In simulation platform enabling hrtimer fence polling
is causing kernel to block/unresposive.
This issue is not observed in sillicon platforms.
As per this CL making following changes.
1. Added flag for kernel Thread polling.
2. Enabling HRTimer fence polling for silicon platform.
only.
Bug 5094695
Change-Id: I011900e9299be71413c92952d447e53abfd34128
Signed-off-by: amitabhd <amitabhd@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3297995
Reviewed-by: Leslin Varghese <lvarghese@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Raghavendra Vishnu Kumar <rvk@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
In Orin and previous chipsets, engine timestamp counter
runs 32 times faster than CNTVCT register, so the actual
timestamp is obtained by right shifting with factor of 5.
In Thor onwards, this shift is not required is for VIC engine.
Jira HOSTX-5905
Change-Id: I69980fdfcf50b15db99b1fbad522aecd571a0f17
Signed-off-by: Mainak Sen <msen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3306825
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Add kernel module parameter to enable a mode where syncpoints
must be freed explicitly (using the free IOCTL) or they will be
left dangling. This ensures that, for particularly locked down
configurations, a syncpoint will be forever left in an expected
state even if the process owning it dies -- for example another
process will not be able to allocate it.
Change-Id: I2f350c710775a296c70910df21e95737a36c6a45
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3284405
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Santosh BS <santoshb@nvidia.com>
1. Add mechanism(s) to trigger syncpoint fence scan when
syncpoint increment UMD called. Following methods are added.
- METHOD-1: Check fence expire in user context when syncpoint
increment UMD API is called.
- METHOD-2: Add tasklet based mechanism that schedule tasklet to scan for
syncpoint fence expiry. This also improve signaling latency.
METHOD-1 is enabled by default, to enable METHOD-2 define MACRO
"HOST1X_EMU_SYNC_INC_TASKLET".
2. Add interface "host1x_syncpt_fence_scan()" that can be called from
client interrupt handler to initiate syncpoint fence scan.
Jira HOSTX-5527
Change-Id: I4d5a0ba9fd67042d824a1df2794b316831001dc4
Signed-off-by: amitabhd <amitabhd@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3267144
Reviewed-by: Raghavendra Vishnu Kumar <rvk@nvidia.com>
Reviewed-by: Sanif Veeras <sveeras@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Leslin Varghese <lvarghese@nvidia.com>
In Linux v6.12, commit 641bb4394f40 ("fs: move FMODE_UNSIGNED_OFFSET to
fop_flags") moves the setting of the FMODE_UNSIGNED_OFFSET into the
various drivers and renames this flag. This change is causing various
tests to fail because opening the DRM device is failing.
Use the presence of the new FOP_UNSIGNED_OFFSET flag in the kernel to
determine if we need to set this flag in the Tegra DRM driver.
Bug 4876974
Change-Id: Ie72a9340187b1c206f0cda6d7e3dab44d603a41c
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3269063
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
The device names allocated by dev_set_name() need be freed
before module unloading, but they can not be freed because
the kobject's refcount which was set in device_initialize()
has not be decreased to 0.
As comment of device_add() says, if it fails, use only
put_device() drop the refcount, then the name will be
freed in kobejct_cleanup().
device_del() and put_device() can be replaced with
device_unregister(), so call it to unregister the added
successfully devices, and just call put_device() to the
not added device.
Add a release() function to device to avoid null release()
function WARNING in device_release(), it's empty, because
the context devices are freed together in
host1x_memory_context_list_free().
Fixes: 8aa5bcb61612 ("gpu: host1x: Add context device management code")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
(cherry picked from commit 55879dad0f3ae8468444b42f785ad79eac05fe5b)
Bug 4876974
Change-Id: I6d9419f429f34022519db428b6a12c361f901c14
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3268195
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
The following upstream commits have been back-ported to the out-of-tree
Tegra DRM driver to align with upstream to help pull in the latest
updates from upstream.
Changes from Linux v6.2:
7ad4384d53c6 drm/tegra: Add missing clk_disable_unprepare() in
tegra_dc_probe()
Changes from Linux v6.3:
3fb1f62f80a1 drm/fb-helper: Remove drm_fb_helper_unprepare() from
drm_fb_helper_fini()
Changes from Linux v6.4:
71ec16f45ef8 drm/tegra: Implement fbdev emulation as in-kernel client
25dda38e0b07 drm/tegra: Initialize fbdev DRM client
1ac45068af9d drm/tegra: Hide fbdev support behind config option
08263e91f260 drm/tegra: Removed fb from struct tegra_fbdev
434434608a91 drm/tegra: Remove struct tegra_fbdev
dc06f4a495df drm/tegra: Include <linux/i2c.h>
f68b63ebc433 drm/tegra: Include <linux/of.h>
437405403ab4 drm/tegra: rgb: Make tegra_dc_rgb_remove() return void
The changes have been back-ported to ensure that they build for all
kernel versions supported and utilise conftest as necessary to handle
kernel differences.
Bug 4876974
Change-Id: I6c8403325a7b4874f3ad1c6a0b1f662c1494c397
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3267798
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
In Linux v6.12, commit 446d0f4849b1 ("drm: Remove struct
drm_mode_config_funcs.output_poll_changed") removes
'output_poll_changed' function pointer from 'drm_mode_config_funcs'
structure and commit b5757a5be2fa ("drm: Remove struct
drm_driver.lastclose") removes 'lastclose' from the 'drm_driver'
structure. This is breaking build the Tegra DRM out-of-tree driver for
Linux v6.12.
Fix this by partially back-porting commit 71ec16f45ef8 ("drm/tegra:
Implement fbdev emulation as in-kernel client") from Linux v6.4 which
removed the use of the 'output_poll_changed' and 'lastclose' function
pointers from the Tegra DRM driver. Note that it is safe to partially
back-port this for now because we don't use the frame-buffer support in
the Tegra DRM driver for Tegra234+ devices.
