information.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/ec34d2ff9b142e1e142e47700714533baf7e659c.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Add management and communication pathways between PF and HW.
Co-developed-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Zhu Yikai <zhuyikai1@h-partners.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Link: https://patch.msgid.link/447e72b1c5f255ea5d79ecf96c8dac58011e604d.1768375903.git.zhuyikai1@h-partners.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
