| Age | Commit message (Collapse) | Author |
|---|
|
The g_skip_frames sensor-op is obsolete, drop it.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-23-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
On raw camera sensors the framerate is controlled through vblank
(and optional) hblank controls.
Having a get_frame_interval makes no sense in this case, drop it.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-22-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
The i2c-client's power-domain has already been powered up when probe()
gets called. So e.g. ACPI resources for regulators and clks have
already been enabled and only the GPIOs need to be set to the on state.
Instead of calling pm_runtime_set_suspended() while the domain is
already powered up and then have detect() do a pm_runtime_get()
to set the GPIOs do the following:
1. Call gc0310_power_on() to only set the GPIOs
2. Set the device's runtime-PM state to active instead of suspended
3. Avoid the device getting suspended as soon as pm_runtime_enable()
gets called by calling pm_runtime_get() before _enable(), this means
moving the pm_runtime_get() / _put() from detect() to probe ()
This fixes power_on() not getting called when runtime-PM is not
enabled in the Kconfig and this keeps the sensor powered-up while
registering it avoiding unnecessary power cycles.
Also modify gc0310_remove() to power-off the device if it is in
active state when gc0310_remove() runs.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-21-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Move the suspend()/resume() functions to above gc0310_detect() and rename
the functions to power_off()/power_on().
No functional changes, this is a preparation patch for reworking
the runtime-pm handling in probe() and remove().
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-20-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Stop having a v4l2_mbus_framefmt mode.fmt driver-data member to store
the fmt for the active-state, instead use sd.active_state fmt.
This also removes the need for gc0310_get_pad_format() since
v4l2_subdev_state_get_format() now will return the correct
v4l2_mbus_framefmt for all whence values.
Instead of switching gc0310_set_fmt() from gc0310_get_pad_format() to
v4l2_subdev_state_get_format() just drop it entirely since there is only
1 fixed mode. Otherwise the new gc0310_set_fmt() would be 100% the same
as v4l2_subdev_get_fmt() after this.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-19-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
v4l2_subdev_pad_ops.get_fmt()
Now that the sd-state's fmt is properly initialized by
internal_ops.init_state(), the driver can be safely switched
to v4l2_subdev_get_fmt().
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-18-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Implement internal_ops.init_state to fill in the v4l2_mbus_framefmt
struct in newly allocated sd-state structs.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-17-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Switch to using the sub-device state lock and properly call
v4l2_subdev_init_finalize() / v4l2_subdev_cleanup() on probe() /
remove().
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-16-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Switch from s_stream() to enable_streams() and disable_streams() pad
operations. They are preferred and required for streams support.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-15-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
is_streaming is only set and never read, drop it.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-14-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Reduce the unnecessary long msleep(100) done on stream start to 10 ms and
move this to gc0310_resume() so that it is also done on the initial
power-up done by gc0310_detect(), which should fix gc0310_detect()
sometimes failing.
While at it switch the sleeps from msleep() / usleep_range() to fsleep().
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-13-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Add a check_hwcfg() function to check if the external clk-freq, CSI
link-freq and lane-count match the driver's expectations.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-12-hansg@kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
When an exposure value > (mode-height + vblank) gets set the sensor will
automatically increase vblank, lowering the framerate.
This is not desirable, limit exposure the maximum exposure to mode-height +
vblank to avoid the unwanted framerate slowdown.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-11-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Add camera orientation and sensor rotation controls using
the v4l2_fwnode_device_parse() and v4l2_ctrl_new_fwnode_properties()
helpers.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-10-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Add support for the vblank and hblank controls, these controls
are mandatory for using the sensor driver with libcamera.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-9-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Add support for the pixelrate control as expected by libcamera,
while at it also add the link-frequency control.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-8-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Add support for the selection API as expected by libcamera.
Note the driver only supports a single fixed resolution and
no cropping, so this is a simple read-only implementation.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-7-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Use V4L2_CID_ANALOGUE_GAIN for gain control, as expected by userspace.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-6-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Switch the GC0310 driver over to the CCI register access helpers.
While at it also add a _REG prefix to all register address defines
to make clear they are register addresses and group register value
defines together with the address definition.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-5-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Currently the sensor is running 30.9 fps, increase vblank
to have it actually run at 30.0 fps.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-4-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Drop the unused GC0310_FOCAL_LENGTH_NUM define, the focal-length
is a property of the sensor-module, not of the raw sensor itself.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-3-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Many functions on the gc0310 driver use a function local variable called
"dev" but these variable's type is not "struct device *" type as one would
expect based on the name. Instead they point to the gc0310 driver data
struct.