To keep the Tegra DRM out-of-tree driver aligned with upstream, the
remaining parts of commit '71ec16f45ef8' will be back-ported in another
change that will be part of a bigger change to align the frame-buffer
support with the upstream driver. For now just fix the build for Linux
v6.12.
Bug 4876974
Change-Id: Idc9b1c198c067ccb59f1337866154c9a8f0a5623
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3267797
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
1. The nvhost interface like syncpoint offset/aperture were
taking "pdev" of Host1x device, but emulated syncpoint driver
implementd the interface with "pdev" of clinet device. This was
done to align all other interface input parameter.
This patch revert the above alignment of intput parameter change.
2. This patch modify SyncpointWait IOCTL handler to return timestamp
of signal observed by KMD to UMD
Bug 4895937
Change-Id: I61ba02992963ba272e5c2514ffe8c24fff510c3a
Signed-off-by: amitabhd <amitabhd@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3226744
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: svcacv <svcacv@nvidia.com>
Reviewed-by: Santosh BS <santoshb@nvidia.com>
Reviewed-by: Raghavendra Vishnu Kumar <rvk@nvidia.com>
- Update Makefile to build the legacy chip related
files only when any of the below configs are set.
CONFIG_ARCH_TEGRA_2x_SOC
CONFIG_ARCH_TEGRA_3x_SOC
CONFIG_ARCH_TEGRA_114_SOC
CONFIG_ARCH_TEGRA_124_SOC
CONFIG_ARCH_TEGRA_132_SOC
CONFIG_ARCH_TEGRA_210_SOC
CONFIG_ARCH_TEGRA_186_SOC
CONFIG_ARCH_TEGRA_194_SOC
- Update drm.c/.h to build legacy chip specific
code portions for the respective configs accordingly.
Jira HOSTX-5727
Signed-off-by: Santosh BS <santoshb@nvidia.com>
Change-Id: I3d2235923011073d82dcfe6b1fd6618ae70d37bb
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3253132
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
- Update Makefile to build the legacy chip specific
files according to the below configs.
CONFIG_ARCH_TEGRA_2x_SOC
CONFIG_ARCH_TEGRA_3x_SOC
CONFIG_ARCH_TEGRA_114_SOC
CONFIG_ARCH_TEGRA_124_SOC
CONFIG_ARCH_TEGRA_132_SOC
CONFIG_ARCH_TEGRA_210_SOC
CONFIG_ARCH_TEGRA_186_SOC
CONFIG_ARCH_TEGRA_194_SOC
- Update dev.c/.h to build legacy chip specific
code portions for the respective configs.
Jira HOSTX-5727
Change-Id: I7dbd23ab1c91c859941c4965123bb60e518063fc
Signed-off-by: Santosh BS <santoshb@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3263215
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
In Linux v6.13, commit 689274a56c0c ("drm: Remove DRM aperture helpers")
removed the DRM aperture helper functions and drivers should simply use
the existing video helpers. Update the Tegra DRM driver to use
aperture_remove_all_conflicting_devices() if available to align with
upstream.
Bug 4991705
Change-Id: Ie4b78a4ea519ffcaaf972417ec2cbe7c436eb808
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3261694
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Add a userspace interface for providing access to the memory syncpoint
shim through dma-buf file descriptors. Applications can pass references
to syncpoints to other devices that only support memory semaphores to
facilitate synchronization with syncpoints.
Only syncpoints allocated through the DRM file descriptor ("owned" by
the application) can be exported read-write. All syncpoints can be
exported read-only.
Jira HOSTX-5722
Bug 4919132
Change-Id: I4f5c007b7f9ba65780c418b4e22c5f59a11ef84d
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3252026
Reviewed-by: Santosh BS <santoshb@nvidia.com>
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Retrieve information for the memory syncpoint shim from the device tree
and add an interface to provide the information to other drivers.
The memory syncpoint shim is a special area on the data backbone,
where each memory page corresponds to a host1x syncpoint. Reading
the page returns the value of the syncpoint, and writing to it
increments the syncpoint by one.
Jira HOSTX-5722
Bug 4919132
Change-Id: Ie892abd754f7543fbc56844dcc40cab3f837e305
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3252025
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Santosh BS <santoshb@nvidia.com>
Firmware requests triggered by userspace action can be executed
in interruptible context, which can cause signals to interrupt
firmware loading.
Ordinarily, one would propagate the error code to userspace and
let it retry. However, request_firmware does not have proper
handling of retryable errors, as sysfs fallbacks will clobber
the original error code. Furthermore, pm_runtime, through which
we would propagate the error code, doesn't support transient
resume failures either.
As such, as a quick fix, change to request_firmware_direct and
retry it directly in situ 10 times.
Bug 4753613
Co-developed-by: Zuyi Hu <zuyih@nvidia.com>
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Change-Id: Idb9a04d4fa80aeffbfd0ad0176cce55c4dedf488
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3202848
(cherry picked from commit a17e81cad18bbdfef0194471524b675d74501054)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3219109
(cherry picked from commit 30c08f369a8f3acd642fcec70814086bd472e734)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nv-oot/+/3253514
GVS: buildbot_gerritrpt <buildbot_gerritrpt@nvidia.com>
Reviewed-by: Santosh BS <santoshb@nvidia.com>
Tested-by: Zuyi Hu <zuyih@nvidia.com>
The change to write the INTF_CRC_CFG register was inadvertently
left out. Correct this.
Bug 4809300
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Change-Id: I208da56d03e90e72ce26dffe7bef1cdeab95fab6