Rename these variables to sensor to make their purpose more clear.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250517114106.43494-2-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Fix a coding style:
"ERROR: that open brace { should be on the previous line"
issue reported in ia_css_vf.host.c:94.
Signed-off-by: Pablo <pablo@pablo.ct.ws>
Link: https://lore.kernel.org/r/20250503200030.5982-1-pablo@pablo.ct.ws
Signed-off-by: Hans de Goede <hansg@kernel.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Replace the duplicate code for calling the special Intel camera sensor GPIO
type _DSM (79234640-9e10-4fea-a5c1-b5aa8b19756f) and mapping GPIOs to
the sensor with a call to int3472_discrete_parse_crs() from the int3472
driver.
Besides avoiding code duplication the int3472 version of the code also
supports more features, like mapping the powerdown GPIO to a regulator on
the mt9m114 which is necessary to make the camera on the Asus T100TA work.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Link: https://lore.kernel.org/r/20250507184737.154747-7-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Commit c7194b21809e ("media: atomisp: On streamoff wait for buffers owned
by the CSS to be given back") added draining of the CSS buffer queue to
the beginning of atomisp_stop_stream().
But it turns out that when telling the CSS to stop streaming it needs at
least 1 buffer queued, because the CSS firmware waits for a frame to be
completed before stopping and without buffers it cannot complete a frame.
At the end of atomisp_stop_stream() it is always safe to return buffer
ownership to the videobuf2-core. Either atomisp_css_stop() has successfully
stopped the stream; or the atomisp_reset() later on which power-cycles
the ISP will definitely have stopped the stream.
Drop the draining of the CSS buffer queue to fix the "stop stream timeout."
error and move the atomisp_flush_video_pipe() call after atomisp_reset(),
passing false for the warn_on_css_frames flag since some buffers still
being marked as owned by the CSS expected on stream off.
Also increase the timeout in destroy_stream(), since this waits for
the last frame to be completed this can take longer then 40 ms. When e.g.
using a framerate of 15 fps, this could take 66ms, make the timeout 200 ms
to be on the safe side.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250505210008.152659-6-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
The atomisp interrupt handling will free the MIPI / CSI-receiver buffers
when processing a frame-completion event if the stop_requested flag is set,
but only in the ISP2400 / BYT, not in the ISP2401 / CHT case.
There are 2 problems with this:
1. Since this is only done in the BYT case the "mipi frames are not freed."
warning always triggers on CHT devices.
2. There are 2 stop_requested flags, ia_css_pipe.stop_requested and
ia_css_pipeline.stop_requested. The ISR checks the ia_css_pipe flag,
but atomisp_css_stop() sets the ia_css_pipeline.stop_requested flag.
So even on BYT freeing the buffers from the ISR never happens.
This likely is a good thing since the buffers get freed on the first
frame completion event and there might be multiple frames queued up.
Fix things by completely dropping the freeing of the MIPI buffers from
the ISR as well as the stop_requested flag always freeing the buffers
from ia_css_uninit().
Also drop the warning since this now always is expected behavior.
Note that ia_css_uninit() get called whenever streaming is stopped
through atomisp_stop_stream() calling atomisp_reset() so the buffers
are still freed whenever streaming is stopped.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250505210008.152659-5-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
atomisp_start_streaming() starts the media pipeline before calling
atomisp_css_start(). On atomisp_css_start() failures stop the pipeline
before returning the error.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250505210008.152659-4-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
When v4l2_subdev_call(sensor, s_stream, 1) fails atomisp_start_streaming()
was not properly returning the buffer ownership back to the videobuf2-core
code, resulting in:
[ 1318.153447] ------------[ cut here ]------------
[ 1318.153499] WARNING: CPU: 0 PID: 4856 at drivers/media/common/videobuf2/videobuf2-core.c:1803 vb2_start_streaming+0xcb/0x160 [videobuf2_common]
...
[ 1318.154551] Call Trace:
[ 1318.154560] <TASK>
[ 1318.154571] ? __warn.cold+0xb7/0x14a
[ 1318.154591] ? vb2_start_streaming+0xcb/0x160 [videobuf2_common]
[ 1318.154617] ? report_bug+0xe0/0x180
[ 1318.154640] ? handle_bug+0x5e/0xa0
[ 1318.154652] ? exc_invalid_op+0x14/0x70
[ 1318.154665] ? asm_exc_invalid_op+0x16/0x20
[ 1318.154697] ? vb2_start_streaming+0xcb/0x160 [videobuf2_common]
[ 1318.154723] ? vb2_start_streaming+0x70/0x160 [videobuf2_common]
[ 1318.154748] vb2_core_streamon+0xa2/0x100 [videobuf2_common]
The sensor streamon call is the last thing that atomisp_start_streaming()
does and it was failing to undo all of the previous steps in general.
Refactor atomisp_stop_streaming() into an atomisp_stop_stream() helper and
call that on sensor streamon failure to properly clean things up.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250505210008.152659-3-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Move atomisp_stop_streaming() above atomisp_start_streaming(), this is
a preparation patch for making atomisp_start_streaming() properly cleanup
if starting the sensor stream fails.
No functional change, only moving a block of code up.
Signed-off-by: Hans de Goede <hansg@kernel.org>
Reviewed-by: Andy Shevchenko <andy@kernel.org>
Link: https://lore.kernel.org/r/20250505210008.152659-2-hdegoede@redhat.com
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Reading the hpd or 5v gpios is something that can sleep, so rework
the code to allow this.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
Ensure that the interrupt function names clearly state for which
gpio they are (cec/hpd/5v). No functional changes.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
If rx-no-low-drive is set, then the CEC pin framework will disable
the detection of situations where a Low Drive has to be generated.
So if this is set, then we will never generate Low Drives.
This helps testing whether other CEC devices generate Low Drive
pulses by ensuring it is not us that is generating them.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
This adds support for inserting 'glitches' after a falling and/or
rising edge. This tests what happens when there are little voltage
spikes after falling or rising edges, which can be caused due to
noise or reflections on the CEC line.
A proper CEC implementation will deglitch this, but a poor implementation
can create a Low Drive pulse in response, effectively making CEC unusable.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
When the EDID is updated, the hotplug detect signal must remain low for
100 ms minimum. Currently these three drivers use that exact minimum,
but some HDMI transmitters need the HPD to be low for a bit longer
before they detect that they need to read the EDID again.
Experience shows that HZ / 7 (= 143 ms) is a good value.
So change HZ / 10 to HZ / 7.
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
|
|
There looks to be an issue in our compression handling when the BO pages
are very fragmented, where we choose to skip the identity map and
instead fall back to emitting the PTEs by hand when migrating memory,
such that we can hopefully do more work per blit operation. However in
such a case we need to ensure the src PTEs are correctly tagged with a
compression enabled PAT index on dgpu xe2+, otherwise the copy will
simply treat the src memory as uncompressed, leading to corruption if
the memory was compressed by the user.
To fix this pass along use_comp_pat into emit_pte() on the src side, to
indicate that compression should be considered.
v2 (Jonathan): tweak the commit message
Fixes: 523f191cc0c7 ("drm/xe/xe_migrate: Handle migration logic for xe2+ dgfx")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Akshata Jahagirdar <akshata.jahagirdar@intel.com>
Cc: <stable@vger.kernel.org> # v6.12+
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://lore.kernel.org/r/20250701103949.83116-2-matthew.auld@intel.com
(cherry picked from commit f7a2fd776e57bd6468644bdecd91ab3aba57ba58)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
This reverts commit fe0154cf8222d9e38c60ccc124adb2f9b5272371.
Seeing some unexplained random failures during LRC context switches with
indirect ring state enabled. The failures were always there, but the
repro rate increased with the addition of WA BB as a separate BO.
Commit 3a1edef8f4b5 ("drm/xe: Make WA BB part of LRC BO") helped to
reduce the issues in the context switches, but didn't eliminate them
completely.
Indirect ring state is not required for any current features, so disable
for now until failures can be root caused.
Cc: stable@vger.kernel.org
Fixes: fe0154cf8222 ("drm/xe/xe2: Enable Indirect Ring State support for Xe2")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20250702035846.3178344-1-matthew.brost@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
(cherry picked from commit 03d85ab36bcbcbe9dc962fccd3f8e54d7bb93b35)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
CIRC_SPACE does not work unless the size argument is a power of 2,
allocate PF queue size on power of 2 boundary.
Cc: stable@vger.kernel.org
Fixes: 3338e4f90c14 ("drm/xe: Use topology to determine page fault queue size")
Fixes: 29582e0ea75c ("drm/xe: Add page queue multiplier")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://lore.kernel.org/r/20250702213511.3226167-1-matthew.brost@intel.com
(cherry picked from commit 491b9783126303755717c0cbde0b08ee59b6abab)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
Our LMEM buffer objects are not cleared by default on alloc
and during VF provisioning we only setup LMTT PTEs for the
actually provisioned LMEM range. But beyond that valid range
we might leave some stale data that could either point to some
other VFs allocations or even to the PF pages.
Explicitly clear all new LMTT page to avoid the risk that a
malicious VF would try to exploit that gap.
While around add asserts to catch any undesired PTE overwrites
and low-level debug traces to track LMTT PT life-cycle.
Fixes: b1d204058218 ("drm/xe/pf: Introduce Local Memory Translation Table")
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Lukasz Laguna <lukasz.laguna@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Piotr Piórkowski <piotr.piorkowski@intel.com>
Link: https://lore.kernel.org/r/20250701220052.1612-1-michal.wajdeczko@intel.com
(cherry picked from commit 3fae6918a3e27cce20ded2551f863fb05d4bef8d)
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
Add HW Steering (HWS) as a secondary option for device steering mode. If
the device does not support SW Steering (SWS), HW Steering will be used
as the default, provided it is supported. FW Steering will now be
selected as the default only if both HWS and SWS are unavailable.
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-11-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Matcher size is dynamic: it starts at initial size, and then it grows
through rehash as more and more rules are added to this matcher.
When rules are deleted, matcher's size is not decreased. Rehash
approach is greedy. The idea is: if the matcher got to a certain size
at some point, chances are - it will get to this size again, so it is
better to avoid costly rehash operations whenever possible.
However, when all the rules of the matcher are deleted, this should
be viewed as special case. If the matcher actually got to the point
where it has zero rules, it might be an indication that some usecase
from the past is no longer happening. This is where some ICM can be
freed.
This patch handles this case: when a number of rules in a matcher
goes down to zero, the matcher's tables are shrunk to the initial
size.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Vlad Dogaru <vdogaru@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-10-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
As a preparation for the following patch that will add support
for shrinking empty matchers, rearrange the code to prevent
forward declaration of functions.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-9-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Track and grow matcher sizes individually for RX and TX RTCs. This
allows RX-only or TX-only use cases to effectively halve the device
resources they use.
For testing we used a simple module that inserts 1M RX-only rules and
measured the number of pages the device requests, and memory usage as
reported by `free -h`.
Pages Memory
Before this patch: 300k 1.5GiB
After this patch: 160k 900MiB
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-8-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Kernel HWS only uses FDB tables and, as such, creates two lower level
containers (RTCs) for each matcher: one for RX and one for TX. Allow
these RTCs to differ in size by converting the size part of the matcher
attribute to a two element array.
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Link: https://patch.msgid.link/20250703185431.445571-7-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Matchers were using the pool abstraction solely as a convenience
to allocate two STE ranges. The pool's core functionality, that
of allocating individual items from the range, was unused.
Matchers rely either on the hardware to hash rules into a table,
or on a user-provided index.
Remove the STE pool from the matcher and allocate the STE ranges
manually instead.
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Link: https://patch.msgid.link/20250703185431.445571-6-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Reduce nesting by adding a couple of early return statements.
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-5-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The bwc layer will use `mlx5hws_rule_skip` to keep track of numbers of
RX and TX rules individually, so export this function for future usage.
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250703185431.445571-4-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
Removing incorrect comment section that is probably some
copy-paste artifact.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Vlad Dogaru <vdogaru@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Link: https://patch.msgid.link/20250703185431.445571-3-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
`flow_source` is not used anywhere in mlx5hws_action_create_dest_array.
Signed-off-by: Vlad Dogaru <vdogaru@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Link: https://patch.msgid.link/20250703185431.445571-2-mbloch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2025-07-03
Vladimir Oltean converts Intel drivers (ice, igc, igb, ixgbe, i40e) to
utilize new timestamping API (ndo_hwtstamp_get() and ndo_hwtstamp_set()).
For ixgbe:
Paul, Don, Slawomir, and Radoslaw add Malicious Driver Detection (MDD)
support for X550 and E610 devices to detect, report, and handle
potentially malicious VFs.
Simon Horman corrects spelling mistakes.
For igbvf:
Kohei Enju removes a couple of unreported counters and adds reporting
of Tx timeouts.
* '10GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
igbvf: add tx_timeout_count to ethtool statistics
igbvf: remove unused interrupt counter fields from struct igbvf_adapter
ixgbe: spelling corrections
ixgbe: turn off MDD while modifying SRRCTL
ixgbe: add Tx hang detection unhandled MDD
ixgbe: check for MDD events
ixgbe: add MDD support
i40e: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()
ixgbe: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()
igb: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()
igc: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()
ice: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()
====================
Link: https://patch.msgid.link/20250703174242.3829277-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
|
|
The shifts and masks for each channel are defined by hardware and
are not something that changes at runtime. Accordingly, describe the
information in an array of const structs and associate elements with
each channel instance, removing the need for the switch and handling of
its default case.
Link: https://patch.msgid.link/20250616-aspeed-lpc-snoop-fixes-v2-10-3cdd59c934d3@codeconstruct.com.au
Signed-off-by: Andrew Jeffery <andrew@codeconstruct.com.au>
|
